Should whole-brain radiotherapy be considered standard of care in newly diagnosed primary central nervous system lymphoma? The G-PCNSL-SG1 randomized phase IV trial

Reporter: Arpi Thukral, MD
The Abramson Cancer Center of the University of Pennsylvania
Last Modified: June 5, 2010

Presenter: E. Thiel
Affiliation: Department of Hematology, Charite Campus Benjamin Franklin, Berlin, Germany


  • Primary central nervous system lymphoma (PCNSL) is a rare form of non-Hodgkin lymphoma (NHL).
    • It is restricted entirely to the brain, leptomeninges, eyes, and rarely the spinal cord.
    • It accounts for approximately 2-4% of all intracranial tumors.
    • It typically presents with focal neurologic symptoms and is characterized by diffuse infiltration of the brain.
  • Previous studies have shown that standard treatment of PCNSL should include high-dose methotrexate (HDMTX).
  • PCNSL is radiosensitive and responds to whole-brain radiotherapy (WBRT); however the role of WBRT in PCNSL remains controversial.
  • Due to the rarity of this disease, most previous trials examining the role of WBRT have been Phase II trials.
    • RTOG 9310 (a phase II single arm trial) showed that chemo followed by RT consolidation provided better outcomes than RT alone compared to historical controls; however, it was also demonstrated that there was high neurotoxicity in patients >60 with combined modality therapy.
  • The role of consolidative WBRT in young patients (<60) is unclear, but appears to improve outcome in previous Phase II trials. Patients who survive and live longer, however, may go on to experience devastating neurotoxicity.
    • This neurotoxicity limits the acceptance of WBRT as standard of care, and many groups already omit WBRT or defer it.
  • The purpose of this study was to conduct a randomized study to define the role of whole brain irradiation (WBI) in patients with PCNSL who have received primary therapy with HDMTX.
    • The question asked by the authors was “Can patients who have undergone primary chemotherapy postpone irradiation with its possible late sequelae until recurrence without incurring losses in progression-free and overall survival?”


  • This study was a Phase IV randomized prospective study in which 551 patients with PCNSCL were enrolled between May 2000 and May 20009.
  • Inclusion criteria:
    • New diagnosis of histologically or cytologically/immunocytologically confirmed primary non-Hodgkin's lymphoma of the CNS
      • Histological diagnosis was performed by preferential stereotactic biopsy of suspicious lesions in the brain or spinal cord.
    • Immunocompetency
    • Life expectancy of at least 2 months
    • Adequate bone marrow reserve with a peripheral granulocyte count of > 1,500/µl and thrombocyte count of > 100,000/µl
    • Creatinine clearance of > 50 ml/min and serum creatinine in the normal range.
    • Written informed consent
  • Exclusion criteria:
    • Manifestation of lymphoma outside of the CNS
    • Active infection
    • HIV positivity
    • Previous treatment of PCNSL other than with corticosteroids, antiepileptics, or diuretics
    • Previous radiotherapy of the brain
    • Concomitant or previous malignant diseases in the last 5 years except for an adequately treated basal cell carcinoma or cervical carcinoma in situ
    • Immunosuppression, concomitant immunosuppressive therapy, or organ transplantation
    • Requirement or use of chemotherapy for another malignant diagnosis.
  • Trial Design:
    • All patients received 6 cycles of HDMTX. They were then randomized to 1 of 4 groups depending on their response to MTX: WBRT (arms A1, B1) or chemotherapy alone (A2, B2).
    • Patients achieving a complete response (CR) received either consolidating WBRT with 45 Gy in 1.5 Gy fractions (arm A1) or no further treatment (A2).
    • Patients without CR received salvage WBRT (B1) or salvage chemotherapy using HD-cytarabine (B2) (4 x 3 g/m2/ 48 h, x 3 weeks).
    • After randomization, all patients were to receive 6 cycles HDMTX (4 g/m2, D1, biweekly) from 1999-2007 and HDMTX plus ifosfamide (1.5 g/m2, D3-D5, biweekly) thereafter.
  • The primary aim of this study was to demonstrate that the omission of WBRT from first-line treatment does not compromise overall survival (OS).
    • The endpoints of the study were overall survival (OS) and progression-free survival (PFS).
  • Statistical considerations:
    • Non-inferiority design
    • 60% power to prove non-inferiority of omission of WBRT in case of HR of 1.2 of WBRT vs. no WBRT


