Rol de la radioterapia adicional para el linfoma primario del sistema nervioso central
Resumen
Antecedentes
Antes de la introducción del agente quimioterapéutico metotrexato, la radioterapia era la única opción de primera línea para el tratamiento de los pacientes con linfoma primario del sistema nervioso central (LPSNC). Ahora que el metotrexato está disponible, se ha cuestionado la función de la radioterapia en el tratamiento del LPSNC. Aunque diversos estudios indican resultados prometedores con respecto a la supervivencia general y sin progresión con la administración de regímenes quimioterapéuticos solos, así como en combinación con la radioterapia, no se han definido regímenes estándar basados en la evidencia.
Objetivos
El objetivo de esta revisión fue evaluar y resumir la evidencia disponible con respecto a la eficacia y la tolerabilidad de la radioterapia además de la quimioterapia en el tratamiento de los pacientes inmunocompetentes con LPSNC.
Métodos de búsqueda
Se hicieron búsquedas en el Registro Cochrane Central de Ensayos Controlados (Cochrane Central Register of Controlled Trials) (CENTRAL) (número 01, 2014), MEDLINE desde enero de 1950 hasta febrero de 2014 y en actas de congresos desde 2005 hasta 2013.
Criterios de selección
Se incluyeron los ensayos controlados aleatorizados (ECA) que compararon quimioterapia más radioterapia con quimioterapia sola en pacientes con LPSNC. Los resultados definidos en esta revisión fueron supervivencia general, supervivencia sin progresión, respuesta al tratamiento, eventos adversos, mortalidad relacionada con el tratamiento y calidad de vida. Se excluyeron los ensayos en los que el régimen de quimioterapia difirió entre los brazos de tratamiento, los ensayos en los que menos del 80% de los participantes presentaban LPSNC o los que reclutaron a pacientes inmunocomprometidos con LPSNC.
Obtención y análisis de los datos
Dos autores de la revisión examinaron de forma independiente los resultados de las estrategias de búsqueda para determinar la elegibilidad para esta revisión. Ambos evaluaron el riesgo de sesgo. Cuando no se disponía de los datos pertinentes, se estableció contacto con el investigador por correo electrónico.
Resultados principales
De los 556 estudios potencialmente relevantes, solamente dos cumplieron los criterios de inclusión. Uno de los estudios se excluyó porque se interrumpió de forma prematura y solo informó resultados preliminares. El único ensayo analizado reclutó a 551 participantes que recibieron quimioterapia de primera línea (metotrexato), seguida de radioterapia cerebral total (RCT) o que recibieron quimioterapia solamente (metotrexato seguido de citarabina en caso de respuesta incompleta). En este ensayo de no inferioridad, la población con intención de tratar consistió en 411 participantes y la población por protocolo en 318 participantes. La posibilidad de riesgo de sesgo de este estudio abierto se calificó como moderada.
La estimación del efecto de quimioterapia más RCT sobre la supervivencia fue similar a la de la quimioterapia sola, pero debido al IC amplio, no es posible descartar la superioridad de cualquiera de las terapias. Esto se aplicó tanto a la población de ITT (CRI 1,01; IC del 95%: 0,79 a 1,30; P = 0,94) como a la población de PP (CRI 1,06; IC del 95%: 0,80 a 1,40; p = 0,71) (evidencia de calidad moderada). Debido al escaso número de participantes y al riesgo de sesgo de detección, se encontró evidencia de baja calidad de una mejoría en la supervivencia sin progresión en los participantes de la población con intención de tratar que recibieron RCT además de quimioterapia (CRI 0,79; IC del 95%: 0,63 a 0,99; P = 0,041). También se observó una mejoría en la SSP con RCT más quimioterapia en los participantes de la población por protocolo, pero el IC fue ligeramente más amplio y el resultado no fue significativo (CRI 0,82; IC del 95%: 0,64 a 1,07; P = 0,14). No se evaluaron la mortalidad relacionada con el tratamiento ni la calidad de vida relacionada con la salud. La neurotoxicidad relacionada con el tratamiento se evaluó clínicamente en 79 participantes, revelando signos de neurotoxicidad en el 49% de los que recibieron quimioterapia más radioterapia y en el 26% de los que recibieron sólo quimioterapia (RR 1,85; IC del 95%: 0,98 a 3,48; P = 0,054) (evidencia de muy baja calidad).
Conclusiones de los autores
En resumen, la evidencia actualmente disponible (un ECA) no es suficiente para concluir que RCT más quimioterapia y la quimioterapia sola tienen efectos similares sobre la supervivencia general en los pacientes con LPSNC. Los resultados indican que el agregado de radioterapia (RCT) a la quimioterapia puede aumentar la supervivencia sin progresión, pero también puede aumentar la incidencia de neurotoxicidad en comparación con la quimioterapia sola (monoterapia con metotrexato). Como la función de la quimiorradioterapia en el tratamiento del LPSNC todavía no está clara, se necesitan ensayos aleatorizados prospectivos adicionales antes de establecer conclusiones definitivas.
PICO
Resumen en términos sencillos
Función de la radioterapia cerebral (rayos X) en el tratamiento del linfoma en el cerebro
Antecedentes: El linfoma primario del sistema nervioso central (LPSNC) es un tipo de cáncer que se presenta en el cerebro o la médula espinal. Es un tipo de linfoma poco frecuente y agresivo. Los pacientes que desarrollan LPSNC sobreviven solamente durante cuatro meses como promedio, si no reciben tratamiento. Durante mucho tiempo, el único tratamiento que mostró algún efecto beneficioso fue la radioterapia cerebral total (RCT), en la que se utilizan los rayos X para destruir las células cancerosas en el cerebro. Sin embargo, varios estudios indican que este método de tratamiento también produce signos de daño al tejido cerebral sano. Desde la introducción del metotrexato, un fármaco potente de quimioterapia que muestra grandes efectos beneficiosos, los expertos han debatido la función de la radioterapia en el tratamiento de los pacientes con LPSNC. La radioterapia podría combinarse con quimioterapia, o no utilizarla en absoluto, especialmente si se consideran sus efectos potencialmente perjudiciales.
Pregunta de la revisión: El objetivo de esta revisión fue encontrar cualquier estudio científico de alta calidad que se centrara en la efectividad y los efectos perjudiciales de la radioterapia en el tratamiento del LPSNC. Una búsqueda amplia en todas las bases de datos relevantes produjo 556 referencias con respecto a este tema. Solamente un estudio cumplió con los criterios estrictos de inclusión, por lo que se analizó detalladamente.
