|Home | About | Journals | Submit | Contact Us | Français|
Childhood cancer survivors exposed to CNS irradiation are at increased risk for neurocognitive deficits; however, limited data exist linking outcomes with region-specific exposure to CNS irradiation. We report associations between region-specific radiation dose and self-reported neurocognitive and health-related quality of life (HRQOL) outcomes in 818 adult survivors of childhood central nervous system (CNS) malignancies from the Childhood Cancer Survivor Study. Survivors were compared with a sibling group and national normative samples to calculate standardized scores. Cumulative radiation dose was calculated for 4 specific brain regions. Logistic regression was used to estimate the association between radiation dose to specific brain regions and outcome measures of functional impairment adjusted for clinical and demographic factors, including sex and age at diagnosis. High radiation dose levels to temporal regions were associated with a higher risk for memory impairment (radiation doses ≥30 to <50 Gy: OR, 1.95; 95% CI, 1.01–3.78; dose ≥50 Gy: OR, 2.34; 95% CI, 1.25–4.39) compared with those with no radiation exposure. No such association was seen with radiation exposure to other regions. Exposure to temporal regions was associated with more social and general health problems, whereas exposure to frontal regions was associated with general health problems and physical performance limitations. Adult survivors of childhood CNS malignancies report higher rates of neuropsychological and HRQOL outcomes, which vary as a function of dose to specific neuroanatomical regions. Survivors with a history of radiation exposure to temporal brain regions are at increased risk for impairment in memory and social functioning.
Central nervous system (CNS) tumors account for 16.6% of all childhood and adolescent malignances and are the second most frequently occurring cancer type in children.1 The most common forms of CNS tumors include astrocytoma (52%), primitive neuroectodermal tumor (PNET; 21%), and ependymoma (9%). Advances in treatment for CNS tumors have resulted in decreased mortality over the past several decades such that 74% of children younger than 20 years of age diagnosed with a CNS malignancy become 5-year survivors.2 Adverse neurocognitive outcomes after cranial irradiation for pediatric CNS tumors include impairments in executive function, memory, attention, and global cognitive abilities.3,4 Emotional distress has been reported to occur more frequently in a subset of survivors of CNS tumors compared with a sibling group.5 Survivors of CNS tumors also experience poorer sociodemographic competence, such as lower educational attainment6 and lower likelihood of marriage7 compared with survivors of other types of cancer. Although recent advances in surgery and radiation therapy have reduced morbidity and improved neurocognitive outcomes, post-treatment functional deficits persist.8–10
In the past, our understanding of the cognitive effects of irradiation in children was limited to patients treated with “whole-brain irradiation” as a component of CNS-directed therapy for acute lymphoblastic leukemia (ALL) and medulloblastoma (MB). Studies in ALL showed the detrimental effects of cranial irradiation for dose levels ≥24 Gy and limited or absent effects when the dose was reduced to ≤18 Gy.11 In patients with MB, the detrimental effects of craniospinal irradiation (CSI) were larger at dose levels of 36 Gy compared with CSI doses ≤23.4 Gy;12 however, more recent studies have not shown a difference.13 Prior to the era of conformal radiation therapy, data were even more limited for patients with localized brain tumors who did not require craniospinal or whole-brain irradiation, but focal CNS radiation.
Previous studies that linked poor outcome with the use of cranial irradiation have been limited in their ability to analyze endpoints based on radiation dose to specific regions of the brain.14 It is reasonable to expect that deficits in executive functioning and memory may be closely associated with radiation dose to frontal and temporal regions, respectively;15,16 however, it has been difficult to evaluate this hypothesis because large-scale studies have not been conducted to correlate region-specific dosimetry with cognitive deficits in long-term survivors of CNS tumors. The Childhood Cancer Survivor Study (CCSS)17,18 includes neurocognitive and Health-Related Quality of Life (HRQOL) outcomes, as well as region-specific dosimetry for over 800 adult survivors of childhood CNS malignancies. These data provide an excellent resource to examine region-specific late effects of radiation. We hypothesized that a dose–response association would exist between irradiation to the frontal and temporal regions of the brain and impairment in the measures of memory and executive functioning and in HRQOL.
