Clinicopathologic and genetic analysis of invasive breast carcinomas in women with germline CHEK2 variants

Checkpoint kinase 2 (CHK2) is a serine/threonine protein kinase encoded by the CHEK2 gene which maintains genomic stability in the setting of DNA damage [1‐3]. CHK2 is activated by the ATM kinase and in turn, interacts with BRCA1, BRCA2, p53, CDC25, and other effectors to initiate DNA repair, apoptosis, and cell cycle arrest [4‐6]. Germline CHEK2 variants were first described in patients without germline TP53 mutations who were thought to meet clinical criteria for Li-Fraumeni syndrome [7, 8], although the association with Li-Fraumeni syndrome has since been negated [9]. Germline CHEK2 variants have now been identified in ~ 1% of the population [9], and pathogenic variants (PVs) are causally linked to multiple types of cancer, including colon cancer and breast cancer (BC), further highlighting the functional relevance of CHEK2 in safeguarding genome integrity [3] [10]. Germline CHEK2 PVs confer a lifetime BC risk ranging from 20 to 44%, depending on family history of BC [11‐13]. Furthermore, individuals with the higher risk c.1100delC allele carry a BC risk of up to two times higher in women and ten times higher in men when compared to non-carriers [14‐16]. Accordingly, clinical guidelines currently recommend enhanced BC screening in individuals with protein truncating CHEK2 variants [17].

Investigations into the phenotype and clinical behavior of CHEK2-associated BCs have been limited. CHEK2-associated BCs tend to occur in younger women, with higher rates of bilateral disease [12], and frequently express estrogen receptor (ER) and/or progesterone receptor (PR) [12, 13]. Interestingly, in contrast to BRCA1/2-related cancers, CHEK2-associated BCs appear to lack the genetic characteristics of homologous recombination DNA repair deficiency [18, 19]. Biallelic CHEK2 inactivation has been observed in some but not all CHEK2-associated BCs [19]. The precise mechanistic relationship between CHEK2 and BC development currently remains uncertain.

In order to characterize BC that develop in the setting of germline CHEK2 variants, we explored the clinicopathologic, histologic, and genetic features of BCs arising in a consecutive series of patients with pathogenic or likely pathogenic germline CHEK2 variants (P/LPVs) at our institution.

In summary, CHEK2-associated invasive BCs tended to arise in younger women with a strong family history of BC and were often multifocal and bilateral. The tumors were mostly immune-poor ER + IDC-NST of predominantly luminal B subtype and were heterogeneous in terms of histologic grade, HER2 status, proliferation index, and multi-parametric molecular testing. NGS overall showed luminal-like genetics and biallelic CHEK2 alteration via LOH as an oncogenic driver in most cases. BCs arising in patients with concurrent germline PVs in the DNA damage genes PALB2 and ATM (but not RAD50) also showed LOH in these genes. Most patients with CHEK2-associated BCs had partial responses to NACT, with few achieving pCR.

Most but not all CHEK2-associated BCs showed biallelic CHEK2 alteration via LOH, which is in keeping with incomplete rates of LOH seen in prior studies [18, 19, 29, 30]. A recent study found high rates of CHEK2 LOH in BCs arising in c.1100delC carriers compared to lower rates in other CHEK2 variants [19]. Consistent with this, we found CHEK2 LOH in all BCs arising in c.1100delC carriers (6/6) but only in 47% (8/17) of BCs arising in other variants. Taken together, our findings and those of others raise consideration that at least some CHEK2 variants may function via haploinsufficiency in driving BC, whereas others (such as c.1100delC) may require bilallelic inactivation. This is in contrast to most other germline BC risk genes, such as BRCA1/2, PALB2, and ATM, PVs of which often show biallelic loss of function in BC [29]. Alternatively, CHK2 inactivation may also be due to non-genetic mechanisms in cases without LOH, and future studies may help address this issue.

Most BCs arising in our population of CHEK2 carriers were ER + IDC-NST, predominantly of luminal B subtype, with a minority of tumors being luminal A or HER2-enriched, and none being triple negative. These findings are consistent with previous reports showing a predominance of ER/luminal BC and HER2 overexpression in this setting [9, 12, 19, 31, 32]. Aside from biallelic CHEK2 inactivation in the majority of cases, the mutational repertoire of the sequenced luminal BCs in our series was otherwise similar to other luminal BCs and included recurrent aberrations in genes such as PIK3CA, GATA3, CCND1, FGFR1, and ERBB2, with a low frequency of TP53 mutations. Our results are thus overall consistent with those of previous studies [18, 19, 31, 33‐35] and highlight CHEK2 as a dominant oncogenic driver in these tumors while also supporting a model in which CHEK2 may act as a facilitator or accelerator of tumorigenesis that is otherwise subtype-specific. This model of CHEK2-facilitated tumorigenesis was previously put forth by Massink et al. in their study describing luminal-like copy number profiles of BCs arising in CHEK2 c.1100delC carriers [18].

