Odds ratios by categories of breast radiation dose estimated from conditional logistic regression and fitted linear dose-response model adjusted for first cancer diagnosis, chemotherapy (yes/no), calendar year of breast cancer diagnosis, and family history of breast/ovarian cancer. The category-specific ORs (95% CIs) were plotted in a linear scale to show they were aligned with the fitted linear model. A, All types (n = 271): OR per 10 Gy, 3.9 (95% CI, 2.5-6.5). B, All types of invasive (n = 201): OR per 10 Gy, 4.5 (95% CI, 2.7-7.3). C, All types of in situ (n = 70): OR per 10 Gy, 2.2 (95% CI, 1.2-6.9). D, Estrogen receptor positive (ER+), invasive (n = 113): OR per 10 Gy, 5.5 (95% CI, 2.8-12.6). E, ER negative (ER-), invasive (n = 41): OR per 10 Gy, 4.8 (95% CI, 1.7-22.3). F, ER unknown (ERnk), invasive (n = 47): OR per 10 Gy, 4.5 (95% CI, 1.8-15.7).
Odds ratios estimated from the fitted linear dose-response model adjusted for first cancer diagnosis, chemotherapy (yes/no), calendar year of breast cancer diagnosis, and family history of breast/ovarian cancer. All doses: OR per 10 Gy, 3.9 (95% CI, 2.5-6.5); less than 1 Gy: OR per 10 Gy, 6.8 (95% CI, 3.9-12.5); 1 to less than 5 Gy: OR per 10 Gy, 4.1 (95% CI, 2.4-7.9); 5 or greater to less than 15 Gy: OR per 10 Gy, 1.8 (95% CI, 1.2-4.2); 15 Gy or more: OR per 10 Gy, 1.4 (95% CI, 1.0-6.4).
eMethods. Statistical Models
eTable 1. Dose-Response Models for Breast Cancer and Breast Dose Radiation in the Childhood Cancer Survivors Study
eTable 2. Results For Testing Joint Effect and Multiplicative and Additive Interaction for Radiation Dose and Anthracyclines
eTable 3. Estimated Odds ratio (OR)a per 10 Gy (and 95% CI) for Subsequent Breast Cancer in Relation to Estimated Radiation Dose to the Breast Cancer Location: Influence Analysis
eTable 4. Distribution of Ovarian Dose Among Cases and Controls by Type of First Cancer
eTable 5. Odds Ratio (OR) per 10 Gy and 95% CI for Subsequent Breast Cancer in Relation to Estimated Dose to the Breast Cancer Location According to Estimated Radiation Dose Delivered to the Ovaries and ER Status
eTable 6. Receipt of Anthracyclines and Alkylating Agents Amongst Controls
eTable 7. Odds ratio (OR) per 100 mg/m2 and 95% CI for Subsequent Breast Cancer in Relation to Anthracycline Dose Among Patients Whose First Cancers Were Associated to Li-Fraumeni Syndrome (LFS) and All Other Patients
eTable 8. Odds ratio (and 95% CI) for Subsequent Breast Cancer in Relation to Specific Anthracycline and Alkylating Agents and Cumulative Doxorubicin Dose (mg/m2): by Breast Cancer Sub-Type
eTable 9. Odds Ratio (and 95% CI) for Subsequent Breast Cancer in Relation to Receipt of Alkylating Agents (yes/no) and Quartile of Cyclophosphamide Equivalent Dose (CED) (mg/m2): by Breast Cancer ER Status
eTable 10. Odds Ratio (OR)a per 10 Gy and 95% CI for Subsequent Breast Cancer in Relation to Estimated Dose to the Breast Cancer Location: Effect Modification
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Veiga LH, Curtis RE, Morton LM, et al. Association of Breast Cancer Risk After Childhood Cancer With Radiation Dose to the Breast and Anthracycline Use: A Report From the Childhood Cancer Survivor Study. JAMA Pediatr. 2019;173(12):1171–1179. doi:10.1001/jamapediatrics.2019.3807
Is the combination of radiotherapy and treatment with anthracyclines associated with an increased risk of breast cancer among survivors of childhood cancer?
