Methods

Trial design: parallel (2 arms)

Location: Institut Gustave Roussy, Villejuif, France

Number of centres: 1

Study duration: February 2005 to September 2006
Trials registry number: none/unknown

Participants

Inclusion criteria: adults with metastatic colorectal adenocarcinoma (grade 3 to 4); life expectancy greater than 3 months; receiving 5FU‐based chemotherapy

Exclusion criteria: HIV; pregnant or lactating; unlikely to comply with interventions; participation in another trial in the previous 12 months (unless regarding chemotherapeutic protocols); undergone a total colectomy; state of subocclusion; chronic inflammatory diseases of the digestive tract; radiation enteropathy

Cancer type: metastatic colorectal adenocarcinoma (grade 3 to 4)

Cancer treatment: 5FU‐based chemotherapy

Age at baseline (years): median 60 (not reported by group)

Gender: not reported

Number randomised: 22 (not reported by group)

Number evaluated: 13 (Group A: 9; Group B: 4)

Interventions

Comparison: TGF‐beta(2) versus placebo

Group A: nutritional supplement of proteins, carbohydrates, fats, vitamins and minerals, with TGF‐beta(2) (2 ng/mg protein); formulas were in powder form, mixed with cool previously boiled water at 0.23 g/mL (100 kcl/100 mL); during each cycle participants received 750 mL to 1000 mL per day plus any other food desired; formula administered for 2 days before, 2 days during, and 3 days after chemotherapy (7 days/cycle)

Group B: same as above without the TGF‐beta(2)

Compliance: "Nine randomised patients who never ate the formula were excluded from the study" (not reported by group)

Duration of treatment: "3 months (test or control formula), for a minimum of one and a maximum of eight cycles of treatment"

Outcomes

• Oral mucositis: WHO 0 to 4 scale (no details reported on who assessed this, or when it was assessed; only reports incidence of any mucositis)

• Chemotherapy‐induced diarrhoea (not an outcome of this review)

Notes

Sample size calculation: not reported

Funding: "This study was funded by Nestec Ltd" ‐ Nestlé (manufacturer of the intervention)

Declarations/conflicts of interest: 6 of the 9 authors were either consultants (1) or employees (5) of Nestlé

Data handling by review authors: reported in additional table

Other information of note: "Due to low accrual of patients (22 patients were enrolled and randomised in 18 months), the study was prematurely stopped"

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "randomly assigned"

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Unclear risk

Quote: "randomly assigned"

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind" and "The test formula differed only by containing an additional..."

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind"

Comment: it is not clear who was blinded. There are subjective elements to the assessment of oral mucositis using this scale, requiring the patient's assessment of pain/soreness and their ability to swallow but, as the participants were unaware of their group allocation, the assessment of oral mucositis can be considered to be blinded

Incomplete outcome data (attrition bias)
All outcomes

High risk

Overall attrition was 41% although it was not reported by group

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (7 arms) dose‐ranging study

Location: Universities of Minnesota and Michigan, USA

Number of centres: 2

Study duration: not reported
Trials registry number: none/unknown

Participants

Inclusion criteria: aged 3 to 65 years; diagnosed with haematological malignancy (including myelodysplastic syndromes); ECOG score of 0 to 2; eligible for allogeneic HSCT after conditioning treatment with chemotherapy with or without TBI

Exclusion criteria: received previous allogeneic HSCT; due to receive a T‐cell‐depleted donor graft; active chronic skin disease; pre‐existent inflammatory bowel disease; uncontrolled (antibiotic‐resistant) bacterial infection; hepatitis; HIV

Cancer type: haematologic: ALL (Group A: 12%; Group B: 3%); AML (Group A: 35%; Group B: 39%); CML (Group A: 10%; Group B: 26%); MDS (Group A: 9%; Group B: 19%); NHL (Group A: 19%; Group B: 3%); Hodgkin's (Group A: 1%; Group B: 0%); Other (Group A: 14%; Group B: 10%)

Cancer treatment: both centres had allogeneic HSCT on day 0 but differed in conditioning regimen and GVHD prophylaxis as follows:

• Minnesota centre (n = 54): conditioning with cyclophosphamide (60 mg/kg per day for 2 days) on days ‐7 and ‐6 and TBI (total dose 13.2 Gy, fractionated as 165 cGy twice daily for 4 days) on days ‐4 to ‐1; GVHD prophylaxis with methotrexate (15 mg/m², IV bolus on day +1, and 10 mg/m², IV bolus on days 3, 6, and 11) and cyclosporine A (starting from day ‐3); Group A: 69%; Group B: 31%

• Michigan centre (n = 46): conditioning with busulfan (1 mg/kg per dose given 4 times daily for 4 days) on days ‐8 to ‐5 and cyclophosphamide (60 mg/kg per day for 2 days) on days ‐3 and ‐2; GVHD prophylaxis with methotrexate (15 mg/m², IV bolus on day +1, and 10 mg/m², IV bolus on days 3, 6, and 11) and tacrolimus or cyclosporine A (starting from day ‐3); Group A: 70%; Group B: 30%

Both centres received G‐CSF (filgrastim) 5 µg/kg per day from 24 hours after HSCT until neutrophil recovery

Age at baseline (years): Group A: median 46 (range 7 to 65); Group B: median 46 (range 7 to 63)

Gender: both groups 58% male

Number randomised: 100 (Group A: 69; Group B: 31)

Number evaluated: 96 (Group A: 65; Group B: 31)

Interventions

Comparison: KGF (palifermin) versus placebo

(4 KGF arms and 3 placebo arms were each combined into a single arm)

Group A: KGF

• (n = 8): 40 µg/kg per day in 6 doses on days ‐11, ‐10, ‐9 and 0, 1, 2 (total dose = 240 µg/kg)

• (n = 10): 60 µg/kg per day in 6 doses on days ‐11, ‐10, ‐9 and 0, 1, 2 (total dose = 360 µg/kg)

• (n = 14): 60 µg/kg per day in 9 doses on days ‐11, ‐10, ‐9 and 0, 1, 2 and 7, 8, 9 (total dose = 540 µg/kg)

• (n = 37): 60 µg/kg per day in 12 doses on days ‐11, ‐10, ‐9 and 0, 1, 2 and 7, 8, 9 and 14, 15, 16 (total dose = 720 µg/kg)

Group B: placebo with matching schedule to either the 6, 9 or 12 dose regimen

Mode of administration not described but presumably IV as in other KGF studies

Compliance: Group A: 20 did not receive all study doses (17 of these were replaced to allow a full assessment of safety); Group B: 2 did not receive all study doses (1 replaced)

Duration of treatment: varied from 13 days (6 doses) to 27 days (12 doses) ‐ see above

Outcomes

• Oral mucositis: WHO 0 to 4 scale (measured 3 times per week during hospitalisation by designated observers, maximum score reported)

• Adverse effects (assessed daily during study period using WHO and NCI‐CTC toxicities scale)

• Incidence and severity of acute GVHD (not an outcome of this review)

• Overall survival (not an outcome of this review)

• Incidence of transplantation‐related toxicity (not an outcome of this review)

• Time to marrow engraftment (not an outcome of this review)

Notes

Sample size calculation: not reported

Funding: government grants from NIH and FDA, and also supported by Amgen (pharmaceutical industry)

Declarations/conflicts of interest: not reported

Data handling by review authors: the data for incidence of mucositis were not reported separately for each dose and therefore it was not possible to include head‐to‐head comparisons of different dosages in this review; the data for incidence of mucositis were presented in subgroups of those that did or did not receive the final methotrexate infusion on day 11 but we used the overall data in our meta‐analyses (the study authors report that there was no difference between these subgroups)

Other information of note: the study authors report a greater decrease in incidence of grade 3 to 4 (severe) oral mucositis due to palifermin in the Minnesota participants (who received a more mucotoxic conditioning regimen) than in the Michigan participants

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "patients...were randomly assigned..."

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Unclear risk

Quote: "patients...were randomly assigned..."

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled"

Comment: it is not clear who was blinded. There are subjective elements to the assessment of oral mucositis using this scale, requiring the patient's assessment of pain/soreness and their ability to swallow but, as the participants were unaware of their group allocation, the assessment of oral mucositis can be considered to be blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Quote: "Data for all patients randomly assigned and who received a transplant were used in all other analyses (intent‐to‐treat)"

Comment: although 18 participants (Group A: 17; Group B: 1) were replaced to allow a full assessment of safety, it seems that the originally randomised participants were included in the analyses

Overall attrition was 4% (Group A: 6%; Group B: 0%) for the oral mucositis incidence outcome. The reasons were unclear but this proportion of attrition is unlikely to have biased the results

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (3 arms)

Location: Europe (Italy, France, the Netherlands, Ireland, Germany, UK, Denmark, Austria, Switzerland, Czech Republic, Sweden, Hungary, Belgium, Finland)

Number of centres: 39

Study duration: December 2006 to February 2009
Trials registry number: NCT00434161

Participants

Inclusion criteria: aged between 18 and 70 years; due to receive high‐dose melphalan; ECOG score of 0 to 2 (or 3, if reason was due to multiple myeloma); at least 2 x 10⁶ CD34+ cells per kg; corrected carbon monoxide diffusing capacity 50% or higher of predicted; absolute neutrophil count at least 1.5 x 10⁹/L and platelets at least 100 x 10⁹/L; total bilirubin 2 mg/dL or lower; aspartate aminotransferase and/or alanine aminotransferase 4 x institutional upper limit of normal or lower

Exclusion criteria: not reported

Cancer type: multiple myeloma

Cancer treatment: 1‐day administration of high‐dose melphalan (200 mg/m²) on day ‐2, followed by auto‐SCT on day 0

Age at baseline (years): Group A: median 55 (range 32 to 69); Group B: median 58 (range 40 to 68); Group C: median 58 (range 41 to 68)

Gender: Group A: 54% male; Group B: 55% male; Group C: 58% male

Number randomised: 281 (Group A: 109; Group B: 115; Group C: 57)

Number evaluated: 281 (Group A: 109; Group B: 115; Group C: 57)

Interventions

Comparison: KGF versus placebo

Group A: KGF (60 µg/kg) daily IV on days ‐6, ‐5, and ‐4, then placebo on days 0 (the day of auto‐SCT), 1, and 2 (total dose = 180 µg/kg)

Group B: KGF (60 µg/kg) daily IV on days ‐6, ‐5, ‐4, 0, 1, and 2 (total dose = 360 µg/kg)

Group C: placebo daily IV on days ‐6, ‐5, ‐4, 0, 1, and 2

Compliance: Group A: 8% discontinued; Group B: 12% discontinued; Group C: 4% discontinued; (point of discontinuation or number of treatments not stated for any group)

Duration of treatment: 6 treatment days (over 9 days)

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed daily by nurses and physicians from day ‐2 to day 32, maximum score reported) (duration of grade 2 to 4 and 3 to 4 oral mucositis also measured but not outcomes of this review)

• Oral pain: OMDQ 5‐point scale for mouth and throat soreness (higher = worse pain) (assessed daily by participants from day ‐2 to day 32, data reported as AUC, not used)

• Quality of life: EQ‐5D 0 (worst imaginable health) to 10 (best imaginable health) scale, incorporating mobility, self‐care, usual activities, pain/discomfort and anxiety/depression (assessed daily by participants from day ‐2 to day 32, mean reported only at day 7 and with no SD, no usable data)

• Normalcy of diet (measured as incidence of TPN) (duration of TPN also measured but not used for analysis in review)

• Adverse events (NCI‐CTC version 3.0 toxicity scale)

• Number of days in hospital

• Number of days of treatment with opioid analgesics (incidence of opioid analgesic use also measured and reported but not an outcome of this review)

• Febrile neutropenia (not an outcome of this review)

• Significant infections (not an outcome of this review)

• Anti‐infective (IV) drug use (not an outcome of this review)

• Blood product use (not an outcome of this review)

• Nonopioid analgesic use (not an outcome of this review)

Notes

Sample size calculation: 275 participants required at 95% power and 5% significance to detect an odds ratio of at least 3.5 between placebo and KGF in grade 2 to 4 oral mucositis

Funding: sponsored by Swedish Orphan Biovitrum (pharmaceutical industry); KGF and placebo manufactured and packaged by Amgen (pharmaceutical industry)

Declarations/conflicts of interest: 2 authors were employees of the sponsors; the remaining authors declared no competing financial interests

Data handling by review authors: we combined the 2 KGF groups to make a single pairwise comparison against placebo; we also made a separate comparison of the 2 KGF regimens

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "Randomization was performed by using an interactive‐voice‐response‐system before planned admission"

Comment: large multicentre trial using high‐tech randomisation method ‐ likely to be done properly

Allocation concealment (selection bias)

Low risk

Quote: "Randomization was performed by using an interactive‐voice‐response‐system before planned admission"

Comment: large multicentre trial using high‐tech randomisation method ‐ likely to be done properly

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled" and "Study drug...packaged...in identical vials"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled"

Comment: it is not clear who was blinded. There are subjective elements to the assessment of lower grades of oral mucositis using the WHO scale, requiring the patient's assessment of pain/soreness and their ability to swallow. Higher grades have more objective elements so may not be affected by potential lack of blinding of the assessor. This would be the same for other subjective and objective outcomes

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately (although quality of life reported with no SD or P values, this does not affect the risk of bias judgement for other outcomes)

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Australia

Number of centres: 23

Study duration: recruitment from September 2006 to April 2010
Trials registry number: ACTRN012605000095662 (mentioned in trial report)

Participants

Inclusion criteria: aged 15 to 60 years with newly diagnosed and previously untreated (except for hydroxycarbamide for high presenting white blood cell count) acute myeloid leukaemia ‐ all subtypes except t(15;17) or variants, or core‐binding factor AML (t(8;21) or inv(16) or variants); ECOG score of 0 to 3; no history of cancer (other than basal cell skin cancer or carcinoma of the cervix in situ, or other localised cancer treated by surgical excision only more than 5 years earlier without evidence of recurrence in the intervening period)

Exclusion criteria: not reported

Cancer type: acute myeloid leukaemia

Cancer treatment: induction chemotherapy consisting of: idarubicin 9 mg/m² daily IV infusion on days 1 to 3; etoposide 75 mg/m² daily IV infusion on days 1 to 7; cytarabine 3 g/m² 12‐hourly IV infusion on days 1, 3, 5, and 7

All participants received G‐CSF (pegfilgrastim) 6 mg subcutaneously on day 8

Age at baseline (years): Group A: mean 46 (SD 12; range 17 to 60); Group B: mean 44 (SD 12; range 16 to 60)

Gender: Group A: 61% male; Group B: 67% male

Number randomised: 160 (Group A: 79; Group B: 81)

Number evaluated: 151 (Group A: 73; Group B: 78)

Interventions

Comparison: KGF (palifermin) versus placebo

Group A: KGF (60 µg/kg) daily IV on days ‐3, ‐2, ‐1 prior to chemotherapy and for 3 days after completion of chemotherapy (total dose = 360 µg/kg)

Group B: same schedule with placebo

Compliance: received all 3 pre‐chemotherapy doses: Group A: 97%; Group B: 100%; received all 3 post‐chemotherapy doses: Group A: 95%; Group B: 96%

Duration of treatment: 6 treatment days (over 14 days)

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed daily by investigators and specifically trained site personnel from the first day of chemotherapy and until the earlier of the date of discharge or day 28 after the start of chemotherapy, maximum score reported) (duration of grade 3 to 4 oral mucositis also measured but not an outcome of this review)

• Adverse events (NCI‐CTC version 2.0 toxicity scale)

• Incidence of severe gastrointestinal toxicities related to the induction chemotherapy (not an outcome of this review)

• Complete response to chemotherapy (not an outcome of this review)

Notes

Sample size calculation: 128 per group required to detect a reduction in grade 3 to 4 mucositis from 22% to 10% at 70% power and 5% significance

Funding: "This study was funded in part from Project Grant 302133 from the National Health and Medical Research Council of Australia" (government); KGF and placebo provided by Amgen (pharmaceutical industry)

Declarations/conflicts of interest: "The authors have no conflicts of interest to declare"

Data handling by review authors: N/A

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "Eligible patients were randomized 1:1 using a block randomization technique and stratification by participating centre to receive placebo or palifermin"

Comment: large multicentre trial using block randomisation and stratification ‐ likely to be done properly

Allocation concealment (selection bias)

Unclear risk

Quote: "Eligible patients were randomized 1:1 using a block randomization technique and stratification by participating centre to receive placebo or palifermin"

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "placebo‐controlled" and "Both investigators and patients were blinded to the randomization outcome"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "placebo‐controlled" and "Both investigators and patients were blinded to the randomization outcome"

Comment: investigators assessed oral mucositis, which would have partly relied on patient's assessment of pain/soreness and their ability to swallow; both investigators and patients were blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Overall attrition was 6% (Group A: 8%; Group B: 4%) for the oral mucositis incidence outcome. The reasons were similar between groups and this proportion of attrition is unlikely to have biased the results

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Australia, Canada and USA

Number of centres: 22

Study duration: September 1999 to May 2001
Trials registry number: none/unknown

Participants

Inclusion criteria: adults with newly diagnosed stage III/IVa or IVb squamous carcinoma of the oral cavity, oropharynx, nasopharynx, hypopharynx, and larynx (or with unknown primary and extensive neck disease) undergoing CRT intended to be curative; Karnofsky performance score of 60 or higher; haemoglobin 10 g/dL or higher; white blood cell count 3.5 x 10⁹/L or higher or absolute neutrophil count 1.5 x 10⁹/L or higher; platelet count 100 x 10⁹/L or higher; serum bilirubin 1.5 mg/dL or lower; serum creatinine lower than 2.0 mg/dL (plus a 24‐hour urinary creatinine clearance 50 mL/min in those aged 60 years or older)

