PORT indicates postoperative radiotherapy.
eTable 1. Demographic, Clinical, and Pathologic Characteristics for Patients Receiving “Low-dose” (18/23.4 Gy) and “High-dose” (36 Gy) Craniospinal Irradiation
eTable 2. Multivariable Logistic Regression Analysis of Predictors of High-dose (36 Gy) Craniospinal Irradiation Among Patients Receiving PORT-Upfront
eTable 3. Multivariable Predictors of Mortality Among Patients Receiving PORT-Upfront
eFigure 1. Consolidated Standards of Reporting Trials Diagram
eFigure 2. Predicted Probability of PORT-Upfront Receipt by Age
Customize your JAMA Network experience by selecting one or more topics from the list below.
Kann BH, Park HS, Lester-Coll NH, et al. Postoperative Radiotherapy Patterns of Care and Survival Implications for Medulloblastoma in Young Children. JAMA Oncol. 2016;2(12):1574–1581. doi:10.1001/jamaoncol.2016.2547
How are postoperative radiotherapy care patterns changing in young children with medulloblastoma, and what are the survival implications?
In this national database analysis of 816 children with medulloblastoma, ages 3 to 8 years, who received postoperative chemotherapy, there was a 15.1% rate of postoperative radiotherapy deferral overall, and deferral rate increased from 2004 to 2012. Postoperative radiotherapy deferral was associated with decreased overall survival in this population.
The practice of postoperative radiotherapy deferral in young children with medulloblastoma seems to be increasing nationally despite postoperative radiotherapy’s association with improved overall survival.
Postoperative radiotherapy to the craniospinal axis is standard-of-care for pediatric medulloblastoma but is associated with long-term morbidity, particularly in young children. With the advent of modern adjuvant chemotherapy strategies, postoperative radiotherapy deferral has gained acceptance in children younger than 3 years, although it remains controversial in older children.
To analyze recent postoperative radiotherapy national treatment patterns and implications for overall survival in patients with medulloblastoma ages 3 to 8 years.
Design, Setting, and Exposures
Using the National Cancer Data Base, patients ages 3 to 8 years diagnosed as having histologically confirmed medulloblastoma in 2004 to 2012, without distant metastases, who underwent surgery and adjuvant chemotherapy with or without postoperative radiotherapy at facilities nationwide accredited by the Commission on Cancer were identified. Patients were designated as having “postoperative radiotherapy upfront” if they received radiotherapy within 90 days of surgery or “postoperative radiotherapy deferred” otherwise. Factors associated with postoperative radiotherapy deferral were identified using multivariable logistic regression. Overall survival (OS) was compared using Kaplan-Meier analysis with log-rank tests and multivariable Cox regression. Statistical tests were 2-sided.
Main Outcomes and Measures
Postoperative radiotherapy utilization and overall survival.
Among 816 patients, 123 (15.1%) had postoperative radiotherapy deferred, and 693 (84.9%) had postoperative radiotherapy upfront; 36.8% of 3-year-olds and 4.1% of 8-year-olds had postoperative radiotherapy deferred (P < .001). On multivariable logistic regression, variables associated with postoperative radiotherapy deferral were age (odds ratio [OR], 0.57 per year; 95% CI, 0.49-0.67 per year) and year of diagnosis (OR, 1.18 per year; 95% CI, 1.08-1.29 per year). On survival analysis, with median follow-up of 4.8 years, OS was improved for those receiving postoperative radiotherapy upfront vs postoperative radiotherapy deferred (5-year OS: 82.0% vs 63.4%; P < .001). On multivariable analysis, variables associated with poorer OS were postoperative radiotherapy deferral (hazards ratio [HR], 1.95; 95% CI, 1.15-3.31); stage M1-3 disease (HR, 1.86; 95% CI, 1.10-3.16), and low facility volume (HR, 1.75; 95% CI, 1.04-2.94).