The CCSS is a retrospective cohort study, with longitudinal follow-up, of 5-year or longer survivors of childhood cancer treated at 26 institutions in the United States and Canada. Eligibility criteria for participation in the CCSS included the diagnosis of cancer before 21 years of age, initial treatment between January 1, 1970 and December 31, 1986, and survival for at least 5 years after diagnosis. Participants were recruited from survivors treated for initial diagnoses of leukemia, CNS malignancy, Hodgkin lymphoma, non-Hodgkin lymphoma, Wilms tumor, neuroblastoma, soft tissue sarcoma, or bone tumors. The cohort and the study design have been described in detail previously.17,18 The CCSS was approved by the respective institutional review boards at the 26 participating centers, and participants provided informed consent.
Collection of baseline data was initiated in 1994: all participants completed a 24-page baseline questionnaire that included information on demographics, personal and family medical history, medical late effects experienced, psychological outcomes and HRQOL, physical limitations, work history, and current living situation. Included in a recent follow-up questionnaire were self-reported assessments of health behaviors, neurocognitive outcomes, and HRQOL. Study questionnaires are available at www.stjude.org/ccss.
The current analysis includes 818 (87.4%) of the 936 eligible CCSS participants diagnosed with CNS malignancies who completed the 2003 follow-up survey (Fig. 1). Patients who experienced paralysis or mental retardation were excluded based on the self-report nature of the neurocognitive and HRQOL outcomes.
To quantify radiation exposure, the brain was divided into 4 anatomic regions: posterior fossa, temporal including the hypothalamic-pituitary axis, frontal, and parieto-occipital (Fig. 2). Individual radiotherapy records were collected for 90% of the patients, and these records gave details of the field sizes and locations and treatment doses for all treatments of the brain. The maximum radiation dose was estimated for each region.19 Right and left sides of the brain were not evaluated separately. During dose quantification, it was assumed that any region received the full-beam dose if at least half of the total region was included in the beam; otherwise, this region was considered to have received scatter dose. Treatment diagrams and photographs taken in the treatment position were reviewed to determine which brain regions were irradiated. When necessary, a written description of the medical record was used to estimate the regions included and the dose administered. Further details of the dosimetry method have been reported previously.20
Neurocognitive functioning was evaluated by the CCSS Neurocognitive Questionnaire (CCSS-NCQ), a scale that has been previously validated in a large sample of cancer survivors.21 The CCSS-NCQ is a 25-item self-report measure designed to evaluate the degree to which the participant has experienced problems with various neurocognitive functions over the past 6 months. Answers are recorded by using a 3-point Likert scale (1, “never a problem”; 2, “sometimes a problem”; 3, “often a problem”), and 4 subscales (Task Efficiency, Emotional Regulation, Organization, and Memory) are derived from the responses.
HRQOL was evaluated with the Medical Outcomes Short-Form-36 (SF-36).22 The SF-36 has been validated for evaluation of HRQOL in cancer survivors.23,24 Items assess general health, well being, and quality of life over the previous month by using a 3- or 5-point Likert scale. The SF-36 yields 8 subscales (Physical Functioning, Role-Physical [ie, functional limitations due to physical problems], Bodily Pain, General Health, Vitality, Social Functioning, Role-Emotional [ie, functional limitations due to emotional problems], and Mental Health) and 2 composite scores (Physical Component Summary and Mental Component Summary).
Primary outcome variables were the CCSS-NCQ and SF-36 subscale scores. The primary predictor was region-specific dose grouped according to the anatomic map in Fig. 2. For the overall population, specific radiation doses were available and categorized by intensity into 4 groups (no radiation, >0 to <30, ≥30 to <50, and ≥50 Gy), corresponding to the typical ranges of treatment doses. Concordance rates for RT dose between the 4 CNS regions ranged between 5% and 25% providing sufficient heterogeneity between regions necessary for region-specific analysis. Descriptive statistics, including frequency distributions, means, and 95% confidence intervals, were calculated for demographic and treatment variables.