Exclusion criteria: previous RT to the head and neck; previous surgery for the primary tumour (not including biopsy); previous CT; allergy to Escherichia coli‐derived products; participation in any other investigational study within the 30 days prior to this study; refusal to use adequate contraception during the study; pregnant or breastfeeding

Cancer type: head and neck: oral (Group A: 12%; Group B: 6%); oropharynx/nasopharynx (Group A: 61%; Group B: 66%); hypopharynx/larynx (Group A: 27%; Group B: 28%)

Cancer treatment:

• Radiotherapy: standard (once daily 2‐Gy fractions, 5 days per week; total 70 Gy over 7 weeks) or hyperfractionated (twice daily 1.25‐Gy fractions with 6‐hour interval, 5 days per week with an 8 to 9 day break after 3 weeks; total 72 Gy over 6.5 weeks)

• Chemotherapy: cisplatin (20 mg/m²/day) as IV bolus injection and 5FU (1000 mg/m²/day) as continuous infusion on the first 4 days of the first and fifth weeks of radiotherapy

Age at baseline (years): Group A: mean 54 (SD 10; range 25 to 80); Group B: mean 56 (SD 10; range 42 to 75)

Gender: Group A: 82% male; Group B: 84% male

Number randomised: 101 (Group A: 69; Group B: 32)

Number evaluated: 97 (Group A: 65; Group B: 32)

Interventions

Comparison: KGF (palifermin) versus placebo

Group A: KGF 60 µg/kg once weekly by IV bolus injection starting on the Friday before CRT began (on the following Monday), then each Friday after completion of RT for 7 weeks, and 2 more doses after completion of CRT i.e. 10 doses in total (total dose = 600 µg/kg)

Group B: same schedule with matching placebo

Compliance: 99 participants (Group A: 67; Group B: 32) received at least 1 dose of their allocated intervention; 69 participants completed the full course (Group A: 47; Group B: 22); mean number of doses (Group A: 8.4; Group B: 9.1)

Duration of treatment: 9 weeks (10 doses)

Outcomes

• Oral mucositis: NCI‐CTC (2.0) 0 to 4 scale (measured weekly by a radiation oncologist for the first 12 weeks, reported as incidence of grade 2 to 4 i.e. moderate to severe, and grade 3 to 4 i.e. severe) (duration, time to onset, and cumulative radiotherapy dose at onset of grade 2+ and 3+ oral mucositis also measured but not outcomes of this review)

• Interruptions to cancer treatment (unscheduled radiotherapy breaks: reported as any breaks and breaks longer than 4 days)

• Normalcy of diet (measured as incidence of supplemental nutrition by a gastrostomy tube)

• Adverse effects (reports collected throughout the 20‐week study)

• Opioid analgesic use (reported as incidence; number of days of treatment with opioid analgesics is an outcome of this review and therefore we did not use these data)

• Dysphagia (not an outcome of this review)

• Xerostomia (not an outcome of this review)

• Antibiotic use (not an outcome of this review)

• Tumour response rate (not an outcome of this review)

• Progression‐free and overall survival (assessed in a longer‐term follow‐up study) (not an outcome of this review)

Notes

Sample size calculation: based on a previous study, 99 participants required to detect a 30% difference in the duration of grade 2 or higher oral mucositis with 80% power provided that the mean duration in the placebo arm was 56 days

Funding: "Supported by Amgen Inc" (pharmaceutical industry)

Declarations/conflicts of interest: multiple and involving: employment or leadership positions with the funders (Amgen); consultant or advisory roles with the funders and other pharmaceutical companies; stock ownership with the funders; honoraria from the funders and other pharmaceutical companies; research funding from the funders and other pharmaceutical companies

Data handling by review authors: the data for incidence of mucositis were presented in subgroups of those that received standard or hyperfractionated RT but we used the overall data in our meta‐analyses; for the interruptions to radiotherapy outcome, we used the data for breaks longer than 4 days as these could be pooled with other studies in this comparison

Other information of note: the study authors report a greater decrease in incidence due to palifermin in the hyperfractionated subgroup than in the standard subgroup (see figure 3A in the study report)

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "Patients were randomly assigned..."

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Unclear risk

Quote: "Patients were randomly assigned..."

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double blinded" and "Palifermin...or matching placebo was administered by intravenous bolus"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double blinded" and "Palifermin...or matching placebo was administered by intravenous bolus"

Comment: it is not clear who was blinded. There are subjective elements to the assessment of oral mucositis using this scale, requiring the patient's assessment of pain/soreness and their ability to swallow but, as the participants were unaware of their group allocation, the assessment of oral mucositis can be considered to be blinded. The other outcomes are objective and therefore unlikely to be affected by any potential lack of blinding of the outcome assessor(s)

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Overall attrition was 4% (Group A: 6%; Group B: 0%) for the oral mucositis incidence outcome. The reasons were unclear but this proportion of attrition is unlikely to have biased the results

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (5 arms) dose‐ranging study

Location: Duke University Medical Centre, Durham, North Carolina, USA

Number of centres: 1

Study duration: not reported
Trials registry number: none/unknown

Participants

Inclusion criteria: premenopausal or perimenopausal patients with histologically confirmed metastatic breast cancer who had chemotherapy for inoperable or metastatic disease; performance status of 0 or 1 (CALGB criteria)

Exclusion criteria: metastatic disease involving the central nervous system; pregnant

Cancer type: stage IV breast

Cancer treatment: AFM regimen (21‐day cycle): 5FU (500 mg/m²/day) continuous infusion on days 1 to 5; adriamycin (25 mg/m²) IV bolus on days 3 to 5; methotrexate (250 mg/m²) IV on day 15 (if oral mucositis less than grade 3)

All participants received G‐CSF (filgrastim) subcutaneously (5 µg/kg/day) on days 7 to 13 and from day 16 until resolution of neutropenia

Age at baseline (years): mean 44 (not reported by group)

Gender: 49 female; 1 male (not reported by group)

Number randomised: 50 (not reported by group)

Number evaluated: 45 (Group A: 36; Group B: 9)

Interventions

Comparison: GM‐CSF (molgramostim) versus placebo

Group A: GM‐CSF

• (n = 9 analysed): 15 mL of mouthwash (0.01 µg/mL) gently swirled in the mouth for 2 minutes before expectorating it; 4 times daily (after mealtimes, even if a meal was not eaten, and at bedtime) after routine oral care procedures; no eating/drinking for 15 minutes after using mouthwash; beginning within 24 hours of the start of 5FU and continued for the 21‐day AFM cycle (total dose = 12.6 µg)

• (n = 9 analysed): same schedule with 0.1 µg/mL mouthwash (total dose = 126 µg)

• (n = 9 analysed): same schedule with 1 µg/mL mouthwash (total dose = 1260 µg)

• (n = 9 analysed): same schedule with 10 µg/mL mouthwash (total dose = 12,600 µg)

Group B: (n = 9 analysed) same schedule with matching placebo mouthwash

Compliance: (not reported by group) mouthwash therapy was discontinued if the participant experienced oral mucositis of grade 3 or above; 30 participants took at least 80% of their prescribed doses; 11 participants discontinued mouthwash therapy within 3 days prior to day 15; 4 participants discontinued mouthwash therapy between day 15 and day 21

Duration of treatment: 21 days (first treatment cycle of AFM)

Outcomes

• Oral mucositis: CALGB 0 to 4 scale (measured on days 1 to 5, 8 to 10, 15 and 22, reported as incidence of grade 3 to 4 i.e. severe) (duration of grade 3 to 4 oral mucositis also measured but not an outcome of this review)

• Adverse effects (assessed during study period)

• Blood measurements (platelet, WBC, granulocyte, lymphocyte) (not an outcome of this review)

• Myelosuppression (not an outcome of this review)

• GM‐CSF plasma concentrations (not an outcome of this review)

Notes

Sample size calculation: this was done but the numbers required are not reported

Funding: "...supported in part by National Cancer Institute (Bethesda, MD) grant number PO1‐47741‐A4"

Declarations/conflicts of interest: not reported

Data handling by review authors: we combined the 4 GM‐CSF groups to make a single pairwise comparison against placebo and, in order to make a head‐to‐head comparison of doses, we grouped the 2 lower doses (0.01 µg/mL and 0.1 µg/mL) together and grouped the 2 higher doses (1 µg/mL and 10 µg/mL) together to make pairwise groups for comparison

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "Patients were randomized by the Duke Cancer Center Protocol Office according to a block randomization scheme and assigned a unique identifier number which designated the GM‐CSF dose level to be received"

Comment: method of sequence generation not fully described but was done by a dedicated specialist centre so was probably done adequately

Allocation concealment (selection bias)

Low risk

Quote: "Patients were randomized by the Duke Cancer Center Protocol Office" and "The patient supply of mouthwash was labelled to correspond with the assigned identifier number and dispensed by the Pharmacy. The patient assignment information was maintained by the Pharmacy"

Comment: the entire randomisation process was performed by third party so the random sequence is unlikely to have been manipulated

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled" and "The patient assignment information was maintained by the Pharmacy"

Comment: it is not clear who was blinded. There are subjective elements to the assessment of oral mucositis using this scale, requiring the patient's assessment of pain/soreness and their ability to swallow but, as the participants were unaware of their group allocation, the assessment of oral mucositis can be considered to be blinded

Incomplete outcome data (attrition bias)
All outcomes

Unclear risk

Overall attrition was 10%. Reasons for attrition fully reported. If all participants would have developed severe oral mucositis or dropped out due to severe oral mucositis and were all from 1 particular group, this would have biased the results. However, attrition was not reported by group, so it is unclear

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Italy

Number of centres: 4

Study duration: May 2007 to June 2011
Trials registry number: EudraCT 2007‐001430‐14 (mentioned in trial report)

Participants

Inclusion criteria: aged 0 to 17 years with leukaemia, lymphoma or solid tumour; due to receive a first autologous PBSCT

Exclusion criteria: not reported

Cancer type: leukaemia/lymphoma (Group A: 22%; Group B: 17%); solid (Group A: 78%; Group B: 83%) (neuroblastoma, Ewing sarcoma/peripheral neuroectodermal tumour, medulloblastoma, Wilms tumour, central nervous system tumour)

Cancer treatment: all participants had autologous PBSCT on day 0 but differed in conditioning regimen as follows:

• Chemotherapy: multiple regimens involving 1 to 4 chemotherapy drugs; the most common regimen was busulfan 16 mg/kg with melphalan 140 mg/m² (Group A: 53%; Group B: 37%)

• Radiotherapy: only 4 participants (2 in each arm) had TBI 12 Gy to 14.4 Gy prior to their chemotherapy

Age at baseline (years): Group A: median 11.1 (range 1.7 to 17.4); Group B: median 11.9 (range 1.6 to 17.2)

Gender: Group A: 66% male; Group B: 59% male

Number randomised: 61 (Group A: 32; Group B: 29)

Number evaluated: 61 (Group A: 32; Group B: 29)

Interventions

Comparison: G‐CSF (pegfilgrastim) versus G‐CSF (filgrastim)

Group A: pegfilgrastim single dose (100 µg/kg; maximum 6 mg) injected on day 3

Group B: filgrastim (5 µg/kg per day; maximum 300 µg per day) injected by 9 or more doses starting on day 3 (total dose = at least 45 µg/kg)

Mode of administration not described but presumably subcutaneously as in other G‐CSF studies

Compliance: all participants received their allocated intervention with no discontinuations

Duration of treatment: Group A: 1 day; Group B: 9 or more days

Outcomes

• Oral mucositis: WHO 0 to 4 scale (reported as incidence of any mucositis and grade 2 to 4 i.e. moderate to severe) (duration of any mucositis also measured but not an outcome of this review)

• Normalcy of diet (measured as incidence of TPN) (duration of TPN also measured but not used for analysis in review)

• Adverse events

• Number of days in hospital (reported as median and range, unable to use data)

• Polymorphonuclear cell recovery (not an outcome of this review)

• Time to platelet engraftment (not an outcome of this review)

• Incidence of febrile neutropenia and proven infection (not an outcome of this review)

• Duration of IV antibiotics (not an outcome of this review)

• Survival (not an outcome of this review)

Notes

Sample size calculation: based on the noninferiority of pegfilgrastim versus filgrastim in speeding the recovery of polymorphonuclear cells

Funding: "The authors have no support or funding to report"

Declarations/conflicts of interest: "The authors have declared that no competing interests exist"

Data handling by review authors: N/A

Other information of note: G‐CSF administration only began after the chemotherapy and PBSCT were completed, by which point oral mucositis may have already begun to develop

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "A computer‐generated randomisation list was drawn up at Data Office Centre of AIEOP in Bologna, Italy, by a statistician not involved in patient management"

Comment: adequate method used

Allocation concealment (selection bias)

Low risk

Quote: "The list was stored by sequentially numbered sealed envelopes that was concealed to investigators until the completion of recruitment. The local investigator ...assigned each eligible patient to randomization list by phoning to AIEOP Data Office Centre"

Comment: ideal method of concealment used

Blinding of participants and personnel (performance bias)
All outcomes

High risk

Treatment regimens were different so blinding not possible

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

It would be possible to blind the outcome assessor for oral mucositis, but it was not mentioned

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: cross‐over (2 arms)

Location: Cancer Centre and Department of Otolaryngology, Veterans General Hospital, Taiwan, Republic of China

Number of centres: 1

Study duration: not reported
Trials registry number: none/unknown

Participants

Inclusion criteria: diagnosed stage IV SCC of head and neck, previously untreated or locally recurrent after previous surgery or radiotherapy or both; ECOG score of 2 or above; adequate bone marrow, liver and renal function

Exclusion criteria: concurrent medical illness; local radiotherapy to oropharynx region in the previous 3 months

Cancer type: head and neck: nasopharyngeal (Group A: 44%; Group B: 27%); tongue (Group A: 22%; Group B: 36%); hypopharynx (Group A: 11%; Group B: 18%); buccal (Group A: 11%; Group B: 9%); tonsillar (Group A: 11%; Group B: 9%)

Cancer treatment: PFL regimen (21‐day cycle): cisplatin (20 mg/m²/day), 5FU (800 mg/m²/day) and leucovorin (90 mg/m²/day) IV for days 1 to 4; cycle repeated every 3 weeks (study consisted of 2 cycles)

Age at baseline (years): Group A: median 44 (range 36 to 62); Group B: median 49 (range 40 to 66)

Gender: Group A: 89% male; Group B: 91% male

Number randomised: 20 (Group A: 9; Group B: 11) ‐ figures for first cycle

Number evaluated: 20 (Group A: 9; Group B: 11)

Interventions

Comparison: GM‐CSF versus no treatment

Group A: GM‐CSF (4 µg/kg) subcutaneously from day 5 to 14 (total dose = 40 µg)

Group B: no treatment

Compliance: not reported

Duration of treatment: 10 days (days 5 to 14 of 21‐day cycle)

Outcomes

• Oral mucositis: RTOG 0 to 4 scale (measured on days 5 to 21 by both physician's objective gross score and participant's subjective functional score, reported as area under the curve and also in the text as incidence of severe gross mucositis i.e. grade 3 to 4) (duration of grade 2 to 4 and 3 to 4 oral mucositis also measured but not an outcome of this review)

• Adverse effects (assessed across both cycles)

Notes

Sample size calculation: not reported

Funding: "..supported in part by Department of Health, Taiwan, Republic of China, research grant no. DOH 83‐HR‐202" and "GM‐CSF (supplied by Schering Plough Corp, Kenilworth, NJ)" (pharmaceutical industry)

Declarations/conflicts of interest: not reported

Data handling by review authors: as stated in the methods section, we would only include first‐period data from cross‐over studies due to potential for period effects (which were reported in this study). The only usable data in this study were reported in the text as incidence of severe gross mucositis for the first cycle

Other information of note: the authors report a significant period effect of GM‐CSF (P < 0.01), whereby the benefits continued into the second cycle

GM‐CSF administration only began after the 4‐day chemotherapy was completed, by which point oral mucositis may have already begun to develop

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "Patients were randomized to receive GM‐CSF or no therapy"

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Unclear risk

Quote: "Patients were randomized to receive GM‐CSF or no therapy"

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

High risk

Comparison with no treatment so blinding not possible

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

It would be possible to blind the outcome assessor, as the data we used were assessed by a physician using an objective scale. However, it was not mentioned

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants appear to be included in the analyses

Selective reporting (reporting bias)

Low risk

Although most of the data were not usable in this review, this does not seem to be due to selective reporting

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: USA

Number of centres: 14

Study duration: recruitment from May 1988 to November 1989
Trials registry number: none/unknown

Participants

Inclusion criteria: newly diagnosed small‐cell lung cancer meeting standard criteria for end‐organ function; ECOG score of 0 to 2

Exclusion criteria: previous radiotherapy; other serious medical illnesses precluding participation

Cancer type: small‐cell lung cancer

Cancer treatment: CAE regimen (21‐day cycle): cyclophosphamide (1000 mg/m²) and doxorubicin (50 mg/m²) on day 1; etoposide (120 mg/m²) on days 1 to 3; all by IV; repeated for up to 6 cycles

Age at baseline (years): Group A: mean 61 (SD 10; range 31 to 78); Group B: mean 62 (SD 8; range 31 to 80)

Gender: Group A: 65% male; Group B: 63% male

Number randomised: 211 (Group A: 101; Group B: 110)

Number evaluated: 195 (Group A: 93; Group B: 102) ‐ figures for first cycle

Interventions

Comparison: G‐CSF (r‐metHuG‐CSF) (filgrastim) versus placebo

Group A: G‐CSF (230 µg/m²) self‐administered subcutaneously on days 4 to 17 (total dose = 3220 µg/m²)

Group B: as above but with placebo

G‐CSF stopped if postnadir neutrophil count exceeded 10 x 10⁹/L after day 12; participants kept receiving their allocated intervention until they experienced fever with neutropenia, then they received unblinded G‐CSF (230 µg/m²) in subsequent cycles; participants in the G‐CSF group who experienced fever with neutropenia were allowed 25% reduction in chemotherapy dosages in subsequent cycles

Compliance: not reported

Duration of treatment: 14 days during a 21‐day cycle

Outcomes

• Oral mucositis: WHO 0 to 4 scale (measured weekly, reported as incidence of any mucositis) (duration and time to onset of oral mucositis also measured but not an outcome of this review)

• Adverse effects (assessed over the 6 cycles)

• Number of days in hospital (reported graphically in secondary trial report with no SD or P value, no usable data)

• Incidence, duration and severity of fever with neutropenia (not an outcome of this review)

• Incidence and duration of antibiotic use (not an outcome of this review)

Notes

Sample size calculation: based on a difference of 20% in the incidence of fever with neutropenia over the 6 cycles

Funding: "The study was designed, coordinated, and analyzed in conjunction with Amgen, the supplier of the G‐CSF"

Declarations/conflicts of interest: not reported but some authors were employed by Amgen (pharmaceutical industry)

Data handling by review authors: for oral mucositis, we only used the data from the first cycle due to the reasons listed above (under 'Interventions')

Other information of note: G‐CSF administration only began after the 3‐day chemotherapy was completed, by which point oral mucositis may have already begun to develop.

