The prevalence of homologous recombination deficiency (HRD) in various solid tumors and the role of HRD as a single biomarker to immune checkpoint inhibitors

Patients with pathologic confirmation of advanced gastrointestinal, GU, or rare cancers at Samsung Medical Center between Oct 2019 and Mar 2020 (n = 501), were prospectively tested for molecular aberrations, including TMB, with the TruSight^™ Oncology 500 assay. All study participants provided written informed consent before study entry. The following clinicopathologic characteristics were collected for all patients: age, sex, primary tumor site, number of metastatic sites, site of metastasis, treatment, and survival. The study protocol was approved (#2020-11-151) by the Institutional Review Board of Samsung Medical Center (Seoul, Korea) and was conducted in accordance with the ethical principles of the Declaration of Helsinki and the Korea Good Clinical Practice guidelines. All patients provided written informed consent before enrollment. Written informed consent included disclosure of information, competency to make a decision, and voluntary nature of the decision for the purpose, benefit, and potential risk of this study.

Samples for analysis were collected from 501 solid tumors and prepared as formalin-fixed paraffin-embedded (FFPE) material. The samples were gathered through biopsy at diagnosis, surgical specimen, or repeat biopsy at the time of disease progression; all were obtained before immunotherapy. The types of samples used in the analysis were as follows: biopsied samples (n = 320, 63.9%) and surgically resected samples (n = 181, 36.1%).

Forty (40) ng of DNA was quantified with the Qubit dsDNA HS Assay (Thermo Fisher Scientific) on the Qubit 2.0 Fluorometer (Thermo Fisher Scientific) and then sheared using a Covaris E220 Focused-ultrasonicator (Woburn, MA, USA) and the 8 microTUBE–50 Strip AFA Fiber V2 following the manufacturer’s instructions. Treatment time was optimized for FFPE material. The treatment settings were as follows: peak incident power (W): 75; duty factor: 15%; cycles per burst: 500; treatment time (s): 360; temperature (°C): 7; and water level: 6. For DNA library preparation and enrichment, the TruSight™ Oncology 500 Kit (Illumina) was used following the manufacturer’s instructions. Post-enriched libraries were quantified, pooled, and sequenced on a NextSeq 500 (Illumina Inc., San Diego, CA, USA). The quality of the NextSeq 500 (Illumina) sequencing runs was assessed with the Illumina Sequencing Analysis Viewer (Illumina). Sequencing data were analyzed with the TruSight^™ Oncology 500 Local App Version 1.3.0.39 (Illumina), a comprehensive tumor profiling assay designed to identify known and emerging tumor biomarkers, including small variants, splice variants, and fusions. The reads were aligned to the reference genome (GRCh37/hg19) using Burrows − Wheeler Aligner-MEM (BWA-MEM) (Li 2013). Poorly mapped reads with a mapping quality (MAPQ) below 20 were removed using Samtools version 1.3.1(Li et al. 2009). Somatic mutations including single-nucleotide variants (SNV) and small insertions and deletions (INDELs) were detected by the Pisces and Psara (Dunn et al. 2019). The rest of pipeline are as follows: CRAFT for copy number variation, TmbRaider for TMB, Hubble for MSI, STAR for RNA alignment, and Manta for fusion calling (Pestinger et al. 2020). Outputs of data, exported from The TSO 500 pipeline (Pestinger et al. 2020) were annotated with Ensembl Variant Effect Predictor (VEP) Annotation Engine, with information from the databases, such as dbSNP, gnomAD genome and exome, 1000 genomes, ClinVar, COSMIC, RefSeq, and Ensembl. The processed genomic alterations were categorized with four-tier system by American Society of Clinical Oncology and College of American Pathologists (Li et al. 2017), annotated with proper reference. The following criteria were used to filter our less significant variants and possible germline variants: (i) variants < 5% allele frequency and < 100 × read depth at the variant were excluded; (ii) variants previously reported to be benign or likely benign in the ClinVar archive (Landrum et al. 2016) were excluded; (iii) variant with a frequency greater than 1% in gnomAD (Karczewski et al. 2020) were excluded.

Importantly, the TruSight^™ Oncology 500 measures homologous recombination deficiency (HRD). The HRD-related genes were as follows: ARID1A, ATM, ATRX, BAP1, BARD1, BLM, BRCA1, BRCA2, BRIP1, CHEK1, CHEK2, FANCA, FANCC, FANCD2, FANCE, FANCF, FANCG, FANCL, MRE11A, NBN, PALB2, PTEN, RAD50, RAD51, and RAD51B. Homologous recombination deficiency was diagnosed if there was at least one HR-related gene mutation.