Conclusions and Relevance
Our national database analysis reveals a higher-than-expected and increasing rate of postoperative radiotherapy deferral in children with medulloblastoma ages 3 to 8 years. The analysis suggests that postoperative radiotherapy deferral is associated with worse survival in this age group, even in the modern era of chemotherapy.
Medulloblastoma (MB) is the most common pediatric brain malignant neoplasm, with an incidence of approximately 6 per million in children ages 1 to 9 years.1,2 Traditionally, postoperative radiotherapy (PORT) to the entire craniospinal axis (CSI), posterior fossa, and postoperative tumor bed has been standard of care.3,4 PORT is associated with numerous deleterious effects in long-term survivors of MB, including neurocognitive decline, endocrinopathies, growth defects, and secondary malignant neoplasm.5-9 Long-term sequelae seem to be dose- and age-dependent, leading to efforts to dose deescalate and defer PORT in young children.5,6,9,10
Historically, CSI dose deescalation was found to be associated with poorer disease control and survival compared with standard-dose CSI to 36.0 Gy.11 However, when combined with the use of cisplatin-based multiagent chemotherapy, CSI dose deescalation has yielded outcomes comparable with those of historical controls treated with standard-dose CSI.12-15 A CSI dose deescalated to 23.4 Gy has since been adopted as a standard approach for average-risk patients with MB, defined as those ages 3 years or older with localized disease, absence of anaplasia, and at least near-total tumor resection.12,14-17 Currently, deescalation to 18.0 Gy is under investigation, along with treatment stratification by molecular subtyping (NCT01878617 and NCT00085735).
Deferral of PORT altogether in favor of adjuvant chemotherapy alone has been studied in multiple prospective trials over the past 2 decades, mostly in children younger than 3 years. While results of these studies have been conflicting,18-24 they have led to the acceptance of PORT deferral for children younger than 3 years to avoid treatment-related morbidity. This strategy, however, is less supported in children 3 years or older.25
Determining the appropriate postoperative treatment for pediatric MB involves complex, multidisciplinary decision-making, balancing treatment efficacy with morbidity, the affected family’s priorities, and institutional preferences.7 In addition, many pediatric patients with MB are treated outside of large multi-institutional trials.26,27 Owing to this ongoing controversy, there may be substantial variation in national PORT treatment patterns, particularly in children 3 years or older. There is also a lack of randomized data comparing survival outcomes with and without PORT in this age group in the era of modern postoperative chemotherapy strategies. In this study, we analyzed recent national practice patterns of PORT for pediatric MB and their implications on patient survival.
We conducted an analysis using the National Cancer Data Base (NCDB) registry data.28 The NCDB captures approximately 70% of all newly diagnosed malignant neoplasms in the United States annually from facilities accredited by the Commission on Cancer. The data used in this study are derived from a deidentified NCDB file. The American College of Surgeons and the Commission on Cancer have not verified and are neither responsible for the analytical or statistical methodology used, nor the conclusions drawn from these data by the investigators. This study was granted an institutional review board exemption by the Yale Human Investigations Committee.
Inclusion criteria were patients ages 3 to 8 years diagnosed from 2004 to 2012 as having histologically confirmed MB without prior cancer diagnosis. We further limited our sample to patients who underwent surgical resection, received adjuvant chemotherapy, and did not have extraneuraxial metastases (stage M4 disease) to avoid ascertainment bias from noncoded adjuvant treatment data and to isolate patients treated with curative intent. We divided the remaining sample into PORT-upfront and PORT-deferred cohorts. We designated patients as “PORT-upfront” if they initiated PORT within 90 days of surgery. Patients who received no PORT or initiated PORT more than 90 days after surgery were designated “PORT-deferred.” A cutoff of 90 days was chosen to reflect the clinical strategy to defer PORT in favor of upfront chemotherapy for at least several cycles, as has been previously investigated.19,21,29 For patients receiving PORT, chemotherapy was designated as concurrent (CCRT) if start time occurred within 14 days of RT start, or sequential (SCRT) otherwise. For the survival analysis, we excluded patients without follow-up or those who died within 2 months of diagnosis to account for immortal time bias.30 We then excluded patients from the PORT-upfront cohort who did not complete PORT with traditional CSI doses of 18.0, 23.4, or 36.0 Gy. The flowchart summarizing patient inclusion and exclusion is found in eFigure 1 in the Supplement.