Performance on each of the outcome measures was dichotomized according to whether or not the survivor reported symptom levels were suggestive of clinical impairment, with impairment defined as a score at or below the 10th percentile for the CCSS-NCQ and below a T-score of 40 for the SF-36. Percentile rankings were referenced to a sibling sample for the NCQ21 and to national standardization samples for the SF-36. The strength of the associations between region-specific radiation dose exposure and impairment on the CCSS-NCQ or SF-36 were examined using logistic regression and reported as odds ratios adjusted for sex, age at diagnosis, and the maximum radiation dose to other regions. The adjusted odds ratios with 95% confidence intervals were estimated compared with no radiation therapy for each region. To test for linear trends in dose gradient, a logistic regression model was examined using a group-linear radiation dose categorization (1, “None (No CRT)”; 2, “>0 to <30 Gy”; 3, “≥30 to <50 Gy”; and 4, “≥50 Gy”) for each outcome measure.
Recognizing the potential for confounding by diagnosis type (eg, most patients with MB would have received craniospinal RT at a standard dose, with minimal variability), we also evaluated the aforementioned outcomes stratified by CNS tumor type (MB and PNET vs astrocytoma and other CNS malignancies). For these analyses, logistic regression was again used to model the outcomes as a function of all 4 regional radiation doses based on a 10-Gy unit increase for each segment, adjusted for age at diagnosis, and sex. Linear trends in 10 Gy dose gradients were examined for outcomes found to be significant in the logistic regression analyses and are presented in Fig. 3A–C. Owing to the paucity of data in the 10–20-Gy gradient, patients receiving >0 to 20 Gy were considered as 1 gradient.
Table 1 includes demographic characteristics of the 818 adult survivors. Of these, 54% were male and 63.3% were younger than age 10 years when the CNS malignancy was diagnosed. Astrocytoma was the primary diagnosis in 65.6% of the cases. Radiation was used in 63.6% of survivors. Mean age at the time of evaluation was 31.3 years (SD = 7.0 years, median = 31.0 years, range 18.3–51.8 years). Mean age at diagnosis did not differ statistically between the MB/PNET and astrocytoma/other CNS tumor groups (P = .09). The survivors diagnosed with MB/PNET received significantly higher cumulative dose of radiation (P < .00001) than those diagnosed with astrocytoma/other CNS tumors. Older age at diagnosis was associated with increased radiation dose to the posterior fossa (P = .01), frontal (P = .004), and parieto-occipital (P = .007) regions (data not shown).
Table 2 presents mean values for neurocognitive outcomes as assessed by the CCSS-NCQ. Complete data for the 25 items on the CCSS-NCQ were available for 669 of the 818 (82%) survivors. The percentage of survivors with impaired Task Efficiency (ie, attention and processing speed in the lower 10th percent of normative data) ranged from 24.0% to 55.2%, and the percentage of survivors with impaired Memory ranged from 24.6% to 55.2%. Irradiation of the temporal region, but not other regions, was significantly associated with Memory problems (radiation doses ≥30 to <50 Gy: OR = 1.95, 95% CI: 1.01–3.78; dose ≥50 Gy: OR = 2.34, 95% CI: 1.25–4.39; Table 3) with a dose–response effect with increasing RT dose across the 4 dose-based groupings (Ptrend = .001). As seen in Fig. 3A, a significant trend was also demonstrated across 10 Gy dose gradients (Ptrend = .003). Irradiation of the temporal region was also associated with poor Task Efficiency with increasing odds of impairment with increasing RT dose in a dose-dependent fashion (Table 3, Ptrend = .03). Again, this trend was also demonstrated across 10 Gy dose gradient exposure (Ptrend = .004; Fig. 3B). In survivors with MB or PNET, irradiation of the temporal region was associated with poor Task Efficiency (OR: 2.95, 95% CI: 1.66–5.22 per 10 Gy dose increase; Table 4) and poor Organization (OR: 2.21, 95% CI: 1.04–4.70 per 10 Gy dose increase). In survivors with astrocytoma and other CNS malignancies, radiation dose to the temporal region was significantly associated with Memory problems (OR: 1.14, 95% CI: 1.03–1.25 per 10 Gy dose increase).