Características de los estudios: Se realizaron búsquedas en todas las bases de datos de estudios relevantes publicados entre enero de 1950 y febrero de 2014. Solo se incluyó un estudio que reclutó a 551 participantes y trató a una mitad con metotrexato seguido de RCT y a la otra mitad con metotrexato solo. Si los participantes del último grupo no respondieron suficientemente al metotrexato solo, se administró otro fármaco, la citarabina. Participantes con un mínimo de 18 años de edad fueron reclutados en 75 centros de Alemania entre mayo de 2000 y mayo de 2009.
Resultados clave: Cuando se analizaron los datos con respecto al efecto de la quimioterapia más RCT o la quimioterapia sola sobre la supervivencia general, los resultados fueron poco precisos y un tratamiento podría haber sido superior al otro. Otro resultado que se consideró además de la supervivencia general fue la supervivencia sin progresión (SSP), un estado en el que la enfermedad no empeora. El agregado de radioterapia a la quimioterapia tuvo un efecto positivo sobre la SSP, ya que extendió ligeramente el período en el cual la enfermedad no progresó, en comparación con el logrado con quimioterapia sola. Los autores de la revisión no analizaron la mortalidad relacionada con el tratamiento.
También se analizó si el tratamiento provocó cualquier daño al tejido cerebral sano durante el tratamiento. No se encontró evidencia de que los síntomas de deterioro de la función cerebral relacionados con el tratamiento fueran más frecuentes en el grupo de participantes que recibieron quimioterapia más radioterapia que en los que recibieron quimioterapia sola.
Calidad de la evidencia: La calidad del cuerpo de evidencia se consideró moderada a baja, ya que solamente se incluyó un ensayo con un escaso número de participantes. Como el estudio incluido no analizó los eventos adversos en todos los participantes, la calidad de la evidencia para el resultado de neurotoxicidad se consideró muy baja.
Conclusión: En resumen, la evidencia actualmente disponible (un ensayo controlado aleatorizado) no es suficiente para concluir que la RCT más quimioterapia y la quimioterapia sola tienen efectos similares sobre la supervivencia general en los pacientes con LPSNC. El agregado de RCT a la quimioterapia puede aumentar la supervivencia sin progresión, pero posiblemente también podría aumentar los niveles de efectos tóxicos sobre el cerebro. Se necesitan ensayos aleatorizados prospectivos adicionales antes de establecer conclusiones definitivas acerca de la función de agregar la radioterapia a la quimioterapia en el tratamiento del LPSNC.
Authors' conclusions
Summary of findings
| Additional radiotherapy for PCNSL | ||||||
| Individuals or population: Participants with PCNSL | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Control (chemotherapy only) | Additional radiotherapy | |||||
| Mortality (instead of OS) | Moderate risk | HR 1.01 | 411 | ⊕⊕⊕⊝ | Instead of overall survival, mortality is reported in this table, for methodological reasons | |
| 750 per 1000 | 753 per 1000 | |||||
| Relapses/death (instead of PFS) Follow up: median 60 months | Moderate risk | HR 0.79 | 411 | ⊕⊕⊝⊝ | Instead of PFS, relapses and deaths are reported in this table, for methodological reasons | |
| 820 per 1000 | 742 per 1000 | |||||
| Treatment‐related mortality | See comment | See comment | Not evaluated | |||
| Adverse events: treatment‐related neurotoxicity | 265 per 1000 | 490 per 1000 | RR 1.85 | 79 | ⊕⊝⊝⊝ | |
| Adverse events: delayed neurotoxicity | see comment | see comment | see comment | 84 | ⊕⊝⊝⊝ | It is not reported whether the participants who received WBR who were integrated into this analysis were from the additional‐WBR group only, or whether any participants from the chemotherapy‐only group who received rescue WBR were also included |
| Quality of life | See comment | See comment | Not evaluated | |||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| GRADE OS = overall survival PCNSL = primary central nervous system lymphoma PFS = progression‐free survival 1 Only one trial included with a wide confidence interval, leading to imprecision 4 Only 84 of 411 participants evaluated | ||||||
Background
Description of the condition
Primary central nervous system lymphoma (PCNSL) is an extranodal high‐grade lymphoma confined to the central nervous system (CNS) compartment. In approximately 90% of cases it is B‐cell derived. It is often impossible to classify the remaining 10%, which can generally be divided into T‐cell, low‐grade and anaplastic lymphomas (Gerstner 2010; Paulus 1999; Rubenstein 2008).
With an incidence of 2/106 persons per year (Haldorsen 2007), PCNSL is a rare but aggressive subtype of non‐Hodgkin lymphoma (NHL), accounting for 2% to 3% of all NHL cases and for 4% of primary brain tumours (Rubenstein 2008). Most individuals with PCNSL are 60 years or older at time of diagnosis (Ferreri 2002; Haldorsen 2007).
Together with immunodeficiency, several autoimmune diseases with an increased incidence of lymphoma in general, such as lupus erythematosus or myasthenia gravis, are currently known to be predisposing conditions for PCNSL (Bhagavathi 2008). The incidence of PCNSL has increased since the 1970s, mainly due to rising numbers of people infected with human immunodeficiency virus (HIV) and, to some degree, improvements in diagnostic tools. However, even adjusting for these factors does not fully account for this increase (Schabet 1999).
PCNSL generally presents as dense clusters of cells surrounding small cerebral blood vessels. The pathogenesis of this phenomenon is not fully understood, though Rubenstein et al. have clarified the signal transduction mechanisms and pathways involved (Rubenstein 2006). PCNSL cells are activated B cells (Camilleri‐Broët 2006) from germinal centres (Montesinos‐Rongen 2008) that express specific forms of microRNA, which sets them apart from the B cells of other forms of lymphoma. PCNSL cells are further characterised by a loss of chromosome 6p21.32‐p25.3, which plays a role in the prevention of this malignancy, and an irregular gain of gene 12q15, which influences apoptosis pathways (Soussain 2009). Current knowledge about the connection between B cells and tumour growth is very limited. However, it has been shown that the specific expression of certain adhesion molecules (for example, ß1‐integrin, matrix metallopeptidase‐2 , matrix metallopeptidase‐9 and intercellular adhesion molecule‐1) may play a role in the process by which B cells penetrate the blood‐brain barrier and their subsequent dissemination (Brunn 2007).
In 65% to 85% of immunocompetent people with PCNSL, the disease presents as a solitary lesion (Bataille 2000; Camilleri‐Broët 1998; Kuker 2005), the average number of lesions being 1.7 per person (Kuker 2005). The location of the lesion varies according to the literature, but spinal cord involvement is rare (Kuker 2005). The size of the tumour may vary. It is generally well demarcated and may show necrotic regions, viable cells generally being found centrally, close to the surrounded blood vessel (Bhagavathi 2008; Kleihues 2000). Individuals with PCNSL may display focal neurological deficits (70% of cases), neuropsychiatric signs (43%), elevated intracranial pressure (33%), seizures (14%) and ocular symptoms (4%) (Bataille 2000). Seizures are less common than in other intracranial neoplasms. A reason for this is the tendency of PCNSL to be centrally located in the white matter of the brain rather than in the epileptogenic areas of grey matter (Gerstner 2010).