All statistical analyses were conducted with SPSS statistics 27. Descriptive statistics are reported as proportion and median. Data are presented as number (%) for categorical variables. Correlations between status of HRD and clinicopathologic features were analyzed by t test, Fisher’s exact test, or one-way analysis of variance (ANOVA), as appropriate. Response categories were assessed according to RECIST 1.1 (Eisenhauer et al. 2009; Schwartz et al. 2016). Objective response rate (ORR) was defined as the percentage of patients with complete response (CR) or partial response (PR). Progression-free survival (PFS) was defined as the interval between the initiation of the treatment and the time of progressive disease (PD). Logistic regression analysis was performed to analyze HRD genes that might be related to treatment response. A Cox regression model was used to analyze the associations of suspecting factors, including HRD and disease progression after ICIs treatment. The Mann–Whitney test was used to compare the difference between HRD and non-HRD. Kaplan–Meier estimates and log-rank tests were used in analysis of all time to event variables, and 95% confidence interval for the median time to event was computed.

Table 1 presents the clinical characteristics of the 501 patients included in this study. The median age of the patients was 59.7 years (range, 21–86), and the majority were male (60.3%). The median age of males was 61 years, while that of female was 58 years. The most frequent tumor type was colorectal cancer (n = 151, 30.1%), followed by gastric cancer (n = 116, 23.2%), sarcoma (n = 60, 12.0%), pancreatic cancer (n = 42, 8.4%), genitourinary (GU) cancer (n = 25, 5.0%), other gastrointestinal (GI) tract cancer (n = 22, 4.4%), melanoma (n = 21, 4.2%), hepatocellular carcinoma (HCC) (n = 12, 2.4%), and rare cancer (n = 4, 0.8%). Among the 501 patients, 65 had been treated with immune checkpoint inhibitors (ICIs). Figure 1 shows the distribution of TMB, MSI, and HR deficiencies. All seven patients with MSI were TMB-high and HR-deficient. On the other hand, of 375 patients with HR deficiency, only 54 were confirmed to be TMB-high or MSI.

Tumors with HRD were observed in 375 of 501 patients irrespective of type. Table 2 presents the status of the HRD and the ratio of patients who received ICI treatment according to tumor type. HR deficiency was observed in 74.9% of patients with various solid tumors including GU cancer (92.0%, 23 of 25), CRC (86.1%, 130 of 151), HCC (83.3%, 10 of 12), pancreatic cancer (76.2%, 32 of 42), biliary tract cancer (75.0%, 36 of 48), gastric cancer (69.0%, 80 of 116), sarcoma (65.0%, 39 of 60), melanoma (57.1%, 12 of 21), other GI tract cancer (AOV cancer, appendiceal cancer, cecal cancer, duodenal cancer, GIST) (50.0%, 11 of 22), and rare cancer (50.0%, 2 of 4). Figure 1 presents the distribution relationship with other biomarkers. All MSI were TMB-high and HR-deficient. However, some TMB-high have no HR deficiency. Figure 2A shows the percentage of confirmed HRD for each tumor type listed in order of high frequency rate. The tumor with the highest frequency of HRD was GU cancer with 92.0% and the lowest frequency was other GI tract cancer (AOV cancer, appendiceal cancer, cecal cancer, duodenal cancer, and GIST) and rare cancer at 50.0%. The distribution of HRD mutations for each cancer type is included in the supplement. (Supplement S1).

We also analyzed the observed genetic variations by HR-related genes. (Fig. 2B) The most frequently reported mutations were BRCA2 (n = 90), ARID1A (n = 77), ATM (n = 71), BARD1 (n = 67). On the other hand, FANCG was observed twice and BAP1 gene was reported only once. Even MRE11A gene was never observed in 501 patients’ NGS results.

Sixty-five patients treated with ICIs at diagnoses as follows: melanoma (95.2%, 20 of 21), HCC (33.3%, 4 of 12), rare cancer (25.0%, 1 of 4), GC (12.2%, 14 of 116), BTC (10.4%, 5 of 48), and sarcoma (5.0%, 3 of 60) (Table 2). We analyzed the correlation between HRD and efficacy to ICIs. Patients with HRD exhibited an objective response rate (ORR) of 27.3% (3 of 11), while patients without HRD achieved an ORR of 39.0% (16 of 41).

Progression-free survival (PFS) after ICIs was 6.5 months (95% CI 0.000–16.175) in patients without HRD and 4.1 months (95% CI 2.062–6.138) in patients with HRD. This difference was not significant (P = 0.441) (Fig. 3). Detailed data on HRD status and progression for each cancer type are included in the supplement (Supplement S2).

Additionally, we conducted Cox proportional hazard analysis for PFS after ICIs (Table 3). TMB was the only meaningful prognostic factor (P = 0.019). Response after immunotherapy was analyzed logistic regression and only TMB was revealed to be statistically significant (P = 0.004). It has been previously reported in our study that TMB by NGS panel is a useful predictor of immunotherapy (Kim et al. 2021).