Association of the CHEK2 c.1100delC variant, radiotherapy, and systemic treatment with contralateral breast cancer risk and breast cancer-specific survival

Breast cancer (BC) patients with a germline CHEK2 c.1100delC variant have an increased risk of contralateral BC (CBC) and worse BC-specific survival (BCSS) compared to non-carriers. We aimed to assess the associations of CHEK2 c.1100delC, radiotherapy, and systemic treatment with CBC risk and BCSS. Analyses were based on 82,701 women diagnosed with invasive BC including 963 CHEK2 c.1100delC carriers; median follow-up was 9.1 years. Differential associations of treatment by CHEK2 c.1100delC status were tested by including interaction terms in a multivariable Cox regression model. A multi-state model was used for further insight into the relation between CHEK2 c.1100delC status, treatment, CBC risk and death. There was no evidence for differential associations of therapy with CBC risk by CHEK2 c.1100delC status The strongest association with reduced CBC risk was observed for the combination of chemotherapy and endocrine therapy [HR(95%CI): 0.66 (0.55–0.78)]. No association was observed with radiotherapy. Results from the multi-state model showed shorter BCSS for CHEK2 c.1100delC carriers versus non-carriers also after accounting for CBC occurrence [HR(95%CI) :1.30 (1.09–1.56)]. In conclusion, systemic therapy was associated with reduced CBC risk irrespective of CHEK2 c.1100delC status. Moreover, CHEK2 c.1100delC carriers had shorter BCSS, which appears not to be fully explained by their CBC risk.


Introduction
Breast cancer (BC) has the highest incidence in women worldwide 1 . One of the germline variants that confer a moderate increased BC risk is the CHEK2 c.1100delC variant [2][3][4] , which is found in approximately 0.7% of the Northern and Western European populations 5 . Overall, carriers of this variant are diagnosed at a younger age than non-carriers 4 and the majority develops BCs that are estrogen receptor (ER)-and progesterone receptor (PR)-positive and human epidermal growth factor receptor 2 (HER2)-negative 3,6 . Although this BC subtype has the most favorable prognosis in the general BC population 7 , CHEK2 c.1100delC carriers have a higher risk of developing contralateral breast cancer (CBC) and worse survival 3,4,6,8,9 compared to non-carriers.
Reasons behind these differences are still unclear. A possible explanation is that CHEK2 c.1100delC carriers have a different response to treatment compared to non-carriers, e.g., their normal tissue might experience more harm from radiotherapy. CHEK2 c.1100delC carriers have a functional de ciency in checkpoint kinase 2 (CHK2), a kinase that controls phosphorylation of downstream factors, such as BRCA1 and BRCA2 10 . This leads to a reduced BRCA1/2 function, impaired DNA repair and increased risk of BC 11 . A recent study showed that the CHEK2 c.1100delC variant also disrupts the apoptosis of BC cells, causing unchecked proliferation and contributing to a poorer prognosis 12 . Radiotherapy has been shown to increase the risk of CBC in the general BC population, especially in younger patients 13 .
Treatment with radiotherapy causes DNA strand breaks, which are less likely to be repaired in CHEK2 c.1100delC carriers 14 . While this might be bene cial for the treatment of the rst primary cancer, which is likely to have lost both functional CHEK2 alleles, and cannot repair DNA strand breaks at all, carriers might be more prone to developing a CBC 15 . One case-only study showed a non-signi cant increased risk for developing CBC after treatment with radiotherapy in CHEK2 c.1100delC carriers versus non-carriers but due to the small study size the effects in the younger population could not be investigated 16 . Only one other small study reported on the association between radiotherapy and CBC risk by CHEK2 c.1100delC status 8 .
On the other hand, less is known about whether the effects of systemic therapy on CBC risk and survival differ by CHEK2 c.1100delC status. A population-based study showed a signi cant decrease in CBC risk following chemotherapy and endocrine therapy in general BC 17 . One single-hospital study also found a decreased risk of CBC after chemotherapy use in CHEK2 c.1100delC carriers, and did not nd evidence for a differential association by CHEK2 c.1100delC status 18 . This study also found no evidence for a differential impact of chemotherapy on survival 18 .
Given this uncertainty, our aim was to assess, within a large international cohort, potential differential associations of treatment given for the rst primary BC (i.e. radiotherapy, chemotherapy and endocrine therapy) by CHEK2 c.1100delC status with CBC risk, and to investigate whether the worse breast cancerspeci c survival (BCSS) so far reported in carriers is explained solely by the increased CBC risk.