Diagnosis of PCNSL relies on imaging procedures, magnetic resonance imaging (MRI) scans of the brain being the gold standard (Abrey 2005). Contrast‐enhanced body‐computed‐tomography (CT) is used to exclude systemic manifestations of lymphoma (Schultz 2010). Staging can be supported by positron emission tomography scans (Kuker 2005). The diagnosis is established by the microscopic evaluation of biopsy material, predominantly from frame‐based biopsies.
Prognosis of PCNSL is poor, with an average survival of four months if untreated and a median overall survival ranging from 14.3 to 55.4 months following treatment (Fine 1993). The choice of therapeutic regimen and the resulting outcome depends highly on the individual's age and clinical performance (Gerstner 2010). The two‐year overall survival (OS) rate has been determined to be approximately 52%, and median failure‐free survival (FFS) is nine months (Ferreri 2003).
Surgical resection is generally irrelevant to the management of PCNSL due to the location and spread of the tumour. Whole brain radiation (WBR) has been the generally used monotherapy for PCNSL in past decades. First steps regarding the use of combined modality chemoradiotherapy in PCNSL were taken using regimens such as cyclophosphamide–adriamycin–vincristine–decadron (CHOD). This regimen yielded poor results, which did not differ from those achieved with radiotherapy alone (Corn 2000). Extensions in progression‐free survival (PFS) and OS were achieved by the introduction of blood‐brain‐barrier‐penetrating chemotherapy, a regimen consisting of methotrexate 2.5 g/m², vincristine, procarbazine and intraventricular methotrexate (12 mg) (DeAngelis 1992; DeAngelis 2002). Since the introduction of methotrexate, a shift towards combined modality chemoradiotherapy has taken place (Ferreri 2011). High‐dose methotrexate is now considered to be the most beneficial single agent and, hence, the basis of any PCNSL treatment (Ferreri 2009; Gerstner 2010).
Description of the intervention
The effect of radiotherapy on PCNSL remains to be defined. It is generally applied to the whole brain to include the multifocal presentation of PCNSL (Schultz 2010) and has shown to increase OS by 8 to 12 months (Gerstner 2010). Although an initial response of 90% has been observed, tumour growth generally progresses after a few months (Nelson 1992).
An opposed‐field arrangement using a 6 to 10 megavolt (MV) photon that focuses equally on the left and right hemisphere is the most common setup. The dose‐corrected inclusion of the meninges and subarachnoid space, as well as the posterior third of the orbits (full inclusion if occular involvement is evident), is highly relevant. The inferior demarcation is the 2nd or 3rd vertebra (Schultz 2010).
Initial research into the use of WBR for the treatment of PCNSL led to the assumption that doses of 50 Gy or higher would produce the best results (Murray 1986). This could not be supported in later studies (Nelson 1992). Doses used in former studies range from 23.4 Gy (Shah 2007) to more than 50 Gy (Bessell 2002; Fisher 2005; Murray 1986). Most of the studies that have included WBR have focused on the varying chemotherapy regimens used, thus allowing only a tentative consideration of the results of radiotherapy.
A well‐described adverse effect of WBC is delayed neurotoxicity, especially in individuals older than 60 years. The clinical manifestation of neurotoxicity commonly includes impairments in muscle coordination (ataxia), subcortical dementia, incontinence, attention and memory. Pathological studies suggest that damage to the neural progenitor cells causes demyelination, neuronal loss, gliosis and rarefaction of the white matter. MRI scans also link this mechanism to white matter and ventricular abnormalities (Gerstner 2010).
How the intervention might work
The underlying mechanism of WBR ‐ and radiotherapy in general ‐ is the creation of free radicals in a specific area of tissue using ionising radiation, causing subsequent DNA damage. The latter affects tumour cells more strongly than healthy cells for two reasons: tumour cells usually lack the repair mechanisms (to a certain degree) that allow healthy cells to survive non‐lethal DNA damage and they have a higher mitotic activity, requiring functioning DNA. Advances in technology allow radiotherapy to focus specifically on tumour tissue, sparing surrounding tissue from most of its adverse effects. Generally the full dose is fractionated. This enables healthy cells to regenerate, allowing the overall dose of radiation administered to be increased. Fractionating the dose also prevents tumour cells from evading the effects of treatment during relatively radioresistant phases of mitosis (Bomford 2002).
The positive effects of combined modality chemoradiotherapy on survival in PCNSL have been well assessed in many studies. In contrast, radiotherapy only, although showing an initial response of 90%, is followed by a remission that lasts for only a few months (Nelson 1999). Hence, research focusing on chemotherapy plus deferred radiotherapy describes radiotherapy only as an inadequate solution (Batchelor 2003; Gerstner 2008). Combined modality chemoradiotherapy may therefore be the most promising regimen currently available for PCNSL.
Why it is important to do this review
The prognosis of PCNSL is poor and the role of radiotherapy, especially WBR, in its treatment remains unclear. Treatment of individuals using WBR alone resulted in a median survival of between 12 and 18 months (Gerstner 2010). Chemotherapy combined with WBR resulted in a PFS of 24 months and an OS of 36.9 months (DeAngelis 2002). It could be shown that methotrexate monotherapy leads to some promising results with an 3‐year overall survival of 32% (Bergner 2012). Current data suggest that for most individuals with PCNSL a combined chemoradiotherapy approach could be the best treatment option (Ferreri 2011; Gerstner 2010; Rubenstein 2008). However, there is a lack of solid data which could serve as a basis for objective judgement. This is mainly due to the fact that study designs and treatment protocols vary greatly, as do results regarding OS and PFS (Laack 2010). An overview that will help to evaluate the effects of radiotherapy as part of the therapy options available for PCNSL is important, especially since the incidence of PCNSL is increasing in line with life expectancy (Laack 2010).
Objectives
The objective of this review was to assess and summarise the evidence available regarding the efficacy and tolerability of radiotherapy in addition to chemotherapy in the treatment of immunocompetent individuals with PCNSL.
Methods
Criteria for considering studies for this review
Types of studies
We considered only randomised controlled trials (RCTs). We included both full text and abstract publications, if sufficient information was available on study design, characteristics of participants, interventions and outcomes. We excluded quasi‐randomised trials and cross‐over trials.
