Treatment, toxicity, and mortality after subsequent breast cancer in female survivors of childhood cancer

Abstract

Childhood cancer survivors, particularly those who received chest radiotherapy, are at high risk for developing subsequent breast cancer. Minimizing long-term toxicity risks associated with additional radiotherapy and chemotherapy is a priority, but therapeutic tradeoffs have not been comprehensively characterized and their impact on survival is unknown. In this study, 431 female childhood cancer survivors with subsequent breast cancer from a multicenter retrospective cohort study were evaluated. Compared with one-to-one matched females with first primary breast cancer, survivors are as likely to be prescribed guideline-concordant treatment (N = 344 pairs; survivors: 94%, controls: 93%), but more frequently undergo mastectomy (survivors: 81%, controls: 60%) and are less likely to be treated with anthracyclines (survivors: 47%, controls: 66%) or radiotherapy (survivors: 18%, controls: 61%). Despite this, survivors have nearly 3.5-fold (95% CI = 2.17-5.57) greater mortality risk. Here, we show survivors with subsequent breast cancer face excess mortality despite therapeutic tradeoffs and require specialized treatment guidelines.

Fig. 1. Cumulative mortality partitioned by cause of death after subsequent breast cancer among childhood cancer survivors.

Mortality probabilities due to breast cancer (BC) are shown relative to other causes in red and blue, respectively (n = 402 overall). Top row panels are stratified by breast cancer disease histology (DCIS/ LCIS: ductal or lobular carcinoma in situ). Middle row panels are stratified by concordance/discordance with temporally matched national breast cancer treatment guidelines. Bottom row panels are stratified by treatment modality.

Fig. 2. Risk factors for all-cause and cause-specific mortality after subsequent breast cancer in childhood cancer survivors.

BC breast cancer, RT radiation therapy, CV cardiovascular, SMN subsequent malignant neoplasm. HRs and corresponding 95% confidence intervals (CIs, shown as error bars) from multivariable Cox proportional hazards regression models evaluating associations between primary and breast cancer treatments and the mortality hazard rate using age as the time scale and adjusting for covariates included above are shown, as well as breast cancer diagnosis year and histology (invasive versus non-invasive) or stage (III/IV versus I/II, invasive carcinomas only) (n = 314 overall). HRs and 95% CIs are annotated for statistically significant variables (two-sided P < 0.05). Primary cancer chest RT dose is provided per 10 Gray and primary anthracycline dose is provided per 100 mg/m².

Fig. 3. Differences in the management of breast cancers in childhood cancer survivors and the general population (controls).

For all panels, up to 344 survivor-control pairs were assessed. Percentages treated with specific modalities are provided in panel A for survivors (blue) and matched controls (grey); percentages for mastectomies and lumpectomies are among those who received surgery for breast cancer, and percentages treated with cyclophosphamide, taxanes, anthracyclines are among those treated with chemotherapy for breast cancer. Mastectomy refers to unilateral or bilateral mastectomy unless otherwise specified. Guideline concordance refers to standard of care for breast cancer consistent with temporal national guidelines. In panel B, the odds of treatment receipt in survivors are compared with matched controls (odds ratios or ORs; point estimates to the right of the dashed line reflect increasing survivor odds of treatment) and corresponding 95% confidence intervals are shown (CIs; shown as error bars). Panel C shows temporal changes in breast cancer treatment decisions for survivors and matched controls; 2007 was the median year of breast cancer diagnosis, each bar reflects the percent treated among all participants with non-missing values for a given treatment within the time window specified in the x-axis.

Fig. 4. All-cause mortality probabilities in childhood cancer survivors and matched controls by breast cancer histology and treatment modality.

In all panels, mortality curves for survivors are shown in blue while those for matched controls are shown in gray, and hazard ratios (HRs) adjusted for guideline-concordant breast cancer treatment comparing mortality risks for survivors and controls are provided along with corresponding 95% confidence intervals (CIs) (n = 241 survivor-control pairs overall). Differences in mortality curves for survivors and matched controls were evaluated with two-sided log-rank tests with robust variance estimation; these p values are shown in the lower left quadrant for each panel.