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "The patients were randomly assigned to chemotherapy followed by study drug (either placebo or G‐CSF)"

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Unclear risk

Quote: "The patients were randomly assigned to chemotherapy followed by study drug (either placebo or G‐CSF)"

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "Placebo was supplied in matching vials for double blinding"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "Placebo was supplied in matching vials for double blinding"

Comment: it is not clear who was blinded. There are subjective elements to the assessment of oral mucositis using this scale, requiring the patient's assessment of pain/soreness and their ability to swallow but, as the participants were unaware of their group allocation, the assessment of oral mucositis can be considered to be blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Overall attrition was 8% (Group A: 8%; Group B: 7%) for the oral mucositis incidence outcome. The reasons were reported and similar between groups, and this proportion of attrition is unlikely to have biased the results

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Ravenna, Italy

Number of centres: 1

Study duration: recruitment from July 1997 to February 2002
Trials registry number: none/unknown

Participants

Inclusion criteria: 14 years of age or older; hospitalised for high‐dose chemotherapy with autologous PBSCT

Exclusion criteria: not reported

Cancer type: breast (Group A: 17.5%; Group B: 27.5%); Ewing's sarcoma (Group A: 41.5%; Group B: 28.5%); osteosarcoma (Group A: 13%; Group B: 14%); NHL (Group A: 15%; Group B: 11.5%); germ cell tumours (Group A: 11%; Group B: 16%); small‐cell lung (Group A: 2%; Group B: 0%); soft tissue sarcoma (Group A: 0%; Group B: 2.5%)

Cancer treatment: high‐dose chemotherapy with autologous PBSCT; chemotherapy regimens were categorised into high risk and low risk and this was used as a stratification factor for randomisation, therefore high‐ and low‐risk participants were equally distributed across groups

All participants received subcutaneous G‐CSF (300 µg/day) until haematopoietic reconstitution

Age at baseline (years): Group A: median 29 (range 15 to 57); Group B: median 29 (range 17 to 61)

Gender: Group A: 59% male; Group B: 55% male

Number randomised: 90 (Group A: 46; Group B: 44)

Number evaluated: 90 (Group A: 46; Group B: 44)

Interventions

Comparison: GM‐CSF versus placebo

Group A: GM‐CSF mouthwash 150 µg/day in 100 cm³ of sterile water taken in 4 doses per day; mouthrinsing performed for 1 minute each time; treatment started on the day after the completion of chemotherapy and continued until bone marrow recovery (absolute neutrophil count > 500/mm³) or resolution of mucositis if still persistent after bone marrow recovery (total dose = variable)

Group B: as above but with placebo (sterile water)

All participants received 0.2% oral chlorhexidine and amphotericin B

Compliance: all but 7 participants regularly completed mouthwashes: 1 participant in the placebo group had none due to persistent vomiting; 6 (4 in placebo group and 2 in GM‐CSF group) started treatment but interrupted it early due to nausea and vomiting

Duration of treatment: variable and dependent on bone marrow recovery/resolution of mucositis

Outcomes

• Oral mucositis: NCI‐CTC 0 to 4 scale (measured daily by the physicians, reported as incidence of mucositis and incidence of grade 3 to 4 i.e. severe) (duration of grade 3 to 4 oral mucositis also measured but not an outcome of this review)

• Oral pain: 0 to 10 VAS (self‐evaluated daily, reported as mean worst score experienced)

• Number of days of treatment with opioid analgesics (also reported as incidence; we did not use these data)

Notes

Sample size calculation: 90 participants required to detect 25% minimal difference in the rate of severe mucositis at 90% power and 5% significance

Funding: no external funding (from correspondence with authors)

Declarations/conflicts of interest: not reported

Data handling by review authors: the data for incidence of mucositis were presented in subgroups of those at low or high risk of mucositis but we used the overall data in our meta‐analyses

Other information of note: there does not appear to be any difference in risk of any or severe mucositis between the low‐ and high‐risk subgroups

GM‐CSF administration only began after the chemotherapy was completed, by which point oral mucositis may have already begun to develop

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote from correspondence with authors: "The randomization list was centralized"

Comment: centralised randomisation method ‐ likely to be done properly

Allocation concealment (selection bias)

Low risk

Quote from correspondence with authors: "The randomization list was centralized"

Comment: centralised randomisation method ‐ likely to be done properly

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, randomized, placebo‐controlled study" and "The color, odor, texture and taste of both solutions were virtually identical"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind, randomized, placebo‐controlled study"

Comment: it is not clear who was blinded. There are subjective elements to the assessment of oral mucositis using this scale, requiring the patient's assessment of pain/soreness and their ability to swallow but, as the participants were unaware of their group allocation, the assessment of oral mucositis can be considered to be blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Department of Hematology and Oncology, University Hospital Freiburg, Germany

Number of centres: 1

Study duration: March 2006 to December 2010
Trials registry number: EudraCT 2008‐001833‐87; DRKS00000043

Participants

Inclusion criteria: adults aged 18 to 75 years with either: high‐grade non‐Hodgkin's lymphoma with high‐risk syndrome (> 2 risk factors according to age‐adapted IPI = international prognostic index) in the first complete remission; Hodgkin's lymphoma in the first recurrence; recurrence of follicular lymphoma; primary therapy of a coat‐cell lymphoma (MCL) in stage II‐IV; due to receive BEAM chemotherapy followed by autologous PBSCT; Karnofsky performance score more than 60%; life expectancy more than 3 months

Exclusion criteria: previous therapy using palifermin; severe concomitant diseases with organ failure; pregnancy, lactation, positive pregnancy test; hypersensitivity to 1 of the trial drugs; severe psychiatric illness; HIV disease or immunologic deficiency; known central nervous system involvement

Cancer type: haematologic: diffuse large‐cell lymphoma (Group A: 33%; Group B: 42%); B‐cell type acute lymphocytic leukaemia (Group A: 10%; Group B: 6%); T‐cell non‐Hodgkin's lymphoma (Group A: 13%; Group B: 11%); follicular/mantle cell lymphoma (Group A: 27%; Group B: 28%); Hodgkin's lymphoma (Group A: 17%; Group B: 14%)

Cancer treatment: prior to receiving autologous PBSCT on day 0, participants received BEAM conditioning regimen from day ‐8 to ‐2: carmustine (BCNU) 300 mg/m²; etoposide 800 mg/m²; cytosine arabinoside 1600 mg/m²; melphalan 140 mg/m²

Age at baseline (years): (ITT population) Group A: median 50 (range 22 to 71); Group B: median 55 (range 22 to 73)

Gender: (ITT population) Group A: 57% male; Group B: 61% male

Number randomised: 73 (Group A: 37; Group B: 36)

Number evaluated: ITT: 66 (Group A: 30; Group B: 36); PP: 54 (Group A: 22; Group B: 32)

Interventions

Comparison: KGF (palifermin) plus best supportive care versus best supportive care alone

Group A: KGF (60 µg/kg) by IV daily for 3 days (days ‐10, ‐9, ‐8) prior to conditioning regimen and autologous PBSCT and then for 3 days after (days 0, 1, 2) (total dose = 360 µg/kg)

Group B: best supportive care ("effective oral hygiene like teeth brushing, oral rinsing") beginning on day ‐8 (the day of hospital admission for BEAM conditioning)

Compliance: Group A: 7/37 withdrew before therapy started, 3/37 had a different conditioning regimen to that specified in the study protocol (unclear if they still received KGF), 5/37 either had no KGF or did not receive all doses; Group B: 4/36 had a different conditioning regimen to that specified in the study protocol (unclear if they still received control intervention)

Duration of treatment: 6 treatment days (over 13 days)

Outcomes

• Oral mucositis: WHO 0 to 4 scale (measured daily during hospital stay by trained nurses, study assistant, or treating physician, maximum score reported) (duration of oral mucositis also measured but not an outcome of this review)

• Normalcy of diet (measured as incidence of TPN)

• Adverse events

• Number of days in hospital (medians reported, unable to use data)

• Number of days of treatment with opioid analgesics (medians reported, unable to use data)

• Survival (not an outcome of this review)

• Febrile neutropenia (not an outcome of this review)

Notes

Sample size calculation: 76 participants required to detect 30% difference in the rate of severe mucositis at 80% power and 5% significance

Funding: Amgen (pharmaceutical industry)

Declarations/conflicts of interest: not reported

Data handling by review authors: data for ITT population used

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "randomly assigned"

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Unclear risk

Quote: "randomly assigned"

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

High risk

Quote: "The randomization result was known to the patient as well as to the practitioners before the start of therapy"

Blinding of outcome assessment (detection bias)
All outcomes

High risk

Quote: "The randomization result was known to the patient as well as to the practitioners before the start of therapy"

Incomplete outcome data (attrition bias)
All outcomes

High risk

Overall attrition was 10% (Group A: 19%; Group B: 0%) for the ITT population. All 7 participants died before therapy started. Although this reason is not related to the outcomes, the balance created by randomisation may have been lost

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (3 arms)

Location: USA

Number of centres: 8

Study duration: not reported
Trials registry number: none/unknown

Participants

Inclusion criteria: aged 18 years or older; due to receive autologous HSCT with a conditioning regimen with a high propensity for producing mucositis (typically 50% incidence of NCI‐CTC grade 3 or 4 when receiving standard mucositis management); Karnofsky performance score of 70%; free of acute or significant chronic dental or periodontal disease at baseline examination

Exclusion criteria: previous HSCT; visible oral ulcerations at screening; pregnant or breastfeeding; childbearing potential or not using adequate contraception; history of allergy to Escherichia coli‐derived products; posterior subcapsular cataract identified at screening; history of thyroid disease prior to receiving chemotherapy (except for hypothyroidism adequately controlled with replacement therapy); history or clinical evidence of active significant acute or chronic diseases that may affect evaluation or interpretation of the effects of the study medication on mucositis; following medications: interleukin 11, topical steroids, sucralfate, hydrogen peroxide, pilocarpine, misoprostol, oral chlorhexidine rinses, or any agent that would affect the assessment of changes in the appearance of mucositis during the study

Cancer type: lymphoma (Group A: 64%; Group B: 71%; Group C: 57%); other haematologic malignancy (Group A: 36%; Group B: 29%; Group C: 43%)

Cancer treatment: prior to receiving autologous HSCT, participants received the following conditioning regimens:

• CBV (cyclophosphamide, etoposide, and carmustine) (Group A: 50%; Group B: 64%; Group C: 43%)

• melphalan monotherapy (Group A: 7%; Group B: 21%; Group C: 14%)

• melphalan combination (Group A: 21%; Group B: 14%; Group C: 28%)

• cyclophosphamide + TBI (total body irradiation) (Group A: 7%; Group B: 0%; Group C: 14%)

• thiotepa + TBI (Group A: 7%; Group B: 0%; Group C: 0%)

• cyclophosphamide + busulfan (Group A: 7%; Group B: 0%; Group C: 0%)

Age at baseline (years): Group A: mean 54 (SD 10); Group B: mean 47 (SD 10); Group C: mean 51 (SD 15)

Gender: Group A: 79% male; Group B: 64% male; Group C: 79% male

Number randomised: 42 (Group A: 14; Group B: 14; Group C: 14)

Number evaluated: 42 (Group A: 14; Group B: 14; Group C: 14)

Interventions

Comparison: KGF‐2 (repifermin) versus placebo

Group A: KGF‐2 (25 µg/kg) by IV daily for 3 days prior to conditioning regimen and autologous HSCT and then for 10 days after (total dose = 325 µg/kg)

Group B: KGF‐2 (50 µg/kg) as above (total dose = 650 µg/kg)

Group C: placebo as above

Compliance: not reported

Duration of treatment: 13 treatment days (over a longer period dependent on conditioning regimen)

Outcomes

• Oral mucositis: NCI‐CTC 0 to 4 scale (incidence of grade 2, 3 or 4, assessed prior to conditioning regimen, on day of HSCT, then 3 times per week until resolution of mucositis)

• Oral mucositis: OMAS 0 to 45 scale (reported as mean worst score and mean 3 worst scores, NCI‐CTC data used for analysis)

• Oral and oropharyngeal pain: 0 (no pain) to 10 (worse pain) scale (assessed on days reported above, reported as mean worst score experienced)

• Normalcy of diet: ability to eat 1 to 4 score, where 1 = normal, 2 = only soft solids, 3 = only liquids, 4 = no solids or liquids (assessed on days reported above, reported as mean worst score)

• Adverse events (assessed from the start of the intervention until 28 days after the final dose, reported as events with a statistically significant difference between groups or that occurred in at least 50% of participants in any group and differed between groups by at least 10%)

• Number of days of treatment with opioid analgesics (assessed as reported for oral mucositis outcome, reported as mean number of days due to mucositis pain, and mean number of days due to all pain; we used the former although acknowledge other studies typically do not specify whether or not they are reporting usage due to mucositis pain)

• Pain on swallowing (not an outcome of this review)

• Laboratory parameters (not outcomes of this review)

• Immunogenicity (not an outcome of this review)

• Electrocardiogram abnormalities, chest x‐ray assessments and ophthalmologic examinations (not outcomes of this review)

Notes

Sample size calculation: not reported

Funding: not reported

Declarations/conflicts of interest: not reported

Data handling by review authors: we combined the 2 KGF‐2 groups to make a single pairwise comparison against placebo and we also made a comparison of the 2 different KGF‐2 dosages against each other

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote from correspondence with authors: "The method to implement the random allocation was by central telephone"

Comment: centralised randomisation method ‐ likely to be done properly

Allocation concealment (selection bias)

Low risk

Quote from correspondence with authors: "The method to implement the random allocation was by central telephone"

Comment: centralised randomisation method ‐ likely to be done properly

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blinded, placebo‐controlled"

Comment: the use of a placebo and identical schedule of treatment for all 3 arms should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blinded, placebo‐controlled"

Comment: it is not clear who was blinded. There are subjective elements to the assessment of oral mucositis using the NCI‐CTC scale, requiring the patient's assessment of pain/soreness and their ability to swallow but, as the participants were unaware of their group allocation, the assessment of oral mucositis can be considered to be blinded. This would be the same for other subjective outcomes. The objective outcomes are unlikely to be affected by any potential lack of blinding of the outcome assessor(s)

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Iran

Number of centres: not reported

Study duration: not reported
Trials registry number: IRCT2013021812510N1 (mentioned in trial report)

Participants

Inclusion criteria: previously untreated acute lymphoblastic leukaemia patients; aged between 5 and 18 years

Exclusion criteria: any other systemic disease; presence of oral mucositis or other oral lesions prior to chemotherapy; history of dermatology or respiratory hypersensitivity; acute lymphoblastic leukaemia recurrence

Cancer type: acute lymphoblastic leukaemia (ALL)

Cancer treatment: induction chemotherapy protocol consisted of standard risk B‐precursor ALL (COG)/dexamethasone, vincristine, L‐asparaginase, intrathecal (methotrexate + ara‐C + hydrocortisone). The intensification protocol was dexamethasone, vincristine, L‐asparaginase/ dexamethasone, cyclophosphamide/6‐thioguanine + cytarabine + intrathecal methotrexate

Age at baseline (years): Group A: mean 8.8 (SD 2.5); Group B: mean 8.4 (SD 2.2); overall range: 5 to 18

Gender: Group A: 49% male; Group B: 49% male

Number randomised: 90 (Group A: 45; Group B: 45)

Number evaluated: 90 (Group A: 45; Group B: 45)

Interventions

Comparison: KGF (palifermin) versus chlorhexidine

Group A: KGF (60 µg/kg) by IV bolus daily for 3 days prior to chemotherapy regimen and then for 3 days after (total dose = 360 µg/kg)