Types of participants
We included trials involving immunocompetent participants with a confirmed diagnosis of PCNSL (by histology, cerebrospinal fluid cytology or vitrectomy in the case of intraocular lymphoma) of all ages, both sexes and all ethnicities. Because PCNSL in individuals with HIV infection or acquired immune deficiency syndrome shows substantial differences with regard to clinical features, course and prognosis of the disease to PCNSL in other types of individual, we therefore excluded studies involving these types of participants. In addition, studies involving participants with PCNSL in the context of other circumstances of immunosuppression were also not eligible.
Due to differences between immunocompetent and immunocompromised individuals with PCNSL we did not expect a study involving both types of participants to be available. If trials consisted of mixed populations with different conditions or involving both immunocompetent and immunocompromised individuals with PCNSL we intended to use data from the subgroup of immunocompetent participants. If subgroup data for these individuals were not provided (after contacting the authors of the trial) we intended to exclude the trial if fewer than 80% of participants had PCNSL and were immunocompetent. However, we did not identify any study involving both types of participants.
Types of interventions
Types of interventions included any single‐agent or multi‐agent chemotherapy (including either standard or high‐dose alkylating agent, antimetabolite, topoisomerase inhibitor, anthracycline, glucocorticoid or monoclonal antibody regimens) administered in addition to radiotherapy (experimental intervention) compared with the same single‐agent or multi‐agent chemotherapy alone (control intervention). Planned supportive and other treatments, as well as routes of administration, had to be identical for participants in the experimental and control study groups.
Types of outcome measures
Primary outcomes
OS, defined as the time from random treatment assignment into the study to death from any cause or to last available follow up.
Secondary outcomes
-
PFS, defined as the time from random treatment assignment into the study to first confirmed progression or relapse, death from any cause or to last follow up
-
Response: complete response (CR), defined as the complete disappearance of the tumour
-
Adverse events (AEs)
-
Treatment‐related mortality (TRM)
-
Quality of life (QoL)
Search methods for identification of studies
Electronic searches
We adapted the search strategies suggested in Chapter 6 of the Cochrane Handbook for Systematic Reviews of Interventions (Lefebvre 2011). We applied no language restrictions so as to reduce language bias.
We searched the following databases:
-
Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library, 31 January 2014 (for search strategy, see Appendix 1);
-
MEDLINE (Ovid) (1950 to 3 February 2014) (for search strategy, see Appendix 2).
We searched the conference proceedings of the annual meetings of the following societies for abstracts published between 2005 and 2013, if not included in CENTRAL:
-
American Society of Hematology;
-
American Society of Clinical Oncology;
-
European Hematology Association.
We searched the following database of ongoing trials:
-
metaRregister of Controlled Trials (http://www.controlled‐trials.com/mrct).
Searching other resources
We handsearched the reference lists of all identified studies, relevant review articles and current treatment guidelines (Marcus 2009; Schlegel 2012).
Data collection and analysis
Selection of studies
Two review authors (JZ, NS) independently screened the results of the search strategies for eligibility for this review by reading the abstracts. In case of disagreement, we obtained the full text of the article. Both authors (JZ, NS) then independently examined the full‐text report to determine eligibility. If no consensus could be reached, we intended to ask a third review author to adjudicate, as suggested in Chapter 7 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), but this was not necessary.
We documented the study selection process in a flow chart as recommended in the PRISMA statement (Moher 2009) showing the total numbers of retrieved references and the numbers of included and excluded studies.
Data extraction and management
Two review authors (JZ, NS) independently extracted the data according to Chapter 7 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We contacted the authors of individual studies for additional information, if required. We used a standardised data extraction form containing the following items.
-
General information: study ID; author; title; journal; publication date; citation and contact details of primary or corresponding authors; sources of funding.
-
Study characteristics: design; objectives and duration of the study; source of participants; number of participating centres; inclusion and exclusion criteria; sample size; treatment allocation; comparability of groups; subgroup analysis; statistical methods; power calculations; compliance with assigned treatment; length of follow up.
-
Participant characteristics: age; sex; ethnicity; setting; number of participants recruited/randomised/evaluated; numbers of participants lost to follow up; additional diagnoses; type and dosage of radiation treatment.
-
Interventions: setting; dose and duration of radiotherapy; type, dosage and duration of chemotherapy (number of cycles); administration route; supportive treatment.
-
Outcomes: OS; PFS; response; AEs; TRM; QoL.
Assessment of risk of bias in included studies
To assess quality and risk of bias we used a questionnaire (validity assessment form) containing the following items, as suggested in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a):
-
sequence generation;
-
allocation concealment;
-
blinding (participants, personnel, outcome assessors);
-
incomplete outcome data;
-
Selective outcome reporting;
-
other sources of bias.
For each criterion, we made a judgement using one of three categories:
-
'low risk': if the criterion was adequately fulfilled in the study, then the study was considered at a low risk of bias for the given criterion;
-
'high risk': if the criterion was not fulfilled in the study, then the study was considered at high risk of bias for the given criterion;
-
'unclear': if the study report did not provide sufficient information to allow for a judgement of 'yes' or 'no', or if the risk of bias was unknown for one of the criteria listed above, then the study was considered at unclear risk of bias for the given criterion.
Measures of treatment effect
For binary outcomes we calculated risk ratios (RRs) with 95% confidence intervals (CIs) for each trial. For time‐to‐event outcomes we extracted hazard ratios (HRs) from published data according to Parmar 1998 and Tierney 2007. We would have calculated continuous outcomes as standard mean differences (SMD), if any had been included.
Dealing with missing data
As suggested in Chapter 16 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b), there are many potential sources of missing data that had to be taken into account: at study level, at outcome level and at summary data level. First, it is important to distinguish between data 'missing at random' and 'not missing at random'.
We contacted the original trial investigator to request missing data, but received no reply. We have addressed the potential impact of missing data on the findings of the review in the Discussion section.
Assessment of heterogeneity
Since only one trial met the inclusion criteria, an assessment of heterogeneity was obsolete. We had intended to assess the heterogeneity of treatment effects between trials using a Chi² test with a significance level at a P value < 0.1. We would have used the I² statistic to quantify possible heterogeneity (I² > 30% moderate heterogeneity, I² > 75 % considerable heterogeneity), as suggested in Chapter 9 of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2011). We had intended to explore potential causes of heterogeneity by sensitivity and subgroup analyses using meta‐regression.
Assessment of reporting biases
In meta‐analyses involving at least 10 trials we had intended to explore potential publication bias by generating a funnel plot, which we would have statistically tested by means of a linear regression test. We would have considered a P value < 0.1 as significant for this test, as suggested in Chapter 10 of the Cochrane Handbook for Systematic Reviews of Interventions (Sterne 2011). However, since only one study was included we did not perform this test.