Study population
We used data from the Breast Cancer Association Consortium (BCAC), selected women of European ancestry, diagnosed with a rst primary invasive BC between 1980 and 2018; exclusion criteria are shown in Fig. 1. The main analyses were based on 82,701 BC patients from 58 BCAC studies (Table S1). All individual studies were approved by the appropriate institutional review boards and/or medical ethical committees. Written informed consent was obtained from all study participants.
Previous analyses investigating the relationship between CHEK2 c.1100delC status, risk of CBC, and mortality have been based on a subset of patients genotyped with Taqman 3, 4 . In particular, the current study includes most carriers from the Weischer et al. study (n = 459) 4 and from the Kriege et al. study (n = 193) 18 , but is based on a larger number of BC patients and includes updated follow-up data.

Data collection
Data included information about CHEK2 c.1100delC status, vital status at last follow-up, CBC occurrence, age and year of diagnosis of the rst primary BC, tumor characteristics of the rst primary BC and CBC, as well as treatment given for the rst primary BC (Tables 1 and 2). In particular, all relevant clinicalpathological and treatment information, as well as outcome information, was collected by individual studies and harmonized by the BCAC Survival, Pathology and Treatment Working Group at the Netherlands Cancer Institute, Amsterdam, the Netherlands, in collaboration with the individual studies before incorporation into the BCAC database (version 13, May 2021). CHEK2 c.1100delC status was obtained from ve different sources: BRIDGES sequencing data 19 , Taqman and iPLEX genotyping 3,4,20 , and imputed genotypes from OncoArray 21 or iCOGS 22 as described in the Supplementary Methods.   Table  S2. Descriptive statistics are shown as mean ± standard deviation (SD) or median and interquartile range (IQR). We used Pearson's χ2 test for categorical data and Kruskal-Wallis test for continuous data to calculate differences in patients' characteristics. The primary study outcomes were time to CBC and BCSS (time to death due to BC).
Hazard ratios (HRs) and 95% con dence intervals (CIs) for the association of treatment given for the rst primary BC (radiotherapy and/or type of systemic treatment) and CHEK2 c.1100delC status with time to CBC were estimated via Cox regression models allowing for delayed entry, strati ed by country and adjusted for age at rst primary BC diagnosis, tumor size, nodal status, grade and ER status. Since ER status is known to violate the proportionality hazards assumption and because the majority of CHEK2 c.1100delC carriers develop ER-positive BC, we performed an additional main analysis restricted to patients diagnosed with a rst primary ER-positive BC. We assumed that patients with unknown CBC status did not develop a CBC during follow-up, and that for CBC cases with unknown time from rst primary BC to CBC diagnosis, CBC occurrence was at last available follow-up.
Time at risk started either three months after rst primary BC diagnosis or at study entry if entry was more than three months after rst primary BC diagnosis, and ended at time of CBC, death or last followup, whichever came rst. We tested for potential differential association of adjuvant and/or neo-adjuvant therapy on CBC risk according to CHEK2 c.1100delC status by including an interaction term between treatment (radiotherapy or systemic treatment) variable and CHEK2 c.1100delC status in the model. CBC risk analyses were strati ed by two follow-up time intervals: i) the rst 5 years after BC diagnosis and ii) starting 5 years after BC diagnosis.
To gain further insight into the relation between CHEK2 c.1100delC status, treatment given for the rst primary BC, CBC risk and death, we used a multi-state model in the framework of the Cox model, with diagnosis of the rst primary BC as initial state, diagnosis of CBC as intermediate (transient) state, and death due to BC, death due to other causes, and death due to unknown causes as absorbing states (Fig. 2), as speci ed in the Supplementary Methods.
The main CBC risk and multi-state analyses were performed on imputed datasets. Complete-case analyses (excluding study subjects with missing values in any of the variables included in the models) were performed as sensitivity analyses. Additional analyses were restricted to: a) patients diagnosed with rst primary BC from 2000 onwards to reduce heterogeneity in treatment regimens; b) patients diagnosed at age 40 or younger to see if the association with radiotherapy was stronger in this subgroup, as reported previously in the general BC population 13 .