Complete data on the SF-36 were available for 811 of the 818 (99%) survivors. The percentage of survivors with impairment (T-scores <40) ranged from 5.0% to 21.1% for physical function, 16.6% to 31.8% for physical limitations, 12.2% to 26.8% for general health, and 22.6% to 36.2% for social function (Table A1). Irradiation of the temporal region was associated with physical limitations at all radiation levels, poor general health at ≥30 to <50 and ≥50 Gy, and poor social functioning at ≥50 Gy (Table 5) in a dose-dependant manner (Fig. 3C). Irradiation of the frontal region was associated with physical limitations and poorer general health. Within the MB/PNET group, irradiation of the temporal region was associated with poorer emotional functioning, whereas radiation dose to the posterior fossa region was associated with poorer general health (Table A2). Within the astrocytoma/other group, irradiation of the temporal region was associated with poorer physical ability, poor general health, and poorer social functioning.
Multiple studies have previously reported that survivors of childhood CNS malignancies treated with whole-brain or craniospinal radiation therapy experience significant neurocognitive deficits, particularly in the areas of memory and executive functioning.8,25–27 However, in this study, we have taken these findings 1 step further by identifying region-specific associations between cranial irradiation exposure and poor outcomes, including memory, task efficiency, and social functioning among adult survivors of childhood CNS malignancies. The identification of radiation dose–response relationships for these associations further increases the strength of the evidence for a causal relationship between poor outcome and region-specific radiation exposure.
Memory functions are largely localized in the temporal regions of the brain.16 Although a previous report had identified memory dysfunction in children with temporal lobe brain tumors,28 survivors in our study demonstrated an increased risk for memory difficulties with increasing dose above ≥30 Gy to the temporal lobe region, whereas no increase in risk was observed with increasing dose to other regions. By controlling for dose to other regions, we ostensibly identified the independent effect of radiation dose to the temporal lobes on memory. We did not identify, as we hypothesized, poor executive function outcomes (ie, Emotional Regulation, Task Efficiency, and Organization) associated with frontal lobe radiation. The relatively small number (n = 63) of participants who received high dose (≥50 Gy) radiation may have limited our ability to identify such an association.
Recognizing that individuals treated for MB and PNET received whole-brain radiation and supplemental focal or regional irradiation of the tumor bed, we analyzed this population separately from survivors of other malignancies. Survivors of MB and PNET reported experiencing difficulties with Task Efficiency and Organization not seen in survivors of astrocytomas and other CNS tumors. Given the dependence of processing speed on white matter integrity,29 and the existence of rich white matter connectivity between the cerebellar region and the frontal brain regions,30 this pattern of deficits in MB survivors is not surprising. Disruption of these white matter tracts in the cerebellar region could result in executive functioning deficits via the loss of supratentorial connections between the cerebellar pathways and the frontal region.
Survivors who received temporal region irradiation also experienced significantly more difficulty in social functioning. It is established that patients with temporal lobe epilepsy demonstrate impairments in facial recognition.31 This is expected because temporal brain regions are involved in the interpretation of emotional expression and facial recognition.32 These potential perceptual limitations, combined with the impact of memory deficits on social communication skills, may account for the increased reliance of social functioning and temporal lobe integrity. Limitations in social functions can contribute to poor HRQOL and persist for several years after treatment for many survivors of CNS malignancies.33 Survivors with a history of radiation therapy to frontal and temporal regions also reported physical limitations and general health difficulties, which can also have an indirect impact on social functioning.