In this case-control study of 271 patients with breast cancer within the North America Childhood Cancer Survivor Study, the combination of anthracyclines and radiotherapy doses to the breast was associated with increased breast cancer risks and was greater than the sum of their effects, consistent with an additive interaction.
The results of this study may help inform surveillance guidelines for childhood cancer survivors and assess potential breast cancer risks from contemporary treatment protocols.
Chest irradiation for childhood cancer is associated with increases in breast cancer risk. Growing evidence suggests that anthracyclines increase this risk, but the outcome of combined anthracycline use and radiotherapy has not been studied.
To evaluate breast cancer risk in childhood cancer survivors following radiotherapy and chemotherapy and assess whether risks varied by estrogen receptor (ER) status.
Design, Setting, and Participants
In a North American hospital-based nested case-control study, a retrospective cohort of 14 358 five-year survivors of childhood cancer, diagnosed from 1970 to 1986 and followed up through December 31, 2016, was analyzed. Cases (n = 271) were defined as women with subsequent breast cancer. Controls (n = 1044) were matched 4:1 to cases by age at first cancer and duration of follow-up (± 2 years). Data analysis was conducted from September 2017 to July 2018.
Radiation dose to breast tumor site and ovaries and cumulative chemotherapy doses, including anthracyclines and alkylating agents.
Main Outcomes and Measures
Odds ratios (ORs) for subsequent breast cancer by ER status.
A total of 271 women served as breast cancer cases (median age at first cancer diagnosis, 15 years [range, 3-20]; median age at breast cancer diagnosis, 39 years [range, 20-57]): 201 invasive (113 ER positive [ER+], 41 ER negative [ER–], and 47 unknown) and 70 in situ breast cancers. The OR for breast cancer increased with increasing radiation dose to the breast (OR per 10 Gy, 3.9; 95% CI, 2.5-6.5) and was similar for ER+ (OR per 10 Gy, 5.5; 95% CI, 2.8-12.6) and ER– (OR per 10 Gy, 4.8; 95% CI, 1.7-22.3) cancers. For women who received ovarian doses less than 1 Gy, the OR per 10 Gy to the breast was higher (OR, 6.8; 95% CI, 3.9-12.5) than for women who received ovarian doses greater than or equal to 15 Gy (OR, 1.4; 95% CI, 1.0-6.4). The OR for breast cancer increased with cumulative anthracycline dose (OR per 100 mg/m2, 1.23; 95% CI, 1.09-1.39; P < .01 for trend), and was 1.49 (95% CI, 1.21-1.83) for ER+ cancer vs 1.10 (95% CI, 0.84-1.45) for ER– cancers (P value for heterogeneity = .47). There was an additive interaction between radiotherapy and anthracycline treatment (P = .04) with the OR for the combined association between anthracycline therapy and breast radiation dose of 10 Gy or more (compared with 0 to less than 1 Gy) of 19.1 (95% CI, 7.6-48.0) vs 9.6 (95% CI, 4.4-20.7) without anthracycline use.
Conclusions and Relevance
This study provides the first evidence to date that the combination of anthracyclines and radiotherapy may increase breast cancer risks compared with use of neither treatment with a similar radiation dose response for ER+ and ER– cancers and possibly higher anthracycline risks for ER+ cancers. These results might help inform surveillance guidelines for childhood cancer survivors.
Treatment advances have improved childhood cancer survival from 62% in 1975 to 85% in 2010.1 Increased survival, however, has raised awareness of the late effects of treatment and shifted attention to maximizing benefits while minimizing risks.2 Breast cancer is a common late effect for female childhood cancer survivors, with chest irradiation being a strong but not sole risk factor.2 Breast cancer risk increases linearly with increasing radiation dose, but reduced ovarian function (eg, due to abdominal/pelvic radiotherapy) can lower this risk.3,4
Changes in treatment and improved understanding of breast cancer causes raise new questions. Chemotherapy, especially anthracyclines, is increasingly being used to treat childhood cancers.5 Anthracycline cardiotoxicity is well established,6 and emerging evidence suggests that these drugs also increase breast cancer risk among patients treated with or without chest radiation.3,6,7 Many patients receive both chemotherapy and radiotherapy, but, to our knowledge, the risk from the combination of anthracyclines and radiotherapy has not been studied. Also, breast cancer is a heterogeneous disease, and subtypes have different causes, treatment, and prognosis. Although radiotherapy has been associated with an increased risk of estrogen-receptor positive (ER+) breast cancer,8,9 neither the radiation dose response nor the association between chemotherapy and ER subtype is known.10
To evaluate breast cancer risk according to radiation dose to the breast and ovaries and use of anthracyclines, we conducted a nested case-control study within the North American Childhood Cancer Survivor Study (CCSS). In the previous CCSS nested case-control study, there were 109 breast cancer cases.3 With 10 years of additional follow-up, there are now 271 eligible cases, which allows us to study novel questions, including radiation risks by breast cancer subtype, and the outcomes of combined radiotherapy and anthracycline therapy.