Results
Patients carrying the CHEK2 c.1100delC variant were diagnosed at a younger age and earlier years. The tumors of carriers were larger at time of diagnosis, were more often lymph node-positive, grade 2, and ERand PR-positive than in non-carriers; also some differences in treatment were observed (Table 1).
Contralateral breast cancer CHEK2 c.1100delC carriers were diagnosed at younger age and in earlier calendar years, both for the rst primary tumor as well as for the CBC. Overall, the characteristics of the CBC were similar between the non-carriers and carriers (Table S3). However, CHEK2 c.1100delC carriers more often had positive nodes at CBC diagnosis than non-carriers (p = 0.02).
CBC risk by treatment and CHEK2 c.1100delC carrier status There was no evidence for a differential association of CHEK2 c.1100delC status by radiotherapy [ Tables 2-3  Adjusted for age at diagnosis, nodal status, size category and grade of rst primary breast cancer. Abbreviations: CT = chemotherapy; ET = endocrine therapy; P-int = P-value for the comparison of a model including an interaction term between CHEK2 c.1100delC status and a speci c treatment (radiotherapy or systemic treatment) with a model without any interaction term.
Results of analyses for patients diagnosed at the age of 40 years or younger or for patients diagnosed from 2000 onwards were in line with the results of the main analyses (Tables S4-S5). Complete-case analyses results were consistent with the corresponding results of the imputed data analyses (Tables S6-S9), except for the association with radiotherapy in patients diagnosed at the age of 40 years or younger. For these patients, radiotherapy was signi cantly associated with increased CBC risk in the complete-case analysis with follow-up starting 5 years after diagnosis of the rst primary BC [Table S7; HR (95%CI): 2.12 (1.06-4.22), P = 0.03]. In addition, interaction terms between treatments and CHEK2 c.1100delC status could not be properly estimated in some of the complete-case analyses, due to insu cient data.
These included, among others, the analysis based on all patients with follow-up starting at 5 years after BC diagnosis; the analysis restricted to patients diagnosed at the age of 40 years or younger and based on the total follow-up; and the analysis restricted to ER-positive BC with follow-up starting 5 years after BC diagnosis (Tables S10-S12  (Table S15). Abbreviations: HR = hazard ratio; CI = con dence interval; BC = breast cancer; CBC = contralateral breast cancer. The models included age at rst primary BC diagnosis, nodal status, tumor size, grade, radiotherapy and systemic treatment given for the rst primary BC as covariates. The model based on all BC patient included ER status of the rst primary BC as additional covariate. Baseline hazards were allowed to vary across country and transition. All the estimates from the model are shown in Tables S13-S14.
Regarding treatment, radiotherapy was associated with a protective association against death due to causes other than BC or unknown causes, while there was no signi cant association with BC-speci c death (Tables S13-S15). Endocrine therapy alone was associated with a signi cantly decreased risk of BC-speci c death (particularly in patients diagnosed with an ER-positive rst primary BC) and with a highly signi cantly decreased risk of death due to unknown causes. The combination of endocrine therapy and chemotherapy was associated with decreased risk of BC death (in patients diagnosed with an ER-positive rst primary BC), with risk of death due to causes other than BC and had the strongest protective association against death due to unknown causes (Tables S14). The corresponding completecase analyses showed similar patterns of association (Tables S16-S18).

