Metastasis Stage, Adjuvant Treatment, and Residual Tumor Are Prognostic Factors for Medulloblastoma in Children: Conclusions From the Children's Cancer Group 921 Randomized Phase III Study

Zeltzer PM, ... Packer RJ.
Abramson Cancer Center of the University of Pennsylvania
Last Modified: November 1, 2001

Share article


Reviewers: John Han-Chih Chang, MD
Source: Journal of Clinical Oncology 1999; Volume 17: pages 832 - 845

Background

In childhood, the most common tumor of the posterior fossa is a medulloblastoma (MB). Most arise in the cerebellar vermis andgrow into the fourth ventricle causing hydrocephalus. In addition to local invasion, there can be spread through the cerebrospinalfluid (CSF). It is classified as a primitive neuroectodermal tumor (PNET). The median age of diagnosis is 5 - 6 years of age, but 20%can occur in infants less than 2 years.

The initial mainstay of treatment is surgical resection. The extent of resection (biopsy only versus subtotal versus total) has beenfound to be a significant factor for outcome favoring those patients in whom a total resection can be performed. Adjuvantpost-operative radiation therapy (RT) and/or chemotherapy (ChT) is the next step since few are cured by surgery alone. In manystudies post-operative RT has been proven to improve survival over surgery alone. In recent years, ChT has come to the forefrontas adjuvant treatment post-operatively. It was suggested that ChT could be used in concert or instead of RT. The InternationalSociety of Pediatric Oncology (SIOP) performed a study of surgery and post-operative RT versus the same treatment with theaddition of vincristine, CCNU and prednisone (VCP regimen). In high-risk patients, there was a significant benefit in survival for theVCP arm of the trial. Similar favorable results for the addition of ChT for high-risk patients were obtained from a study conducted bythe Children's Cancer Study Group (CCSG) in association with the Radiation Therapy Oncology Group (RTOG). The PediatricOncology Group (POG) also ran a randomized trial of postoperative RT versus RT plus MOPP ChT. Again survival appeared to beimproved with the addition of ChT.

Thus, the goal is the optimization of ChT. "What kind" and "how much" are the questions that needed to be addressed. Recentphase II trials have touted the success of the 8-in-1 ChT regimen in response and survival data. The VCP regimen from the SIOPstudy has been a benchmark. This study from the Children's Cancer Group (CCG) attempted to answer the nagging questions ofwhat ChT agents would be the most optimal.

Patients and Methods

Since it was considered a standard ChT regimen, VCP was added to RT in the control arm, while the 8-in-1 regimen with delayed RTwas given in the experimental arm. Two hundred and three patients with the diagnosis of MB or posterior fossa PNET wererandomized. Eligible patients could have Chang Classification M1 - 4 or T3b 4 disease. Age cutoffs were older than 1.5 and lessthan 21 years. The CCG also enrolled patients that had at least 1.5 cm2 of residual disease present on CT or MRI.

Surgical resection or biopsy was performed on all patients. Staging was based on neurosurgeon's report, postoperative CT, MRI ormyelography, spinal fluid for cytology and bone marrow.

Arm 1 ChT consisted of 8 vincristine (1.5mg/m2) weekly injections during RT, followed by 8 6-week cycles of weekly vincristine(1.5mg/m2) x 3 weeks, CCNU (lomustine 100 mg/m2) on day 1 and prednisone on days 1 - 14. Arm 2 ChT consisted of 2 courses of8-in-1 treatment, which is composed of vincristine (1.5 mg/m2), methylprednisolone (300 mg/m2 x 3 per day), CCNU (75mg/m2),hydroxyurea (1.5 g/m2), procarbazine (75 mg/m2), cisplatin (60 mg/m2), cyclophosphamide (300 mg/m2) and cytarabine (300 mg/m2) allgiven over the course of 24 hours. The 2 courses were separated by 14 days and subsequent RT began 14 days after the secondcycle. After the RT, 8 more cycles of 8-in-1 was delivered every 6 weeks.

In both arms, the RT to the primary site was delivered prior to full craniospinal irradiation (CSI) to allow for some bone marrowrecovery in arm 2. Those patients 3 years of age or older received a total of 54 Gy to the primary site and 36 Gy to the CSI axis. Thosethat were younger received only a total of 45 Gy to the primary site and 23.4 Gy to the CSI axis. Those that had spinal metastaseswere also given 14.4 - 18 Gy total boost to areas of bulk disease. Based on thedesign of the trial, patients in arm 2 were scheduled tohave a delay of approximately 28 days in their RT relative to arm 1.