As described previously,11,12 the CCSS is a retrospective cohort of 14 358 childhood cancer survivors diagnosed between 1970 and 1986 at 26 centers in the United States and Canada. Patients were diagnosed before age 21 years with leukemia, central nervous system cancer, Hodgkin lymphoma, non-Hodgkin lymphoma, neuroblastoma, bone or soft-tissue sarcoma, or kidney cancer (Wilms tumor) and survived at least 5 years. The research protocol was approved by human subject committees at participating institutions. Participants or parents of children younger than 18 years provided informed consent; when younger participants reached 18 years, a new informed consent was obtained. Participant data are maintained securely in the CCSS Coordinating Center at St Jude Children’s Research Hospital, Memphis, Tennessee, and data sets provided for analysis are deidentified.
Treatment data for the first cancer were abstracted from medical records according to a standardized protocol.12 Baseline and 5-year follow-up as well as self-administered questionnaires were used to collect information, including height, weight, family history of cancer, reproductive history, and diagnoses of new cancers up to December 31, 2016. Breast cancers in persons who had died were ascertained through family members and the National Death Index. All cancer diagnoses were confirmed by pathologic reports (if available), medical records, or death certificates.
Case patients had invasive or in situ female breast cancer diagnosed 5 or more years after their first cancer diagnosis (n = 283). Twenty-one women presented with bilateral synchronous breast cancers; priority to be included in the analysis was given to invasive rather than in situ tumors or to the larger tumor when both were invasive or in situ. Four to 5 patients serving as controls were randomly selected for each case (n = 1133), matched on age at first cancer (±2 years) and duration of follow-up (±2 years). Nine controls and 1 case were excluded because of an error in the control selection. We further excluded 11 cases (3.9%) and 36 controls (3.9%) because of unknown radiotherapy status. Because 11 cases from the set were excluded, the respective 44 controls were also excluded. This process resulted in 271 included cases and 1044 controls.
We retrieved ER and progesterone receptor status from pathologic reports for 154 of the 201 invasive breast cancers (76.6%), but only 27 of the 70 in situ cases (38.6%). Most of the 113 ER+ invasive cancers were progesterone receptor positive (80.5%) and 90.2% of 41 ER negative (ER–) cancers were progesterone receptor negative. Because of the study period, ERBB2 (formerly HER2) status was available for only 45% of the cases, and in total there were only 21 triple-negative breast cancers (20% of those with known status). Our subtype analyses, therefore, focused on ER+ vs ER– invasive breast cancers.
Radiotherapy records included dates of therapy, beam energy, field locations and size, blocking, and total prescribed dose by field. As described previously,3,13 breast doses in Gray (unit of absorbed dose of ionizing radiation) were determined to breast tumor location for each case and corresponding location in matched controls. The proximity of each radiation field to the breast tumor location was determined by reviewing photographs, diagrams, notes on blocking, and field shaping. Breast dose was calculated according to the assigned proximity, set as either in-beam (blocked/unblocked), on the field edge, on the block edge, near the field edge, or out of beam. If tumor location was unknown but a relatively uniform dose was delivered across the breast, a mean dose to the breast was calculated. For approximately 12% of cases and controls, breast dose could not be adequately estimated because the tumor location was unknown and the breast dose was highly nonuniform. Dose quality scores were assigned based on radiotherapy information and specificity of breast tumor location. We included all patients with estimated breast dose and a sensitivity analysis compared results by organ-dose quality score.