Limitations to the current study should be considered. The self-report medium may not be the most sensitive or specific method to determine neurocognitive deficits.34 However, recent studies have shown that self-reported neurocognitive data correlate with neuroimaging in older adults and children.35,36 As a next step in validating these measures, efforts are underway to compare these self-report measures with established neuropsychological batteries. An additional potential limitation is that we did not obtain complete participation from all survivors eligible for participation in the CCSS follow-up questionnaire. Also, although radiation exposure is certainly associated with poor outcomes, this study cannot control for the independent effects of both direct injury from the tumor itself and surgical procedures. However, the dose–response effect of radiation increases the strength of a causal relationship for radiation. Additionally, the inability to perform separate dosimetry estimates to the hypothalamic-pituitary axis and the temporal lobe may have resulted in deficits in social functioning due to hypothalamic injury being attributed to the temporal lobe. Finally, as noted by previous CCSS investigations,2 therapeutic regimens have changed over the years and some current regimens include increased use of chemotherapy with lower craniospinal doses and smaller boost volumes, which may make our results less generalizable to more recent survivors of childhood CNS malignancies.
Despite these limitations, our study provides valuable, region-specific information on the neurocognitive and HRQOL-related outcomes for adult survivors of pediatric brain malignancies after radiation therapy. Since the recent advent of conformal radiation therapy for the treatment of brain tumors in children, investigators have similarly sought to correlate cognitive outcomes with dose to functional subunits of the brain. In a study that included 86 children with ependymoma, Merchant et al.37 showed that the fractional volume of brain that received dose within specified intervals could be used to estimate IQ. The best models were developed using the dosimetry of supratentorial brain volumes, and the study demonstrated the importance of primary tumor location and correcting for factors, including age at the time of irradiation. Similar models have been developed for patients with craniopharyngioma,38 low-grade glioma,10 and MB.39 One poignant finding is that for most models, both high and low doses contribute to a decline in IQ; however, the volume that receives the highest dose always has the greatest effect.
In particular, these data highlight the importance of radiation dose to the temporal region by identifying associations with poor memory and social function outcomes. These findings accentuate the need for more targeted delivery modalities, such as conformal or intensity-modulated radiation therapy or proton therapy that can limit dose to normal brain volumes, and potentially lower rates of poor functional outcome. As these survivors are at-risk for experiencing difficulties with memory and social functioning, early intervention aimed at preventing the development of deficits and curbing the progression of deficits is warranted. Computerized and face-to-face interventions have been successful at effecting positive changes in memory with individuals with mild cognitive impairment.40,41 These interventions, however, have not yet been used with brain tumor survivors. Therefore, the validation of use of these strategies within the brain tumor survivor population is needed. Participation in social skills training opportunities at a young age and as young adults may facilitate development of social skills for CNS tumor survivors.42,43 Continued follow-up of the CCSS population will further identify the patterns of neurocognitive change as this population ages through middle adulthood.
Conflict of interest statement. None declared.
This work was supported by the National Cancer Institute (U24-CA55727, to L.L.R., Principal Investigator). Support to St. Jude Children's Research Hospital also provided by the Cancer Center Support (CORE) grant (CA 21765) and the American Lebanese-Syrian Associated Charities (ALSAC).