Group B: chlorhexidine (concentration not reported) mouthwash used for 1 minute once daily for 3 days prior to chemotherapy regimen and then for 3 days after

Compliance: not reported

Duration of treatment: 6 treatment days (over an unspecified longer period)

Outcomes

• Oral mucositis: WHO 0 to 4 scale (measured after 1 and 2 weeks and reported separately for each time point)

• Adverse events

Notes

Sample size calculation: not reported

Funding: not reported

Declarations/conflicts of interest: "There is no conflict of interest in relation to this study"

Data handling by review authors: we report the data at 2 weeks as they represent the maximum oral mucositis score experienced better than those at 1 week

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "The patients were randomly assigned to the palifermin or control group by using the table of random numbers"

Comment: adequate method used

Allocation concealment (selection bias)

Unclear risk

Quote: "The patients were randomly assigned to the palifermin or control group by using the table of random numbers"

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

High risk

Comparison with chlorhexidine so blinding not possible

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "Each patient was evaluated for oral lesions one and two weeks after the chemotherapy completion by the same specialist who was blind to the type of treatment" and "This limited use of chlorhexidine was to prevent the adverse effects like tooth discoloration and temporally taste changes"

Comment: grade 1 on this scale would require the unblinded participant's assessment of soreness but other aspects of the scale are more objective and were assessed by a blinded assessor. Also, blinding may not have been broken by staining/discolouration due to limited use of chlorhexidine

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Australia, Canada, and Europe (Austria, France, Germany, Italy, Spain, UK)

Number of centres: 38

Study duration: recruitment from January 2005 to August 2007
Trials registry number: NCT00131638; 2004‐002016‐28 (EudraCT number)

Participants

Inclusion criteria: more than 18 years old; resected for pathohistologically documented high‐risk stage 2 to 4B SCC of the oral cavity, oropharynx, hypopharynx, or larynx; ECOG score of 0 to 2; at least 2 of 9 areas of the oral or oropharyngeal mucosa due to receive at least 50 Gy RT

Exclusion criteria: tumours of the lips, paranasal sinuses, salivary glands, or unknown primary site; metastatic disease; history of chronic pancreatitis or acute pancreatitis within the last year; prior RT to the head and neck region or prior chemotherapy; previous treatment on this study or with other KGFs

Cancer type: head and neck: oropharynx (Group A: 47%; Group B: 48%); oral cavity (Group A: 32%; Group B: 27%); larynx (Group A: 11%; Group B: 15%); hypopharynx (Group A: 10%; Group B: 10%); other (Group A: 1%; Group B: 1%)

Cancer treatment: after R0 or R1 resection:

• radiotherapy: standard fractionation of once daily 2‐Gy fractions, 5 days per week; total 60 Gy (for R0 resection) over 6 weeks, or 66 Gy (for R1 resection) over 7 weeks, both with allowable range of ± 15%

• chemotherapy: cisplatin (100 mg/m²) by IV after appropriate hydration on days 1 and 22 (for R0 resection), or days 1, 22 and 43 (for R1 resection)

Age at baseline (years): Group A: mean 56 (SD 8); Group B: mean 57 (SD 9)

Gender: Group A: 85% male; Group B: 80% male

Number randomised: 186 (Group A: 92; Group B: 94)

Number evaluated: 186 (Group A: 92; Group B: 94)

Interventions

Comparison: KGF (palifermin) versus placebo

Group A: KGF (120 µg/kg) 3 days prior to start of, and then once per week during radiochemotherapy, i.e. 7 doses for those with R0 resection, 8 doses for those with R1 resection (total dose = 840 µg/kg or 960 µg/kg respectively)

Group B: same schedule with placebo

Mode of administration not described but presumably IV as in other KGF studies

Compliance: 78% of participants in KGF group completed all planned doses compared to 86% in placebo group

Duration of treatment: 7 or 8 treatment days (over 7 or 8 weeks), depending on R0 or R1 resection respectively

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed twice weekly by trained evaluators during radiochemotherapy and then until either mucositis had reduced to grade 2 or lower or week 15, whichever occurred first, maximum score reported) (duration and time to onset of grade 3 to 4 oral mucositis also measured but not outcomes of this review)

• Interruptions to cancer treatment: incidence of 5 or more missed consecutive RT fractions; incidence of chemotherapy delays/discontinuations

• Oral pain: OMWQ‐HN 0 (no soreness) to 4 (extreme soreness) scale for mouth and throat soreness

• Normalcy of diet (measured as incidence of supplemental feeding by TPN, PEG, nasogastric tube, or IV hydration) (broken down by overall supplemental feeding and also where due to oral mucositis; we used the latter although acknowledge other studies do not specify reason for supplemental feeding)

• Adverse events: reported as those with a difference in incidence of at least 5% between arms

• Use of opioid analgesics (total dose reported but not an outcome of this review)

• Xerostomia (not an outcome of this review)

• Weight change (not an outcome of this review)

• Laboratory assessments (not an outcome of this review)

• Survival (not an outcome of this review)

Notes

Sample size calculation: assuming 60% of placebo group would develop grade 3 to 4 mucositis, 90 per group required to detect a reduction of at least 25% at 90% power and 5% significance

Funding: "This study was supported by Amgen" (Amgen also named as sponsor on trials registry ‐ pharmaceutical industry)

Declarations/conflicts of interest: some authors had both employment or leadership positions and stock ownership within Amgen

Data handling by review authors: N/A

Other information of note: study originally randomised participants to 3 arms (180 µg/kg once per week for 7 weeks, 180 µg/kg once per week for 4 weeks followed by placebo for the next 3 doses, or placebo throughout) but, after 1 serious adverse event of respiratory insufficiency reported in 1 of the first 10 participants, the data monitoring committee decided to restart the study using 120 µg/kg doses, excluding the 17 randomised participants from the efficacy assessments. The arm with KGF for 4 weeks followed by placebo was stopped due to slow recruitment, after enrolment of 38 participants, and the results analysed in a separate appendix

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "Random assignment was made by a centralized interactive voice response system"

Comment: large multicentre trial using high‐tech randomisation method ‐ likely to be done properly

Allocation concealment (selection bias)

Low risk

Quote: "Random assignment was made by a centralized interactive voice response system"

Comment: large multicentre trial using high‐tech randomisation method ‐ likely to be done properly

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, randomized, placebo‐controlled"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind, randomized, placebo‐controlled"

Comment: it is not clear who was blinded. There are subjective elements to the assessment of lower grades of oral mucositis using the WHO scale, requiring the patient's assessment of pain/soreness and their ability to swallow. Higher grades have more objective elements so may not be affected by potential lack of blinding of the assessor. This would be the same for other subjective and objective outcomes

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Tehran University of Medical Sciences, Tehran, Iran

Number of centres: 1

Study duration: February 2014 to March 2015
Trials registry number: IRCT2015042518842N8 (mentioned in trial report)

Participants

Inclusion criteria: aged 18 years or older with non‐Hodgkin's lymphoma, Hodgkin's disease or multiple myeloma; due to receive autologous HSCT; adequate cardiac, pulmonary, renal and hepatic function

Exclusion criteria: Karnofsky performance score less than 70%; participation in another study using an unlicensed product

Cancer type: haematologic: non‐Hodgkin's lymphoma (Group A: 25%; Group B: 25%); Hodgkin's disease (Group A: 23%; Group B: 23%); multiple myeloma (Group A: 53%; Group B: 53%)

Cancer treatment: prior to receiving autologous HSCT, participants received the following conditioning regimens:

• Hodgkin's/non‐Hodgkin's: high‐dose combination chemotherapy (carboplatin 750 mg/m² IV daily for 2 days, etoposide 300 mg/m² IV daily for 2 days, cytarabine 300mg/m²/dose IV 2 doses in each day for 2 days, and melphalan 140 mg/m² IV for 1 day)

• Multiple myeloma: high‐dose melphalan (100 mg/m² IV daily for 2 days)

Age at baseline (years): Group A: mean 43 (SD 14); Group B: mean 45 (SD 16)

Gender: Group A: 55% male; Group B: 48% male

Number randomised: 80 (Group A: 40; Group B: 40)

Number evaluated: 80 (Group A: 40; Group B: 40)

Interventions

Comparison: Erythropoietin (recombinant human) versus placebo

Group A: 50 IU/mL erythropoietin mouthwash in aqueous vehicle (sodium benzoate, sodium citrate, citric acid, sodium hydroxide, sugar and distilled water) supplied in glass bottle stored at 4°C, 15 mL 4 times daily, starting from the first day of conditioning chemotherapy until 14 days after HSCT or until discharge from hospital (i.e. neutrophil recovery), whichever occurred first, oral intake not permitted for 1 hour following mouthwashing

Group B: same schedule with placebo (aqueous vehicle‐only)

All participants received oral hygiene care in addition to 20 drops of nystatin every 3 hours, mouthwashes containing 10 mL chlorhexidine 0.02% plus 10 mL diluted povidone iodine every 3 hours

Compliance: "However, it was a limitation of our study that EPO mouthwash administration might be affected by patients' low compliance" (no data reported)

Duration of treatment: variable and dependent on neutrophil recovery

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed daily by single trained pharmacist starting from the first day of conditioning chemotherapy and then until either 21 days after HSCT or mucositis had resolved, whichever occurred first, maximum score reported) (duration and time to onset reported but not outcomes of this review)

• Number of days in hospital

• Use of opioid analgesics (incidence reported but not an outcome of this review)

• Blood measurements (not an outcome of this review)

• Incidence and duration of fever (not an outcome of this review)

Notes

Sample size calculation: 40 per group required assuming a 30% decrease in incidence of grade 2 to 4 mucositis at 5% significance and 80% power

Funding: "There was no applicable funding source for the clinical trial"

Declarations/conflicts of interest: "The authors have no conflict of interests to report"

Data handling by review authors: there is a discrepancy in the incidence of grade 2 to 4 mucositis between Figure 2 and Table 2 (the latter has 1 extra event per group). However, this does not change the effect estimate. We have used the data in Table 2 as it reports numbers of participants along with percentages

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "Patients were randomly allocated...in a blocked randomization schedule" and "Both patient randomization and drug preparation were performed in the pharmaceutical laboratory of Pharmacy Department"

Comment: method of random sequence generation not described but done by university hospital pharmacy and therefore probably done adequately

Allocation concealment (selection bias)

Low risk

Quote: "Both patient randomization and drug preparation were performed in the pharmaceutical laboratory of Pharmacy Department"

Comment: not explicitly described but pharmacy‐controlled randomisation should have ensured concealment of the random sequence from those recruiting participants

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, randomized, placebo‐controlled" and "The study participants, the attending physician and the outcome assessor were all blind to the treatment assignment" and "There were no differences in colour, flavour, taste or container of the study drug and the placebo"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind, randomized, placebo‐controlled" and "The study participants, the attending physician and the outcome assessor were all blind to the treatment assignment"

Comment: all parties were blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: USA (16 sites) and Australia (4 sites)

Number of centres: 20

Study duration: December 2005 to November 2008
Trials registry number: NCT00189488 (mentions obsolete number in trial report: NCT00964899)

Participants

Inclusion criteria: aged 18 years or older with haematologic malignancy (including myelodysplastic syndromes) and due to receive allogeneic SCT (marrow or PBPC) after a conditioning regimen; Karnofsky performance score of 70% or more; related donor or HLA‐matched unrelated donor identical at 6/6 HLA‐A, ‐B and ‐DRB1 loci (molecular typing of class I and class II for unrelated donors)

Exclusion criteria: other malignancies; prior SCT; previous use of KGF; active infection or oral mucositis; congestive heart failure (NYHA class III or IV); use of a T‐cell depleted graft for GVHD prophylaxis; inadequate renal, liver or pulmonary function; pregnant or breastfeeding; refusal to use adequate contraception during study; participation in another investigational device or drug trial in previous 30 days

Cancer type: haematologic: leukaemia (Group A: 71%; Group B: 79%); myelodysplastic syndrome (Group A: 16%; Group B: 12%); non‐Hodgkin's lymphoma (Group A: 12%; Group B: 8%); multiple myeloma (Group A: 0%; Group B: 1%); Hodgkin's disease (Group A: 1%; Group B: 0%)

Cancer treatment: prior to receiving allogeneic SCT on day 0, participants received 1 of the following conditioning regimens from day ‐11 to ‐2:

• cyclophosphamide plus TBI with or without etoposide

• TBI plus etoposide

• melphalan plus TBI > 11 Gy

• busulfan plus cyclophosphamide

• busulfan plus melphalan (fully ablative doses)

• fludarabine plus melphalan (fully ablative doses)

Participants received methotrexate (with a calcineurin inhibitor ‐ either cyclosporine or tacrolimus) for GVHD prophylaxis on days 1, 3 and 6 (planned), and on day 11 (if toxicity allowed) at doses of 15 mg/m², 10 mg/m², 10 mg/m² and 10 mg/m² respectively

Age at baseline (years): Group A: median 42 (range 18 to 62); Group B: median 44 (range 18 to 64)

Gender: Group A: 52% male; Group B: 63% male

Number randomised: 155 (Group A: 77; Group B: 78)

Number evaluated: 155 (Group A: 77; Group B: 78)

Interventions

Comparison: KGF (palifermin) versus placebo

Group A: KGF (60 µg/kg) by IV bolus daily for 3 days prior to start of conditioning therapy, then a single 180 µg/kg dose after conditioning, but often 1 or 2 days before SCT (total dose = 360 µg/kg)

Group B: same schedule with placebo

Compliance: received at least 1 dose: Group A: 99%; Group B: 96%; received all doses: Group A: 92%; Group B: 88%

Duration of treatment: 4 treatment days (over roughly 14 days)

Outcomes

• Oral mucositis: WHO 0 to 4 scale (reported as incidence of grade 2 to 4 i.e. moderate to severe, and grade 3 to 4 i.e. severe, assessed daily by trained evaluators from day ‐11 (first day of conditioning) and then until hospital discharge or day 28, whichever occurred first) (duration also measured but not an outcome of this review)

• Normalcy of diet (measured as incidence of TPN)

• Adverse events

• Use of opioid analgesics (incidence reported but not an outcome of this review)

• Incidence and severity of acute GVHD (not an outcome of this review)

Notes

Sample size calculation: based on GVHD (not met due to early stopping)

Funding: "This study was supported by research funding from Amgen Inc. Jonathan Latham of PharmaScribe, LLC received funding from Amgen Inc. to provide assistance with the preparation of the manuscript. Xuesong Guan of Amgen Inc. provided assistance with statistical analyses" (pharmaceutical industry)

Declarations/conflicts of interest: some authors were employees and stockholders of Amgen and some received compensation from Amgen for consultation

Data handling by review authors: N/A

Other information of note: planned sample size was 200 participants but the study was stopped due to slow recruitment

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "Subjects were randomly assigned..."

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Unclear risk

Quote: "Subjects were randomly assigned..."

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled"

Comment: it is not clear who was blinded but grades 2 to 4 on the WHO scale are sufficiently objective and unlikely to be affected by any lack of blinding of the assessors

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Saga Medical School, Saga, Japan

Number of centres: 1

Study duration: not reported
Trials registry number: none/unknown

Participants

Inclusion criteria: breast cancer patients

Exclusion criteria: not reported

Cancer type: breast: primary advanced (Group A: 2; Group B: 5); inflammatory (Group A: 4; Group B: 1); recurrent (Group A: 1; Group B: 1)

Cancer treatment: preoperative IA high‐dose adriamycin (10 mg to 40 mg every 2 to 3 days to a total dose of 70 mg to 170 mg)

Age at baseline (years): Group A: mean 53 (SD 11; range 38 to 69); Group B: mean 52 (SD 10; range 45 to 69)

Gender: all female

Number randomised: 14 (Group A: 7; Group B: 7)

Number evaluated: 14 (Group A: 7; Group B: 7)

Interventions

Comparison: G‐CSF versus no treatment

Group A: G‐CSF (125 µg) by daily subcutaneous injection until leukocyte counts > 8000/mm³; timing in relation to chemotherapy not specifically reported, but the group was further divided into 2 subgroups where one (n = 4) received G‐CSF during/as an adjunct to the chemotherapy, and the other (n = 3) received G‐CSF afterwards (after the leukocyte counts were likely to drop below 2000/mm³)

Group B: no treatment

Compliance: not reported

Duration of treatment: variable and dependent on leukocyte recovery (to > 8000/mm³)

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed once 1 to 7 days prior to chemotherapy and then every day afterwards by a single experienced examiner, reported as incidence of grade 2 to 4) (duration of grade 2 to 4 also measured but not an outcome of this review)

• Blood measurements (not an outcome of this review)

• Alopecia (not an outcome of this review)

• Adult respiratory distress syndrome (not an outcome of this review)

• Fever (not an outcome of this review)

Notes

Sample size calculation: not reported

Funding: "G‐CSF (Neutrogin) was provided by Chugai Pharmaceutical"

Declarations/conflicts of interest: not reported

Data handling by review authors: the data for incidence of mucositis were presented in subgroups of those receiving G‐CSF during or after chemotherapy but we used the overall data in our meta‐analyses

Other information of note: both cases of mucositis were in the subgroup who received G‐CSF after chemotherapy. Oral mucositis may have already begun to develop in this subgroup

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "randomized into two groups"

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Unclear risk

Quote: "randomized into two groups"

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

High risk

Comparison with no treatment so blinding not possible

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

It would be possible to blind the outcome assessor, as the data we used were assessed by an examiner looking for erythema and ulcers. However, it was not mentioned