Patient and facility characteristics included age at diagnosis (continuous), year of diagnosis (continuous), sex, histologic subtype (desmoplastic vs large cell vs classic/not otherwise specified), race (white vs nonwhite), facility volume (high-volume vs low-volume), Chang-modified M-stage (M0 vs M1-3), resection extent (no macroscopic residual vs macroscopic residual vs residual unknown), city environment (metropolitan vs nonmetropolitan), distance from facility (<20 miles vs ≥20 miles), and insurance status (private vs nonprivate). High-volume facilities were defined a priori as facilities that treated at least 10 cases of MB during the study time period. Facilities that treated fewer than 10 cases were defined as low-volume. A threshold of 10 cases was chosen to identify facilities that averaged more than 1 case per year. PORT treatment details, including modality, regional CSI dose, and use of concurrent chemotherapy, were examined in the PORT-upfront group.
Demographic, clinicopathologic, temporal, and treatment details were compared between patients in the PORT-upfront and PORT-deferred cohorts using χ2 and Wilcoxon rank-sum tests. Multivariable logistic regression was used to determine independent factors associated with PORT deferral. An all-inclusive logistic regression model was constructed to include all variables captured. Additional logistic regression models were constructed using forward and backward stepwise methods with entry threshold probability P < .05 and removal threshold probability P < .10. Models were compared for goodness-of-fit with Hosmer-Lemeshow tests. Model-predicted probability of PORT receipt by age was plotted. Clinical characteristics were also compared between patients who received high-dose CSI (36.0 Gy) and low-dose CSI (18.0 or 23.4 Gy), by univariable and multivariable analyses.
Overall survival (OS) was evaluated using time from diagnosis until death. The Kaplan-Meier estimator with log-rank test was used to compare OS between PORT-upfront and PORT-deferred cohorts. Multivariable Cox proportional hazards models were used to calculate adjusted hazard ratios (HRs) and their 95% CIs). Three model versions were constructed. The first was all-inclusive to variables that were clinically relevant and/or significantly associated with OS on univariable analysis. The second and third models were constructed using forward and backward stepwise selection, respectively, with entry threshold probability P < .05 and removal threshold probability P < .10. Schoenfeld residuals were calculated for each model to test that the proportional hazards assumption was not violated. We conducted sensitivity analyses for survival using PORT-deferred cutoffs of 60, 120, and 180 days to check for differences in outcome. We also conducted a sensitivity analysis for the PORT-upfront cohort, including patients who received any CSI dose from 18.0 Gy to 39.6 Gy.
All tests were 2-sided, and P <.05 was considered statistically significant. All analyses were performed using Stata SE statistical software (version 13.0; Stata Corp).
We identified 816 patients who met inclusion criteria (eFigure 1 in the Supplement). Patients were treated at 193 unique facilities, among which 27 (14.0%) were high volume. The maximum number of cases treated at 1 facility was 27, with high-volume facilities treating 271 patients (33.2%).
Overall, 693 patients (84.9%) had PORT upfront, and 123 patients (15.1%) had PORT deferred (Table 1). Of patients with PORT deferred, 74 patients (60.2%) had no PORT recorded as part of the initial course of management, and 49 patients (39.8%) received PORT more than 90 days after surgery. PORT deferral increased from 2004 to 2012 (odds ratio [OR], 1.15 per year; 95% CI 1.07-1.24 per year), from a low of 7.3%% in 2005 to a high of 27.1% in 2012 (Figure 1). PORT deferral decreased with patient age (OR, 0.63 per year; 95% CI, 0.55-0.71 per year), ranging from 36.8% of 3-year-old patients to 4.1% of 8-year-old patients having PORT deferred. On univariable analysis, a higher proportion of patients with stage M1-3 disease had PORT deferred compared with those with M0 disease (21.2% vs 12.2%; P < .001). There was no difference in the distribution of patients with macroscopic residual, no macroscopic residual, or unknown residual tumor between PORT cohorts (P = .72).
