ATM (OMIM 607585) encodes the PI3K-related serine/threonine protein kinase comprising 3,056 amino acids, ataxia-telangiectasia mutated. This protein kinase primarily resides in the nucleus of dividing cells and has long been recognized as a key upstream modulator of the response to DNA double-stranded breaks, as well as oxidative and other genotoxic stresses. A critical function of ATM involves the recruitment and cooperation with DNA-damage sensing proteins such as BRCA1 in the face of double-strand breaks (1). ATM also plays a critical role in the DNA damage response through the phosphorylation of downstream substrates including CHK2 and p53, which then stimulate the cell-cycle checkpoint arrest and cellular apoptosis pathways, respectively (1, 2).

Biallelic pathogenic loss-of-function variants in ATM have long been associated with the genomic instability syndrome, ataxia-telangiectasia. Manifestations of the disorder include cerebellar ataxia, oculocutaneous telangiectasia, immunodeficiency, radiosensitivity, premature aging, and a predisposition to cancer development, primarily of lymphoid origin (3).

It was later recognized that women who are heterozygous for pathogenic loss-of-function variants in ATM have an increased risk for breast cancer, leading ATM to become regarded as a moderate penetrance breast cancer susceptibility gene, conferring a 2.3-fold increased risk for breast cancer compared to the general population (4, 5).

The risk of cancer among individuals heterozygous for ATM pathogenic variants has since been demonstrated for a broader range of malignancies. ATM pathogenic variant carriers appear to be at moderate-to-high increased risk for pancreatic, prostate, and gastric cancers and at low-to-moderate increased risk for ovarian and colorectal cancer as well as melanoma (6–9).

The association of ATM with cancer susceptibility has directly influenced clinical practice, from the creation of gene panels for molecular testing to cancer surveillance guidance and treatment recommendations. Here we present six novel cases of individuals with carcinomas of the gallbladder, uterus, duodenum, kidney, and lung as well as a sarcoma, all harboring germline ATM pathogenic variants, which suggests further expansion of the ATM-associated cancer susceptibility phenotype.

We reviewed the patients seen in our hereditary cancer clinic who had undergone comprehensive multigene panel testing and were found to be heterozygous for a germline pathogenic variant in ATM. A total of thirty-one patients met these criteria, none of which had a pathogenic variant in any other hereditary cancer gene. Twenty-five of these patients had a personal or family history of cancer that is consistent with the currently described spectrum of ATM-associated malignancies. Notably, the remaining six patients (19.4%) had a personal history of cancers not currently associated with ATM hereditary cancer susceptibility. Only one of these six patients was identified to harbor variants of uncertain significance in other hereditary cancer genes ( Supplementary Table 3 ). To evaluate for genotype-phenotype correlations, we compared the pathogenic ATM variants in the 6 atypical cancer cases to the 25 typical cancer cases, though no clear differences were evident ( Figure 1 ).

The personal and family history of cancer among the six patients with malignancies atypical for ATM-associated hereditary cancer susceptibility were then evaluated in detail ( Table 1 ). We observed that two of the six patients had a personal history of known ATM-associated cancers. Patient 1 had a history of breast adenocarcinoma, and Patient 6 had a history of melanoma. All six patients had at least one 1^st or 2nd-degree relative with ATM-associated cancer including breast, pancreatic, prostate, gastric, and colorectal adenocarcinomas. Four of the six patients meet current NCCN criteria for germline genetic testing based on either their personal history of cancer or family history of cancer ( Supplementary Table 4 ).

Here we describe six unrelated patients found to be heterozygous for germline, loss-of-function variants in ATM and diagnosed with at least one malignancy that is not currently associated with the ATM cancer susceptibility syndrome, including gallbladder carcinoma, uterine carcinoma, renal cell carcinoma, duodenal adenocarcinoma, lung adenocarcinoma, and a vascular sarcoma. These atypical cancers were diagnosed in a significant proportion of patients with ATM-associated hereditary cancer susceptibility in our cohort. Nevertheless, our review of the literature indicates that the prevalence of germline ATM pathogenic variants among individuals diagnosed with these atypical cancers appears to be less than 1% in most cases.

Exploration of somatic alterations in The Cancer Genome Atlas (TCGA) datasets demonstrated there was a high frequency of pathogenic alterations in ATM relative to other tumor suppressors associated with hereditary cancer syndromes, suggesting that ATM loss may contribute to oncogenesis within these cancer types. Furthermore, our analysis reveals that alterations in ATM tend to co-occur with pathogenic alterations in BRCA1 and CHEK2 while being mutually exclusive of TP53. This latter finding underscores the shared functional pathway between ATM and TP53 and reinforces prior work showing that DNA damage creates selection pressure for the inactivation of p53 that can be abrogated through the loss of ATM (39, 40). More broadly, these pathway interactions illustrate how germline ATM pathogenic variants may influence the acquisition of somatic alterations throughout cancer initiation and progression.

The cases reported in this series exemplify how the tumor spectra associated with cancer susceptibility syndromes are broadening through the clinical use of pan-cancer panels. The germline pathogenic variants in ATM detected in all six of our patients were discovered through comprehensive hereditary cancer panels. This testing approach is supported by studies demonstrating significantly higher detection rates compared to panels oriented specifically to the genes that are most relevant to the personal or family history of cancer (41–43). The heterogeneity in cancers documented among our patients with germline ATM variants and their families further emphasizes the value of pan-cancer panels, and certainly the importance of including ATM in multi-gene panels testing for hereditary cancer predisposition.