|Region||Dose level (Gy)||Physical function||Role physical|
|N||Mean||95% CI||% Impaired||N||Mean||95% CI||% Impaired|
|>0 to <30||153||48.3||46.5–50.1||20.9||151||48.1||45.9–50.3||25.2|
|≥30 to <50||110||48.3||46.3–50.4||20.9||110||46.8||44.2–49.4||31.8|
|>0 to <30||90||50.3||48.1–52.4||16.7||88||48.4||45.4–51.3||23.9|
|≥30 to <50||174||49.8||48.3–51.2||16.7||175||47.7||45.7–49.8||27.4|
|>0 to <30||269||49.4||48.2–50.7||17.1||267||47.7||46.0–49.4||27.0|
|≥30 to <50||199||48.7||47.3–50.2||19.6||200||47.4||45.4–49.4||29.0|
|>0 to <30||247||48.4||47.1–49.8||21.1||245||47.6||45.9–49.3||28.6|
|≥30 to <50||185||49.6||48.1–51.0||15.7||186||48.0||46.0–50.0||27.4|
|Body pain||General health|
|>0 to <30||152||51.6||49.9–53.3||15.13||153||48.51||46.7–50.3||20.3|
|30 to <50||110||52.2||50.3–54.1||15.45||112||46.82||44.8–48.9||26.8|
|>0 to <30||89||51.8||49.6–53.9||14.61||91||50.02||48.1–51.9||13.2|
|≥30 to <50||175||52.4||50.9–53.9||13.14||177||48.44||46.9–50.0||20.3|
|>0 to <30||267||52.1||50.9–53.4||11.99||271||48.14||46.8–49.5||19.6|
|≥30 to <50||200||52.2||50.8–53.7||14.00||201||48.09||46.6–49.6||22.9|
|>0 to <30||245||51.9||50.6–53.2||12.65||249||48.03||46.6–49.5||20.9|
|≥30 to <50||186||52.2||50.6–53.7||15.59||187||48.73||47.2–50.3||21.4|
|>0 to <30||114||44.4||42.1–46.7||43.9||152||43.4||41.2–45.6||36.2|
|≥30 to <50||97||43.2||40.8–45.7||43.3||111||44.6||42.2–47.1||29.7|
|>0 to <30||75||43.8||41.3–46.3||38.7||90||44.1||41.3–46.9||30.0|
|≥30 to <50||152||45.2||43.2–47.1||37.5||174||45.5||43.6–47.4||28.7|
|>0 to <30||218||43.5||41.9–45.2||45.0||269||43.4||41.6–45.1||34.6|
|≥30 to <50||171||45.2||43.3–47.1||36.8||199||44.2||42.4–46.1||32.7|
|>0 to <30||200||44.1||42.3–45.8||43.0||247||43.3||41.6–45.1||34.0|
|≥30 to <50||161||44.8||42.9–46.8||37.3||185||44.9||43.0–46.8||31.4|
|Role emotional||Mental health|
|>0 to <30||152||48.4||46.0–50.8||15.8||115||55.0||52.9–57.0||9.6|
|≥30 to <50||110||45.8||42.8–48.9||23.6||97||53.5||51.6–55.4||8.3|
|>0 to <30||89||48.2||45.2–51.2||15.7||75||55.3||53.0–57.6||8.0|
|≥30 to <50||174||46.9||44.5–49.3||19.5||152||55.2||53.6–56.7||8.6|
|>0 to <30||267||46.8||44.9–48.7||19.5||219||54.5||53.1–55.9||9.1|
|≥30 to <50||199||46.0||43.6–48.3||21.6||171||55.1||53.6–56.7||9.4|
|>0 to <30||245||47.3||45.3–49.2||18.8||201||54.9||53.5–56.3||9.0|
|≥30 to <50||185||47.1||44.8–49.5||18.9||161||55.1||53.6–56.7||9.9|
Abbreviation: SF-36, Medical Outcomes Short-Form-36.