Dose estimation also considered breast size and development at the time of treatment. As photographs or clinical notes were not available for most patients, breast development was determined using the median age of entry into a Tanner stage, taking into account race14 and age at menarche. Tanner stage 1 represents sexual immaturity (prepubertal: no breast development) and stage 5 indicates full maturity (full breast development). For dosimetric purposes, categories of Tanner stages were nipple/bud (Tanner 1 or 2), underdeveloped (Tanner 3), and developed (Tanner 4 or 5).
Cumulative dose to the breast tumor site was calculated, including all treatments received up to 5 years before breast cancer diagnosis to account for the minimum 5-year latency period for radiation-related breast cancer.15 Doses to the left and right ovaries were estimated for the same period and the minimum ovarian dose was used if doses differed.
We summarized receipt of chemotherapy, anthracyclines, and alkylating agents prior to the cutoff date as ever/never and also for specific agents most commonly received (doxorubicin, daunorubicin, cyclophosphamide, procarbazine, and mechlorethamine). We calculated cumulative anthracycline dose (sum of doxorubicin and daunorubicin dose) and alkylating agent exposure using cyclophosphamide-equivalent dose (milligrams per meters squared).16
We used multivariable conditional logistic regression in Epicure, version 2.00.02 (Risk Science International) to estimate odds ratios (ORs) and 95% CIs for breast cancer in relation to radiation dose categories and chemotherapy.17 For continuous radiation dose, we fitted a linear radiation dose-response model, as there was no evidence of downward curvature at higher doses or upward curvature at lower doses (eMethods and eTable 1 in the Supplement). An influence analysis was conducted to evaluate the changes in dose response when omitting first cancers, and a sensitivity analysis restricting to cases where breast cancer was the first subsequent neoplasm (n = 260 cases). We estimated the OR per 10 Gy and 95% CI from the linear dose-response model. We analyzed chemotherapy for cumulative anthracycline dose and cyclophosphamide-equivalent dose as well as for commonly used agents. Trend tests were based on the continuous variable.
We conditioned models on study matching factors (age at first cancer and follow-up duration) and adjusted for first cancer type, chemotherapy (yes/no), calendar year of follow-up (1978-1994, 1995-1999, and 2000-2016), and first-degree family history of breast and/or ovarian cancer (yes/no), all considered significant breast cancer risk factors in the multivariable analysis (P < .05).
We evaluated effect modification of the radiation dose response by ovarian dose and other factors using interaction terms and goodness of fit evaluated by likelihood ratio tests. We assessed the joint effect of radiation (continuous dose) and anthracyclines (yes/no) by comparing the goodness of fit and the Akaike information criterion18 of models with a multiplicative or additive joint effect with models including an interaction term (eMethods and eTable 2 in the Supplement). We used case-case analysis to compare ORs for invasive and in situ breast cancer and ER subtypes. Statistical tests were 2-sided and were based on an α level of .05. Data analysis was conducted from September 2017 to July 2018.
Among the 271 women serving as breast cancer cases, the median age at first cancer diagnosis was 15 years (range, 3-20) and median age at breast cancer diagnosis was 39 years (range, 20-57); cases were more likely to be Hodgkin lymphoma survivors and have received radiotherapy than controls (Table 1). Compared with women with ER+ invasive breast cancers, women with ER– invasive cancers were more likely to be leukemia or soft-tissue sarcoma survivors, and less likely to be Hodgkin lymphoma survivors.
The OR for overall breast cancer increased linearly with increasing radiation dose to breast cancer location (Figure 1). The OR per 10 Gy was 3.9 (95% CI, 2.5-6.5), and it was significantly increased even for breast doses lower than 5 Gy (OR, 1.7; 95% CI, 1.0-3.0) compared with 0 Gy. An influence analysis did not show evidence that omitting any type of first cancer changed the dose-response relationship (eTable 3 in the Supplement).