Data synthesis
Since only one study was included in the review, we did not perform any meta‐analyses. We had intended to perform analyses according to the recommendations of Chapter 9 of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2011). We would have used aggregated data for analysis. For statistical analysis, we had intended to enter data into the Cochrane statistical package Review Manager (RevMan) 5.1 (RevMan 2011). One review author would have entered data into the software and a second review author would have checked it for accuracy. We had intended to perform meta‐analyses using a fixed‐effect model (for example, the generic inverse variance method for survival data outcomes and the Mantel‐Haenszel method for dichotomous data outcomes). We would have used the random‐effects model for sensitivity analyses. For future updates of the review we will use both a fixed‐effects and a random‐effects model and report results from both models.
We have created a 'summary of findings Table for the main comparison' giving the absolute risk of the following patient‐relevant outcomes: mortality, relapses and deaths, TRM, AEs and QoL for each group (intention‐to‐treat (ITT) population) using GRADE criteria, as recommended in Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2011)
Subgroup analysis and investigation of heterogeneity
Since only one study was included in the review, subgroup analysis and the investigation of heterogeneity were not applicable.
We had intended to perform analyses on the following subgroups, if appropriate:
-
age (<60 years and ≥60 years);
-
chemotherapeutic agents: different types and dosages of chemotherapeutic agents;
-
stage of disease;
-
irradiation: different doses.
Sensitivity analysis
Since only one study was included in the review, sensitivity analysis was not applicable.
We intended to perform sensitivity analysis to assess how sensitive our results were to reasonable changes.
-
Quality components, including full text publications/abstracts, preliminary results versus mature results.
-
Random‐effects modelling.
Results
Description of studies
Results of the search
The search of the literature produced 556 potentially relevant references. Of these 546 were excluded during the initial screening process. The remaining 10 were evaluated in detail. The BMPD study, accounting for eight references, was excluded because the trial was abandoned prematurely and only vague preliminary data in summarised form with no specific data per arm were obtainable. The validity of the preliminary data in terms of methodology could not be verified. Thus, we included only the G‐PCNSL‐SG‐1 trial, reported in two publications, in this review. See flow diagram (Figure 1) for more details of the search results.
Results of the search
Included studies
Only one study met the predefined criteria and was included (G‐PCNSL‐SG‐1). All data were extracted from the full‐text version.
Design
The G‐PCNSL‐SG‐1 trial was a multi‐centre, two‐arm, randomised, non‐inferiority trial with a 1:1 randomisation ratio. Participants were enrolled between May 2000 and May 2009 and randomised to a study arm in blocks of six.
Sample sizes
G‐PCNSL‐SG‐1 recruited 551 participants and divided them into two trial arms. Participants received high‐dose methotrexate with (N = 273) or without subsequent WBR (N = 278). A non‐inferiority design with a power of 60% and a HR of 1.2 was chosen to demonstrate the non‐inferiority of chemotherapy without WBR compared with chemotherapy plus WBR. To successfully demonstrate non‐inferiority, a minimum number of 151 participants was necessary per arm.
Location
The trial involved 75 centres throughout Germany.
Participants
A total of 551 participants of 18 years of age or older were included and assigned to receive first‐line chemotherapy. Before and during first‐line chemotherapy, 17 participants failed to meet inclusion criteria and were excluded, 8 refused treatment, 66 died, 27 were not available for follow‐up investigation and the response status to chemotherapy was unknown in 22. Thus, 203 participants (of 273 originally) in the additional‐WBR arm and 208 (of 278) in the chemotherapy‐only arm finished first‐line chemotherapy and had a known response status.
PCNSL was confirmed histologically, cytologically or via immunocytochemistry. Further inclusion criteria were: no history of cytostatic chemotherapy, no involvement of extra‐CNS tissue and a minimum of two months further life expectancy. Individuals were excluded if they were HIV‐positive or suffered from an active infection. A limited Karnofsky performance score (KPS) of less than 50% for reasons other than PCNSL led to exclusion, as did a KPS of less than 30% as a direct consequence of PCNSL (G‐PCNSL‐SG‐1).
Interventions
All enrolled participants were randomised to one of two trial arms: chemotherapy (4 g/m² methotrexate intravenously (iv) on day one; after August 2006, 1.5 g/m² ifosfamide iv on day three, four and five of the same 14‐day cycle was added) plus subsequent WBR (45 Gy, administered in 30 fractions à 1,5 Gy) or chemotherapy alone. Hence, participants (N = 409) accrued in the first six years (2000 to 2006) received methotrexate only and were then treated according to trial arm. After 2006, due to an assumption by the investigators that methotrexate alone may not be sufficient, subsequently accrued participants (N = 128) were treated with methotrexate plus ifosfamide. All participants, irrespective of treatment arm, were primarily treated according to a chemotherapy regimen. After this initial intervention, allocation to treatment arms defined further treatment: complete responders in the additional‐WBR arm received consolidating radiotherapy, whereas complete responders in the chemotherapy‐only arm received no further treatment. If a CR was not achieved after chemotherapy, participants assigned to the additional‐WBR arm received rescue WBR, whereas those in the chemotherapy‐only arm received high‐dose cytarabine (G‐PCNSL‐SG‐1).
Outcomes
Primary outcome measure
The primary outcome was OS, based on a non‐inferiority design to demonstrate no decrease in OS when WBR was omitted.
Secondary outcome measures
The secondary outcomes were: CR rate after first‐line chemotherapy, WBR or cytarabine; PFS; toxic effects of therapy; neurotoxicity, assessed clinically and via MRI orCT. Sequential analyses by Mini‐Mental State Examination score were planned but never carried out.
Funding
German Cancer Aid provided funding but was otherwise not involved and had no access to any data.
Conflicts of interest
No conflicts of interest were mentioned and we have no reason to suspect otherwise.
Excluded studies
The BMPD study met all our predefined inclusion criteria to the degree of being considered for inclusion. Its aim was to treat individuals with PCNSL with the BMPD regimen initially, comprising: BCNU (bis‐chloroethylnitrosourea), methotrexate (intrathecally and iv), procarbazine and dexamethasone. Following this, participants who achieved a CR were randomly assigned to one of two study arms to receive either WBR or a further course of the same chemotherapy regimen. Participants without a CR would receive idarubicin and ifosfamide until a CR was achieved and then undergo randomisation. Non‐responders automatically received rescue WBR. Of 56 initially included participants only 23 were randomised. Of the 11 participants then randomised to receive WBR, 4 refused further treatment. The study was subsequently cancelled prematurely.
Only vague preliminary data were attainable and we could not verify their validity. Efforts to obtain any relevant data by contacting the lead researcher, A Korfel, remain unsuccessful. Hence, the BMPD study was excluded from this review despite meeting the formal inclusion criteria.