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Seoul National University Hospital, Seoul, South Korea

Number of centres: 1

Study duration: recruitment from March 2009 to August 2014
Trials registry number: NCT00845819 (mentioned in trial report)

Participants

Inclusion criteria: aged 18 years or older with haematologic malignancy; due to receive intensive chemotherapy followed by autologous or allogeneic HSCT; normal oral cavity (grade 0 mucositis); ECOG score of 0 to 2

Exclusion criteria: received chemotherapy, radiotherapy or surgery within previous 3 weeks; history of allergy to the intervention or similar drugs; participation in other clinical trials within the previous 4 weeks with the potential to affect study results

Cancer type: haematologic: multiple myeloma (Group A: 57%; Group B: 55%); lymphoma (Group A: 36%; Group B: 38%); other (Group A: 7%; Group B: 7%)

Cancer treatment: prior to receiving autologous (Group A: 96%; Group B: 96%) or allogeneic HSCT, participants received the following conditioning regimens:

• high‐dose melphalan (Group A: 57%; Group B: 57%)

• mitoxantrone‐etoposide‐cytarabine‐melphalan (Group A: 15%; Group B: 23%)

• busulfan‐etoposide‐cytarabine‐melphalan (Group A: 19%; Group B: 16%)

• other (Group A: 9%; Group B: 4%)

Age at baseline (years): Group A: median 53 (range 18 to 65); Group B: median 51 (range 19 to 65)

Gender: Group A: 49% male; Group B: 54% male

Number randomised: 138 (Group A: 69; Group B: 69)

Number evaluated: 136 (Group A: 67; Group B: 69)

Interventions

Comparison: EGF (recombinant human) versus placebo

Group A: EGF (50 µg/mL) daily by oral spray, applied twice daily, sprayed (6 sprays per application) over the entire oral mucosa and then swallowed, no oral intake for 30 minutes afterwards; starting on first day of conditioning therapy and continuing until absolute neutrophil count recovered more than 1000 µL for 3 days and mucositis had resolved

Group B: placebo as above

Compliance: median patient compliance rate: Group A: 93% (range 35% to 100%); Group B: 92% (range 18% to 100%)

Duration of treatment: variable and dependent on neutrophil recovery/resolution of mucositis

Outcomes

• Oral mucositis: NCI‐CTC (version 3.0) 0 to 4 scale (assessed daily during study period by researchers, reported as incidence of grade 2 to 4 i.e. moderate to severe, and grade 3 to 4 i.e. severe) (duration and time to onset reported but not outcomes of this review)

• Oral pain: mouth and throat soreness 0 to 10 scale (reported as AUC median/range and only for those who had grade 2 to 4 mucositis ‐ data not usable)

• Quality of life: modified OMDQ (reported as AUC median/range and only for those who had grade 2 to 4 mucositis ‐ data not usable)

• Normalcy of diet (use of total parenteral nutrition)

• Adverse events (NCI CTC version 3.0)

• Number of days in hospital (listed as an outcome but not reported anywhere in the results ‐ data not usable)

• Number of days of treatment with opioid analgesics (reported as median/range and only for those who had grade 2 to 4 mucositis ‐ data not usable)

• Incidence of febrile neutropenia (not an outcome of this review)

• Blood infections (not an outcome of this review)

• Antibiotic use (not an outcome of this review)

• Clinical laboratory measurements (not outcomes of this review)

Notes

Sample size calculation: 62 participants per group required to detect 27% difference in incidence of grade 2 to 4 mucositis with 80% power and 5% significance

Funding: multiple government grants; Daewoong Pharmaceutical Company (Seoul, Korea) only supplied interventions but provided no further funding and had no involvement with data collection, analysis or manuscript writing

Declarations/conflicts of interest: none apparent

Data handling by review authors: N/A

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "Patients were randomly assigned...using a computer‐generated randomization protocol, by the Medical Research Collaborating Center, Seoul National University Hospital"

Comment: adequate method used

Allocation concealment (selection bias)

Low risk

Quote: "Patients were randomly assigned...using a computer‐generated randomization protocol, by the Medical Research Collaborating Center, Seoul National University Hospital"

Comment: although concealment not explicitly mentioned, use of centralised/third party randomisation ‐ likely to be done properly

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "placebo‐controlled, double‐blind" and "clinicians, patients, and investigators responsible for assessing outcomes and analyzing data were masked to treatment assignments"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "placebo‐controlled, double‐blind" and "clinicians, patients, and investigators responsible for assessing outcomes and analyzing data were masked to treatment assignments"

Comment: all parties were blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Only 2 of 138 randomised participants were not included in the analysis

Selective reporting (reporting bias)

Low risk

Although most of the data were not usable in this review, this does not seem to be due to selective reporting

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Canada, USA and Europe (Hungary, Poland, Austria, Germany, Italy, Czech Republic)

Number of centres: 46

Study duration: recruitment started August 2005, 4‐month follow‐up finished September 2007
Trials registry number: NCT00101582; 2005‐000213‐35 (EudraCT number)

Participants

Inclusion criteria: newly diagnosed, unresected stage 3 to 4B SCC of the oral cavity, oropharynx, nasopharynx, hypopharynx, or larynx; no evidence of secondary malignancy; at least 2 of 9 areas of the oral or oropharyngeal mucosa due to receive more than 50 Gy RT

Exclusion criteria: not reported

Cancer type: head and neck: oropharynx (Group A: 59%; Group B: 54%); oral cavity (Group A: 5%; Group B: 10%); larynx (Group A: 17%; Group B: 10%); hypopharynx (Group A: 15%; Group B: 23%); nasopharynx (Group A: 4%; Group B: 3%)

Cancer treatment:

• Radiotherapy: standard fractionation of once daily 2‐Gy fractions, 5 days per week; total 70 Gy over 7 weeks

• Chemotherapy: cisplatin (100 mg/m²) by IV infusion on days 1, 22 and 43

Age at baseline (years): Group A: mean 56 (SD 9); Group B: mean 55 (SD 8)

Gender: Group A: 84% male; Group B: 85% male

Number randomised: 188 (Group A: 94; Group B: 94)

Number evaluated: 188 (Group A: 94; Group B: 94)

Interventions

Comparison: KGF (palifermin) versus placebo

Group A: KGF (180 µg/kg) by IV bolus over 30 to 60 seconds, 3 days prior to start of, and then once per week during radiochemotherapy, i.e. 8 doses (total dose = 1440 µg/kg)

Group B: same schedule with placebo

Compliance: 93% (SD 19%) of planned KGF doses were administered compared to 96% (SD 14%) in placebo group

Duration of treatment: 8 treatment days (over 8 weeks)

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed twice weekly by trained evaluators during radiochemotherapy and then until either mucositis had reduced to grade 2 or lower or week 15, whichever occurred first, maximum score reported) (duration and time to onset of grade 3 to 4 oral mucositis also measured but not outcomes of this review)

• Interruptions to cancer treatment: incidence of 5 or more missed consecutive RT fractions; incidence of chemotherapy delays/discontinuations

• Oral pain: OMWQ‐HN 0 (no soreness) to 4 (extreme soreness) scale for mouth and throat soreness (assessed twice weekly by trained evaluators during radiochemotherapy)

• Normalcy of diet (measured as incidence of supplemental feeding by TPN, PEG, nasogastric tube, or IV hydration) (duration of supplemental feeding also reported but not used for analysis)

• Adverse events: NCI‐CTC (version 3.0) reported separately for those related to study drugs

• Use of opioid analgesics (total dose reported but not an outcome of this review)

• Xerostomia (not an outcome of this review)

• Survival (not an outcome of this review)

• Laboratory assessments (not an outcome of this review)

• Antipalifermin antibodies (not an outcome of this review)

Notes

Sample size calculation: assuming 60% of placebo group would develop grade 3 to 4 mucositis, 90 per group required to detect a reduction of at least 25% at 90% power and 5% significance

Funding: "Supported by Amgen" (Swedish Orphan Biovitrum named as sponsor on trials registry, Amgen named as collaborator ‐ both pharmaceutical industry)

Declarations/conflicts of interest: some authors had both employment or leadership positions and stock ownership within Amgen; some authors had received research funding from Amgen

Data handling by review authors: N/A

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "A centralized randomization system assigned patients to either palifermin or placebo in a 1:1 ratio"

Comment: large multicentre trial using centralised randomisation method ‐ likely to be done properly

Allocation concealment (selection bias)

Low risk

Quote: "A centralized randomization system assigned patients to either palifermin or placebo in a 1:1 ratio"

Comment: large multicentre trial using centralised randomisation method ‐ likely to be done properly

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "placebo‐controlled, double‐blind"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "placebo‐controlled, double‐blind"

Comment: it is not clear who was blinded. There are subjective elements to the assessment of lower grades of oral mucositis using the WHO scale, requiring the patient's assessment of pain/soreness and their ability to swallow. Higher grades have more objective elements so may not be affected by potential lack of blinding of the assessor. This would be the same for other subjective and objective outcomes

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (5 arms)

Location: UK

Number of centres: 12

Study duration: recruitment from August 1989 to July 1990
Trials registry number: none/unknown

Participants

Inclusion criteria: adults due to receive BMT after conditioning

Exclusion criteria: myeloid malignancies

Cancer type: mixed haematologic and solid (not reported by group): Hodgkin's disease (29%); non‐Hodgkin's lymphoma (33%); multiple myeloma (Group A: 20%); ALL (15%); solid (3%)

Cancer treatment: prior to receiving BMT (autologous 84%; allogeneic 16%), participants received a conditioning regimen which consisted of chemotherapy only (71%) or with TBI (29%)

Age at baseline (years): median 36 (range 17 to 64) (not reported by group)

Gender: 69% male (not reported by group)

Number randomised: 121 (Group A: 96; Group B: 25); Group A represents 4 arms with different dosages

Number evaluated: 121 (Group A: 96; Group B: 25)

Interventions

Comparison: G‐CSF versus placebo

Group A: G‐CSF (2 µg/kg, 5 µg/kg, 10 µg/kg or 15 µg/kg) by 30‐minute IV daily starting from the day after BMT transplant and continuing until neutrophil count was > 1.0 x 10⁹/L for 3 consecutive days or until day 28, whichever occurred first

Group B: as above but with placebo

Compliance: not reported

Duration of treatment: variable and dependent on neutrophil recovery

Outcomes

• Oral mucositis: unspecified scale (frequency of measurement not mentioned, no usable data)

• Adverse events

• Number of days in hospital (median reported, no usable data)

• Neutropenia‐related outcomes (not outcomes of this review)

• Antibiotic use (not an outcome of this review)

• Fever (not an outcome of this review)

• Sepsis (not an outcome of this review)

• Blood product use (not an outcome of this review)

Notes

Sample size calculation: not reported

Funding: "Financial support for this trial was provided by Chugai Rhone Poulene" (pharmaceutical industry)

Declarations/conflicts of interest: 1 author was employed by the funders

Data handling by review authors: oral mucositis reported narratively in additional table

Other information of note: G‐CSF administration only began after the conditioning and BMT transplant were completed, by which point oral mucositis may have already begun to develop (although time scale of conditioning/BMT transplant not reported)

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "Patients were randomised...in blocks of five by a computer‐generated randomisation schedule"

Comment: adequate method used

Allocation concealment (selection bias)

Unclear risk

Quote: "Patients were randomised...in blocks of five by a computer‐generated randomisation schedule"

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "single blind...vehicle‐controlled"

Comment: the use of a placebo (the 'vehicle') should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

High risk

Quote: "single blind...vehicle‐controlled"

Comment: the quote implies that outcome assessment was not blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

High risk

Oral mucositis not mentioned in the methods section as 1 of the study end points. It is only mentioned in the results narratively as there being no difference between any group, but with no data or P value

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Vita‐Salute San Raffaele University Hospital, Milan, Italy

Number of centres: 1

Study duration: conducted from April 2009 to January 2015
Trials registry number: none/unknown

Participants

Inclusion criteria: children aged 7 years or older with B‐cell lineage acute lymphoblastic leukaemia; scheduled to receive autologous HSCT after a conditioning regimen; Karnofsky performance score of 70 or more; were to have at least 1.5 x 10⁶ CD34+ cells reinfused per kilogram available for transplant; adequate cardiac, pulmonary, renal and hepatic function

Exclusion criteria: not reported

Cancer type: B‐cell lineage acute lymphoblastic leukaemia

Cancer treatment: prior to receiving autologous HSCT (on day 0), participants received a conditioning regimen which consisted of TBI delivered in 8 fractions over 3 days (‐3 to ‐1) with at least 6 hours between fractions, followed by chemotherapy on day ‐1 (type and dose of radiotherapy and chemotherapy not reported)

Age at baseline (years): Group A: median 11 (range 7 to 16); Group B: median 11 (range 7 to 16)

Gender: Group A: 52% male; Group B: 44% male

Number randomised: 60 (Group A: 30; Group B: 30)

Number evaluated: 54 (Group A: 27; Group B: 27)

Interventions

Comparison: KGF (palifermin) versus placebo

Group A: KGF (60 µg/kg) by IV, on days ‐6 (3 days prior to start of conditioning regimen), ‐5 and ‐4, and on days 0 (the day of HSCT), 1, and 2 after transplant (total dose = 360 µg/kg)

Group B: same schedule with placebo

Compliance: not reported

Duration of treatment: 6 treatment days (over 9 days)

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed daily by same clinician from day ‐7 to day 28, or until severe mucositis had reduced to grade 0, 1, or 2, data in text and figure 2 do not agree with data in table 3) (duration of grade 2 to 4 and 3 to 4 oral mucositis also measured but not an outcome of this review)

• Oral pain: OMDQ 5‐point scale for mouth and throat soreness (higher = worse pain) (assessed daily by participant, reported as AUC, not used)

• Normalcy of diet (methods states incidence of supplemental feeding by TPN but only duration is reported, yet text states enteral and table states parenteral, not used for analysis) (patient‐reported difficulty eating and drinking also assessed daily, both using OMDQ 0 (no difficulty) to 5 (unable to do) scale, but the means are reported as whole numbers, data not used)

• Adverse events: NCI‐CTC (version 4.0)

• Opioid analgesic use (reported as quantity per day; number of days of treatment with opioid analgesics is an outcome of this review but only medians were reported, and therefore we did not use these data)

• GVHD incidence and severity (not an outcome of this review)

• Fever with neutropenia (not an outcome of this review)

• HSV incidence (not an outcome of this review)

• Candidiasis incidence (not an outcome of this review)

• Superinfections incidence (not an outcome of this review)

• Blood measurements (not an outcome of this review)

Notes

Sample size calculation: numbers required not reported but was estimated at 80% power and 5% significance

Funding: "This work was performed with Departmental funding only"

Declarations/conflicts of interest: the authors report that they have no conflict of interest (supplemental material on journal website)

Data handling by review authors: we emailed the lead author June 2017 for clarification of the oral mucositis data but, until we receive a response, we are unable to use those data

Other information of note: unclear definitions of ulcerative (should be grade 2 to 4) and severe (should be grade 3 to 4): "...ulcerative OM (WHO grades
3 and 4), incidence and duration of severe OM (WHO grades 3 and 4)..."

