Leveraging Real-World Genomic Data to Characterize Immunotherapy Response in Cancers with Germline CHEK2 Cancer Risk variants

Background
CHK2 is a DNA damage response (DDR) protein critical for maintaining genomic stability. Germline CHEK2 variants occur at 1-2% in the general population and confer elevated risk for multiple cancers. Loss-of-function mutations in DDR genes can sensitize tumors to immunotherapy through effects on tumor immunogenicity and immune signaling. We examined the relationship between germline CHEK2 variants and immune biomarkers across cancer types.

Methods
We used the Tempus Lens Platform (Tempus AI, Inc., Chicago, IL) to query the de-identified, multimodal Tempus database for patients (pts) with tumor-normal matched samples, Tempus xT (DNA, 648 genes) sequencing, at least 30% tumor purity, and any cancer diagnosis. Pts with known mismatch repair (MMR) deficiency or pathogenic/likely pathogenic germline variants (PGVs) in Tempus germline reportable genes other than CHEK2 were excluded to minimize confounding effects on immunogenicity. PD-L1 expression was assessed by IHC. Tumor immune microenvironment composition was evaluated using quanTIseq. Pt characteristics were compared using Chi-squared, Fisher’s exact, or Wilcoxon rank sum tests, as appropriate. Time-to-event analyses were restricted to pts with non-small cell lung cancer (NSCLC), with the index date defined as initiation of first-line (1L) treatment. Risk set adjusted real-world time to next treatment (rwTTNT) and real-world overall survival (rwOS) were analyzed by Kaplan-Meier and Cox proportional hazards models.

Results
The evaluable cohort included 89,326 pts (median age at diagnosis 65; 50% female) of whom 1,323 (1.5%) harbored CHEK2 PGVs. Pts with CHEK2 PGVs were more likely to self-identify as White (94% vs. 79%; p < 0.001) but were similar in age and sex distribution. No clinically significant differences in tumor mutational burden (TMB), PD-L1 expression, microsatellite instability, or immune cell infiltration were observed between pts with or without CHEK2 PGVs. ERBB2 amplification was enriched in pts with CHEK2 PGVs, and TP53 and RB1 somatic variants were enriched in pts without CHEK2 PGVs. Treatment response data were available for 3,567 (rwTTNT) and 3,588 (rwOS) pts with NSCLC, including 39 pts with CHEK2 PGVs. In univariable analyses, pts with CHEK2 PGVs had modestly prolonged rwTTNT (median not reached vs. 12.95 months; HR = 0.52; p = 0.039) and rwOS (median 32.35 vs. 18.64 months; HR = 0.78; p = 0.26). In multivariable analyses, including 1L ICI, PD-L1, and TMB, this trend continued for both rwTTNT and rwOS. However, no significant interaction terms were observed.

Conclusions
CHEK2 PGVs were not associated with known biomarkers of immunotherapy response. The observed prolonged TTNT in pts with NSCLC and CHEK2 PGVs highlights the need for further studies to clarify how tumors arising from CHEK2 PGVs differ biologically and prognostically from those with intact CHK2 function.