  • 551 patients were initially enrolled on the study, however, only 526 patients received at least one course HDMTX-based chemotherapy. (25 patients were deemed ineligible)
  • Of these 526 patients, 66 died on HDMTX, 60 dropped out, and eventually only 411 entered the post-HDMTX phase. Of these, 318 were treated per protocol (PP population).
  • Patient characteristics:
    • Median age = 61 years
    • Male predominance (183 males vs. 135 females)
    • 8.2% of patients had disease present in CSF
  • PP population:
    • Median OS: 32.4 months in the chemotherapy+WBRT arm (A1+B1, n=154) and 37.1 months in the chemotherapy alone arm (A2+B2, n=164) (p=0.70)
    • Median PFS: 18.3 and 12 months (p=0.13) for WBRT group vs. no WBRT group, respectively.
  • PP population with CR:
    • OS was not statistically different between the 2 groups.
    • Median PFS was 36.3 months in the chemotherapy+WBRT arm (n=56) and 21.3 months in the chemotherapy alone arm (n=96) (p=0.038)
  • PP population without CR:
    • OS was not statistically different between the 2 groups.
    • Median PFS was 5.6 months in the chemotherapy+WBRT arm (n=98) and 3.0 months in the chemotherapy alone arm (n=68) (p=0.003)
  • Intent-to-treat population:
    • OS was not statistically different between the 2 groups.
    • Median PFS was15.5 months in the chemotherapy+WBRT arm and 9.9 months in the chemotherapy alone arm (p=0.041)
  • On subset analysis examining age (< or > 60 years) as a prognostic factor, OS was 28.4 months vs. 14.2 months, respectively (p<0.005).
  • On subset analysis examining KPS (performance status (< or > 70) as a prognostic factor, OS was 31.5 months vs. 9.8 months, respectively (p<0.005).
  • Late neurotoxicity was examined in PP patients with CR by chi2 analysis. On clinical evaluation, neurotoxicity was seen in 48.9% of patients in the WBRT arm vs. 26% in the no WBRT arm. On neuroradiologic evaluation, neurotoxicity was seen in 72.5% of patients in the WBRT arm vs. 41.7% in the non-WBRT arm.

Author's Conclusions

  • This is the first phase IV PCNSL study and largest trial ever performed world-wide.
  • No significant difference was seen for OS when WBRT was omitted from HDMTX-based chemotherapy in patients with newly diagnosed PCNSL.
  • PFS was prolonged on subset analyses confirming the role of WBRT for disease control. However, there is a lack of OS benefit and salvage treatment may be as effective as up-front treatment.
  • Age and KPS were most important prognostic factors seen.
  • Late neurotoxicity was more frequent in patients with WBRT.

Clinical Implications

  • Previously, the standard treatment for PCNSL has included HDMTX followed by WBRT (usually to doses of 45Gy). However, due to the concern of side effects of WBRT (mainly long-term neurotoxicity), its role has been controversial in the existing literature.
  • The authors conclude that WBRT is not needed in patients with PCNSL since an OS benefit was not seen with WBRT, even though a PFS benefit was seen. In addition, they justify omitting WBRT using their conclusions of increased neurotoxicity with WBRT.
    • However, one can argue that from a quality of life standpoint, it may be better for a patient to live without progression of disease, even if OS is not improved.
    • In terms of the clinical implications of these results, the major question becomes how one should balance the PFS benefit with the risk of long-term neurotoxicity. Are patients with PCNSL living long enough to experience neurotoxicity? And, if so, does the risk of neurologic issues from progression of disease outweigh the risk of toxicity from WBRT?
  • This study had many limitations questioning the validity of its results. These include:
    • The study had overall poor results in terms of response to chemotherapy compared to previously published data for PCNSL.
    • Phase IV trials are traditionally defined as post-marketing surveillance trials, and this study does not fit this description.
    • The study population was very small for a non-inferiority study, and results may not be widely applicable due to this limited study population.
    • No confidence intervals were shown for results by authors.
    • Response to chemo should have been taken into consideration for treatment decisions. In this study, patients with CR or non-CR were equally likely to receive radiation.
      • Additionally, the authors did not explicitly state the number of patients who received delayed WBRT when progression occurred, and this could certainly have impacted patient outcomes.
    • 551 pts were enrolled but 140 pts (25%) were removed from analysis including intention to treat (ITT) patients. Additionally, only 58% of pts were treated per protocol. ITT data was only for 411 patients.
    • Of the PP patients, the number of patients in the 2 arms was not balanced.
      • In the CR group, there were 56 patients in the WBRT arm vs. 96 patients in the no WBRT arm. In the non-CR group, there were 98 patients in the WBRT arm vs. 68 patients in the no WBRT arm.
      • Despite upfront randomization, there is a discrepancy in the number of patients in each arm, and this may have introduced bias.
    • It is unclear from the authors how neurotoxicity was measured. They mention the use of clinical evaluation and neuroadiologic evaluation, however objective criteria for measurement were not mentioned. Kaplan-Meier data may have been more suitable for reporting of toxicity data as compared to the chi2 test used by the authors. Furthermore, MTX can also cause neurotoxicity and this was not examined as a confounding variable.
  • Although the authors have made a valiant attempt to test the omission of WBRT in the treatment of PCNSL, this study has many flaws that limit its applicability to practice. The results presented here should thus be interpreted with caution.
  • For these reasons, these data can only be hypothesis generating at this time and need to be confirmed in future prospective trials.
  • A recent trial by Shah et. al. from MSKCC published in JCO examined a reduced WBRT dose of 24 Gy vs. 45 Gy for patients with PCNSL. They demonstrated that reduced-dose WBRT was not associated with neurocognitive decline, and disease control to date is excellent.
    • It may be beneficial to aim future trials at dose reduction of WBRT, as opposed to omission, given that PFS is improved with WBRT.


We Need to Know When to Stop
by Rodney Warner, JD
September 26, 2013

Stay informed with the latest information from OncoLink!   Subscribe to OncoLink eNews
View our newsletter archives