Discussion
The main goal of this study was to assess potential differential associations of treatment by CHEK2 c.1100delC status with CBC risk, and to investigate if the poorer survival in CHEK2 c.1100delC carriers may be explained alone by the occurrence of CBC. The Breast Cancer Association Consortium provided a unique resource of 963 carriers of this single CHEK2 variant to study this question in more detail.
These data did not support the hypothesis of differential associations of treatment with CBC risk by CHEK2 c.1100delC status. As expected, systemic therapy was found to decrease CBC risk, with the strongest association in the rst ve years after rst primary BC diagnosis, when endocrine therapy is likely to be ongoing 17,23 . Overall, we did nd that the combination of endocrine therapy with chemotherapy resulted in the largest reduction in CBC risk, which has been previously reported 17 . The lack of evidence for a differential association of systemic therapy with CBC risk by CHEK2 c.1100delC status suggests that carriers experience a similar bene cial effect as non-carriers. This is in line with previous studies in CHEK2 c.1100delC carriers 18, 24, 25 .
Also, we did not nd a signi cant association of radiotherapy with CBC risk. This lack of association is in contrast with previous studies in sporadic BC patients, which showed that radiotherapy is a contributor to CBC risk, especially when treatment was administered at a younger age 13, 26-28 . One explanation for this might be the change of radiation techniques over time. However, analyses restricted to patients diagnosed from the year 2000 onwards, when treatment regimens were expected to be more homogeneous, showed similar results as were found in the main analyses. Therefore, although observational -and non-randomized -studies like the present cannot rebut hypotheses of causality, these changes are unlikely to be the reason behind the lack of association between radiotherapy and CBC risk in our study.
In line with previous studies 3, 4 we found a greater than two-fold increased risk of CBC in CHEK2 c.1100delC carriers compared to non-carriers. This is consistent with the reported increase in risk of a rst primary BC 2,19 , suggesting that genetic variants that predispose to the development of a rst primary BC will also predispose to the development of a CBC. We also observed a shorter BCSS in CHEK2 c.1100delC carriers compared to non-carriers, after accounting for CBC occurrence, age at diagnosis of the rst primary BC and tumor characteristics. This suggests that the shorter BCSS in CHEK2 c.1100delC carriers versus non-carriers is partly explained by a component other than the established prognostic factors.
Moreover, CHEK2 c.1100delC carriers were on average diagnosed in earlier calendar years compared to non-carriers. Therefore, carriers probably received less e cacious chemotherapy and endocrine therapy compared to non-carriers, which could have affected survival.
The main strengths of our study are the large sample size, including information about tumor pathology, treatment, time to CBC and survival, and a median follow-up of over 9 years. In addition, the use of a There are some limitations to our study that need to be acknowledged. Between studies there was minor heterogeneity in the de nition of stage, grade and cut-offs for ER, PR, and HER2 status, which would have affected both carriers and non-carriers to a similar extent and is unlikely to have impacted our conclusions. Many of the variables related to tumor characteristics and treatment had large proportions of missing values. Complete-case analyses have less power to detect the associations of interest and might be biased if case data are not missing completely at random 29 . We addressed the missing data problem by employing multiple imputation 29 , which should provide unbiased estimates, provided data are missing at random and that imputation models are correctly speci ed. Analyses restricted to complete-case data yielded results that were mostly consistent with the results based on imputed data. In addition, in some complete-case analyses the number of CHEK2 c.1100delC carriers was too low to properly estimate the interaction terms. This underlines the importance of the analyses based on imputed data, which avoids losses in the number of cases and events in the analyses. We also did not consider type of chemotherapy or endocrine therapy in the analyses, nor had we information about ovarian function suppression. Moreover, information about the occurrence of primary ipsilateral BCs was very limited and could not be properly accounted for in our analyses. However, based on the available information, there was no difference in the proportion of ipsilateral BC between CHEK2 c.1100delC carriers and non-carriers (0.6% in both groups) and is unlikely to have had a major impact on our BCSS results. An additional limitation was the lack of information on cause of death for about 25% of those who had died. This would result in a loss of power to detect associations with BCSS in case most of the deaths of unknown causes were due to BC. However, this would, at worst, dilute our results rather than leading to false-positive signi cant associations with BCSS. Finally, while we accounted for several established BC prognostic factors in our analyses, we cannot exclude the presence of residual bias affecting to some extent our results. An example of such bias is known as "indication bias", which applies to the presence of an indication which causes or affects the outcome of interest 30 . This could explain some of the unexpected results for the association of radiotherapy and systemic treatment with death-related outcomes, in case treatment decisions are in uenced by the presence/absence of certain conditions or morbidities in such a way that patients receiving the treatment are less likely to die from other causes than BC. While indication bias could have affected the treatment-related effects on mortality, it is less likely to be an issue for the association of CHEK2 c.1100delC status and treatment with CBC risk and survival.
In conclusion, the results of our study did not provide evidence for differential associations with radiation or systemic therapy by CHEK2 c.100delC status on CBC risk. This suggests that associations with these treatments on CBC risk are similar between carriers and non-carriers. Furthermore, we con rmed the

Data availability statement
The datasets analyzed during the current study are not publicly available due to protection of participant privacy and con dentiality, and ownership of the contributing institutions, but may be made available in an anonymized form via the corresponding author on reasonable request and after approval of the involved institutions. To receive access to the data, a concept form must be submitted, which will then be reviewed by the BCAC Data Access Coordination Committee (DACC); see http://bcac.ccge.medschl.cam.ac.uk/bcacdata/.

Ethics statement
All individual studies were approved by the appropriate institutional review boards and/or medical ethical committees. Written informed consent was obtained from all study participants.