Results

As mentioned above, 203 patients of the 212 entered met eligibility criteria. Based on the eligibility parameters set above, this is apopulation of high-risk patients. The 5-year progression-free survival (PFS) and overall survival (OS) were 54% and 55%,respectively for the entire group. Figure 3 depicts graphically that the VCP regimen was superior to the 8-in-1 regimen in PFS at5-years (63% versus 45%).

Further analysis was then performed on 188 patients in whom pathology was confirmed with a central review. The others had eitherdiscordant pathology or were not available for review. Seven-year estimates of OS and PFS were 55% and 56%, respectively. Again,arm 1 with VCP was superior in 5-year PFS to arm 2 (64% versus 47%). Of those that relapsed, 50% had died within 6 months of theirfirst relapse. Two-year OS for those that relapsed was 9%.

The analysis of prognostic factors begins with patient age. Those 1.5 to less than 3 years of age per protocol had received less dose(9 Gy less) than their older counterparts. The PFS is also much worse if younger than 3 years of age. No difference was seenbetween ages 1.5 - 1.9 years versus 2.0 - 2.9 years. The 5-year PFS in ages 1.5 - 2.9 years was 32% versus 56 - 61% for those 3 - 21years (figure 5). Subsequent analyses were then performed with patients 3 years old and older.

Chang M stage was determined to be a statistically significant prognostic factor in those children older than 3 years. The 5-year PFS's for M0, M1 (microscopic cells in the CSF) and M2 - 4 (gross nodular seeding in the cerebellum, cerebral or spinalsubarachnoid space, 3rd or 4th ventricles, or extraneural axis) were 70%, 57% and 40%, respectively. Chang T stage was not astatistically significant prognostic factor when corrected for M stage.

Extent of surgical resection did seem to be a prognostic factor for the M0 subset of patients (n = 86) older than 3 years. The PFS at5-years of those who had less than 1.5 cm2 residual disease on CT or MRI was 78%. This greatly exceeds the 54% 5-year PFS ofthose that had 1.5 cm2 or more residual.

Toxicity data demonstrated much more increased toxicity (grades 3 and 4) in the 8-in-1 arm over the VCP arm. Hematologic toxicityfor each cycle of the 8-in-1 was two to three times that of the three-drug regimen.

The 8-in-1 regimen did have a built-in delay in RT of 28 days later than the VCP regimen. The actual results revealed a median delayof an additional 4.5 days. In both treatment arms over 96% received the specified RT dose. Data reported by the group stated thatthe delay in the XRT did not change outcome, though no specific numbers were given.

Discussion and Critique

The results mentioned above have demonstrated the superiority of VCP and RT over the 8-in-1 regimen with delayed RT. Despitethe promising phase II trials with the 8-in-1 regimen in patients with recurrent and primary brain malignancies, this randomized trialhas been effective in showing that more may not necessarily be better in this case. It has really highlighted the importance of phaseIII randomized trials to prove efficacy. One of the explanations of the VCP regimen superiority may stem from the delay of the RT inarm 2 causing the worse outcome. The authors also argue that perhaps the vincristine that was given during the RT was better thanadministration before or after the RT. Another argument presented, along the same lines, was that total vincristine dose was reducedin the 8-in-1 regimen as contrasted with the VCP regimen.

Prognostic factors have been a controversial issue in MB. This study attempted to stratify for risk factors prospectively in order toconfirm their impact on outcome. It seems that they have discovered increasing Chang M stage as a strong prognosticator of poorPFS. Greater residual disease on CT or MRI scan appears to also point to a worse PFS. Age also continues to be a prognosticator ofPFS. Those of age 3 or older tend to have a better overall outcome than those younger than 3. The confounding issue remains thatthe younger patients receive less dose of RT. If one postulates that the delay in RT may have caused the increased relapse rate inarm 2, lower dose may also be a factor. Unfortunately, it may be difficult to justify a higher dose based on the neurological sequelaefrom increased dose in the younger children.

Overall, the authors presented a very well designed trial that appears to answer some of the very important questions about ChToptimization and prognosis. However, it has continued to propagate the questions of how to integrate ChT and RT and what is thesafe yet therapeutic dose of RT that can be given to children less than 3 years of age.