An estimated 228,150 individuals in the United States were diagnosed with lung cancer in 2019, with non-small cell lung cancer (NSCLC) accounting for approximately 85% of these patients [1]. As the availability of targeted therapies for patients with NSCLC has increased, the importance of histologic subtypes of NSCLC has waned, while categorization of these tumors by genomic and protein expression has become impactful in determining the therapy predicted to have the highest clinical benefit.

The increasing prominence of diagnostic assays used to identify predictive biomarkers accompanying specific therapies, known as companion diagnostics (CDx), is evidenced by the acceleration of oncologic drug approvals with a CDx requirement in recent years [2]. The 2019 National Comprehensive Cancer Network (NCCN) guidelines highly recommend testing of patients with NSCLC for EGFR and BRAF mutations; ALK, ROS1, and NTRK rearrangements; and programmed death-ligand 1 (PD-L1) expression levels; and recommends RET rearrangement; ERBB2 and MET mutations; MET amplifications [3]. All of these genomic alterations are potentially targetable based on positivity of these alterations. For examples, EGFR mutations have been shown to have sensitivity to EGFR tyrosine kinase inhibitors (TKI) such as erlotinib, and gefitinib; BRAF mutations have been shown to be sensitive to dabrafenib plus trametinib in metastatic nonsquamous NSCLC; ALK and ROS1 rearrangements have been shown to respond well to numerous tyrosine kinase inhibitors (TKI) such as crizotinib and certinib; NTRK fusions have been responsive to larotrectinib and entrectinib; and responsiveness to multiple therapies have also been found in numerous clinical studies of patients with RET rearrangement; ERBB2 and MET mutations; and MET amplifications [3]. In addition, although KRAS is typically not a biomarker used to identify patients eligible for targeted therapy, KRAS mutations have been associated with lack of response for EGFR TKIs and promising targetable therapies to KRAS mutations such as G12C are being investigated [4, 5].

The immunotherapy biomarker landscape is increasingly complex and changing rapidly. In NSCLC, DAKO 22C3 was the first assay approved as a CDx for pembrolizumab in NSCLC at a tumor portion score (TPS) of 50, and the CDx cut-off has since been lowered to a TPS of ≥1 [6]. In addition, in May 2020, DAKO 28-8 was approved as a CDx for opdivo and VENTANA SP142 was approved as a CDx for atezolizumab in NSCLC. Other relevant biomarkers for immunotherapy eligibility include MSI (microsatellite instability) testing and tumor mutational burden testing (TMB), both of which can be measured by comprehensive genomic profiling (CGP). For MSI testing, microsatellite instability-high (MSI-H) status or loss of MMR (mismatch repair) markers by immunohistochemistry allows a patient to be potentially eligible for pembrolizumab based on MSI being a pan-tumor companion diagnostic for pembrolizumab [7]. In June 2020, the FDA approved the FoundationOne®CDx Assay for pembrolizumab at a TMB ≥ 10 mutations/Mb cut-off in solid tumors based on the KEYNOTE-158 clinical trial [8, 9]. Adding to the complexity are potential resistance biomarkers such as STK11, KEAP1, and EGFR for checkpoint inhibitors (CPIs) [10–12].

In this study, we performed a survey of biomarkers used in NSCLC real-world samples assessed by comprehensive genomic profiling (CGP) combined with PD-L1 immunohistochemistry (IHC) to understand the landscape of the genomic and protein expression biomarkers in NSCLC. In addition, we examined the relationship between PD-L1 and TMB to better understand the interplay of these biomarkers in the context of immunotherapy.

The use of CGP (TMB^{>10mutations/Mb}) and PD-L1 IHC^TPS≥50% in NSCLC identified at least one positive NCCN recommended biomarker for 70.5% (6,661/9,450) of patients in this cohort. Among the remaining patients without a positive NCCN recommended biomarker (n = 2,789), 86.7% (2,419) had an alteration which could be considered actionable or allow them to be potentially eligible for a biomarker-associated clinical trial. In sum, this means that 96.1% (9,080/9,450) of patients tested had a positive result in at least one actionable or potentially actionable biomarker. For immunotherapy eligibility, by examining TMB^{>10mutations/Mb} and PD-L1^TPS≥50%, an additional 2,271 patients were potential immunotherapy candidates when compared to PD-L1^TPS≥50% alone. These findings underscore the utility of combining CGP with IHC to expand potential eligibility for biomarker-associated therapies and clinical trial enrollment to most patients with NSCLC. Further investigation is necessary to determine if certain combinations of PD-L1 tumor cell expression and TMB status (i.e. PD-L1^TPS≥50% and TMB^{>10mutations/Mb}) are associated with better response to immunotherapy when compared to a single biomarker test and its correlation with potential resistant immunotherapy biomarkers such as STK11, KEAP1, and EGFR.

In the PD-L1^TPS≥50% disease subset, BRAF mutations, MET mutations, MET amplifications, and KRAS mutations were significantly enriched. As there are available targeted therapies for BRAF mutations and MET genomic alterations, more clinical studies need to be conducted to assess the efficacy of combining immunotherapy with BRAF and/or MET inhibitors. Consistent with the literature, we saw evidence of high correlation between KRAS and PD-L1 expression [18]. As mentioned above, KRAS mutations have been associated with lack of response for EGFR TKIs and due to the paucity of KRAS targeted therapies in the past, KRAS mutant tumor have been hard to treat with targeted therapies. However, KRAS G12C is now a biomarker with promising targeted therapies (ClinicalTrials.gov Identifier: NCT03600883, NCT04006301, NCT03785249) and so immunotherapy combined with KRAS inhibitors should be explored further. In the PD-L1^TPS<50% cohort, EGFR mutations, ERBB2 mutations, STK11 mutations, and KEAP1 mutations were enriched. EGFR, STK11, and KEAP1 mutations are known potential resistance mutations for immunotherapy and the increased prevalence of these mutations in the PD-L1^TPS<50% cohort could suggest a reduced response to immunotherapy. In addition, while Guisier et al. demonstrated no decrease in immunotherapy response in ERBB2 mutated NSCLC in a small cohort of patients, larger clinical studies need to be performed to further evaluate the role of ERBB2 mutations in immunotherapy response [19].

One limitation of this study is that as a reference laboratory, clinical information including treatment history is not available for our patient cohort. However, in general, the samples we receive represent real-world samples often acquired from later stage diseases and that should be taken into consideration when interpreting the data. In addition, since we are a US based laboratory, most of samples are from patients of European ancestry. The rates of the NCCN guideline highly recommended biomarkers in our cohort were similar to the literature. For example, in our cohort, the rate of EGFR mutations was 14.0% (1,322/9,450), while in the literature the rate of EGFR mutations is approximately 15% for patients of European ancestry [20]. Of note, our PD-L1 expression rates were similar to certain pembrolizumab clinical trials rates of PD-L1 expression. For example, our TPS ≥ 50% prevalence was 30.5% compared to the 30.2% of the KEYNOTE-024 study [13].

In conclusion, these findings highlight the improved clinical utility of combining CGP with IHC to expand the biomarker-guided therapeutic options available for patients with NSCLC, relative to single biomarker testing alone.