Risk of bias in included studies
See the 'risk of bias' section in the 'Characteristics of included studies' table and Figure 2.
Risk of bias summary: review authors' judgements about each risk of bias item for G‐PCNSL‐SG‐1.
Allocation
In G‐PCNSL‐SG‐1, participants were allocated en bloc (six at a time) by the Biostatistics Centre (Department of Biostatistics and Clinical Epidemiology, Charité Berlin, Berlin, Germany) using a computer program designed for this purpose. Upon enrolling a participant the investigator contacted the centre via fax, receiving a reply fax disclosing the allocation of the participant to a study arm. Because of the thorough methodology described above, we considered the study to be at low risk of allocation and selection bias.
Blinding
Neither participants nor physicians were blinded regarding allocation to trial arm. A placebo version of radiotherapy was judged to be unfeasible. Any blinding of outcome assessors was not possible as the assessing physician was simultaneously responsible for the treatment of the participant (G‐PCNSL‐SG‐1). Thus, we considered the study to be at high risk of performance and detection bias.
Incomplete outcome data
The only included study, G‐PCNSL‐SG‐1, revealed several methodological limitations. Protocol was violated in 93 cases at the discretion of the treating physician (i.e. application of rescue‐WBR for patients in the chemotherapy‐only arm). A high number of participants were unavailable during follow‐up investigations, which limits the data, as does the high number of participants for whom response status is unknown (49 of 526; 9%). Further limiting factors include the introduction of ifosfamide mid‐trial and the publication of preliminary results.
The possibility of attrition bias regarding the assessment of neurotoxic effects is increased by the limited sample size and treatment cross‐over. Participants in the chemotherapy‐only group who received rescue WBR after an incomplete response to chemotherapy were seemingly included in the analysis of neurotoxicity as recipients of WBR. The integration of participants from both arms into the analysis of neurotoxicity renders irrelevant the prospective two‐arm randomisation with regard to this outcome and severely limits the relevance of the data regarding neurotoxicity. This is especially true since participants requiring rescue WBR were likely to be suffering from severe PCNSL or to be in a reduced physical condition. The study authors cite insufficient funding as one reason for these limitations (G‐PCNSL‐SG‐1).
We therefore considered the study to be at high risk of attrition bias.
Selective reporting
We considered the study to be at low risk of selective reporting bias. All endpoints mentioned in the clinical trials registry entry (NCT00153530) have been reported.
Other potential sources of bias
We identified no other potential sources of bias.
Effects of interventions
See: Summary of findings for the main comparison Additional radiotherapy for PCNSL
One RCT, G‐PCNSL‐SG‐1, was included. Hence, no meta‐analysis of data was possible. In G‐PCNSL‐SG‐1, 25 participants were excluded prior to the commencement of chemotherapy (13 from the additional‐WBR arm and 12 from the chemotherapy‐only arm) due to participants refusing treatment or failing to meet inclusion criteria. This diminished the number of included participants from 551 to 526.
The ITT population included all participants who met all inclusion criteria and were treated with first‐line chemotherapy, and whose response was documented. The per‐protocol (PP) population included those participants who received the treatment according to the protocol of their specific group allocation (excluding non‐responders from the chemotherapy‐only arm who did not receive rescue cytarabine therapy after an initial incomplete response). The only included study recorded a median follow‐up of 50.7 months for the ITT population (N = 526). Median follow‐up in the PP population was 53.7 months.
Primary outcome: OS
A difference in OS between the additional‐WBR arm and the chemotherapy‐only arm could not be demonstrated, either in the ITT population (HR 1.01, 95% CI 0.78 to 1.30; P = 0.94) or the PP population (HR 1.06, 95% CI 0.80 to 1.40; P = 0.71). In the PP population, mean OS in the additional‐WBR arm (N = 154) was 32.4 months (range 25.8 to 39.0 months), that in the chemotherapy‐only arm (N = 164) was 37.1 months (range 27.5 to 46.7 months). Median OS was 21.5 months (95% CI 17.8 to 25.1) in the ITT population and 35.3 months (95% CI 29.6 to 41.1) in the PP population (G‐PCNSL‐SG‐1).
Stratification of data regarding response to first‐line chemotherapy did not reveal any statistically significant differences. Among participants achieving a CR the results were similar for both the ITT population (HR 1.18, 95% CI 0.78 to 1.78; P = 0.43) and the PP population (HR 1.15, 95% CI 0.73 to 1.80; P = 0.56). Results were also similar for participants without a CR , both in the ITT population (HR 0.85, 95% CI 0.62 to 1.16; P = 0.31) and the PP population (HR 0.74, 95% CI 0.51 to 1.06; P = 0.10) (G‐PCNSL‐SG‐1).
Secondary outcomes
PFS
Analysis of data for the ITT population showed a statistically significant increase in PFS in the additional‐WBR arm in contrast to the chemotherapy‐only group (HR 0.79, 95% CI 0.63 to 0.99; P = 0.041) (see Figure 4). This significant increase was not seen in the PP population (HR 0.82, 95% CI 0.64 to 1.07; P = 0.14). G‐PCNSL‐SG‐1 reported a PFS of 18.3 months (range 11.6 to 25.0 months) in participants receiving additional WBR and 11.9 months (range 7.3 to 16.5 months) in participants receiving chemotherapy‐only in the PP population. Participants in the additional‐WBR arm demonstrated a two‐year PFS of 43.5% (95% CI 35.3 to 51.7). Two‐year‐PFS was 30.7% (95% CI 23.1 to 38.3) for those in the chemotherapy‐only arm of the trial.
The relationship between the ITT and PP populations was reversed when the participants achieving a CR after first‐line chemotherapy only were included in the analysis. In this analysis, the increase in PFS seen in the additional‐WBR group in the ITT population was not significant (HR 0.79, 95% CI 0.54 to 1.14; P = 0.21) whereas that in the PP population was (HR 0.64, 95% CI 0.42 to 0.98; P = 0.04) (G‐PCNSL‐SG‐1). For patients without a CR to first‐line chemotherapy, WBR had a statistically significant positive effect on PFS in both the ITT (HR 0.69, 95% CI 0.52 to 0.92; P = 0.011) and PP populations (HR 0.60, 95% CI 0.42 to 0.85; P = 0.004) (G‐PCNSL‐SG‐1).
Response
The differentiation of response into overall, complete (CR) and partial response was available only for the response to first‐line chemotherapy. "Partial response" was not assessed there after. Of 203 participants assigned to additional WBR, 85 (42%) achieved a CR to first‐line chemotherapy. Of the 208 participants assigned to chemotherapy alone, 96 (46%) achieved a CR.