Moreover, our cases illustrate the way NCCN Guidelines may significantly limit the detection of hereditary cancer susceptibility, as many cancer types, particularly those with a low prevalence, do not currently qualify individuals for genetic testing. In our series, neither Patient 4 nor Patient 6 met guidelines for hereditary cancer testing based on their personal or family history of cancer, illustrating circumstances in which ATM germline variants may go undetected. This was underscored in a recent study where germline genetic testing in an unselected pan-cancer patient population found that over half of the patients who tested positive for pathogenic variants were not eligible for hereditary cancer testing based on current guidelines (44). Improving the detection of ATM-associated hereditary cancer is clinically meaningful as it can enable cascade genetic testing for family members who are at risk, facilitate appropriate cancer surveillance for those affected, and prompt the use of precision-based therapies.

Although no specific ATM genotype-phenotype correlations were discerned in this study, we did observe that the ATM c.1564_1565delGA variant was present in two patients with uterine cancer, and the ATM p.V2716A variant was present in two patients with lung adenocarcinoma. In addition, we found the germline ATM variant c.6679C>T identified in Patient 5 with epithelioid hemangioendothelioma was also seen in a case of endometrioid uterine cancer, and the germline ATM variant c.4236+1G seen in Patient 6 with lung adenocarcinoma was also present in a case of renal cell carcinoma. Meanwhile, none of the pathogenic germline ATM variants identified in our six patients overlapped among the somatic loss-of-function ATM variants present in the cancer sequencing datasets. Larger sample sizes will be needed to systematically investigate whether specific germline ATM variants confer a greater risk for atypical ATM cancers.

A limitation of this study is that we cannot exclude the possibility that our patients’ malignancies were incidental to, rather than caused by the pathogenic variant in ATM, as the incidence of heterozygosity for pathogenic variants in ATM has been estimated in gnomAD for European populations to be 0.48% (45). A segregation analysis of our patients’ families may have strengthened the association between atypical cancers and pathogenic variants in ATM, as this would have revealed whether the gallbladder carcinoma in the maternal aunt of Patient 1, the leukemias in the maternal uncle and grandfather of Patient 4, or the anaplastic astrocytoma in the paternal nephew of Patient 6 were also associated with pathogenic variants in ATM. Moreover, we were not able to confirm whether ATM loss of heterozygosity occurred in our six patients, which could have provided further evidence that ATM is driving carcinogenesis based on the two-hit hypothesis (46). Yet, this does not appear to be a reliable finding as the frequency of biallelic two-hit events across all cancers where a germline pathogenic variant is present has been estimated at 38.3%, with a recent study finding that ATM somatic second hits occurred in 11.3% in a wide spectrum of cancers including breast, pancreatic, bladder, uterine and lung cancers, among others (15, 47). A case-control analysis for each of the atypical-ATM cancer types could have also provided compelling evidence of an association. This has been previously performed in two cohorts of patients with lung adenocarcinoma, demonstrating a significant enrichment for germline pathogenic variants in ATM in cases compared to controls (48, 49). Unfortunately, the feasibility of case-control studies is limited by the rarity of many of these cancers, as well as the low penetrance of ATM in association with these atypical cancers.

Nevertheless, our finding of relatively frequent somatic ATM alterations across many of the atypical cancer types has been corroborated by an analysis showing that ATM was among the most frequently mutated genes in the DNA-damage response pathway across 33 distinct types of cancers (50). This evidence has also been bolstered by a recent study demonstrating a high prevalence of germline pathogenic variants in ATM relative to other hereditary cancer genes across a spectrum of cancers lacking testing guidelines, including bladder, brain, esophagus, and head and neck cancers (15). Furthermore, our identification that pathogenic ATM alterations in these atypical cancers are associated with the co-occurrence of alterations in BRCA1 and CHEK2, and mutual exclusivity with TP53, suggests that cancers with a germline pathogenic variant in ATM may depend on specific molecular alterations that confer targetable vulnerabilities. This has already been documented in patients with metastatic prostate cancer harboring a germline pathogenic variant in ATM sensitizing to PARP inhibition, as well as in patients with pancreatic cancer where monoallelic pathogenic variants in ATM produce susceptibility to combined therapies that synergistically target the DNA-damage response pathway (51, 52).

In conclusion, this study combines detailed clinical phenotyping with a comprehensive review of relevant germline ATM reports and large-scale tumor sequencing data to propose that germline pathogenic variants in ATM may be associated with the development of cancers of the gallbladder, duodenum, uterus, kidney, and lung as well as sarcoma cancers. Consideration should be given to a broadening of the ATM-cancer susceptibility syndrome phenotype within hereditary cancer testing guidelines, as this has the potential to improve the detection of affected patients and facilitate the use of more effective cancer treatments.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material .

The studies involving human participants were reviewed and approved by University of Miami institutional research board (IRB #20081166). Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

NB: Conceptualization, methodology, data curation, formal analysis, project administration, visualization, writing–original draft, and writing–review and editing. RS-S: Conceptualization, investigation, resources, data curation, writing–review and editing. MH: Investigation, resources, data interpretation, writing–review and editing. DP: Investigation, resources, and writing–review and editing. DS: Investigation, resources, writing–review and editing. MT: Conceptualization, methodology, formal analysis, investigation, writing–original draft, and writing–review and editing, visualization, supervision. All authors contributed to the article and approved the submitted version.