Adjuvant anti‐VEGF therapy for overall survival and relapse‐free survival in patients with resected non‐metastatic colorectal cancer
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
To assess the efficacy and safety of the use of anti‐angiogenic agents (VEGF inhibitors) in the adjuvant setting after resection in patients diagnosed with resectable colorectal cancer.
This would permit a summary of the safety and effectiveness of anti‐VEGF therapies overall, as well as in pre‐planned prognostic subgroups, and identify areas where further research could be directed.
Background
Description of the condition
Incidence
Colorectal cancer is a leading cause of death worldwide, with nearly 1.4 million new cases diagnosed in 2012 (Ferlay 2015). Approximately 54% of these newly diagnosed cases occur in more developed countries. Despite ongoing advances in both surgical and medical therapies, colorectal cancer continues to be a major cause of morbidity and mortality, and has an enormous impact on the families of people with colorectal cancer and society’s health costs (Ferlay 2015).
Impact
Colorectal cancer stage (that is, its extent of spread) is the major determinant of prognosis following diagnosis of colorectal cancer, and patients diagnosed with stage I disease have a 83% chance of remaining alive at five years after resection (Edge 2010; Hari 2013), compared with patients with stage III disease (who have extensive regional lymph node involvement) who have 28% chance of five‐year survival. Patients with stage IV disease — where the cancer has moved away from the primary site, often to the lungs or liver — are described as having metastatic disease. Whilst a small number of these patients can have surgery for their metastases, generally most are not candidates for surgery with curative intent and are instead treated with chemotherapy to prolong survival and to improve or maintain their quality of life. Frequency, five years disease free survival (DFS) and five‐year overall survival OS is presented in Table 1, as reported by Howlander 2015 and Hari 2013.
As seen in Table 1, three out of four patients diagnosed with colorectal cancer are eligible for tumour resection at initial diagnosis (stages I to III). However, even after surgery, significant numbers of patients experience recurrence. This is especially the case for patients diagnosed with stage III disease; if no further treatment is given, more than 40% of these patients will have disease recurrence within five years (André 2009). Local recurrences may be surgically removed, but this can lead to considerable morbidity and cost. Metastatic recurrence is incurable in most cases.
| Stage | Frequency | Metastases | 5‐ year disease‐free survival | 5‐ year overall survival | Usual recommended treatment after surgery |
| I | 39% (I and II) | No | 95% | 83% | Observation |
| II | 39% (I and II) | No | 85% | 70% | Consider adjuvant chemotherapy |
| III | 36% | No | 69% | 58% | Recommend adjuvant chemotherapy (e.g. FOLFOX) |
| IV | 20% | Yes | (0%) | 13% | Surgery not possible in most cases |
Description of the intervention
As many patients with resected colorectal cancer still carry a significant risk of recurrent disease, adjuvant chemotherapy (chemotherapy given after curative surgery) is often initiated to minimize the probability of recurrence by eradicating microscopic metastases before they have a chance to fully develop. Whilst the decision regarding adjuvant chemotherapy must be individualized to the patient, considering their general health and well‐being, therapy recommendations are generally dependent on the stage of the colorectal cancer at diagnosis. For stage I disease, observation is recommended. For stage II disease, some patients may be recommended for chemotherapy depending on their individual risk of relapse and their general health. The patients diagnosed with stage III disease have higher chances of recurrence compared to those with stage I/II disease; adjuvant chemotherapy (such as FOLFOX, which is the combination of 5‐fluorouracil, leucovorin and oxaliplatin, capecitabine, or 5‐fluorouracil alone) has been proven to decrease the risk of recurrence in these patients. However, the above treatments reduce, but do not eliminate, all risk, and leave some patients with significant recurrence risk at the end of therapy. Research has focused on how to improve these chemotherapy regimens to further reduce the risk of recurrence particularly for high‐risk patients.