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "A centralized randomization system assigned patients either to palifermin or conventional treatment in a 1:1 ratio......The statistician gave randomization list to the pharmacy, so the patient and the clinical research team (who assessed outcomes) were blinded to the study treatment"

Comment: randomisation 'system' used and done by a statistician, likely to be done properly

Allocation concealment (selection bias)

Low risk

Quote: "A centralized randomization system assigned patients either to palifermin or conventional treatment in a 1:1 ratio......The statistician gave randomization list to the pharmacy, so the patient and the clinical research team (who assessed outcomes) were blinded to the study treatment"

Comment: centralised randomisation with pharmacy assigning participants to groups

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled" and "The statistician gave randomization list to the pharmacy, so the patient and the clinical research team (who assessed outcomes) were blinded to the study treatment. The pharmacy provided the research team with the blinded study medication"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "The statistician gave randomization list to the pharmacy, so the patient and the clinical research team (who assessed outcomes) were blinded to the study treatment. The pharmacy provided the research team with the blinded study medication"

Comment: outcome assessors were clearly blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Overall attrition was same in both groups (10%) with the same reason given

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Vita‐Salute San Raffaele University Hospital, Milan, Italy

Number of centres: 1

Study duration: conducted from April 2010 to January 2014
Trials registry number: none/unknown

Participants

Inclusion criteria: children aged 9 to 15 years with B‐cell lineage acute lymphoblastic leukaemia; scheduled to receive allogeneic HSCT after a conditioning regimen

Exclusion criteria: not reported

Cancer type: B‐cell lineage acute lymphoblastic leukaemia

Cancer treatment: prior to receiving allogeneic HSCT, participants received a conditioning regimen of a total of 12 Gy TBI delivered in 8 fractions of 1.5 Gy twice daily over 4 days

Age at baseline (years): Group A: median 12 (range 8 to 15); Group B: median 12 (range 8 to 15)

Gender: Group A: 50% male; Group B: 55% male

Number randomised: 46 (Group A: 24; Group B: 22)

Number evaluated: 46 (Group A: 24; Group B: 22)

Interventions

Comparison: KGF (palifermin) versus placebo

Group A: KGF (60 µg/kg) by IV, for 3 days prior to start of conditioning regimen and for 3 days after completion (total dose = 360 µg/kg)

Group B: same schedule with placebo

Compliance: not reported

Duration of treatment: 6 treatment days (over unclear period of time ‐ not fully described)

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed by the clinical research team at baseline, days 1, 2, 3 (it is not clear when this is in relation to the cancer treatment or study treatment) and at the end of the cycle ‐ "All of these results were confirmed at the last follow‐up (60 days)" ‐ it is not clear how they have been confirmed if, for example, OM has resolved by 60 days; it is not clear whether or not there were multiple cycles, incidence of each grade does not add up or make sense or match the data in table 3) (duration of grade 2 to 4 and 3 to 4 oral mucositis also measured but not an outcome of this review)

• Normalcy of diet (duration of supplemental feeding by TPN, not used for analysis) (patient‐reported difficulty eating and drinking assessed using OMDQ 0 (no difficulty) to 5 (unable to do) scale, but the means are reported as whole numbers, data not used)

• Adverse events

• Opioid analgesic use (number of days of treatment with opioid analgesics (morphine) reported but unclear if mean/median and no SD or P value reported; unable to use data; also reported as quantity per day but not an outcome of this review)

Notes

Sample size calculation: not reported

Funding: "This work has been performed with departmental funding only"

Declarations/conflicts of interest: "The authors certify that there is no conflict of interest with any financial organization regarding the material discussed in the manuscript"

Data handling by review authors: we emailed the lead author June 2017 for clarification of the oral mucositis data but, until we receive a response, we are unable to use those data

Other information of note: "Patients who had severe (grade 3 or 4) OM during blinded cycles received open‐label palifermin at the same dosages as in the other group. The research team assessed patients for OM at baseline before the cycle; at days 1, 2, 3 and at the end of the transplant cycle." ‐ nowhere else in the report suggests that there were multiple cycles of treatment; unclear definitions of ulcerative mucositis i.e. described correctly in one section as grades 2 to 4, but described in another section as grades 1 to 4

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "No stratification was performed and two distinct computer‐generated a randomization lists. The statistician gave both randomization lists to the pharmacy, so the patient and the clinical research team (who assessed outcomes) were blinded to the study treatment"

Comment: adequate method done by a statistician (although it is unclear why there were 2 lists)

Allocation concealment (selection bias)

Low risk

Quote: "No stratification was performed and two distinct computer‐generated a randomization lists. The statistician gave both randomization lists to the pharmacy, so the patient and the clinical research team (who assessed outcomes) were blinded to the study treatment"

Comment: randomisation 'system' used and done by a statistician, likely to be done properly

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled" and "The statistician gave both randomization lists to the pharmacy, so the patient and the clinical research team (who assessed outcomes) were blinded to the study treatment. The pharmacy provided the research team with the blinded study medication"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "The statistician gave both randomization lists to the pharmacy, so the patient and the clinical research team (who assessed outcomes) were blinded to the study treatment. The pharmacy provided the research team with the blinded study medication"

Comment: outcome assessors were clearly blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Turku University Central Hospital and Helsinki University Central Hospital, Finland

Number of centres: 2

Study duration: recruitment from November 1994 to August 1996
Trials registry number: none/unknown

Participants

Inclusion criteria: scheduled to receive total target dose of at least 56 Gy to the oropharyngeal mucosa

Exclusion criteria: prior chemotherapy or radiotherapy; concurrent use of anticholinergic drugs; autoimmune thrombocytopenic purpura; WHO performance status higher than 2

Cancer type: head and neck: mobile tongue (Group A: 30%; Group B: 20%); oral cavity, other (Group A: 40%; Group B: 25%); oropharynx (Group A: 20%; Group B: 15%); nasopharynx (Group A: 5%; Group B: 10%); supraglottic larynx (Group A: 5%; Group B: 15%); hypopharynx (Group A: 0%; Group B: 15%)

Cancer treatment:

• conventional radiotherapy (Group A: 50%; Group B: 50%): 1.9 Gy to 2 Gy daily fractions, 5 fractions per week

• hyperfractionated radiotherapy (Group A: 50%; Group B: 50%): 1.6 Gy fractions twice per day with minimum 6 hours between fractions, to a total dose of 38.4 Gy; planned break of 9 to 12 days during which the mucosa healed in order to allow further doses to a total of 56 Gy to 73 Gy; overall treatment time was 5 to 6 weeks

Total target dose: Group A: median 65 Gy (range 56 to 68); Group B: median 66 Gy (range 58 to 73); oral surgery prior to RT: Group A: 20%; Group B: 30%; oral surgery after radiotherapy: 23 (not reported by group)

Age at baseline (years): Group A: mean 63 (SD 10; range 47 to 87); Group B: mean 59 (SD 13; range 33 to 79)

Gender: Group A: 60% male; Group B: 55% male

Number randomised: 40 (Group A: 20; Group B: 20)

Number evaluated: unclear; results presented as percentages

Interventions

Comparison: GM‐CSF (molgramostim) plus sucralfate versus sucralfate alone

Both groups rinsed mouth with sucralfate (1 g) suspension for 1 minute before swallowing, 6 times per day, starting after 1 week of RT and continued until the end of RT (including weekends and planned/unplanned treatment breaks). Both groups advised to rinse their mouths with saline solution in between the sucralfate rinses if required

Group A: after cumulative radiation dose of 10 Gy, GM‐CSF (150 µg to 300 µg ‐ dependent on body weight) by daily subcutaneous injection until the last day of RT; not given over weekends or during planned/unplanned treatment breaks; mean total dose = 3398 µg (range 300 µg to 7200 µg)

Group B: no other treatment

Oral pain was treated with anti‐inflammatory analgesics and with local anaesthetic mouthwashes (lidocaine hydrochloride, Xylocain 0.5%)

Compliance: not reported

Duration of treatment: variable and dependent on duration of RT

Outcomes

• Oral mucositis: 0 to 2 scale where 0 = no mucositis, 1 = erythema and edema without ulcers but food intake/use of dental prosthesis was not affected, and 2 = one or more ulcers or bleeding or food intake/use of dental prosthesis was affected (assessed weekly during RT and 1 and 6 months after completion of RT, reported at start of RT and weekly for 4 weeks after; no usable data)

• Interruptions to cancer treatment (only 2 due to mucositis but not reported by group; data not usable)

• Oral pain: 0 to 10 VAS and also 1 to 4 scale where 1 = no pain, 2 = mild, 3 = moderate, and 4 = severe (assessed daily at midday during RT) (data not usable ‐ 1 to 4 scale reported as "no difference" and VAS reported on a graph with no SD or P values)

• Adverse events (due to GM‐CSF, sucralfate, or both)

• Saliva flow rates (not an outcome of this review)

• Salivary lactoferrin (not an outcome of this review)

• Blood measurements (not an outcome of this review)

• Body weight (not an outcome of this review)

• Overall survival (not an outcome of this review)

Notes

Sample size calculation: not reported

Funding: GM‐CSF and sucralfate were both provided by pharmaceutical industry (Schering‐Plough Corporation and Orion‐Farmos Pharmaceuticals respectively)

Declarations/conflicts of interest: not reported

Data handling by review authors: no usable data

Other information of note: GM‐CSF administration began after cumulative radiation dose of 10 Gy, by which point oral mucositis may have already begun to develop

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "assignment to the treatment groups was carried out via a phone call to the randomization center located at the Finnish Cancer Registry, Helsinki"

Comment: use of a specialist randomisation centre

Allocation concealment (selection bias)

Low risk

Quote: "assignment to the treatment groups was carried out via a phone call to the randomization center located at the Finnish Cancer Registry, Helsinki"

Comment: third party/remote randomisation would have ensured concealment of the random sequence from those recruiting participants

Blinding of participants and personnel (performance bias)
All outcomes

High risk

Blinding not possible as no placebo was used

Blinding of outcome assessment (detection bias)
All outcomes

Unclear risk

It would be possible to blind the outcome assessor for oral mucositis as the scale was fairly objective, but it was not mentioned. It would not be possible to blind oral pain assessment as this was done by the participant who was not blinded (however, we could not use the pain data so there was no bias for this outcome)

Incomplete outcome data (attrition bias)
All outcomes

Unclear risk

It is not clear how many participants were included in the analyses

Selective reporting (reporting bias)

High risk

Poor reporting of oral mucositis and oral pain

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Royal Marsden Hospital, London, UK

Number of centres: 1

Study duration: recruitment from September 1997 to October 2000
Trials registry number: NCT00004256

Participants

Inclusion criteria: histologically proven T1 N0 or T2 N0 glottic carcinoma; due to receive RT with a 16‐fraction 3‐week regimen; WHO performance status 0 or 1

Exclusion criteria: renal or hepatic impairment; serious infections requiring antibiotics; taking corticosteroids or likely to require them; known allergy to GM‐CSF

Cancer type: laryngeal

Cancer treatment: accelerated radiotherapy: once‐daily fractions of 3.125 Gy, to a total dose of 50 Gy in 16 fractions over 21 days

Age at baseline (years): Group A: median 60 (range 48 to 79); Group B: median 65 (range 32 to 70)

Gender: Group A: 93% male; Group B: 86% male

Number randomised: 29 (Group A: 15; Group B: 14)

Number evaluated: 29 (Group A: 15; Group B: 14)

Interventions

Comparison: GM‐CSF (molgramostim) versus no treatment

Group A: GM‐CSF (150 µg) by daily subcutaneous injection, starting on day 14 of RT and continuing for 14 days i.e. for the last week of RT and for 1 week after completion of RT (total dose = 2100 µg)

Group B: no treatment

Compliance: 2 participants did not complete their course of GM‐CSF

Duration of treatment: 14 days

Outcomes

• Oral mucositis: RTOG 0 to 4 scale (measured weekly by 1 of 2 independent observers using physician's objective criteria (see Appendix 9), reported graphically over time and by maximum score experienced in the text) (time to mucositis healing also measured but not an outcome of this review)

• Normalcy of diet: use of feeding tubes

• Adverse events

• Analgesic use ("No differences were detected in...analgesic usage" ‐ we do not know if this is number of days or whether or not opioids)

• Dysphagia and odynophagia (not outcomes of this review)

• Candida infection (not an outcome of this review)

• Laryngeal edema (not an outcome of this review)

• Moist or dry desquamation (not an outcome of this review)

• Weight change (not an outcome of this review)

• Skin erythema (not an outcome of this review)

Notes

Sample size calculation: 17 per group needed at 90% power and 5% significance to detect reduction from an anticipated 60% incidence of severe mucositis to 10%

Funding: not reported

Declarations/conflicts of interest: not reported

Data handling by review authors: N/A

Other information of note: imbalance in stage distribution, with more T2 patients in the GM‐CSF group. Consequently more of this group were treated with larger fields

GM‐CSF administration only began after 14 days of radiotherapy, by which point oral mucositis may have already begun to develop

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "were randomly assigned to the active or control arms"

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Unclear risk

Quote: "were randomly assigned to the active or control arms"

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

High risk

Blinding not possible as no placebo was used

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "At each visit one of two independent observers, blinded to group allocation, scored mucositis..."

Comment: a physician's objective version of the RTOG scale was used. Use of feeding tube is also an objective outcome

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Analgesic use reported as "No differences" with no data, however mucositis fully reported

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (7 arms)

Location: USA

Number of centres: 10

Study duration: not reported
Trials registry number: none/unknown

Participants

Inclusion criteria: 18 years or older; metastatic colon or rectal adenocarcinoma; scheduled to receive palliative 5FU and leucovorin chemotherapy; ECOG performance status 0 to 2 (ambulatory at least 50% of waking hours); life expectancy of at least 4 months; free of lesions and no recent history (within 30 days before baseline examination) of oral ulceration, herpes simplex, oral candidiasis, severe gingivitis, or the presence of active or chronic mucositis, xerostomia, or diarrhoea; absolute neutrophil count ≥ 1.5 x 109/L; platelet count ≥ 100 x 109/L; serum creatinine ≤ 2.0 mg/dL; serum bilirubin ≤ 2.0 mg/dL; serum aspartate amino transferase less than 5 times the upper limit of normal; absence of other serious concurrent medical illness

Exclusion criteria: received chemotherapy, radiotherapy, or other investigational drugs within 4 weeks of enrolment (6 weeks for chemotherapy with mitomycin or nitrosoureas); any unresolved adverse event from previous therapy; major surgery within 2 weeks before study entry; history of insulin‐dependent diabetes mellitus; pregnant or breastfeeding; of child‐bearing potential and not using adequate contraceptive precautions; previous hypersensitivity reaction to leucovorin calcium or Escherichia coli‐derived material

Cancer type: metastatic colorectal

Cancer treatment: (palliative) leucovorin 20 mg/m² by rapid IV injection, followed immediately by 5FU 425 mg/m² by rapid IV injection for 5 consecutive days on days 4 to 8 of each 28‐day cycle

Age at baseline (years): Group A: mean 62 (SD 10; range 44 to 84); Group B: mean 66 (SD 13; range 41 to 86)

Gender: Group A: 57% male; Group B: 59% male

Number randomised: 81 (Group A: 54; Group B: 27)

Number evaluated: 81 (Group A: 54; Group B: 27)

Interventions

Comparison: KGF (recombinant human) versus placebo

Group A: KGF

• (n = 7): 1 µg/kg per day by IV bolus on days 1 to 3 (total dose = 3 µg/kg)

• (n = 8): 10 µg/kg as above (total dose = 30 µg/kg)

• (n = 10): 20 µg/kg as above (total dose = 60 µg/kg)

• (n = 11): 40 µg/kg as above (total dose = 120 µg/kg)

• (n = 8): 60 µg/kg as above (total dose = 180 µg/kg)

• (n = 10): 80 µg/kg as above (total dose = 240 µg/kg)

Group B: placebo as above

Compliance: 3 participants stopped KGF treatment due to adverse reactions involving the skin (1 in 60 and 2 in 80 µg/kg group)

Duration of treatment: 3 days

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed on days 1, 4, 8, 15 and 28, reported as incidence of grade 2 to 4; first cycle data only) (duration of mucositis also measured but not an outcome of this review)

• Oral pain: self‐assessed daily questionnaire including mouth and throat pain assessed on both a 5‐point ordinal scale (reported graphically, no usable data) and a 0 to 10 VAS (reported as AUC, not used)

• Adverse events: WHO 0 to 4 scale (reported as events of any grade that differed by at least 10% between any KGF group and placebo; events were not reported if they occurred in less than 10% of KGF group; also reported as incidence of grade 3 to 4 events)

• Blood measurements (not an outcome of this review)

Notes

Sample size calculation: not reported

Funding: "Supported by a grant from Amgen, Inc" (pharmaceutical industry)

Declarations/conflicts of interest: not reported

Data handling by review authors: we combined the 6 KGF groups to make a single pairwise comparison against placebo; in order to make a head‐to‐head comparison of doses we grouped the 3 lower doses (1 µg/kg, 10 µg/kg and 20 µg/kg) together and grouped the 3 higher doses (40 µg/kg, 60 µg/kg and 80 µg/kg) together to make pairwise groups for comparison

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "This study was a multicenter, randomized, double‐blinded, placebo‐controlled, phase I study"

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Unclear risk

Quote: "This study was a multicenter, randomized, double‐blinded, placebo‐controlled, phase I study"

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blinded, placebo‐controlled"

Comment: the use of a placebo should have ensured that blinding was successful; incidence of adverse events does not appear to differ enough between groups to cause unblinding

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blinded, placebo‐controlled"

Comment: it is not clear who was blinded but the outcome is grade 2 to 4 WHO‐scale mucositis and it is unlikely that this would be affected by any lack of blinding

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: USA and Canada

Number of centres: 7

Study duration: 1 November 1990 to 1 July 1993
Trials registry number: none/unknown

Participants

Inclusion criteria: any age; undergoing allogeneic BMT (from HLA‐identical sibling) for haematological malignancy

Exclusion criteria: due to receive methotrexate as part of GVHD prophylaxis; chronic lymphocytic leukaemia; chronic myelogenous leukaemia in blast crisis; acute lymphoblastic leukaemia or acute myelogenous leukaemia with failure at first induction or progressed beyond a second relapse; previously received cytokines; HIV; life expectancy less than 7 days

Cancer type: haematologic malignancy: lymphoid malignancy (Group A: 21%; Group B: 16%); acute myeloid malignancy (Group A: 34%; Group B: 27%); chronic myeloid malignancy (Group A: 32%; Group B: 30%); multiple myeloma (Group A: 6%; Group B: 5%); myelodysplastic syndrome (Group A: 6%; Group B: 11%); aplastic anaemia (Group A: 2%; Group B: 11%)

Cancer treatment: prior to receiving allogeneic BMT (from HLA‐identical sibling), participants received the following conditioning regimens: busulfan and cyclophosphamide with TBI (Group A: 9%; Group B: 16%) or without TBI (Group A: 32%; Group B: 34%); busulfan, cyclophosphamide, cytosine arabinoside, and methotrexate with TBI (Group A: 23%; Group B: 13%) or without TBI (Group A: 8%; Group B: 11%); cyclophosphamide and steroid with TBI (Group A: 0%; Group B: 2%); cyclophosphamide, cytosine arabinoside, and steroid with TBI (Group A: 23%; Group B: 18%); etoposide with TBI (Group A: 0%; Group B: 2%); etoposide, cytosine arabinoside, and cyclophosphamide with TBI (Group A: 2%; Group B: 0%); etoposide, cytosine arabinoside, cyclophosphamide, and asparaginase with TBI (Group A: 0%; Group B: 2%); etoposide, methotrexate, cytosine arabinoside, and steroid with TBI (Group A: 4%; Group B: 4%)

All participants received cyclosporine and methylprednisolone for GVHD prophylaxis