A total of 68 of the incomplete responders in the chemotherapy‐only group received high‐dose cytarabine therapy, which led to a CR in 17 participants (25%). A total of 131 of the participants in both groups without a CR were treated with rescue WBR (98 from the additional‐WBR arm and 33 from the chemotherapy‐only arm ‐ the latter at the discretion of their treating physician, thus causing violation of protocol) ‐ 59 (45%) achieved a CR. Combining the incomplete responders from both arms into the analysis of CR after WBR limits the comparability between the efficacy of radiotherapy and cytarabine as rescue regimens for gaining a CR.
AEs
Treatment‐related neurotoxicity was analysed only in participants in the PP population who had achieved and sustained a CR for at least three months in the follow‐up period. Clinically assessed treatment‐related neurotoxic effects were analysed for 79 participants, of which 45 were from the additional‐WBR group and 34 from the chemotherapy‐only group. A total of 53 of 79 participants reached a CR directly after initial chemotherapy, 22 after subsequent WBR and 4 after subsequent cytarabine therapy. Neurotoxicity was found clinically in 22 participants (49%) in the additional‐WBR arm and 9 participants (26%) in the chemotherapy‐only arm after 20.4 months and 32.4 months median follow‐up, respectively (RR 1.85, 95% CI 0.98 to 3.48; P = 0.054) (see Figure 3).
Forest plot of comparison: 1 Chemotherapy plus radiotherapy versus chemotherapy only, outcome: 1.1 Treatment‐related neurotoxicity.
The trial investigated long‐term neurotoxic effects using MRI or CT. Delayed neurotoxic effects were analysed in a group of 84 participants (median follow up: 51.4 months). A total of 56 of these participants had reached a CR following initial chemotherapy, 25 following subsequent WBR and 3 following subsequent therapy with cytarabine. Of 49 participants who received WBR, 35 (71%) fulfilled the criteria of neurotoxicity. Of 35 participants who did not receive WBR, only 16 (46%) fulfilled the criteria of neurotoxicity. As it is not reported whether the participants receiving WBR who were integrated into the analysis were from the additional‐WBR group only, or whether any participants from the chemotherapy‐only group that received rescue WBR were also included, the relevance of the analysis is diminished.
TRM
No TRM data specifically relating to the chemotherapy plus WBR or the chemotherapy regimens was described.
QoL
No data were available for the analysis of this outcome. The only included study, G‐PCNSL‐SG‐1, did not analyse QoL.
Discussion
The insights of this review are limited by the fact that only one study, G‐PCNSL‐SG‐1, met the inclusion criteria and was eligible for inclusion.
Summary of main results
We found one trial, including 551 participants, that evaluated WBR in addition to chemotherapy in individuals with PCNSL.
-
OS: no significant difference between the group receiving chemotherapy plus WBR and the group receiving chemotherapy only could be demonstrated.
-
PFS: a significant increase in PFS in the group receiving chemotherapy plus WBR, but not in the chemotherapy‐only group, could be demonstrated in the analysis of the ITT population, but not the PP population.
-
Neurotoxicity: there was no evidence to suggest that WBR in addition to chemotherapy increased the incidence of treatment‐related clinical neurotoxicity compared with chemotherapy alone.
-
Data regarding TRM and QoL were not reported.
Overall completeness and applicability of evidence
The results of this systematic review should be interpreted taking into consideration of the following.
-
Only one trial, evaluating 551 participants, was included. The review is therefore not adequately powered to detect small differences, especially in outcomes with few events.
-
The trial investigated long‐term neurotoxic effects using MRI or CT. It is not specified whether the participants who received WBR who were integrated into this analysis were from the additional‐WBR group only, or whether any participants from the chemotherapy‐only group that received rescue WBR were also included. This is highly relevant, as 33 participants from the chemotherapy‐only group received rescue WBR at the discretion of their treating physician. An inclusion of these participants into the analysis of neurotoxicity after WBR would influence data and distort any results.
-
An ongoing trial that may increase our understanding of the role of WBR in PCNSL is PRECIS, a randomised, controlled, prospective trial started in 2008. Its protocol is designed to compare the effect of first‐line high‐dose methotrexate chemotherapy followed by WBR to that of first‐line high‐dose methotrexate and haematopoietic stem cell rescue in individuals with PCNSL under the age of 60 years. Another ongoing trial, IELSG32, randomises individuals with PCNSL to one of three different chemotherapeutic interventions. Either WBR or ASCT will be administered to participants achieving a CR. The trial started in November 2009 and is currently still in the recruitment phase. This also applies to RTOG‐1114, a phase II clinical trial that is designed to treat individuals with PCNSL with rituximab, methotrexate, vincristine and procarbazine followed by either cytarabine or WBR plus cytarabine.
Quality of the evidence
Overall, we judge the potential risk of bias of the included trial as moderate. The included trial is reported as an adequately randomised and an open‐label study. The open‐label design (physicians and participants not blinded) could lead to performance or detection bias. In the included study, the blinding of participants as well as the blinding of physicians in the context of radiotherapy was impossible. Consequently, we judge the risk of bias to be high for this study. We judge the potential risk of attrition bias as high, because not all participants randomised were analysed accordingly.
We judge the quality of the evidence body as moderate for the outcome OS, because of the small number of events, leading to wide CIs and, hence, imprecision in the results. We judge the quality of evidence for the outcome PFS to be low, due to imprecision and performance or detection bias (participants, physicians and outcome assessors not blinded). For the outcome AEs, we judge the quality of the evidence body as very low, due to serious imprecision and non‐blinding.
Potential biases in the review process
We are not aware of any obvious flaws in our review process. Moreover, we carried out all the relevant processes in duplicate to avoid bias.
We performed a sensitive search strategy and searched the relevant data from international lymphoma congresses manually, in order to minimise retrieval bias. We therefore assume to have identified all relevant RCTs concerning the review question. The number of included trials, one, was too low to generate a funnel plot to explore potential publication bias. Moreover, as this type of intervention, radiotherapy, is usually evaluated in investigator‐initiated trials, there is no manufacturer or company available from which to request missing data. Additionally, for an intervention such as radiotherapy there is no need for a trial to be registered in advance in a clinical trial registry, as such registries apply mainly to RCTs of drugs. All these points could have introduced publication bias.
