Allele-specific HLA LOH in solid tumors: distinct patterns by tumor type and potential prognostic relevance

Abstract

T-cell receptors (TCRs) recognize antigens derived from fragments of somatically expressed proteins that are degraded by the proteasome and presented by specific human leukocyte antigen (HLA) molecules. Recent therapeutic advances using the TCR as a tumor-targeting moiety have focused attention on HLA loss of heterozygosity (LOH) as a potential resistance mechanism. Allele-specific LOH, rather than allele-agnostic, is particularly pertinent, but rarely evaluated. Using a real-world dataset comprising 78,418 cases, we demonstrate that allele-specific LOH occurs at a relatively low frequency (<10%) across all patients with cancer. We observed a modest increase in allele-specific HLA LOH in cancers harboring associated neoantigen driver mutations (eg, KRAS or TP53 mutations), but the overall frequency remained consistently low. Furthermore, using an orthogonal dataset, we integrated clinical outcomes with HLA LOH and identified distinct impacts on overall survival in colorectal cancer (CRC) and non-small cell lung cancer (NSCLC) cohorts. For instance, A*02:01-specific LOH was linked to worse survival in CRC (HR 0.5355, 95% CI 0.2991 to 0.9589, p=0.0094) but showed a trend toward improved survival in NSCLC (HR 1.249, 95% CI 0.7778 to 2.005). These findings underscore the relevance of allele-specific HLA LOH assessments and reveal nuanced differences in its clinical implications, which should be accounted for in the optimization of TCR-based immunotherapies.

Keywords: Adoptive cell therapy - ACT; Colorectal Cancer; Human leukocyte antigen - HLA; Immunotherapy; Non-Small Cell Lung Cancer.

Figure 1. Gene-specific and allele-specific HLA LOH in solid tumors. (A) Differences in HLA LOH between classical HLA class I genes in the indicated solid tumor types in the FMI dataset; sample size for each group is shown. (B) Germline homozygosity rates of any allele type at each HLA class I gene across the FMI dataset. (C) Allele-specific HLA LOH frequencies in relation to sample sizes across all indications as well as within CRC, NSCLC, and PC. Simple linear regression line, ±95% prediction bands are shown. HLA-A, B, and C are color-coded; dashed blue line indicates allele-agnostic LOH. (D) Allele-specific HLA-A*02:01, A*03:01 and A*11:01 LOH in relation to allele-agnostic HLA LOH across indications with n≥300. Green dashed lines indicate half of allele-agnostic HLA LOH; blue line indicates allele-agnostic LOH across all indications. Total sample size for each ontology is shown. CNS, central nervous system; CRC, colorectal cancer; CUP, unknown primary carcinoma; FMI, Foundation Medicine, Inc; GIST, gastrointestinal stromal tumor; HLA, human leukocyte antigen; LOH, loss of heterozygosity; NSCLC, non-small cell lung cancer.

Figure 2. HLA LOH in relation to KRAS and TP53 driver mutations in solid tumors. (A and B) Allele-agnostic and allele-specific HLA LOH of frequent HLA-A allele plus presented (A) KRAS and (B) TP53 neoantigen pairs across all cancer types and within common indications in the FMI dataset. Sample sizes for each subgroup are indicated on the graphs. Given the variability between indications and potential functional relevance, statistical significance was assessed within specific ontologies; *p≤0.05, **p≤0.01, ***p≤0.001. CRC, colorectal cancer; FMI, Foundation Medicine, Inc; HLA, human leukocyte antigen; LOH, loss of heterozygosity; NSCLC, non-small cell lung cancer.

Figure 3. Impact of HLA LOH on outcomes in the GENIE/MSK dataset. (A and B) Kaplan-Meier curves showing impact of allele-agnostic HLA LOH on OS probabilities in (A) CRC and (B) NSCLC from the time of diagnosis and advanced disease; plots were truncated at 15 and 10 years, respectively. (C and D) HRs (±95% CIs) of allele-agnostic and allele-specific HLA LOH in (C) CRC and (D) NSCLC from the time of diagnosis and advanced disease. An HR below 1 indicates a trend toward poorer survival, while an HR above 1 indicates a trend toward better survival. Diagnosis versus advanced allele-specific HRs were compared using paired t-test. *p≤0.05, **p≤0.0, ***p≤0.001. (E and F) Kaplan-Meier curves showing impact of representative (A*02:01 and A*11:01) allele-specific HLA LOH on OS probabilities in (E) CRC and (F) NSCLC from the time of diagnosis and advanced disease; plots were truncated at 15 and 10 years, respectively. CRC, colorectal cancer; HLA, human leukocyte antigen; LOH, loss of heterozygosity; NSCLC, non-small cell lung cancer; OS, overall survival.