The development of metastatic recurrence is strongly influenced by vascular epidermal growth factor (VEGF) and thus drugs that specifically target this pathway have been investigated to complement the use of chemotherapy. Angiogenesis — the formation of new blood vessels — is thought to be one of the primary mechanisms behind tumour growth and particularly metastatic growth. VEGF plays a dominant role in the angiogenesis pathway, by promoting growth of new vessels and maintenance of existing vascular endothelial cells. It has been investigated in the pathogenesis of bowel cancer and is associated with increased cancer‐specific mortality (Bendardaf 2008).
Several agents targeting VEGF have been developed. Bevacizumab is a humanized monoclonal antibody directed against VEGF (Presta 1997). Aflibercept is a recombinant fusion protein which inhibits the action of VEGF (Van Cutsem 2012). In addition to the above, multiple tyrosine kinase inhibitors have been developed which target VEGF, including regorafenib (Grothey 2013) and brivanib (Siu 2013). Regorafenib has been incorporated into the treatment of metastatic colorectal cancer progressing after several lines of chemotherapy (Grothey 2013).
How the intervention might work
Inhibition of the VEGF pathway may decrease the formation of new blood vessels, thus affecting the supply of nutrients and oxygen to the tumour, and potentially leading to reduced tumour growth, both in distant sites and in the area local to the resected primary tumour. As tumour recurrence is dependent on tumour cells growing or metastasizing, or both, anti‐VEGF agents may therefore reduce the recurrence rate, or at least slow down the development of tumour recurrence.
Anti‐VEGF agents may also lead to the normalization of tumour‐related blood vessels, which are usually disorganized and chaotic (Jain 2006). These disorganized blood vessels may impede the normal flow of blood, and thus decrease the amount of chemotherapy administered to tumour cells. If these vessels were to normalize in terms of organization or diameter, or both, there may be more even and consistent distribution of chemotherapy, thereby potentially increasing the effectiveness of adjuvant chemotherapy.
Why it is important to do this review
To prevent recurrence of the disease, adjuvant chemotherapy has been used in patients with resected metastatic colorectal cancer for more than 20 years. However, many of these patients will still experience recurrence, ultimately causing their death. Several randomized phase III trials have investigated whether the addition of anti‐VEGF agents (such as bevacizumab) to chemotherapy improves the overall survival or relapse‐free survival. Both the AVANT trial, de Gramont 2012, and the NSABP C‐08 trial, Allegra 2011, investigated the addition of bevacizumab to chemotherapy in the adjuvant setting. Unfortunately, neither demonstrated improved overall survival or relapse‐free survival with the addition of bevacizumab.
Despite the disappointing results in the adjuvant setting (i.e. administered to patients with stage I to III who have undergone surgery), anti‐VEGF agents have improved overall survival when given to patients with metastatic disease (Giantonio 2007; Saltz 2008; Van Cutsem 2011), and have become part of the standard treatment for metastatic colorectal cancer (NCCN 2015). However, the use of anti‐VEGF agents is associated with specific side effects related to the VEGF pathway including increased rates of hypertension, proteinuria and wound breakdown. Wound breakdown is a concern in the adjuvant setting, as it may result in re‐operation, prolonged hospital stay or even death. Other major complications described in trials include gastrointestinal perforation, fistula/abscess and bleeding (de Gramont 2012).
This discrepancy of benefits between the metastatic setting, where benefit is consistent (Wagner 2009), and the adjuvant setting, where no definite benefit has been demonstrated, remains a main scientific conundrum that has raised many questions to the role of angiogenesis and the mechanism of anti‐VEGF therapy in the respective settings.
Objectives
To assess the efficacy and safety of the use of anti‐angiogenic agents (VEGF inhibitors) in the adjuvant setting after resection in patients diagnosed with resectable colorectal cancer.
This would permit a summary of the safety and effectiveness of anti‐VEGF therapies overall, as well as in pre‐planned prognostic subgroups, and identify areas where further research could be directed.