Concurrent germline P/LPVs in multiple hereditary BC risk genes (so-called double heterozygotes) are rare, and data regarding cancer risk and clinical implications are scarce. The majority of reported double heterozygotes in the literature include BRCA1 and/or BRCA2, with only few studies and case reports describing double heterozygosity exclusively involving other BC risk genes, such as CHEK2, ATM, PALB2, NBS1, and BLM [36‐44]. Given the rarity of these events, conclusions regarding cancer risk, cancer phenotype, patient surveillance, and clinical management of these cases are limited. It has been suggested that BRCA1/2 and CHEK2 double heterozygotes do not have increased BC risk or a different BC phenotype beyond what is expected for BRCA1/2 variants [39, 40, 44]. Whether a similar principle holds for double heterozygotes involving CHEK2 and non-BRCA1/2 BC risk genes is not known due to paucity of data. Taken as a whole, BCs arising in double heterozygotes do not appear to present at younger age or with increased bilaterality compared to single heterozygotes [39]. Of the five double heterozygotes involving CHEK2 with ATM, PALB2, RAD50, or MUTYH in our study, we noted a younger age of BC presentation than the cohort median in three cases (CHEK2 c.433delC/MUTYH c.536A > G, age 34; CHEK2 c.1100delC/PALB2 c.2827_2830del, age 25; CHEK2 c.1100delC/ATM c.237del, age 29), and bilateral and multifocal tumors in only one patient (CHEK2 c.707 T > C/RAD50 c.2517dupA). We therefore find no compelling evidence of a synergistic effect compared to CHEK2 carriers alone, but note that the small number of such cases in our series significantly limits meaningful conclusions. There is similarly little data on BC risk and phenotype in CHEK2 homozygotes or compound heterozygotes, although available data suggests that these patients may have higher BC risk, present at younger age, and more often have a second BC diagnosis compared to heterozygotes [45‐47]. Our study included two CHEK2 homozygotes (c.1100delC and c.499G > A), who presented with BC at 45 and 39 years of age, respectively, and have not developed additional BCs at time of study.

Ours is one of few studies assessing the response of CHEK2-associated BCs to NACT. Although decreased post-treatment tumor size and Ki67 proliferative index reflected treatment effect in all evaluable cases, only three patients in our series achieved pCR. Two tumors with pCR were HER2 + (Luminal B HER2 + and HER2 enriched). These results are not surprising, given the overall lower pCR rates of ER + /luminal tumors [48, 49] as well as the absence of robust homologous recombination DNA repair deficiency in CHEK2-associated BC [18, 19, 30]. Indeed, our results are similar to a prior study of poor NACT responses in eight CHEK2 patients with BC, in which responses to NACT were also found to be worse in CHEK2 carriers when compared to non-carriers [49]. Larger studies are required to confidently compare NACT sensitivities of CHEK2-associated ER + BCs to other BCs. We speculate that the absence of significant immune cell infiltration may also impact response rates to NACT.

Guidelines recommend annual mammography with consideration for magnetic resonance imaging starting at age 40 in patients with CHEK2 P/LPVs. However, unlike for patients with germline BRCA1 or BRCA2 alterations, there are currently no guidelines regarding risk-reducing mastectomy in CHEK2 carriers, and decision making is generally based on individualized and family risk [17]. There is a paucity of data evaluating recurrence rates after breast conservation versus unilateral or bilateral mastectomy in CHEK2 carriers with BC, although some studies have suggested a higher risk of contralateral BC [32, 50, 51]. In our study, only 11 patients with BC opted for breast conservation, and none experienced local recurrence. On the other hand, we found higher than expected rates of bilateral and multifocal tumors compared to the general population, which is consistent with prior studies [52] and has implications arguing against breast conservation. Aside from CHEK2 germline status and type of CHEK2 variant (i.e. c.1100delC vs other), patient age at presentation, family history, and patient preferences, should also be considered when planning surgical intervention and surveillance for these patients.