Distinct genomic and immunophenotypic features of solid-predominant versus nonsolid-predominant stage I lung adenocarcinomas and association with disease recurrence after surgical resection

8514

Background: Compared to lung adenocarcinomas (LUAD) with nonsolid-predominant histology (lepidic, acinar, papillary, micropapillary), those with predominantly solid features have a higher risk of disease recurrence after surgical resection. However, little is known about the genomic landscape and immunophenotype of solid vs nonsolid stage I LUAD. Methods: We collected clinicopathologic data from patients with resected stage I NSCLC (AJCC 8th Edition), which underwent next-generation sequencing to identify genomic alterations and tumor mutational burden (TMB). A subset of these samples also had multiplexed immunofluorescence for CD8+, FOXP3+, PD-1+, and PD-L1 to determine differences in tumor immune cells subsets according to histologic subtype. Disease free-survival (DFS) was compared in patients based on their predominant histologic subtype (solid vs nonsolid). Results: Among 658 LUADs, 11.4% (N = 75) had solid-predominant and 88.6% (N = 583) nonsolid-predominant histology. After a median follow-up of 50 months from the time of surgery, 145 patients (22.0%) experienced recurrence. Compared to nonsolid-predominant LUAD, those with solid predominance had a significantly lower prevalence of activating EGFR, BRAFV600E, and METex14 mutations as well as ALK/ RET/ ROS1 rearrangements (9.3% versus 31.6%, P < 0.001), no difference in KRASG12C frequency (24% versus 16.8%, P = 0.14), a higher TMB (median 12.2 versus 7.2 mutations/megabase; P < 0.001), and a shorter median DFS from the time of surgical resection (43.2 months versus not reached, HR: 3.3 [95% CI: 2.2-4.9], P < 0.001). The detrimental effect of solid-predominant LUAD in DFS remained significant after adjusting for other factors such as tumor stage, surgery type, smoking status, and TMB (HR: 2.66 [95% CI: 1.71-4.11], P < 0.001]. Among LUADs profiled by multiplex immunofluorescence, compared to tumors with nonsolid-predominant subtype (N = 197), those with solid predominance (N = 23) had significantly higher numbers of CD8+, FOXP3+, PD-1+ immune cells, and PD-1+ CD8+ T cells, both intratumorally (P < 0.001) and at the tumor-stroma interface (P < 0.001). Solid-predominant subtype was also associated with a higher median PD-L1 expression level on tumor (5% versus 1%; P = 0.01) and immune cells (16% versus 7%, P = 0.02). Conclusions: Among patients with surgically-resected stage I LUAD, solid-predominant histology was associated with distinct genotypic and immunologic characteristics. These findings may aid in identifying patients at greater risk of recurrence after surgery.

Impact of STK11 copy loss on clinical outcomes to PD-(L)1 blockade in non–small cell lung cancer

9059

Background: STK11 is among the most commonly altered genes in non-squamous lung cancers. While STK11 mutation is associated with diminished efficacy of immune checkpoint inhibition (ICI), particularly in KRAS mutated tumors, it is not known whether STK11 copy deletion influences outcomes to ICI. Methods: Patients with advanced non-squamous non-small cell lung cancer (NSCLC) treated with ICI whose tumors underwent genomic profiling were included. Clinical outcomes to ICI were analyzed according to KRAS mutation and STK11 deletions. STK11 copy number variations (CNVs) were determined using an internal informatic pipeline. Kaplan-Meier methodology was used to estimate event-time distributions. Results: Of 559 patients with non-squamous NSCLCs (Nsq-NSCLC), 40.4% (N = 226) had a KRAS mutation ( KRASmut), 18.4% (N = 103) had an oncogenic STK11 mutation ( STK11mut), and 22.5% had either single (N = 123), or bi-allelic (N = 3) deletion ( STK11del). Given that 32.5% of STK11del cases had a concurrent oncogenic STK11 mutation in our cohort, to isolate the impact of STK11del on ICI outcomes we excluded samples with STK11mut from this analysis. In all comers with NSCLC, STK11del had no impact on objective response rate (22.2% versus 23.8%, P = 0.8), progression-free (PFS, HR 0.90, P = 0.30), and overall survival (OS, HR 0.96, P = 0.79) to ICI. When we examined the impact of STK11del on clinical outcomes to PD-(L)1 blockade among KRASmut cases we found that STK11del was associated with a numerically lower ORR (13.3% versus 30.0%, P = 0.12), and a significantly shorter PFS (HR 0.57, P = 0.018) compared to cases without STK11del. No difference in OS were observed between these groups (HR 0.77, P = 0.39). Among KRASwt NSCLCs, STK11del cases had a similar ORR (22.6% versus 22.9%, P = 0.99), PFS (HR 0.92, P = 0.63), and OS (HR 1.18, P = 0.32) to PD-(L)1 inhibition compared to cases without STK11del. Among KRASmut but not KRASwt NSCLCs, cases with STK11del had significantly lower PD-L1 expression compared to those without STK11 deletions (27.5% versus 70%, P = 0.01). Conclusions: STK11 deletion is associated with low response rate and short progression-free survival among KRAS mutant NSCLCs. Future analyses will incorporate additional cases to increase sample size and immunopathologic findings to assess impact of mono and bi-allelic deletion on protein expression.