The dose response was attenuated for in situ breast tumors compared with invasive cancers (OR per 10 Gy, 2.2; 95% CI, 1.2-6.9 vs 4.5; 95% CI, 2.7-7.3), although not significantly (P > .50) (Figure 1B and C). A high proportion (n = 56) of in situ breast cancers were diagnosed in Hodgkin lymphoma survivors (80.0%), and only 4 cases were diagnosed in patients who did not receive radiotherapy. The dose response was similar for ER+ and ER– invasive breast cancers (P = .94 for heterogeneity), as well as those with unknown ER– status (Figure 1D, E, and F). In sensitivity analyses, the overall dose response was similar for the highest and the lowest-quality radiation dose data (OR per 10 Gy, 4.1; 95% CI, 2.6-7.0 vs 3.6; 95% CI, 2.2-6.4; P = .55 for heterogeneity). A similar dose response was also observed when restricted to breast cancers that were the second primary cancer (OR per 10 Gy, 3.3; 95% CI, 2.0-6.2).
Ovarian radiation dose was a strong modifier of the risk associated with breast dose (P < .01). Most patients (901 [68.5%]) received no or low doses to the ovaries (<1 Gy) (eTable 4 in the Supplement) and for these patients, the OR per 10 Gy to the breast was 6.8 (95% CI, 3.9-12.5) (Figure 2). High doses to the ovaries (≥15 Gy) resulted in a lower odds of breast cancer associated with radiation to the breast (OR per 10 Gy, 1.4; 95% CI, 1.0-6.4). This protective effect of ovarian radiation was evident for ER+ and ER– invasive breast cancers (eTable 5 in the Supplement). Most of the patients who received 15 Gy or more to the ovaries reported acute ovarian failure (48 [62.3%]), defined as never menstruating by age 18 years or menopause within 5 years of their first cancer diagnosis. Although only 17 of women (23.2%) who received ovarian irradiation of 5 to less than 15 Gy reported acute ovarian failure, they had substantially reduced—although still elevated—odds of breast cancer (Figure 2).
Anthracyclines were the most commonly used agents in bone cancer followed by non-Hodgkin lymphoma and leukemia survivors and increased with year of first cancer diagnosis (eTable 6 in the Supplement). Odds for breast cancer increased with increasing cumulative anthracycline dose (OR per 100 mg/m2, 1.23; 95% CI, 1.09-1.39; P < .01 for trend), with a slightly higher dose response for ER+ than ER– cancers (OR per 100 mg/m2, 1.49; 95% CI, 1.21-1.83 vs 1.10; 95% CI, 0.84-1.45; P = .47 for heterogeneity) (Table 2). A similar anthracycline dose response was observed for patients whose first cancers were associated with Li-Fraumeni syndrome (leukemia, central nervous system cancer, and non-Ewing sarcoma) and all other patients (P = .22 for heterogeneity) (eTable 7 in the Supplement). There was also evidence of a dose response for doxorubicin (P = .02 for trend), the most commonly received anthracycline drug (257 [24.6%] of controls) but not daunorubicin (69 [6.6%] of controls) (eTable 8 in the Supplement). There was an additive interaction between radiotherapy and anthracyclines (P = .04) with the OR for breast cancer for anthracyclines and a breast radiation dose of 10 Gy or more, compared with 0 to less than 1 Gy, of 19.1 (95% CI, 7.6-48.0) vs 9.6 (95% CI, 4.4-20.7) without anthracyclines (Table 3).
Almost half (49.8%) of the controls received alkylating agents, most commonly cyclophosphamide (365 [35.0%]) and/or procarbazine (212 [20.3%]) (Table 1; eTable 8 in the Supplement). There was no clear association between breast cancer risk and alkylating agents and no dose response for cyclophosphamide-equivalent dose overall (P = .51 for trend) or for ER+ invasive cancers (P = .99 for trend) (eTable 9 in the Supplement). There was some evidence of an increased risk of ER– invasive cancers with alkylating agents (OR, 2.1; 95% CI, 0.9-4.9) and for the highest quartile of cyclophosphamide-equivalent dose (>13 955 mg/m2) (OR, 3.1; 95% CI, 0.7-12.7) vs no alkylating agents. This possible increase was significant for procarbazine (OR, 5.2; 95% CI, 1.0-27.2), but not for cyclophosphamide (OR, 1.6; 95% CI, 0.6-4.7) or mechlorethamine (OR, 0.2; 95% CI, 0.03-1.5) (eTable 8 in the Supplement).