Methods
Criteria for considering studies for this review
Types of studies
All relevant parallel‐group randomized controlled trials (RCTs) or quasi‐RCTs, with no restrictions regarding language or date of publication, will be eligible for inclusion in the review. Cluster‐RCTs will be eligible if we identify them from our literature searches. We will exclude trials that evaluate the use of anti‐VEGF therapy prior to surgery. We will exclude case reports, letters, editorials, commentaries and reviews.
Types of participants
Patients with a histological diagnosis of colorectal carcinoma whose disease has been completely resected (Stage I, II or III). We will include patients of all ages in the analyses, although we note that the major trials identified restrict enrolment to adults aged 18 years or over.
Types of interventions
Eligible interventions include therapies that include anti‐VEGF as one of its targets of action. Agents may be intravenous or oral, and any dose/duration/frequency investigated by the trial will be eligible for inclusion. Trials that evaluate the use of anti‐VEGF agents alone (anti‐VEGF therapy versus observation) and in combination with chemotherapy (anti‐VEGF therapy plus chemotherapy versus the same chemotherapy alone) will be eligible for inclusion in the review.
Types of outcome measures
Primary outcomes
-
Overall survival, from time of randomisation in the study.
-
relapse‐free survival, from time of randomisation in the study.
Secondary outcomes
-
Disease‐specific survival, from time of randomisation in the study.
-
toxicity/adverse events of anti‐VEGF therapy (e.g. bowel perforation, hypertension, thromboembolic events, bleeding);
-
rates of discontinuation of therapy due to adverse events.
Search methods for identification of studies
Electronic searches
We will search the following electronic databases without any language restrictions:
-
the Cochrane Central Register of Controlled Trials (CENTRAL) (latest issue) (Appendix 1);
-
Ovid MEDLINE (1950 to date) (Appendix 2);
-
Ovid Embase (1974 to date) (Appendix 3);
-
ClinicalTrials.gov;
-
the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP);
-
the Database of Abstracts of Reviews of Effects (DARE).
Searching other resources
We will handsearch the proceedings of major conferences (ASCO, ASCO GI, ESMO, World GI) that are two years prior to the date of the electronic search for relevant trials. We will also search OpenGrey (www.opengrey.eu) for additional relevant trials. We will check the reference lists of any identified systematic reviews/meta‐analyses on this topic for additional studies.
Data collection and analysis
Selection of studies
We will download all titles and abstracts retrieved by electronic searching to a reference management database and will remove duplicates. Two review authors will independently screen titles and abstracts for inclusion. We will code all the potentially eligible studies as either 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. We will retrieve the full‐text study reports/publications. Two review authors will independently screen the full‐text articles and identify studies for inclusion. We will include all eligible studies irrespective of whether the study reported the measured outcomes. We will resolve any disagreement through discussion or, if required, we will consult a third review author. We will list the excluded studies with the reasons for exclusion in the 'Characteristics of excluded studies' tables. We will collate multiple reports of the same study so that each study, rather than each report, is the unit of interest in the review. In addition, we will provide any information we can obtain about ongoing studies. We will present a PRISMA flow chart to illustrate the steps we use for the identification and selection of studies, and we will specify the number of studies considered at each step. We will present the reasons for inclusion or exclusion of each study in a 'Characteristics of included studies' table or a 'Characteristics of excluded studies' table.
We will include studies that only report results in abstract form in the 'Studies awaiting assessment' section.
Data extraction and management
Two review authors will independently extract data for the outcomes described.
Relevant data for extraction include the following.
-
Study identifiers:
-
study ID, author, year of publication, and citation used for data extraction;
-
-
characteristics of study:
-
inclusion criteria, exclusion criteria, number of participants
-
Baseline information of participants
-
Dose and other regimen details of interventions and controls, length of follow‐up
-
Risk of bias assessment
-
Financial disclosures of principal and senior authors; funders of the trial;
-
-
outcomes:
-
overall survival: hazard ratios (HRs), 95% confidence intervals (CIs), P value
-
relapse‐free survival: HRs, 95% CIs, P value, five‐year survival rate
-
disease‐specific survival: HRs, 95% CIs, P value
-
toxicity: rates of Grade 3‐4 toxicities as defined by NCI CTCAE v4.03: overall toxicity, bowel perforation, hypertension, thromboembolic events, bleeding, discontinuation of therapy (as number of events over number of evaluated patients in each study arm.