|HRQOL (SF-36) outcomes||Factors||Medulloblastoma and PNET odds ratio (95% CI)||Astrocytomas and other CNS odds ratio (95% CI)|
|Sex (F vs M)||1.10 (0.43–2.83)||1.55 (0.97–2.48)|
|Age at Diagnosis||1.04 (0.93–1.17)||1.02 (0.97–1.07)|
|Posterior fossa||0.68 (0.34–1.34)||1.05 (0.93–1.17)|
|Temporal||1.30 (0.68–2.51)||1.20 (1.07–1.35)|
|Frontal||0.95 (0.49–1.84)||1.14 (0.99–1.31)|
|Parieto-occipital||1.23 (0.63–2.40)||0.89 (0.77–1.04)|
|Sex (F vs M)||0.97 (0.47–1.97)||1.41 (0.97–2.05)|
|Age at Diagnosis||1.06 (0.97–1.16)||1.01 (0.98–1.05)|
|Posterior fossa||1.05 (0.62–1.76)||1.07 (0.97–1.18)|
|Temporal||0.99 (0.62–1.58)||1.09 (0.99–1.19)|
|Frontal||0.79 (0.47–1.32)||1.06 (0.94–1.20)|
|Parieto-occipital||1.33 (0.79–2.21)||0.95 (0.84–1.07)|
|Sex (F vs M)||2.35 (0.92–6.01)||1.59 (0.99–2.58)|
|Age at Diagnosis||1.12 (1.00–1.26)||1.01 (0.96–1.06)|
|Posterior fossa||0.79 (0.38–1.63)||0.92 (0.81–1.06)|
|Temporal||1.56 (0.82–2.95)||0.98 (0.87–1.11)|
|Frontal||0.91 (0.41–1.98)||1.10 (0.94–1.29)|
|Parieto-occipital||0.96 (0.46–1.98)||1.04 (0.89–1.21)|
|Sex (F vs M)||1.17 (0.52–2.61)||1.24 (0.82–1.88)|
|Age at Diagnosis||1.13 (1.02–1.25)||0.99 (0.95–1.04)|
|Posterior fossa||0.49 (0.27–0.88)||1.02 (0.92–1.13)|
|Temporal||1.16 (0.66–2.05)||1.15 (1.04–1.28)|
|Frontal||0.87 (0.49–1.53)||1.04 (0.91–1.18)|
|Parieto-occipital||1.71 (0.96–3.05)||0.98 (0.86–1.12)|
|Sex (F vs M)||2.80 (1.33–5.88)||1.90 (1.34–2.71)|
|Age at Diagnosis||1.04 (0.95–1.14)||1.01 (0.97–1.04)|
|Posterior fossa||0.85 (0.50–1.42)||0.99 (0.90–1.09)|
|Temporal||1.20 (0.74–1.97)||1.07 (0.97–1.17)|
|Frontal||0.87 (0.51–1.48)||0.97 (0.86–1.09)|
|Parieto-occipital||1.17 (0.68–2.02)||1.09 (0.97–1.22)|
|Sex (F vs M)||1.38 (0.70–2.74)||0.76 (0.53–1.07)|
|Age at Diagnosis||0.99 (0.90–1.08)||0.94 (0.91–0.97)|
|Posterior fossa||0.74 (0.45–1.21)||0.96 (0.88–1.05)|
|Temporal||1.57 (0.98–2.50)||1.10 (1.01–1.20)|
|Frontal||1.02 (0.61–1.71)||0.93 (0.83–1.05)|
|Parieto-occipital||0.88 (0.53–1.47)||1.06 (0.95–1.19)|
|Sex (F vs M)||2.32 (1.04–5.17)||1.34 (0.91–1.98)|
|Age at Diagnosis||1.06 (0.96–1.17)||1.00 (0.96–1.04)|
|Posterior fossa||0.62 (0.33–1.17)||1.03 (0.93–1.14)|
|Temporal||1.98 (1.12–3.49)||1.00 (0.90–1.10)|
|Frontal||0.78 (0.44–1.38)||0.95 (0.82–1.09)|
|Parieto-occipital||1.12 (0.64–1.98)||1.02 (0.90–1.16)|
|Sex (F vs M)||0.51 (0.14–1.90)||0.73 (0.41–1.30)|
|Age at Diagnosis||1.16 (0.99–1.35)||1.00 (0.94–1.06)|
|Posterior fossa||0.64 (0.29–1.43)||0.99 (0.85–1.16)|
|Temporal||0.80 (0.34–1.88)||1.03 (0.88–1.19)|
|Frontal||0.84 (0.36–1.97)||0.94 (0.77–1.15)|
|Parieto-occipital||1.55 (0.66–3.67)||1.02 (0.85–1.23)|
aMale is the referent group for sex and age is modeled as a continuous variable.
bRadiation dose used as a continuous variable with odds ratios estimated based on a 10-unit Gy increase for each segment.