On multivariable analysis including all variables captured, independent predictors of PORT deferral were age (OR, 0.57 per year; 95% CI, 0.49-0.67 per year) and year of diagnosis (OR, 1.18 per year; 95% CI, 1.08-1.29 per year) (Table 2). Stepwise and all-inclusive models were in agreement (Table 2). After adjusting for all variables captured, estimated predicted probability of receipt of PORT upfront increased with age, from 69% for 3-year-olds to 97% for 8-year-olds (eFigure 2 in the Supplement).
Of the 693 patients with PORT upfront, 443 patients received documented CSI regional doses of 18.0 Gy (n = 34), 23.4 Gy (n = 261), or 36.0 Gy (n = 148) (eTable 1 in the Supplement). Of the excluded 250 patients, 153 patients were coded either erroneously or as unknown, 67 patients had doses that fell outside the 18.0 to 36.0 Gy range, and 30 patients had doses that fell between 18.0 and 36.0 Gy, but were not exactly 18.0, 23.4, or 36.0 Gy. A total of 382 patients (86.2%) received CCRT, and 61 patients (13.8%) received SCRT. Patients more likely to receive CSI to 36.0 Gy vs 23.4 or 18.0 Gy were male (P = .01), had large cell histologic findings (P < .001), and had stage M1-3 disease (P < .001). On multivariable analysis, variables independently associated with use of 36 Gy CSI were year of diagnosis (OR, 1.15 per year; 95% CI, 1.03-1.28 per year), stage M1-3 disease (OR, 70.29; 95% CI, 20.60-239.84 per year), and large cell histologic findings (OR, 6.43; 95% CI, 2.28-18.10) (eTable 2 in the Supplement). Among all PORT-upfront patients, proton use increased during the time period with 2 patients (2.2%) treated in 2004 and 14 patients (18.0%) treated in 2012 with protons (data not shown).
For the survival analysis, 474 patients were included (eFigure 1 in the Supplement). Eighty-three patients (17.5%) had PORT deferred, and 391 patients (82.5%) had PORT upfront. The median follow-up was 4.8 years. At the time of data extraction, 89 deaths (18.8% of patients) had occurred. Of the patients with PORT upfront, 328 (85.2%) were alive. Of the patients with PORT deferred, 57 (68.7%) were alive. Median survival times were not met in either cohort. Five-year estimated OS was improved with PORT upfront vs PORT deferred (82.0% vs 63.4%; P < .001) (Figure 2). On univariable analysis, other variables associated with OS were stage M1-3 disease (HR, 1.97; 95% CI, 1.17-3.31) and older age (HR, 0.87 per year; 95% CI, 0.77-0.99 per year). On multivariable survival analysis, PORT deferral was independently associated with poorer OS in all models. For the Cox regression model that included all analyzed variables, variables independently associated with poorer OS were PORT deferral (HR, 1.79; 95% CI, 1.03-3.11) and treatment at a low-volume facility (HR, 1.85; 95% CI, 1.10-3.13) (Table 3). Both forward and backward stepwise selection yielded identical models that demonstrated independent associations for poorer OS with PORT deferral (HR, 1.95; 95% CI, 1.15-3.31), stage M1-3 disease (HR, 1.86; 95% CI, 1.10-3.16), and low-volume facility (HR, 1.75; 95% CI, 1.04-2.94) (Table 3). Among patients with PORT deferred, there was no statistically significant difference in estimated 5-year survival for those who received PORT after 90 days compared with those who received no PORT (68.9% vs 53.9%; P = .28). Additional sensitivity analyses analyzing PORT-deferral cutoffs of 60, 120, and 270 days and analysis of patients including CSI doses from 18 to 39.6 Gy in the PORT-upfront cohort demonstrated similar survival outcomes on univariable and multivariable analysis (data not shown).