-
toxicity: rates of Grade 5, deaths from therapy.
-
Two review authors will extract data onto separate electronic data collection forms and will then compare them to ensure accuracy. We will resolve any disagreements by consensus with a third review author (NP).
Assessment of risk of bias in included studies
We will base the 'Risk of bias' assessments on susceptibility to bias using the revised Cochrane 'Risk of bias' tool, Higgins 2011, for the following domains:
-
random sequence generation;
-
allocation concealment;
-
blinding of participants and personnel;
-
blinding of outcome assessment;
-
incomplete outcome data;
-
selective reporting bias; and
-
other biases (baseline imbalance, significant protocol deviations, inappropriate influence of funders, recruitment bias in cluster‐RCTs).
We will judge each domain as either at low, uncertain or high risk of bias according to the criteria set out in the 'Risk of bias' tool (see Appendix 4; Higgins 2011).
We will determine the risk of bias for a study by the highest 'Risk of bias' assessment for a key domain, as per the 'Risk of bias' tool. Where there are disagreements regarding assessment of a key domain in a trial, which we are unable to resolve via discussion, a third review author (NP) will adjudicate on the final assessment of risk of bias in the particular domain.
Measures of treatment effect
We will measure the outcomes of interest as follows:
-
overall survival: HR with 95% CIs;
-
relapse‐free survival: HR with 95% CIs;
-
disease‐specific survival: HR with 95% CIs;
-
disease‐specific mortality at five years: odds ratio (OR) with 95% CIs;
-
toxicity measures (detailed above): each to be analyzed as a dichotomous outcome, with OR and absolute risk difference with 95% CIs;
-
rates of discontinuation of therapy: to be analysed and presented as a dichotomous outcome, with ORs and absolute risk difference with 95% CIs.
Unit of analysis issues
We will consider unit‐of‐analysis issues in the included trials. As multiple group comparisons are usual across treatment arms in clinical trials, the main anticipated issue would be trials that involve multiple intervention groups and avoidance of double inclusion (particularly of the placebo group). We will resolve this as per the Cochrane guidelines (Higgins 2011a), ideally by combining groups to create a single pair‐wise comparison. In the case of data for the same outcome being evaluated at different time points by different trials, we will define three timeframes (short‐term, medium‐term and long‐term) as separate outcomes in order to avoid unit of analysis errors (Deeks 2011). If we identify cluster‐RCTs that meet the inclusion criteria of the review, we will seek expert statistical advice to minimize potential unit‐of‐analysis issues.
Dealing with missing data
We will contact the authors of the included studies for additional or updated information that is not available from the published trials. If unavailable, we will estimate the log‐HR and its variance directly or indirectly according to the methods outlined by Parmar 1998.
Assessment of heterogeneity
We will assess the clinical and methodological heterogeneity in each included trial and across trials combined in meta‐analysis. Where the difference between interventions tested or populations is sufficiently great to prevent meaningful synthesis, we will separate studies in terms of analysis and will present the results separately.
We will assess statistical heterogeneity using the I² statistic in combination with the strength of evidence from the Chi² test of heterogeneity based on Cochrane’s Q statistic. For this Chi² test, we will use a P value of 0.10 to assess presence of heterogeneity. We will consider I² statistic values of greater than 30% to represent a degree of heterogeneity worthy of further investigation. If a sufficient number of studies are available, we will explore heterogeneity via subgroup analysis (or meta‐regression) with regard to differing characteristics of the clinical studies. The purpose of this exploration will be to identify plausible causes of heterogeneity, such as stage or presence/absence of microsatellite instability.
