Temporal trends of subsequent central nervous system malignancies among survivors of childhood cancer

Abstract

Background: It is not known whether temporal changes in childhood cancer therapy have reduced risk of subsequent malignant neoplasms of the central nervous system (CNS), a frequently fatal late effect of cancer therapy.

Methods: Five-year survivors of primary childhood cancers diagnosed between 1970 and 1999 in the Childhood Cancer Survivor Study with CNS subsequent malignant neoplasms were identified. Cumulative incidence rates and standardized incidence ratios were compared among survivors diagnosed between 1970-1979 (n = 6223), 1980-1989 (n = 9680), and 1990-1999 (n = 8999). Multivariable models assessed risk factors for CNS subsequent malignant neoplasms.

Results: A total of 157 CNS subsequent malignant neoplasms (1970s, 52; 1980s, 63; 1990s, 42) were identified, excluding meningiomas, which were most often malignant gliomas. The proportion of survivors receiving any cranial radiotherapy exposure was reduced over time (1970s, 77.0%; 1980s, 54.3%; 1990s, 33.9%), while the proportion receiving more than 35 Gy cranial radiotherapy showed a smaller reduction (11.4%, 10.8%, and 8.5%, respectively). Twenty-year cumulative incidence and standardized incidence ratios for CNS subsequent malignant neoplasms by treatment decade were 0.32% (95% confidence interval = 0.18% to 0.46%) and 6.6 (95% CI = 5.0 to 8.7); 0.55% (95% CI = 0.41% to 0.70%) and 8.3 (95% CI = 6.6 to 10.4); and 0.43% (95% CI = 0.31% to 0.55%) and 9.2 (95% CI = 7.0 to 12.0), respectively, with no statistically significant decreases between eras. Multivariable analyses showed increased risk for cranial radiotherapy dose levels more than 10 Gy and for primary diagnoses of medulloblastoma and/or primitive neuro-ectodermal tumor (hazard ratio [HR] = 18.7, 95% CI = 9.2 to 37.9) and astrocytoma (HR = 10.1, 95% CI = 5.3 to 19.5). Three-year cumulative incidence of death after CNS subsequent malignant neoplasms, by treatment decade, were 76%, 74%, and 73%, respectively.

Conclusion: CNS subsequent malignant neoplasm incidence has not decreased despite fewer survivors exposed to CNS-directed radiotherapy. CNS subsequent malignant neoplasm remains a substantial source of mortality for affected patients.

Figure 1.

Flow diagram of eligible Childhood Cancer Survivor Study participants and exclusion criteria applied for this analysis. ^aOther reasons include incarceration, unavailable proxy reporter, incomplete data entry, and language barrier. Meningioma as a subsequent malignant neoplasm is complicated by potential for multifocal and asynchronous lesions; the temporal epidemiology of meningioma as a subsequent malignant neoplasm in the Childhood Cancer Survivor Study cohort is being reported separately. This analysis focuses on the development of malignant central nervous system subsequent malignant neoplasms. A total of 16 patients who are documented to have received supratherapeutic radiation doses (>100 Gy) are excluded from analysis. A sensitivity analysis was performed, indicating that this exclusion does not impact model results. CRT = Cranioradiotherapy.

Figure 2.

Radiation dose exposures across treatment eras. A) Percent of survivors exposed to radiotherapy dose categories. For each treatment era, the proportion of patients receiving no CNS radiotherapy increased from 23.0% to 66.1% (P_trend < .001). For above 0 to 10 Gy, 38.9%, 22.3%, and 14.0% were exposed in their respective treatment eras. The proportion of patients receiving the highest CNS radiotherapy dose level (>35 Gy) decreased to a smaller degree from 11.4% in the 1970s, to 10.8% in the 1980s, to 8.5% in the 1990s. B) Percent of survivors exposed to radiotherapy dose categories, stratified by primary childhood cancer diagnosis group (for each group, P_trend < .001). Primary CNS tumors account for most of the highest radiotherapy dose exposures. Survivors with primary leukemia saw a 57.7% increase in the proportion receiving no radiotherapy from the 1970s to the 1990s, while survivors with primary CNS tumors saw a 26.0% increase in the proportion receiving no radiotherapy across this timeframe. CNS = central nervous system.

Figure 3.

Cumulative incidence and standardized incidence ratios of CNS subsequent malignant neoplasm across treatment eras. A) For all patients with CNS subsequent malignant neoplasm, the 20-year cumulative incidence of subsequent malignant neoplasm increased from the 1970s to the 1980s (P = .02) and did not statistically change from the 1980s to the 1990s. The standardized incidence ratio of SMN did not statistically change across treatment eras. B) When stratified by primary childhood cancer diagnosis, there is an overall increase in the 20-year cumulative incidence of CNS subsequent malignant neoplasm for those with primary CNS tumors (1990s vs 1970s, P = .004). There is a reduction in the 20-year cumulative incidence of CNS subsequent malignant neoplasm for those with primary leukemias treated in the 1990s when compared with the 1980s (P = .004) but not when compared with the 1970s (P = .54). C) When stratified by primary childhood cancer diagnosis, there are no changes in the standardized incidence ratios across treatment eras. CNS = central nervous system.

Figure 4.

Risk factors for development of CNS subsequent malignant neoplasm. Multivariate models assessed demographic and treatment-related risk factors for development of CNS subsequent malignant neoplasm. Shown are forest plots for the association of primary childhood cancer diagnosis, radiotherapy dose exposures, and chemotherapy agent exposures with risk for CNS subsequent malignant neoplasm. A) Primary childhood cancer diagnoses leukemias, lymphomas, and CNS tumors increased risk for CNS subsequent malignant neoplasm, relative to non-CNS solid tumors. B) All radiation dose levels above 10 Gy are associated with increased risk for CNS subsequent malignant neoplasm. Additionally shown is the stratified analysis for the most abundant primary tumor group, primary leukemias. C) In the unstratified analysis adjusting for demographic factors and radiation exposure, epipodophyllotoxin exposure is associated with increased risk for CNS subsequent malignant neoplasm. CI = confidence interval; CNS = central nervous system; IT = intrathecal; SMN = subsequent malignant neoplasm.

Figure 5.

CNS subsequent malignant neoplasm confers poor prognosis. Among survivors with CNS subsequent malignant neoplasm, the 3-year overall survival following CNS subsequent malignant neoplasm attainment is approximately 30% across each treatment era (log-rank test, statistically nonsignificant). CNS = central nervous system; SMN = subsequent malignant neoplasm.