Age at baseline (years): Group A: mean 32; Group B: mean 34

Gender: Group A: 60% male; Group B: 54% male

Number randomised: 109 (Group A: 53; Group B: 56)

Number evaluated: 109 (Group A: 53; Group B: 56)

Interventions

Comparison: GM‐CSF (yeast‐derived recombinant human) versus placebo

Group A: GM‐CSF (250 µg/m²) by 4‐hour IV infusion starting on day 0 (the day of BMT) to day 20 (total dose = 5250 µg/m²)

Group B: as above with placebo

Compliance: Group A: 13 participants (25%) stopped their intervention early: 11 due to adverse events (1 rash, 7 bone pain, 2 acute dyspnoea, 1 seizure); 2 (with no apparent toxicity) due to participant or physician request; Group B: 9 participants (16%) stopped their intervention early: 6 due to adverse events (2 rash, 2 bone pain, 1 elevated liver enzymes, 1 infections); 2 (on day 19) due to miscalculation of the number of days to receive the intervention; 1 (on day 9) due to severe mucositis and veno‐occlusive disease of the liver

Duration of treatment: 21 days

Outcomes

• Oral mucositis: assessment methods are not clear but reported on 0 to 4 scale which is probably WHO (assessment frequency and duration also unclear, reported in text as incidence of grade 2 to 4 and grade 3 to 4)

• Adverse events: 0 to 4 scale which is probably WHO (reported as incidence of any grade of event and incidence of grade 3 to 4 events with a greater than 10% frequency)

• Number of days in hospital (reported as median values; unable to use data)

• GVHD incidence and severity (not an outcome of this review)

• Veno‐occlusive disease of the liver (not an outcome of this review)

• Blood measurements (not an outcome of this review)

• Infection: bacterial, fungal and viral (not an outcome of this review)

• Duration of IV antibiotics (not an outcome of this review)

• Fever (not an outcome of this review)

• Survival: overall, disease‐free, relapse (not an outcome of this review)

Notes

Sample size calculation: not reported

Funding: "..product provided by Immunex" (pharmaceutical). Also mention of "sponsoring company" which is likely to be Immunex

Declarations/conflicts of interest: not reported but 4 authors employed by Immunex

Data handling by review authors: N/A

Other information of note: GM‐CSF administration only began after the conditioning regimen, by which point oral mucositis may have already begun to develop

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "Assignment to treatment was made via a randomization schema prepared by Almedica Corporation"

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Low risk

Quote: "Blinded numbered vials containing placebo or rhGM‐CSF were provided to each participating centre. The pharmacists, principal investigators, patients, support care personnel and sponsoring company were blinded to the study medication for the entire course of the study"

Comment: probably done

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind placebo‐controlled" and "The pharmacists, principal investigators, patients, support care personnel and sponsoring company were blinded to the study medication for the entire course of the study"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind placebo‐controlled" and "The pharmacists, principal investigators, patients, support care personnel and sponsoring company were blinded to the study medication for the entire course of the study"

Comment: everyone involved in the study was blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

5 participants in the placebo group received cytokines off study during the first 42 days after BMT but unlikely to bias the results in a meaningful way

Methods

Trial design: parallel (3 arms)

Location: Russia

Number of centres: 9

Study duration: recruitment from August 2006 to May 2007
Trials registry number: none/unknown

Participants

Inclusion criteria: 18 years or older with colorectal cancer (stage I to IV); undergoing chemotherapy as primary cancer therapy; experienced WHO grade 2 or higher oral mucositis during first cycle of chemotherapy, but recovered prior to enrolment (i.e. grade 0); ECOG score of 0 to 2

Exclusion criteria: pregnant (or risk of pregnancy) or lactating; scheduled to receive RT to the head and neck; received other investigational drugs within 14 days of the start of the study; evidence of alcohol and drug abuse; pre‐existing conditions such as active fungal or herpetic infection

Cancer type: colorectal (stage I to IV)

Cancer treatment: all participants received 5FU (97% with leucovorin) but 1 received capecitabine; 6 participants (3 in low‐dose ITF, 3 in placebo) also received oxaliplatin; doses or regimens comparable between groups

Age at baseline (years): Group A: median 62 ; Group B: median 56 ; Group C: median 59

Gender: Group A: 42% male; Group B: 33% male; Group C: 48% male

Number randomised: 99 (Group A: 33; Group B: 33; Group C: 33)

Number evaluated: 99 (Group A: 33; Group B: 33; Group C: 33)

Interventions

Comparison: ITF (recombinant human) versus placebo

Group A: ITF (10 mg/mL) oral spray, 300 µL (3 sprays) to the oral mucosa 8 times daily, starting on the first day (day 1) of the second chemotherapy cycle for total 14 days (total dose = 336 mg)

Group B: ITF (80 mg/mL) oral spray as above (total dose = 2688 mg)
Group C: placebo oral spray as above

Compliance: patient‐reported compliance was 97%

Duration of treatment: 14 days

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed on days 0, 1, 3, 5, 7, 10, 12, 14, and 21 ± 2, maximum score reported) (duration of mucositis also measured but not an outcome of this review)

• Oral mucositis: OMAS 0 to 45 scale (assessed on days reported above; WHO data used for analysis)

• Adverse events (assessed on days reported above)

• Patient daily self‐assessment: including discolouration, mouth and throat pain, preference for solid/semi‐solid food, analgesic use, and dysphagia (no usable data for the outcomes relevant to this review)

Notes

Sample size calculation: based on previous study, 80% power to detect 40% difference in incidence of WHO grade 2 or above

Funding: The GI Company (pharmaceutical)

Declarations/conflicts of interest: some authors had roles with the funding pharmaceutical company (and other companies) both compensated and uncompensated

Data handling by review authors: we combined the 2 ITF groups to make a single pairwise comparison against placebo; we also made a separate comparison of the 2 ITF dosages

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "The patients were randomly assigned"

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Unclear risk

Quote: "The patients were randomly assigned"

Comment: insufficient information to determine whether or not the random sequence was adequately concealed

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled"

Comment: it is not clear who was blinded. There are subjective elements to the assessment of oral mucositis using this scale, requiring the patient's assessment of pain/soreness and their ability to swallow but, as the participants were unaware of their group allocation, the assessment of oral mucositis can be considered to be blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: USA and Australia

Number of centres: 15

Study duration: not reported
Trials registry number: none/unknown

Participants

Inclusion criteria: 18 years or older with advanced (Duke's D) colon or rectal adenocarcinoma; scheduled to receive bolus 5FU and low‐dose leucovorin (Mayo regimen) as first‐line or subsequent therapy; normal oral cavity examination (no ulceration, herpes simplex, candidiasis, or severe gingivitis); ECOG score of 0 to 2; life expectancy of 4 months or more; absolute neutrophil count of 1.5 x 10⁹/L or higher; platelet count of 100 x 10⁹/L or higher; normal renal and hepatic function

Exclusion criteria: previous radiotherapy or chemotherapy within 4 weeks or major surgery within 2 weeks of study day 1; insulin‐dependant diabetes; known allergy to leucovorin; known hypersensitivity to Escherichia coli‐derived material

Cancer type: advanced colorectal (Duke's D)

Cancer treatment: low‐dose leucovorin (20 mg/m²) by IV immediately followed by 5FU (425 mg/m²) by rapid IV once daily for 5 consecutive days on days 4 to 8 of each 28‐day cycle; 2 cycles (5FU dose could be decreased by 20% during cycle 2 if moderately severe toxicities occurred in cycle 1)

Age at baseline (years): Group A: mean 65 (SD 11.2); Group B: mean 65 (SD 11.1)

Gender: Group A: 57% male; Group B: 72% male

Number randomised: 65 (not reported by group)

Number evaluated: 64 (Group A: 28; Group B: 36)

Interventions

Comparison: KGF (palifermin) versus placebo

Group A: KGF (40 µg/kg) by IV for 3 consecutive days (days 1 to 3 of each 28‐day cycle) before the start of chemotherapy (total dose = 120 µg/kg)

Group B: as above with placebo

Compliance: not reported (only reports "palifermin was generally well tolerated")

Duration of treatment: 3 days

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed on days 0, 1, 4, 8, 12, 15, and 28, maximum score reported)

• Oral pain: patient daily self‐assessment using OMDQ 5‐point scale for mouth and throat soreness (higher = worse pain, reported as AUC, not used)

• Adverse events

• Diarrhoea (related to the cancer therapy ‐ not an outcome of this review)

• Laboratory assessments (not an outcome of this review)

• Neutropenia (not an outcome of this review)

• Antibody assessments (not an outcome of this review)

• Survival (not an outcome of this review)

Notes

Sample size calculation: not reported

Funding: "Supported by Amgen Inc" (pharmaceutical industry)

Declarations/conflicts of interest: all authors were linked to the funding pharmaceutical company in terms of employment, consultancy, stock ownership, honoraria, and receipt of research funding

Data handling by review authors: data were reported separately for the chemotherapy cycles 1 and 2 ‐ we used the data for cycle 1

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "Patients were randomly assigned (by center and prior chemotherapy)"

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Unclear risk

Quote: "Patients were randomly assigned (by center and prior chemotherapy)"

Comment: insufficient information to determine method of random sequence generation

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled"

Comment: it is not clear who was blinded. There are subjective elements to the assessment of oral mucositis using this scale, requiring the patient's assessment of pain/soreness and their ability to swallow but, as the participants were unaware of their group allocation, the assessment of oral mucositis can be considered to be blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Only 1 randomised participant was not included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: Helsinki University Central Hospital, Finland

Number of centres: 1

Study duration: recruitment from October 1999 to April 2001
Trials registry number: none/unknown

Participants

Inclusion criteria: undergone radical surgery for head and neck cancer and scheduled to receive postoperative RT to a total dose of 50 Gy or more to the oral and oropharyngeal mucosa

Exclusion criteria: prior chemotherapy or RT; chronic autoimmune or inflammatory disease; WHO performance score > 2

Cancer type: head and neck: mobile tongue (Group A: 33%; Group B: 37%); floor of mouth (Group A: 19%; Group B: 16%); tonsil (Group A: 29%; Group B: 32%); oral cavity other (Group A: 19%; Group B: 16%)

Cancer treatment: mean time from radical surgery to start of RT: Group A: 39 days (range 20 to 73); Group B: 38 days (range 20 to 56); conventionally fractionated RT to a total dose of 50 Gy to 60 Gy in 2‐Gy daily fractions, 5 times weekly (on weekdays) for 5 to 6 weeks, to the primary site and locoregional lymph nodes

Age at baseline (years): Group A: median 56 (range 43 to 87); Group B: median 60 (range 24 to 72)

Gender: Group A: 57% male; Group B: 79% male

Number randomised: 40 (Group A: 21; Group B: 19)

Number evaluated: 40 (Group A: 21; Group B: 19)

Interventions

Comparison: GM‐CSF versus sucralfate

Group A: starting after a cumulative RT dose of 10 Gy (after 1 week of RT); mouth was cleaned with water, then GM‐CSF (37.5 µg per 25 mL rinse) mouthwash rinsed around the mouth for 3 minutes then swallowed, 4 times daily after meals, on RT‐days (weekdays), until the end of RT (total dose dependent on duration of RT)

Group B: as above but with sucralfate (1 g per 25 mL rinse) (total dose dependent on duration of RT)

Compliance: not reported (only reports "both mouthwashes were well tolerated, and none of the patients reported any adverse effects related to the mouthwashes")

Duration of treatment: 4 to 5 weeks (dependent on duration of RT)

Outcomes

• Oral mucositis: RTOG 0 to 4 scale (assessed before RT, weekly during RT, and at 1, 2, and 4 weeks after RT, reported graphically as mean score over time but authors provided maximum incidence data on request)

• Interruptions to cancer treatment (RT interruptions due to OM)

• Oral pain: 0 (no pain) to 10 (worse pain) VAS (assessed as reported above, reported graphically as mean score over time – no mean maximum score and no SDs, unable to use data)

• Normalcy of diet: use of PEG feeding tube

• Adverse events

• Incidence of hospitalisation (number of days in hospital is an outcome of this review and therefore we did not use these data)

• Opioid analgesic use (reported as incidence; number of days of treatment with opioid analgesics is an outcome of this review and therefore we did not use these data)

• Laboratory parameters (not outcomes of this review)

• Use of antimycotic agents (not an outcome of this review)

• Use of antibiotics (not an outcome of this review)

• Weight loss (not outcomes of this review)

Notes

Sample size calculation: not reported

Funding: not reported

Declarations/conflicts of interest: not reported

Data handling by review authors: we used oral mucositis incidence (maximum score experienced per patient) data provided by the authors; all participants experienced mucositis (grade 1 or above) as they had already received a week of RT before the intervention started, therefore we felt an analysis of incidence of any grade of mucositis should not be included

Other information of note: GM‐CSF/sucralfate administration only began after 1 week of RT, by which point oral mucositis may have already begun to develop

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "Randomization was done using computer‐generated random digits"

Comment: adequate method used

Allocation concealment (selection bias)

Low risk

Quote: "they were assigned to a treatment group by way of a telephone call to the randomization office"

Comment: third party/remote randomisation would have ensured concealment of the random sequence from those recruiting participants

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind" and "Both solutions looked alike, and neither the investigators nor the patients were aware of the contents of the solutions given. The drug vials were marked with a study code that prevented identification of the allocation group"

Comment: adequate methods to ensure blinding

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blinded, placebo‐controlled"

Comment: nobody involved in the study was aware of group allocation

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: USA

Number of centres: not reported

Study duration: recruitment from January 1995 to April 1996
Trials registry number: none/unknown

Participants

Inclusion criteria: 18 years or older with histologically proven head and neck malignancy; mucosa of the oropharynx and/or oral cavity to be included in RT portal; Karnofsky performance score of 60 or more; no known hypersensitivity to E coli‐derived proteins; haemoglobin and platelet counts 100 x 10⁹/L or higher; absolute neutrophil count higher than 1.5 x 10⁹/L; normal blood pressure; agreement to bring dentition to acceptable level prior to RT (teeth with periapical abscesses or severe periodontal disease extracted, followed by a 10‐day healing period allowance)

Exclusion criteria: lactating/pregnant/females not taking effective form of contraception; scheduled to receive chemotherapy or radiosensitising drugs during the planned RT; history of myeloid malignancy; underlying collagen vascular disease; active rheumatoid arthritis

Cancer type: head and neck: nasopharynx (Group A: 0; Group B: 1); oropharynx (Group A: 2; Group B: 1); tongue (Group A: 2; Group B: 1); larynx (Group A: 2; Group B: 1); other/unknown (Group A: 2; Group B: 2)

Cancer treatment: radiotherapy in 1.8 Gy to 2 Gy standard daily fractions from Monday to Friday, to total dose 50 Gy or more

Age at baseline (years): not reported

Gender: Group A: 100% male; Group B: 83% male

Number randomised: 14 (Group A: 8; Group B: 6)

Number evaluated: 14 (Group A: 8; Group B: 6)

Interventions

Comparison: G‐CSF (r‐metHuG‐CSF) (filgrastim) versus placebo

Group A: G‐CSF (starting at 3 µg/kg and titrated to keep neutrophil count between 10 x 10⁹/L and 30 x 10⁹/L) by daily subcutaneous injection, starting on the first day of RT (weekdays), until the end of RT (total dose dependent on duration of RT and neutrophil counts)

Group B: placebo as above

Compliance: not reported (only reports "Filgrastim was well tolerated")

Duration of treatment: variable and dependent on duration of RT

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed on weekly by single examiner, reported as incidence of any mucositis and grade 3 or higher)

• Oral mucositis: Hickey 0 to 3 scale (not used)

• Interruptions to cancer treatment (delays or dose reductions, no usable data)

• Weight change (not an outcome of this review)

• Blood measurements (not an outcome of this review)

Notes

Sample size calculation: 54 required to detect 30% decrease in incidence of grade 2 or 3 mucositis at 80% power at the 5% significance level

Funding: "Financial support was provided through a grant from Amgen Inc" (pharmaceutical industry)

Declarations/conflicts of interest: not reported

Data handling by review authors: interruptions to cancer treatment outcome only reported narratively

Other information of note: study was stopped after an interim analysis. Authors state that "owing to administrative obstacles, completion of the trial is not possible", rather than due to previously stated early stopping rules

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "The patients were randomized equally between the two treatment groups" and "The study material and randomization list were held by the UCLA Pharmacy for the duration of the study"

Comment: method of random sequence generation not described but done by UCLA Pharmacy and therefore probably done adequately

Allocation concealment (selection bias)

Low risk

Quote: "Amgen Inc...prepared and packaged all drug and placebo in identical containers, with the only designator being the randomization number. The study material and randomization list were held by the UCLA Pharmacy for the duration of the study"

Comment: pharmacy‐controlled randomisation would have ensured concealment of the random sequence from those recruiting participants

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled" and "Amgen Inc...prepared and packaged all drug and placebo in identical containers, with the only designator being the randomization number"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled" and "A single examiner...who was blinded to the patients' randomization status...scored mucositis on a weekly basis"

Comment: both participants and outcome assessor were blinded

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

Although the study reports an interim analysis only, the study was stopped due to administrative reasons and is unlikely to introduce bias

Methods

Trial design: parallel (2 arms)

Location: USA

Number of centres: 13

Study duration: recruitment from March 2001 to October 2002
Trials registry number: NCT00041665

Participants

Inclusion criteria: 18 years or older; Karnofsky performance score of 70 or more; scheduled to undergo autologous HSCT after conditioning regimen of fractionated TBI plus etoposide and cyclophosphamide for haematological cancers; adequate cardiac, pulmonary, renal, and hepatic function