Agreements and disagreements with other studies or reviews
The lack of additional trials meeting the inclusion criteria demonstrates the limited number of prospective, randomised studies focusing on PCNSL. Regarding the role of radiotherapy, G‐PCNSL‐SG‐1 is the only one identified that evaluates WBR in addition to chemotherapy. Conclusions from this trial must be drawn tentatively due to several cases of protocol violation and potential sources of bias. The current consensus is that methotrexate is an effective, possibly the most effective, chemotherapeutic agent (Batchelor 2003; DeAngelis 2002; Gerstner 2010; Illerhaus 2008; Laack 2010) for the treatment of PCNSL. WBR leads to delayed neurotoxicity, especially in individuals aged over 60 years (Batchelor 2003; Laack 2010). It may also play a very important role in PFS, especially in those individuals not achieving a CR after chemotherapy alone. The role of WBR in OS remains to be clarified, but may be of less significance than its role in PFS (Batchelor 2003; Bessell 2002; DeAngelis 2002; Ekenel 2007; Gavrilovic 2006; Gerstner 2010; Illerhaus 2008; Laack 2010; Montemurro 2007). G‐PCNSL‐SG‐1 is the first prospective, randomly controlled, multi‐centre trial to verify (to a certain degree) these insights into the treatment of PCNSL.
A further relevant trial would have been BMPD, the latest preliminary report of which dates from 2004. The study design was aimed to define the role of WBR in the treatment of PCNSL. It recruited 56 participants who were randomised to treatment with and without WBR after achieving a CR following first‐line high‐dose methotrexate chemotherapy. The small number of participants combined with a high dropout rate led to the premature abandonment of the trial (BMPD).
While this review and its only included study, G‐PCNSL‐SG‐1, focus on the role of radiotherapy in the treatment of PCNSL, an alternative approach was chosen by Ferreri 2009 in a trial comparing methotrexate with methotrexate plus cytarabine. Due to different endpoint definitions, the outcomes of these two trials cannot easily be compared. The results of Ferreri 2009 demonstrate that a multi‐chemotherapy regimen may offer an alternative for the treatment of PCNSL to single‐agent chemotherapy or chemotherapy plus radiotherapy. In this prospective, randomised phase II trial, 3.5 g/m² methotrexate were administered to all included participants with PCNSL (N = 79) on day one of a three‐week cycle that was repeated four times. The intervention group received an additional 2 g/m² cytarabine on days two and three of all cycles. WBR followed four weeks after the final cycle of chemotherapy in all participants below the age of 60 years, whereas the administration of WBR was left to the discretion of the treating institution for participants aged over 60 years. The primary endpoint, complete remission following chemotherapy, was seen in significantly more participants receiving methotrexate plus cytarabine than in those receiving methotrexate only (18 of 39 participants (46%) versus 7 of 40 participants (18%), respectively; P = 0.006).
The methotrexate regimen used in G‐PCNSL‐SG‐1 (4 g/m² every two weeks, repeated six times) differed slightly from that used by Ferreri 2009 and achieved a higher percentage of CRs (35% versus 18%, respectively), but this percentage was still inferior to that seen following the combined therapy of methotrexate and cytarabine (46%).
Following the administration of WBR in Ferreri 2009, which was received by 54 of the 79 participants, CR rates reached 64% in the intervention arm and 30% in the control arm, suggesting ‐ in line with the findings from G‐PCNSL‐SG‐1 ‐ that radiotherapy is an effective treatment for achieving a CR. No adequate data regarding the neurotoxic effects of WBR were published. Methotrexate plus cytarabine led to an improvement in three‐year FFS compared with methotrexate alone: 38% versus 21%, respectively (P = 0.01). Ferreri 2009 reported a high rate of haematological toxicity in the intervention group: the dose of chemotherapy had to be reduced in 17 of 40 participants due to thrombocytopenia, severe neutropenia or anaemia, but in only one participant in the control arm. These levels of toxicity, though described as "acceptable" by the authors, make this exact regimen an unlikely choice for standard PCNSL therapy. The trial does, however, serve as a promising foundation for further research and the refinement of agents and dosages.
Abrey 2011 reviewed both G‐PCNSL‐SG‐1 and Ferreri 2009, coming to the conclusion that the omission of WBR is a feasible option, especially in older individuals who are prone to neurotoxicity. The author concludes that a methotrexate‐based multi‐chemotherapeutic regimen approach is currently the best option for individuals with newly diagnosed PCNSL. Regarding WBR, Abrey 2011 concludes: "... WBRT [whole body radiotherapy] may be omitted in patients achieving a CR to methotrexate‐based chemotherapy; however, the patient should be aware that this may result in a higher risk of tumour recurrence" (Abrey 2011).
Reviewing the role of WBR in the treatment of PCNSL, Graber 2011 notes that most trials providing insights into the role of WBR in PCNSL treatment are of poor methodology. The authors conclude that WBR improves PFS, but these benefits must be weighed against the risk of neurotoxicity, especially in older individuals. For these individuals, WBR should be applied only as salvage therapy and after thorough consideration (Graber 2011).
Results of the search
Risk of bias summary: review authors' judgements about each risk of bias item for G‐PCNSL‐SG‐1.
Forest plot of comparison: 1 Chemotherapy plus radiotherapy versus chemotherapy only, outcome: 1.1 Treatment‐related neurotoxicity.
Comparison 1 Chemotherapy plus radiotherapy versus chemotherapy only, Outcome 1 Treatment‐related neurotoxicity.
| Additional radiotherapy for PCNSL | ||||||
| Individuals or population: Participants with PCNSL | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Control (chemotherapy only) | Additional radiotherapy | |||||
| Mortality (instead of OS) | Moderate risk | HR 1.01 | 411 | ⊕⊕⊕⊝ | Instead of overall survival, mortality is reported in this table, for methodological reasons | |
| 750 per 1000 | 753 per 1000 | |||||
| Relapses/death (instead of PFS) Follow up: median 60 months | Moderate risk | HR 0.79 | 411 | ⊕⊕⊝⊝ | Instead of PFS, relapses and deaths are reported in this table, for methodological reasons | |
| 820 per 1000 | 742 per 1000 | |||||
| Treatment‐related mortality | See comment | See comment | Not evaluated | |||
| Adverse events: treatment‐related neurotoxicity | 265 per 1000 | 490 per 1000 | RR 1.85 | 79 | ⊕⊝⊝⊝ | |
| Adverse events: delayed neurotoxicity | see comment | see comment | see comment | 84 | ⊕⊝⊝⊝ | It is not reported whether the participants who received WBR who were integrated into this analysis were from the additional‐WBR group only, or whether any participants from the chemotherapy‐only group who received rescue WBR were also included |
| Quality of life | See comment | See comment | Not evaluated | |||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| GRADE OS = overall survival PCNSL = primary central nervous system lymphoma PFS = progression‐free survival 1 Only one trial included with a wide confidence interval, leading to imprecision 4 Only 84 of 411 participants evaluated | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Treatment‐related neurotoxicity Show forest plot | 1 | 79 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.85 [0.98, 3.48] |
