Adjuvant therapy for high-risk medulloblastoma: more is better?

See the article by Dufour et al., pp. 1163–1172.

The majority of older children diagnosed with medulloblastoma (MB) are now expected to become long-term survivors with treatment combining maximal safe surgical resection, adjuvant craniospinal irradiation (CSI), and chemotherapy.

Using clinical risk stratification, patients with average-risk MB (defined by greater than near total resection and the absence of metastatic disease) have an expected 5-year overall survival of around 85%.^1,2 Patients with high-risk (HR) MB (metastatic and or with less than near total resection) have a survival around 60%-70% according to most recent clinical trials.^3–6 Some groups have elected to incorporate large cell anaplastic (LCA) histology into the HR clinical risk group, irrespective of the metastatic status, or extent of resection at diagnosis, based on the premise that the poor survival associated with this histology might be improved with more intensive treatment. This variation of risk group definition may limit the direct comparison of outcome between clinical trials for patients with HR MB.

Our growing understanding of the underlying molecular landscape of MB brings further information to refine risk stratification but also adds more complexity to rationalize treatment intensification.

Dufour et al. report an encouraging survival (5-year progression-free survival [PFS] of 76%, CI 63-83) for their HR MB population using a rather intensive regimen combining conventional chemotherapy, 2 cycles of myeloablative chemotherapy, and CSI followed by maintenance chemotherapy with 6 cycles of temozolomide.⁷ Their experience contributes to the prospective data needed to decipher the interplay between treatment intensity and molecular subgrouping in HR MB. However, given the relatively limited size of their cohort, the survival figure associated with this strategy needs to be cautiously interpreted. For a recent illustration, the survivals reported for HR MB in the SJMB03 and SJMB96 trials were 58.7% and 70%, respectively, despite the use of the same treatment regimen and risk stratification, when SJMB03 enrolled twice as many HR patients than SJMB96.^2,3

All the recently published or presented clinical trials with molecularly characterized HR MB confirm the excellent survival of WNT patients, irrespective of the regimens used. These concurring findings are now opening the door to the possibility of a careful de-escalation of therapy for selected HR WNT patients, as in the current investigation of average-risk WNT MB, for whom “less will likely be better” in terms of acute and long-term side effects.

Similarly, in those trials for HR MB using CSI and various intensities of conventional chemotherapy and/or high-dose chemotherapy, the survival for group 4 patients follows that of WNT patients and is superior to that observed in the group 3 and SHH patients as reported by Dufour and collaborators.⁷ In the PNET HR+5, no event was described at 5 years for group 4 patients (all metastatic at diagnosis), leading the authors to suggest a possible treatment effect where the intensity of such strategy may overcome the negative effect of HR features.⁷ Leary and collaborators recently presented the results of the COG ACNS0332 trial for HR MB, also describing a favorable outcome for the patients with group 4 MB (5-year event-free survival [EFS] of 77% ± 5%) using 36 Gy CSI and prolonged conventional chemotherapy.⁸ As in the PNET HR+5 trial, the 103 HR group 4 patients enrolled in SJMB03 all had metastatic disease. Their 5-year PFS following 36 Gy CSI and 4 courses of non-myeloablative high-dose chemotherapy and stem cell rescue was 68.1% (±6.6%).³ A similar outcome was reported with CSI and 4 cycles of maintenance chemotherapy in the HIT2000 protocol.⁵ The PNET5+ regimen should likely be evaluated on a larger number of patients with group 4 MB to confirm the benefit of this intensive regimen observed on their initial 15 patients.

In trying to determine who could benefit from more, the COG ACNS0332 study recently provided some interesting findings, demonstrating, in a randomized manner, that the addition of concomitant carboplatin during CSI was exclusively beneficial to HR group 3 patients, increasing their 5-year PFS to 73% (±8%). This result contrasts considerably with the 40.6% (±8.7%) and 50% (28%-72%) survivals reported for the HR group 3 patients in the SJMB03 and HIT2000 trials, respectively. With this particular protocol, intensification with carboplatin appeared to be better for a subset of molecularly characterized HR patients. The incorporation of carboplatin as a radiosensitizer into different chemotherapy backbones for group 3 HR MB would warrant further investigations.

The PNET HR+5 study also points out some of the limitations associated with intensification strategies for HR patients. In this protocol, the adjuvant CSI planned after high-dose chemotherapy was to start at the latest 150 days after initial surgery, but 34% of their patients had a delay greater than 150 days with some outlying patients delayed up to 210 days. Furthermore, a quarter of the patients eligible for maintenance therapy could not receive temozolomide due to persistent thrombocytopenia. As rightfully discussed by the authors, the absence of relapse in the group of the patients who did not receive maintenance begs the question of the contribution of maintenance in the improved outcome they reported. Of note, the randomization of isotretinoin for HR MB in COG ACNS0332 was suspended after futility analysis showed no difference in EFS between patients receiving isotretinoin maintenance vs none.⁸ While the benefit of maintenance therapy still has to be demonstrated, the choice of agents to introduce has to take into account the intensity of prior adjuvant treatment.

While the field of pediatric MB is now moving toward integrated clinical and molecular risk stratification, the possibly confounding impact of factors such as LCA histology, molecular events like TP53 mutation in SHH subgroup or MYC/MYCN amplifications in non-WNT non-SHH subgroups, and finally, the variations of treatment intensity remain in question. Some of the answers may come from international pooled data analyses of the prospective trials for children with molecularly characterized HR MB, to more accurately predict those patients for whom more therapy will truly be better.