Incidence of Germline ATM Variants in a Consecutive Clinical Cohort of CLL Patients

Background

The pathogenesis of chronic lymphocytic leukemia (CLL) remains unknown, but first-degree relatives of affected patients (pts) have a 3-to-8-fold increased risk of developing CLL, suggesting an inherited component. Previously, we performed an exome-wide comparison of rare germline variants (vts) among CLL pts compared to controls and identified a 1.8X increased risk of any rare germline ATM vt in CLL pts. Certain vts had much higher relative risk; for example, ATM p.L2307F was associated with a 10X increased risk of CLL (Tiao Leukemia 2017). ATM p.F858L and p.P1054R also imparted a 2X increased risk of CLL in a candidate gene association study (Rudd Blood 2006). Most ATM missense vts have not been fully characterized and are classified as vts of uncertain significance (VUS) in clinical testing. We observed that ATM missense vts were common in our CLL clinic and sought to formally test the hypothesis of association by examining the frequency of ATM missense vts identified by next-generation-sequencing (NGS) routinely sent on hematologic malignancy pts seen at our institution. These pts served as a population with well-characterized diagnoses whose genetics had been uniformly determined by a CLIA-certified lab, allowing us to directly ask whether ATM VUS are enriched in CLL compared to other hematologic malignancies, and whether CLL characteristics differ between pts with and without ATM VUS.

Methods

The Brigham and Women's Hospital Rapid Heme Panel (RHP) is an NGS assay that interrogates 95 cancer-related genes, including the entire coding sequence of ATM. RHP results were aggregated for all 999 pts who had the test sent from the DFCI Lymphoma clinic (excluding T-PLL) between 7/1/2014 and 5/25/2018. RHP data were also aggregated for 876 pts seen by 3 physicians over the same time in the DFCI Leukemia clinic (acute leukemia and other myeloid disorders). Only vts with a total population allelic frequency (PAF) of <1% are included in RHP results, restricting our analysis to less common vts. Statistical analysis was performed in R.

Results

Out of 384 pts with CLL or monoclonal B lymphocytosis, 99 pts (25.8%, 95% CI 21.5-30.5%) had at least one ATM vt. 71 different ATM vts were seen, with 10 (14%) deemed pathogenic (nonsense mutations, internal insertions or deletions). All remaining 61 vts were missense vts, and the majority were likely germline: 43 are in the gnomAD germline vt database, and of the other 18, 8 had allele frequencies in blood between 40 and 60%. The most common vts were p.S707P (n = 11 pts), p.L2307F (n=9), p.F858L (n=8), and p.D1853V (n=8), all of which are known to be germline.

ATM missense vts were more frequent in CLL (97 pts with missense vts out of 384 total, 25.3%) than in non-CLL lymphomas (91 out of 615, 14.8%, p=0.00006). No other individual lymphoma type had a higher frequency of ATM missense vts than CLL. Missense vts were also more frequent in CLL than in pts seen in the myeloid malignancy clinic (143 out of 876, 16.3%, p=0.0003). When restricting analysis to missense vts with PAFs reported in gnomAD, we found a higher percentage of extremely rare vts (PAFs <0.001%) in CLL pts (7 out of 42 vts, 16.7%) compared to the myeloid malignancy group (2 out of 55, 3.6%, p=0.04); the non-CLL lymphoma pts had a similar frequency of extremely rare vts (5 out of 39, 12.8%, p=0.12). Consistent with a biologic role for these vts, 32% of CLL pts with missense vts had 11q-deleted disease, while only 10.5% of pts without missense vts had 11q-deleted disease (p=3*10^-6). Pts with missense vts were also significantly less likely to have TP53-aberrant disease, 10.9% vs. 20.7% (p=0.04), consistent with a mutually exclusive role for ATM and TP53. There was no difference in IGHV mutation status between CLL pts with and without missense vts (56.6% vs 53.4% unmutated, respectively).

Conclusion

One in 4 CLL pts has an ATM missense vt; these are more frequent in CLL than in other lymphoid or myeloid disorders. Most of these are germline and have previously been classified as VUS. CLL pts with missense vts have a higher frequency of 11q-deleted disease. These results highlight an important role for germline ATM missense vts as contributors to the inherited risk for developing CLL and emphasize the importance of analyzing VUS and constitutional data generated by NGS assays, the latter typically not reported by most laboratories.