Assessment of reporting biases
We will construct funnel plots if more than 10 studies are eligible. We will plot the intervention effects (ORs/HRs) against the study size on a logarithmic scale. Where appropriate, we will perform assessment of asymmetry as per the tests proposed by Harbord 2006 or Rücker 2008, as dictated by the observed degree of heterogeneity between studies. Where available, we will search study protocols for evidence of reporting bias.
Data synthesis
We will synthesize the above endpoints using a random‐effects model and using the generic inverse‐variance model if there are more than five studies for the relevant endpoint. If we identify fewer than five studies, we will use a fixed‐effect model. We will use intention‐to‐treat analyses wherever possible.
We will use random‐effects meta‐analysis throughout the proposed meta‐analyses to incorporate heterogeneity (statistical or clinical, or both, where present) amongst studies.
Subgroup analysis and investigation of heterogeneity
Where possible, we will perform subgroup analyses according to:
-
stage (I/II versus III);
-
for stage III disease: Nodal staging (N1 (three or fewer nodes involved); N2 (four or more nodes involved);
-
type of anti‐VEGF agent (monoclonal antibody versus tyrosine kinase inhibitor or other);
-
duration of anti‐VEGF therapy (using meta‐regression);
-
presence of lymphovascular space invasion (yes versus no);
-
microsatellite instability (yes versus no).
We will present forest plots for each of the subgroups if possible.
If a sufficient number of studies are available in reporting outcomes by the above subgroups, we will perform meta‐regression in order to compare the effects of anti‐VEGF therapy on the subgroups. Where present, we will explore heterogeneity, both with reference to study characteristics and to the above subgroups.
Sensitivity analysis
We will perform sensitivity analyses to test the robustness of the review with regard to the following:
-
exclusion of studies where there was no allocation concealment, or where the method of randomization was unclear, or where the masking was unclear;
-
exclusion of quasi‐RCTs.
We will perform assessments of the effect of publication bias on the meta‐analyses using the non‐parametric trim and fill method (Duval 2000; Taylor 1998).
Where clinically plausible (a clinical reason can be identified for an extreme result of one trial compared to the other trials), we will exclude outlier studies from the meta‐analysis as part of sensitivity analyses. We may introduce further sensitivity analyses if we identify potential issues during the course of the review.
Quality of the evidence
We will use the Grading of Recommendations, Assessments, Development and Evaluation (GRADE) approach to assess the quality of evidence of the listed outcomes. and present the results in 'Summary of findings' tables (Schünemann 2011). GRADE will incorporate the outcomes of overall survival, relapse‐free survival, overall Grade 3‐4 toxicity and Grade 5 toxicity.
The GRADE system classifies the quality of evidence in one of four grades:
-
high: further research is very unlikely to change our confidence in the estimate of effect;
-
moderate: further research is likely to have an impact on our confidence in the estimate of effect and may change the estimate;
-
low: further research is very likely to have an important impact on our confidence on the estimate of effect and is likely to change the estimate;
-
very low: any estimate of effect is very uncertain.
The quality of evidence can be downgraded by one (serious concern) or two levels (very serious concern) for the following reasons: risk of bias, inconsistency (unexplained heterogeneity, inconsistency of results), indirectness (indirect population, intervention, control, outcomes) and imprecision (wide CIs, single trial, contradictive effect estimates). The quality can also be upgraded by one level due to a large summary effect.
| Stage | Frequency | Metastases | 5‐ year disease‐free survival | 5‐ year overall survival | Usual recommended treatment after surgery |
| I | 39% (I and II) | No | 95% | 83% | Observation |
| II | 39% (I and II) | No | 85% | 70% | Consider adjuvant chemotherapy |
| III | 36% | No | 69% | 58% | Recommend adjuvant chemotherapy (e.g. FOLFOX) |
| IV | 20% | Yes | (0%) | 13% | Surgery not possible in most cases |