Of the 391 patients in the survival analysis with PORT upfront, 287 (73.4%) received CCRT, and 104 (26.6%) received SCRT. On univariable analysis, CCRT was associated with improved OS vs SCRT (HR, 0.59; 95% CI, 0.36-0.99), and CSI dose to 23.4 or 18 Gy was associated with improved OS vs 36 Gy (HR, 0.53; 95% CI, 0.31-0.85), though these associations did not persist on multivariable analysis (eTable 3 in the Supplement).
Our national database analysis reveals a surprisingly high and increasing rate of PORT deferral in children with MB age 3 to 8 years, despite the OS benefit associated with upfront PORT for this population in the era of modern chemotherapy.
Our analysis suggests that, in young children with MB, it is predominantly patient age, not clinicopathologic risk, that drives PORT deferral. There were substantial increases in PORT use with each increasing year of age from 3 to 8 years, especially from ages 3 to 5 years. This trend could be explained by the concern about age-dependent, long-term toxic effects from CSI and cerebellar RT, particularly in very young children.5,6,9 M-stage, resection status, and histologic findings did not seem to predict for PORT deferral after adjusting for age.
National trends favoring PORT deferral over time, particularly in young children, may be attributable to several factors. As more literature and knowledge accumulates regarding the long-term risks of CSI radiation, there may be more inclination to defer PORT. In addition, there were several trials conducted from 2004 to 2005 that studied multiagent chemotherapy in place of PORT as a valid treatment option for very young children.20-23 Although these studies were primarily focused on children younger than 3 years, it is possible these treatment strategies are being extrapolated to children several years older. Of note, there are several active protocols studying adjuvant chemotherapy without PORT in children up to 5 years old with average-risk and high-risk MB (NCT02017964 and NCT02025881).
Of note, treatment of MB in the study population was spread out among almost 200 unique facilities, with most patients being treated at institutions that treat a low volume of MB annually. It is possible that institutional bias and experience level may be affecting postoperative MB decision-making. While there was no association found between facility volume and use of PORT-upfront, we did observe an association with higher facility volume and improved survival. The correlation between improved survival and facility volume has been demonstrated for other malignant neoplasms and may be related to the complex nature of MB treatment, both in terms of multidisciplinary coordination and treatment delivery.31
The estimated 5-year survival rate in patients treated with PORT in our analysis (82.0%) is comparable with those in multiple other prospective studies analyzing adjuvant MB regimens.12,14,17,32 Meanwhile, the observed 18.6% absolute reduction in 5-year OS associated with PORT deferral can serve as benchmark for adjuvant treatment decision-making. Importantly, the survival benefit of PORT was maintained after adjusting for multiple potential confounding factors.
There are no prospective studies, to our knowledge, comparing survival with or without PORT over the past decade in children with MB age 3 years or older. The aforementioned single-arm studies20-23 from 2004-2005 evaluating PORT deferral with multiagent adjuvant chemotherapy in children predominantly younger than 3 years produced varying results. Rutkowski et al23 reported that 5-year OS was 66% among all patients, but was 93% among those with stage M0/M1 disease and no residual tumor, leading to the conclusion that a chemotherapy-only approach is promising in patients without initial metastases. However, Geyer et al21 found a more modest 5-year OS of 43% for patients with PORT deferred. One prospective trial, Grill et al22 studied PORT deferral with adjuvant chemotherapy for pediatric MB in a patient population that included patients age 3 and older. This study demonstrated a 5-year OS of 73% in patients with complete resection and M0 disease, but only 13% in patients with M+ disease.22 Chi et al20 included children up to 10 years old and reported a 3-year OS of 60% in a cohort of patients treated with intensive adjuvant chemotherapy alone. In addition, a recent retrospective series including adults and children (n = 66) observed that PORT deferral was associated with poorer OS.33 Although it is challenging to compare results of different studies owing to the heterogeneity of the study populations, these results approximate our analysis’ 5-year estimated OS of 63.4% in PORT-deferred patients.16
For patients who did receive PORT upfront, our analysis suggests that CSI dose was driven by known clinicopathologic risk factors, M-stage, and large-cell histologic findings, as suggested by current Children’s Oncology Group (COG) guidelines and protocols (NCT00085735). Another validated risk factor, macroscopic residual tumor, did not seem to affect choice of CSI dose, although we were unable to completely evaluate the effect of this variable given a large number of uncoded entries for this variable in the NCDB.