Exclusion criteria: not reported

Cancer type: haematologic: NHL (Group A: 68%; Group B: 65%); Hodgkins (Group A: 20%; Group B: 22%); multiple myeloma (Group A: 10%; Group B: 8%); leukaemia (Group A: 2%; Group B: 5%)

Cancer treatment: prior to autologous HSCT, participants received the following conditioning regimen:

• Radiotherapy: prior to chemotherapy, TBI of total 12 Gy in 6, 8, or 10 fractions over 3 or 4 days, with at least 6 hours between fractions

• Chemotherapy: IV etoposide (60 mg/kg) the day after the last RT fraction (4 days before HSCT) and cyclophosphamide (100 mg/kg) 2 days later (2 days before HSCT)

All participants received G‐CSF (filgrastim) 5 µg/kg per day from HSCT until neutrophil recovery

Age at baseline (years): Group A: median 48 (range 18 to 69); Group B: median 49 (range 19 to 68)

Gender: Group A: 56% male; Group B: 68% male

Number randomised: 214 (Group A: 107; Group B: 107)

Number evaluated: 212 (Group A: 106; Group B: 106)

Interventions

Comparison: KGF (palifermin) versus placebo

Group A: KGF (60 µg/kg) by IV for 3 consecutive days starting 3 days before RT, followed by 3 more doses after HSCT starting on the day of HSCT (total dose = 360 µg/kg)

Group B: placebo as above

Compliance: 212 participants received at least 1 dose of their allocated intervention and 205 participants (Group A: 103; Group B: 102) "completed the study" (it is not clear whether or not this means that they received all doses)

Duration of treatment: 6 treatment days over 13 to 14 days

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed daily by trained assessors starting 8 days before HSCT and for 28 days after HSCT or until severe mucositis had returned to grade 0 to 2, maximum score reported) (duration of grade 3 to 4 oral mucositis also measured but not an outcome of this review)

• Oral mucositis: RTOG 0 to 4 scale (assessed as above but only used for reporting duration of severe mucositis)

• Oral mucositis: WCCNR 0 to 3 scale (assessed as above but only used for reporting duration of severe mucositis)

• Oral pain: patient daily self‐assessment (day ‐12 to 28) using OMDQ 5‐point scale for mouth and throat soreness (higher = worse pain, reported as AUC, not used)

• Quality of life: physical, functional, emotional, and social/family well‐being domains of the FACT general questionnaire (days ‐12, ‐1, 7, 10, 14, and 28, reported as AUC, not used)

• Normalcy of diet (measured as incidence of TPN)

• Adverse events: 1 (mild) to 5 (fatal) scale

• Number of days of treatment with opioid analgesics (reported as median and range, unable to use data)

• Dysphagia (not an outcome of this review)

• Febrile neutropenia (not an outcome of this review)

• Infections (not an outcome of this review)

• Survival (not an outcome of this review)

• Laboratory results (not an outcome of this review)

Notes

Sample size calculation: 210 participants required to detect a mean difference in duration of severe oral mucositis of at least 3 days (SD 6.6) at 90% power and 5% significance

Funding: "Funded by Amgen" (pharmaceutical industry)

Declarations/conflicts of interest: the majority of authors were linked to the funding pharmaceutical company in terms of employment, consulting fees, lecture fees, receipt of research funding, and ownership of equity

Data handling by review authors: N/A

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Unclear risk

Quote: "patients were randomly assigned in a 1:1 ratio"

Comment: insufficient information to determine method of random sequence generation

Allocation concealment (selection bias)

Unclear risk

Quote: "patients were randomly assigned in a 1:1 ratio"

Comment: insufficient information to determine method of random sequence generation

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "placebo‐controlled, double‐blind"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "placebo‐controlled, double‐blind"

Comment: it is not clear who was blinded. There are subjective elements to the assessment of lower grades of oral mucositis using the WHO scale, requiring the patient's assessment of pain/soreness and their ability to swallow. Higher grades have more objective elements so may not be affected by potential lack of blinding of the assessor

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Only 2 randomised participants (1 per group) were not included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: USA

Number of centres: 1

Study duration: recruitment from January 1992 to December 1996
Trials registry number: none/unknown

Participants

Inclusion criteria: 18 years or older with histologically proven AJCC stage II to IV squamous cell carcinoma of the head and neck; undergone gross complete resection with negative pathological surgical margins and medically fit to begin RT within 8 weeks of surgery; Karnofsky performance score of 80% or more; adequate haematologic and serum metabolic laboratory indices

Exclusion criteria: nasopharyngeal cancer; concurrent active malignancy other than localised; nonmelanoma skin cancer; previous RT to head and neck; previous chemotherapy; positive serum β‐human chorionic gonadotropin in women of procreative potential; need for tube feeding at the start of RT

Cancer type: stage II to IV squamous cell carcinoma of the head and neck: hypopharynx (Group A: 11%; Group B: 5%); larynx (Group A: 16%; Group B: 23%); oral cavity (Group A: 26%; Group B: 27%); oropharynx (Group A: 21%; Group B: 32%); unknown primary (Group A: 26%; Group B: 14%)

Cancer treatment: postoperative external beam RT in once‐daily fractions of 1.8 Gy, 5 days per week, to total dose of 63 Gy at primary site and involved neck (54 Gy to regional lymphatics at risk for subclinical metastasis; spinal cord dose limited to 45 Gy)

Age at baseline (years): Group A: median 67 (interquartile range 59 to 75); Group B: median 61 (interquartile range 54 to 67)

Gender: Group A: 79% male; Group B: 73% male

Number randomised: 41 (Group A: 19; Group B: 22)

Number evaluated: 40 (Group A: 19; Group B: 21)

Interventions

Comparison: G‐CSF versus placebo

Group A: G‐CSF (3 µg/kg) by daily subcutaneous injection, 7 days per week, starting 3 days before the start of RT and continued to end of RT (total dose = dependent on duration of RT)

Group B: placebo as above

Compliance: not reported

Duration of treatment: variable and dependent on duration of RT

Outcomes

• Oral mucositis: unknown 0 to 3 scale, where 0 = none, 1 = erythema, 2 = patchy mucositis, and 3 = confluent mucositis (assessed twice weekly by the treating radiation oncologist, maximum score reported)

• Interruptions to cancer treatment: RT interruptions

• Normalcy of diet (use of PEG feeding tube, as defined by 10% or higher weight loss from pre‐RT baseline)

• Adverse events (NCI‐CTC)

• Blood measurements (not an outcome of this review)

• Survival (not an outcome of this review)

Notes

Sample size calculation: planned sample sample size was 66 per group to detect absolute difference of 20% in PEG tube use (from 10% in one group to 30% in the other) at 80% power and 5% significance

Funding: "This study was supported by NIH grant #CA69913" (government)

Declarations/conflicts of interest: "No benefits in any form have been or will be received from a commercial party related directly or indirectly to the subject of this article"

Data handling by review authors: although the scale for assessing oral mucositis does not completely match the 0 to 4 scales (such as WHO, etc), it was possible to dichotomise it for use in the 'any mucositis' meta‐analysis, and we also used grade 3 (confluent mucositis) as incidence of severe mucositis

Other information of note: planned to enrol 132 participants but stopped due to slow recruitment

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "Patients were randomized...by randomly permuted blocks...A randomization list was prepared by the Memorial Sloan‐Kettering Cancer Center Biostatistics Service and held by the pharmacy"

Comment: specialist centre used

Allocation concealment (selection bias)

Low risk

Quote: "A randomization list was prepared by the Memorial Sloan‐Kettering Cancer Center Biostatistics Service and held by the pharmacy. Investigators did not have access to this list, ensuring that allocation could not be predicted before registration or changed afterwards"

Comment: central/remote randomisation would have ensured concealment of the random sequence from those recruiting participants

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled" and "All patients and treating physicians were blinded to treatment group assignment"

Comment: the use of a placebo should have ensured that blinding was successful

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled" and "All patients and treating physicians were blinded to treatment group assignment"

Comment: the treating physician was the outcome assessor

Incomplete outcome data (attrition bias)
All outcomes

Low risk

Only 1 randomised participant was not included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: University of Texas MD Anderson Cancer Center, USA

Number of centres: 1

Study duration: recruitment from December 2005 to February 2008; last follow‐up in May 2008
Trials registry number: NCT00267046 (mentioned in trial report)

Participants

Inclusion criteria: aged 15 to 65 years with sarcoma; due to start chemotherapy at the centre; Karnofsky performance score of 80% or more; adequate bone marrow, renal, and hepatic function

Exclusion criteria: history of pelvic RT; clinically significant cardiac disease; undergone surgery within the previous 2 weeks

Cancer type: sarcoma

Cancer treatment: doxorubicin (total dosage 90 mg/m²) administered by continuous IV infusion over 72 hours, and ifosfamide (total dosage 10 g/m²) administered by 3‐hour IV infusion for 4 days; participants with osteosarcoma (Group A: 2; Group B: 1) received the same dosage of doxorubicin but with intra‐arterial cisplatin (120 mg/m²); all participants received G‐CSF (pegfilgrastim) the day after chemotherapy; cycle repeated every 21 days; planned 6 cycles

Age at baseline (years): Group A: median 42 (range 17 to 63); Group B: median 39 (range 15 to 64)

Gender: Group A: 53% male; Group B: 50% male

Number randomised: 48 (Group A: 32; Group B: 16)

Number evaluated: 48 (Group A: 32; Group B: 16)

Interventions

Comparison: KGF (palifermin) versus placebo

Group A: KGF (180 µg/kg) by IV as single dose 3 days before the start of each chemotherapy cycle

Group B: placebo as above

Compliance: the proportion of participants that completed all 6 blinded cycles (i.e. they took their allocated intervention) was higher in Group A (63%) than Group B (31%)

Duration of treatment: 1 day per 3‐week cycle; planned total 6 days of intervention over 18 weeks

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed for each cycle before chemotherapy, days 10, 12, 14 and at the end of the cycle, reported as incidence of grade 2 to 4 and grade 3 to 4, reported separately for blinded phase (first 2 cycles) and then the open‐label phase) (duration of mucositis also measured but not an outcome of this review)

• Oral pain: 0 to 10 (10 being worst) scale (assessed by questionnaire for each cycle before chemotherapy, days 10, 12, 14 and at the end of the cycle, reported for blinded phase, reported as median maximum score; unable to use data)

• Quality of life: 1 to 7 (7 being worst) scale (assessed by questionnaire for each cycle before chemotherapy, days 10, 12, 14 and at the end of the cycle, reported for blinded phase, reported as median maximum score; unable to use data)

• Normalcy of diet: eating and drinking assessed separately on 0 to 4 (4 being most difficult) scales (assessed by questionnaire for each cycle before chemotherapy, days 10, 12, 14 and at the end of the cycle, reported for blinded phase, reported as median maximum score; unable to use data)

• Adverse events

• Opioid analgesic use (reported as quantity per cycle; number of days of treatment with opioid analgesics is an outcome of this review and therefore we did not use these data)

• Multiple patient‐reported outcomes (overall health, sleeping, dysphagia, talking, brushing teeth, throat pain, rectal soreness) (not outcomes of this review)

• Blood and laboratory measurements (not outcomes of this review)

• Weight loss (not an outcome of this review)

• Survival (not an outcome of this review)

Notes

Sample size calculation: 48 participants required to detect absolute difference of 50% (from 26% in Group A to 76% in Group B) in grade 2 to 4 mucositis at 88% power and 5% significance

Funding: "Amgen provided the palifermin and partial funding for the study" (pharmaceutical industry)

Declarations/conflicts of interest: principal investigator is a member of the Amgen board

Data handling by review authors: we only report data from the first 2 cycles (blinded phase) as very few participants received placebo in the remaining cycles

Other information of note: not reported

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Quote: "Two distinct computer‐generated randomization lists were prepared by the University of Texas M.D. Anderson Cancer Center, Department of Biostatistics, one for the 20 patients who consented to pharmacokinetic sampling and the other for the 28 patients who did not. For the pharmacokinetics cohort, the treatment allocation ratio was 4 patients receiving palifermin to 1 receiving placebo, in blocks of 5; for the other cohort, the ratio was 4 patients receiving palifermin to 3 receiving placebo, in blocks of 14"

Comment: adequate method used

Allocation concealment (selection bias)

Low risk

Quote: "The statistician provided both randomization lists to the pharmacy, so the patient and the clinical research team (who assessed outcomes) were blinded to the study treatment. At patient enrolment, the research team notified the pharmacy, which assigned the patient the next sequential slot and treatment from the appropriate randomization list on the basis of whether he or she had consented to pharmacokinetic sampling. The pharmacy provided the research team with the blinded study medication. Upon completion of the study, pharmacy provided the statistician with the 2 randomization lists, including individual patient treatment assignments, for analysis"

Comment: pharmacy‐controlled randomisation would have ensured concealment of the random sequence from those recruiting participants

Blinding of participants and personnel (performance bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled" and "the patient and the clinical research team (who assessed outcomes) were blinded to the study treatment" and "Blinding might not have been maintained because of adverse effects of palifermin"

Comment: some personnel may not have been blinded due to increased adverse effects, although adverse effects from palifermin may have been difficult to isolate from adverse effects of cancer treatment

Blinding of outcome assessment (detection bias)
All outcomes

Low risk

Quote: "double‐blind, placebo‐controlled" and "the patient and the clinical research team (who assessed outcomes) were blinded to the study treatment" and "Blinding might not have been maintained because of adverse effects of palifermin" and "patients were assessed at each cycle by both research and clinical teams, including those without direct knowledge of the protocol"

Comment: it seems unlikely that lack of blinding would affect outcomes as the higher grades of oral mucositis assessed in this study are more objective; also, some were assessed by personnel without knowledge of the protocol

Incomplete outcome data (attrition bias)
All outcomes

Low risk

All randomised participants were included in the analyses

Selective reporting (reporting bias)

Low risk

Data for outcomes of this review were reported appropriately

Other bias

Low risk

No other sources of bias are apparent

Methods

Trial design: parallel (2 arms)

Location: the Netherlands

Number of centres: 1

Study duration: recruitment from May 1997 to August 1999
Trials registry number: none/unknown

Participants

Inclusion criteria: malignant disease and due to have myeloablative treatment followed by autologous or allogeneic BMT or PBSCT

Exclusion criteria: not reported

Cancer type: haematologic: lymphoma (Group A: 28%; Group B: 50%); acute leukaemia (Group A: 28%; Group B: 28%); CML (Group A: 22%; Group B: 6%); multiple myeloma (Group A: 22%; Group B: 17%)

Cancer treatment: prior to autologous or allogeneic BMT or PBSCT, participants received the following conditioning regimens:

• BEAM (Group A: 28%; Group B: 44%): carmustine (BCNU) 300 mg/m² on day ‐6; etoposide (VP16) and cytosine arabinoside (Ara‐C) 200 mg/m² of each on days ‐5, ‐4, ‐3, ‐2; melphalan 140 mg/m² on day ‐1

• CYTBI (Group A: 56%; Group B: 44%): cyclophosphamide 60 mg/kg on days ‐5 and ‐4; TBI 4.5 Gy on days ‐2 and ‐1

• BUCY (Group A: 17%; Group B: 11%): busulfan 4 mg/kg on days ‐7, ‐6, ‐5, ‐4; cyclophosphamide 60 mg/kg on days ‐3 and ‐2

Lymphoma patients received BEAM; other patients received CYTBI or BUCY; nearly all participants received autologous or allogeneic PBSCT

Age at baseline (years): Group A: median 47 (range 25 to 63); Group B: median 48 (range 18 to 62)

Gender: Group A: 61% male; Group B: 39% male

Number randomised: 36 (Group A: 18; Group B: 18)

Number evaluated: 36 (Group A: 18; Group B: 18)

Interventions

Comparison: GM‐CSF versus placebo

Group A: GM‐CSF (300 µg daily dose) gel, 5 mL twice daily (early in the morning and before going to sleep) kept in the mouth for as long as possible and then swallowed; no oral intake for 1 hour afterwards; starting on day 1 (the day after the day of transplant) and continued until neutrophil recovery (typically up to 14 days after transplant)

Group B: placebo as above

All participants followed hospital's standard protocol for mouth care: toothbrushing after meals, rinsing with 0.9% saline, and if there was inflammation, 0.12% chlorhexidine rinse 6 times daily

Compliance: 8 participants (Group A: 4; Group B: 4) did not complete the study; they all cited the main reason being nausea and the taste of the gel

Duration of treatment: variable and dependent on neutrophil recovery

Outcomes

• Oral mucositis: WHO 0 to 4 scale (assessed daily by dentists, reported in text as incidence of grade 3 to 4)

• Oral mucositis: 8 (all 8 sites normal) to 24 (all 8 sites severely affected) oral assessment score (assessed daily by dentists, reported graphically as median score over 14 days, unable to use data)

• Oral pain: 0 (no pain) to 100 (worst pain) VAS (assessed daily by participant, reported graphically as median score over 14 days, unable to use data)

• Normalcy of diet: incidence of total parenteral nutrition (started when patients were unable to eat for longer than 3 days)

• Number of days in hospital (reported as median, unable to use data)

• Opioid analgesic use (reported as incidence; number of days of treatment with opioid analgesics is an outcome of this review and therefore we did not use these data)

• Blood measurements (not an outcome of this review)

• Fever (not an outcome of this review)

• Infection (not an outcome of this review)

• Antibiotic use (not an outcome of this review)