There was no evidence that type of first cancer, age at radiation exposure, age at menarche, or menopausal status modified the radiation dose-response relationship for breast cancer overall (eTable 10 in the Supplement). The exception was radiotherapy after menarche, which was associated with a significantly lower dose response (P = .01 for heterogeneity). This difference was reduced and nonsignificant, however, when we restricted analysis to patients receiving ovarian doses <1 Gy (P = .53 for heterogeneity).
To our knowledge, this is the largest study of treatment-related breast cancer after childhood cancer. The large sample size, ER status, and detailed treatment records enabled assessment of several new hypotheses. We present novel evidence for a similar radiation dose response for ER+ and ER– breast cancer and possibly higher anthracycline risk for ER+ than ER– breast cancer. To date, we also provide the first evidence that the combination of anthracyclines and radiotherapy may markedly increase breast cancer risks and is greater than the sum of their individual effects, consistent with an additive interaction.
Our radiation dose-response estimate seems to be consistent with the earlier CCSS case-control study (n = 109 cases)3 and the case-control study from the French/British childhood cancer survivors (n = 16 cases).19 With 271 breast cancers, we had more statistical power to quantify the potential modifying effect of the ovarian dose and showed the higher breast cancer risk in women receiving less than 1 Gy to the ovaries compared with 15 Gy or more. It is well established that breast irradiation in childhood carries higher risks than adulthood exposures.15 Most of our patients were exposed within a narrow period (age 10-20 years), and we did not find evidence of more radiosensitive periods within this age range.
There have been 2 reports of breast cancer risks from anthracyclines.6,7 The Dutch childhood cancer survivors study (n = 45 cases) reported a dose-response with doxorubicin of similar magnitude to that reported herein.7 In a previous CCSS analysis of patients who did not receive chest irradiation, the investigators also reported a dose response for anthracyclines (n = 47 breast cancers).6 This restriction meant that they could not evaluate the interaction with chest radiotherapy, and neither study evaluated breast cancer subtypes. We found some possible indication that anthracycline risks may be higher for ER+ than for ER– cancers, which warrants further investigation.
Previous studies of Hodgkin lymphoma survivors have found that alkylating agents, primarily procarbazine, reduce breast cancer risks, likely owing to ovarian ablation.20,21 The studies of patients with childhood cancer either found no reduction or possibly increased risks.6,7 We did not find any possible association with breast cancer, even following high cumulative doses of alkylating agents (cyclophosphamide-equivalent dose >10 000 mg/m2). It is unclear why there would be qualitatively different findings across these studies, but the effect of the change from MOPP (mechlorethamine, oncovin, procarbazine, and prednisone) to ABVD (adriamycin [doxorubicin], bleomycin, vinblastine, and dacarbazine) regimens in Hodgkin lymphoma treatment and possible risk of ER– breast cancers warrants further investigation.
To our knowledge, collection of tumor subtype information allowed us to evaluate the association with treatment for the first time. A previous study reported higher risks for ER– than for ER+ breast cancer after Hodgkin lymphoma compared with the general population using US Surveillance, Epidemiology, and End Results registry data.9 There was also a recent case series that characterized breast cancer subtypes in childhood and young adulthood cancer survivors.22 The investigators reported a higher proportion of triple-negative cancers (29%) compared with the general population, but did not have detailed treatment data or a comparison population.
In the past few decades, radiotherapy doses have been lowered for many childhood cancers, and the volume of tissue irradiated has been reduced. It is surprising therefore that there has not been clear evidence that these decreases have translated into lower breast cancer risks.20,23 One possibility that our results suggest is that this increased breast cancer risk could be associated with the concurrent increase in the use of anthracycline therapy.5 The National Wilms’ Tumor Study also reported increased second cancer risks (predominantly sarcomas) after the combination of abdominal radiotherapy and doxorubicin.24 It has been suggested that breast cancer risks associated with anthracyclines may be influenced by a subset of patients who have increased breast cancer susceptibility due to Li-Fraumeni syndrome.7 However, recent studies reported that the prevalence of TP53 pathogenic variants was only 0.7% among childhood acute lymphoblastic leukemia survivors25 and 0.3% among 5-year childhood cancer survivors in the St Jude Lifetime Cohort Study,26 suggesting it will be rare also in our population.