While there was a clear survival detriment for PORT deferral, PORT dose deescalation from 36.0 Gy CSI to 23.4 Gy or 18.0 Gy was not an independent predictor of inferior OS. This finding is consistent with results of prior single-arm studies.14,15 The safety of 18.0 Gy CSI is currently being studied in the current standard-risk COG protocol (ACNS0331 [NCT00085735]).
There are several limitations to this study, including the possibility of miscoding and the inherent selection bias characteristic of retrospective cohort studies. Although we used multivariable regression to adjust for confounding factors, such as age, M-stage, histologic findings, and facility volume, there are limits to the breadth and granularity of variables captured in the NCDB. The NCDB does not capture performance status of children, which may affect both the decision to defer PORT and OS. To help control for this, we analyzed only patients who received postoperative chemotherapy. In addition, we were unable to evaluate the survival impact of residual tumor cross-sectional area greater than 1.5 cm2, a validated risk factor, owing to the absence of this information in the NCDB. However, in using a surrogate marker, “macroscopic residual tumor,” we found a balanced distribution across our treatment cohorts and no impact of this variable on OS. It is unlikely that these data would have affected the independent association of PORT with OS, although its absence limited our ability to analyze the impact of PORT by traditional average-risk and high-risk groupings. Furthermore, a recent, large multi-institutional analysis34 challenged the prognostic value of residual tumor, finding no independent association of resection extent with survival after adjusting for molecular subtype. Molecular subtypes were unavailable in the NCDB, although their use may soon help identify patient subgroups who could safely be managed with PORT dose deescalation or deferral. Another limitation of the study was the inability to identify exact chemotherapeutic agents. However, since 2004 chemotherapeutic regimens would likely have remained relatively similar across the study population, and differences in regimen would be unlikely to confound the observed survival benefit of PORT.
Using a large national database, we found an increasing rate of PORT deferral for MB in children age 3 to 8 years over time, especially among younger patients. The analysis suggests that PORT deferral is associated with worse survival in this age group, even in the modern era of chemotherapy. Further study is needed to determine reasons for PORT deferral and to identify subgroups of patients who may be safely managed with PORT deferral, perhaps with the aid of molecular subtyping.
Corresponding Author: Benjamin H. Kann, MD, Yale University School of Medicine, 35 Park St, LL509, New Haven, CT 06519 (firstname.lastname@example.org).
Accepted for Publication: May 25, 2016.
Correction: This article was corrected online September 15, 2016, to fix an error in the Results section. It was corrected online October 27, 2016, to fix a typo in the Methods section.
Published Online: August 4, 2016. doi:10.1001/jamaoncol.2016.2547
Author Contributions: Dr Kann had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All information and materials in the manuscript are original.
Study concept and design: Kann, Park, Yeboa, Benitez, Bindra, Roberts.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Kann, Bindra.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Kann, Park, Lester-Coll.
Administrative, technical, or material support: Yeboa.
Study supervision: Yeboa, Khan, Bindra, Marks, Roberts.
Conflict of Interest Disclosures: Dr Park received honoraria and travel reimbursement from Varian Medical Systems, unrelated to the current work. Dr Lester-Coll has received funding from Elekta, unrelated to the current work. Dr Yeboa has received travel reimbursement from Eli Lilly, unrelated to the current work. Dr Khan has received consulting fees from Elekta and Vertex Pharmaceuticals, unrelated to the current work, and research funding from Elekta and Cianna Medical, unrelated to the current work. Dr Roberts has received travel funding for a IBA-sponsored workshop on proton radiotherapy for pediatric cancers, unrelated to the current work. No other disclosures are reported.
Create a personal account or sign in to: