Males have a higher incidence of medulloblastoma (MB) than females, but the effect of gender on survival is unclear. Studies have yielded conflicting results, possibly due to small sample sizes or differences in how researchers defined MB. We aimed to determine the effect of gender on survival in MB using a large data set and strict criteria for defining MB.
A sample of 1,226 subjects (763 males and 463 females) was identified from 1973 to 2002, using the Surveillance Epidemiology and End Results (SEER-9) registry. MB was strictly defined to exclude non-cerebellar embryonal tumors (primitive neuro-ectodermal tumors). Because children <3 years of age are known to have worse survival, patients were stratified by age <3 years at diagnosis (95 males, 82 females) and >3 years (668 males, 381 females).
Overall, there was no significant difference in survival between males and females (log rank P = 0.22). However, among subjects >3 years, females had significantly greater survival than males (log rank P = 0.02). In children <3 years, there was a non-significant trend toward poorer survival in females (median survival: males 27 months, females 13 months; log rank P = 0.24). This interaction between age group and gender was statistically significant (P = 0.03).
Medulloblastoma (MB) is the most common malignant brain tumor in children. While there are emerging biologic data that help predict prognosis, there are still conflicting conclusions about the effect of many basic clinical factors, such as gender, on outcome 1. It is well established that MB incidence is higher in males than females 2, but it is unclear if gender affects prognosis.
While some studies found that males have poorer survival than females 3, 4, others found no gender difference in survival 5–7. Previous studies have been limited by small sample size, with some series examining patients only from a single institution 3, likely contributing to these conflicting conclusions. Also, some studies misclassified cerebral primitive neuroectodermal tumor (PNET) as MB 5, although recent studies have established definitively that these entities are molecularly distinct 8. We aimed to determine the effect of gender on survival in MB, using a large data set and strict criteria for defining MB.
Subjects were identified from the Surveillance Epidemiology and End Results (SEER-9) registry, which encompasses data collected from nine cancer registries across the United States from 1973 to 2002. All patients listed in the registry with a diagnosis of MB were included in the study (Fig. 1). MB was strictly defined to exclude non-cerebellar embryonal tumors (PNETs). This “strict” classification of MB encompassed International Classification of Disease (ICD-O-3) histology codes 9470, 9471, 9472, and 9474 (which includes desmoplastic nodular medulloblastoma, medullomyoblastoma and medulloblastoma not otherwise specified) and was limited to tumor with primary locations listed as the cerebellum, brain stem, overlapping lesions, or those not otherwise specified (site codes C71.6–C71.9). Because infratentorial neoplasms recorded as PNETs are congruous with MB, PNETs (histology code 9473) were also included in the strict classification only if the primary location of the tumor was in the cerebellum or brainstem (site codes C71.6, C71.7).
Survival was defined as time from diagnosis until death due to all causes. Kaplan–Meier estimates of survival were calculated, and the log rank test was used to determine survival differences between males and females. In addition, because children less than 3 years of age often do not receive craniospinal irradiation and have worse survival 9, patients were stratified a priori by age into two categories for additional analysis: infants (age at diagnosis less than 3 years), and older children or adults (age at diagnosis 3 years or older). Chi-square tests were used to test for differences in the proportion of males and females diagnosed with MB. Cox proportional hazards regression was used to calculate mortality ratios and to test for the interaction between gender and age. All statistical analyses were performed using Statistical Package for the Social Sciences (SPSS) version 15.0 (Chicago, IL).
To compare our results with other reports, which often combine MB and PNET, a second analysis with a broad definition of MB was also used. This definition included ICD-O-3 histology codes 9470–9474 and included tumors occurring at any site. Thus this broad definition included all of the subjects in the strict group, with the addition of supratentorial PNETs. Results reported are from the subgroup of patients identified using the strict definition of MB unless otherwise specified.
A sample of 1,226 subjects was identified using the strict definition of MB, including 177 infants <3 years and 1,049 children and adults ≥3 years. The characteristics of these patients are summarized in Table I. The mean age was 13.9 years (SD = 13.4 years), with a range of 0–86 years. More than seventy percent of patients were younger than 18 years of age at diagnosis. Approximately 85% of the tumors were histologically classified as MB and 92% of the tumors were located in the cerebellum, with only 3% indicated as overlapping or not otherwise specified (Table I).
Overall, approximately 62% of the patients identified were male (Table I). In the subgroup of patients older than 3 years of age at diagnosis, the ratio of males to females was nearly 2:1, whereas in the subgroup of patients younger than 3 years of age, the male to female ratio was nearly 1:1 (Table II). This was a statistically significant difference in proportions (P = 0.01). Findings were unchanged (results not shown) when the analysis was restricted to tumors coded only as MB and located in the cerebellum.
Table II. Gender Distribution by Age Subgroup
Age <3 years
Age ≥3 years
The overall median survival time for the patients was 108 months and the 5-year survival rate was 57.3%. Among infants, the median survival was 19 months and, among older children and adults, the median survival time was 121 months. Survival results are summarized in Table III. Overall, there was no significant survival difference between males and females (log rank P = 0.22), although there was a trend toward a female survival advantage (Fig. 2). Median survival time for females was 152 months and median survival time for males was 90 months.
Table III. Comparison of Survival Between Males and Females Using a Strict (MB) and a Broad (MB + PNET) Definition of Medulloblastoma
In the subgroup of patients 3 years of age or older, there was a significant female survival advantage (log rank P = 0.02), with a median survival time for females of 211 months and a median survival time for males of 93 months (Fig. 3). The hazard ratio comparing males to females was 1.26, indicating that, among patients 3 years of age and older, there is a 26% higher mortality rate in males than females.
In the subgroup of patients younger than 3 years of age at diagnosis, the females no longer had a significant survival advantage over the males (P = 0.24, hazard ratio 0.80, favoring males). Median survival time for females was 13 months and median survival time for males was 27 months (Fig. 4).
When testing for an interaction with age as a continuous variable, there was no significant interaction between gender and age (P = 0.889). However, when age was considered as a categorical variable (less than 3 years vs. 3 years or older), as had been defined a priori, there was a statistically significant interaction between age and gender (P = 0.03).
Findings were unchanged (analyses not shown) when subjects with tumors coded as medulloblastoma in the brainstem were excluded (n = 65) or analysis was restricted to only tumors coded as medulloblastoma and located in the cerebellum (n = 1161).
Broad Definition of MB
The analysis was repeated with the broad definition of MB, which included supratentorial PNETs. Similar results and patterns were found (Table III). For MB + PNET analysis, there was also overall a non-significant trend toward female survival advantage (log rank P = 0.35), with a significant female survival advantage in patients 3 years of age and older (log rank P = 0.03), and a non-significant trend toward a male survival advantage in patients younger than 3 years of age (log rank P = 0.13).
Using a large sample from a population-based cancer registry, we did not find a significant overall survival difference by gender in subjects diagnosed with MB. There was, however, a significant interaction with age. Females less than 3 years of age had no survival difference and a trend toward poorer survival, whereas females 3 years of age and over had significantly better survival.
We did not demonstrate a similar interaction between age and gender when age was treated as a continuous variable. Thus, it does not appear that, as age increases, there is an increasing difference in survival between males and females. Instead, our results indicate that there appears to be a distinct difference in the effect of gender in these two age categories, with a survival advantage among females in the older group and a slight, although not significant, survival disadvantage among females younger than 3 years of age.
Our results support previous research by Weil et al. 3, which concluded that females with MB have a better outcome than males. This study, which is the most often cited study on gender and survival in MB, was a single institution consecutive series of 109 patients. Using a larger data set, we found similar results, but only among patients 3 years of age and older.
In contrast, many previous studies reported no overall difference in survival by gender. These include a study of 63 patients by Preston-Martin et al. 6, which did not distinguish MB from PNET, and another report of a sample of several hundred children, which included neuroglial tumors in the analysis 7. In addition to the smaller sample sizes, it is possible that an interaction between age and gender that was not accounted for also contributed to a lack of overall significant survival difference in these studies. Also, while our study demonstrated similar findings between the MB only and MB + PNET groups, varying definitions of MB used in previous studies may account for some of these conflicting results.
Among children younger than 3 years of age, we did not find a significant survival difference. While this lack of a difference might further support the notion of an interaction of age with gender, we also interestingly noted a trend toward improved survival for males less than 3 years of age in our study. This reversal could relate to underlying differences in MB biology and clinical behavior between these two populations 9, 10, but could also be due to chance and needs to be confirmed in further studies.
It is unclear why males older than 3 years of age may have worse survival than females, but is likely due to a combination of factors. Possibilities include an innate effect of gender on MB biology, varying responses to treatment, or hormonal differences. In patients younger than 3 years, a potential reason that females do not have a survival advantage over males could be because of an inherent difference in the tumors of these youngest patients. A recent study used gene expression profiles to identify different subgroups of MB, including a subgroup that showed a significant up-regulation of the Sonic Hedgehog pathway 11. This subgroup was comprised mainly of patients less than 3 years of age. While this study only contained 46 samples, it does suggest the possibility that MB in very young patients is biologically different than in older patients.
While our data are retrospective and could be subject to miscoding or other misclassification, our study utilized a large population-based sample with increased power and minimizes ascertainment bias. It is also possible that some of the tumors coded as medulloblastoma were the heretofore unrecognized entity atypical teratoid rhabdoid tumor, perhaps occurring disproportionately at the brainstem or cerebellum. However, repeat analyses, as above, excluding medulloblastomas located in the brainstem or restricting to medulloblastoma only in the cerebellum yielded no different results, and exclude differential misclassification. We did not have data on therapies provided to these patients, but have no reason to believe that gender influenced treatment.
All cases considered, gender did not appear to play a role in explaining outcome. However, when considering the groups age <3 years and >3 years, there was clearly interaction between gender and age in regards to outcome. Further exploration of the relationship between gender and age in MB outcome might detect differences relevant to emerging pathologic subtypes of MB or other underlying molecular biology.
Ms. Curran was supported by the Stanford University Medical Scholars Program and the Child Neurology Foundation.