Early animal studies suggested that anthracyclines are highly potent in producing malignant transformation and mutation in mammalian cell systems in vitro.27-29 Although the mechanism of action is poorly understood, the high binding affinity of anthracyclines for DNA with subsequent disruption of DNA-template functions may play a role.28 Anthracyclines can cause radiation recall, which is an inflammatory skin reaction confined to previously irradiated areas trigged when chemotherapy agents are administered.30 This phenomenon suggests a biological interaction between radiation and anthracyclines and highlights the need for further research into their combined effect.
This study has limitations, including small numbers for some subgroups, and therefore limited power for some analysis (eg, effect modification). Because of the calendar period, ERBB2 (formerly HER2) status was available for only 21 cases, which prevented analyses of more refined subtypes. Treatment information for recurrent disease or intervening new primary cancers might be incomplete and there may be errors in the estimated breast doses. Under standard statistical assumptions, dose error that is nondifferential with respect to outcome will bias dose-response relationships toward the null.31 When we restricted our analysis to patients with the highest quality dose estimates, however, the risk estimate increased by only 5%. Recommendations that women with a history of chest irradiation start breast cancer screening at a younger age were introduced in the mid-2000s, which could have introduced bias in more recent years. A CCSS survey of women treated with chest irradiation conducted in 2005-200632 reported that only 36% of women younger than 40 years had received screening in the previous 2 years, compared with 76% of those older than 40 years, suggesting that screening bias was likely limited.
The study strengths include large sample size, long-term follow-up, detailed treatment records, individualized radiation dosimetry, and questionnaire data on reproductive and hormonal factors.
In this large case-control study, breast cancer risk was associated with radiation dose to the breast tumor site and ovaries following childhood cancer. The results suggest that the combination of anthracyclines and even moderate radiation doses to the breast (≥10 Gy) can increase breast cancer risks. Our results may help inform risk management for patients with childhood cancer treated in the past, as well as potential breast cancer risk associated with current treatment protocols.
Accepted for Publication: June 19, 2019.
Corresponding Author: Lene H. Veiga, PhD, Radiation Epidemiology Branch, Division of Cancer Epidemiology & Genetics, National Cancer Institute, 9609 Medical Center Dr, Bethesda, MD 20892 (email@example.com).
Published Online: October 28, 2019. doi:10.1001/jamapediatrics.2019.3807
Author Contributions: Dr Veiga 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.
Concept and design: Oeffinger, Leisenring, Whitton, Armstrong, Inskip, Berrington de Gonzalez.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Veiga, Curtis, Weathers, Whitton, Berrington de Gonzalez.
Critical revision of the manuscript for important intellectual content: Veiga, Curtis, Morton, Withrow, Howell, Smith, Oeffinger, Moskowitz, Henderson, Arnold, Gibson, Leisenring, Neglia, Turcotte, Whitton, Robison, Armstrong, Inskip, Berrington de Gonzalez.
Statistical analysis: Veiga, Curtis, Whitton, Inskip, Berrington de Gonzalez.
Obtained funding: Withrow, Moskowitz, Robison, Armstrong.
Administrative, technical, or material support: Arnold, Leisenring, Neglia, Robison, Armstrong.
Supervision: Leisenring, Neglia, Inskip.
Conflict of Interest Disclosures: Ms Smith reported receiving grants from St Jude Children’s Research Hospital and other support from the National Cancer Institute (NCI) Radiation Epidemiology Branch during the conduct of the study, and other support from Memorial Sloan Kettering outside the submitted work. Dr Moskowitz reported receiving grants from the National Institutes of Health (NIH) during the conduct of the study. Dr Gibson reported receiving grants from the NIH during the conduct of the study. Dr Leisenring reported receiving grants from the NIH during the conduct of the study. Dr Armstrong reported receiving grants from the NIH during the conduct of the study. No other disclosures were reported.
Funding/Support: This work was supported by the Intramural Research Program of the NCI/NIH, grants CA55727 (principal investigator [PI]: Dr Armstrong) and R01CA136783 (PI: Dr Moskowitz) from the NCI, and grant P30 CA008748 from the Memorial Sloan Kettering Cancer Center Core Grant). Support to St Jude Children’s Research Hospital was also provided by Cancer Center Support grant CA21765 and the American Lebanese-Syrian Associated Charities.
Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.