Activating MET kinase domain mutations define a novel molecular subtype of non–small cell lung cancer that is clinically targetable with the MET inhibitor elzovantinib (TPX-0022)

9124

Background: In non-small cell lung cancer (NSCLC), MET exon 14 skipping (METex14) mutations and MET amplification can be targeted with MET inhibitors. Here, we describe a novel, actionable molecular subtype of NSCLC characterized by activating MET-tyrosine kinase domain (MET-TKD) mutations in the absence of METex14 mutations. Methods: Clinicopathologic and genomic data were abstracted from NSCLC cases included in a multi-institutional cohort of tumors that underwent genomic profiling in the GENIE v10, China PanCancer, and the International Cancer Genome Consortium/ The Cancer Genome Atlas (ICGC/TCGA) datasets. External validation of the prevalence of MET-TKD mutations was performed on an independent cohort of NSCLC tissue and liquid samples from the Foundation Medicine genomic database. Results: Among 14,099 NSCLC samples in the multi-institutional cohort, 71 (0.5%) harbored MET-TKD mutations without concurrent METex14 mutations: 55 of these had uncertain pathogenic significance and 16 had known oncogenic potential, including MET H1094Y/R, D1228H/N/V, N1100S, H1106D, V1188I, and M1250T, in order of decreasing prevalence. In a separate cohort of 91,515 NSCLC samples from the Foundation Medicine database, MET-TKD mutations lacking concurrent METex14 mutations were identified in 799 (0.9%) samples, including H1094Y, L1195V, D1228H/N, M1250T and others. Among 60 NSCLC samples harboring MET-TKD mutations without concurrent METex14 mutations with complete genomic data in the multi-institutional cohort, 36 (60%) had concurrent driver alterations in KRAS, EGFR, ROS1, BRAF, HER2, or RET, while 24 (40%) had no concurrent oncogenic drivers. Among patients with available demographic data in the multi-institutional cohort, those with MET-TKD-mutant NSCLC (N = 70) were significantly younger than patients with METex14-mutant NSCLC (N = 353) (median age 63 [range 30-86] vs 73 [range 44-88], p < 0.0001), and there was no significant difference in sex or self-reported race. Confirmed partial responses to the MET tyrosine kinase inhibitor elzovantinib (TPX-0022) were observed in two patients with MET-TKD-mutant NSCLC and no other detectable driver mutations: a 64-year-old man with MET H1094Y-mutant NSCLC, and an 80-year-old man with MET F1200I-mutant NSCLC. Conclusions: Potentially actionable MET-TKD mutations lacking concurrent METex14 mutations represent a novel genomic subtype in 0.5-0.9% of NSCLC, and occur in the absence of other known drivers in a subset of cases.

Three-year outcomes and correlative analyses in patients with non–small cell lung cancer (NSCLC) and a very high PD-L1 tumor proportion score (TPS) ≥ 90% treated with first-line pembrolizumab

9043

Background: Although 1st-line PD-1 monotherapy has improved survival in advanced NSCLC with a PD-L1 TPS ≥50%, responses occur in ̃45% of patients (pts). We previously showed that among pts treated with 1st-line pembrolizumab, clinical outcomes were significantly improved in those with a PD-L1 TPS of ≥90% compared to a TPS of 50-89%. Here, we report the 3-year survival analysis to 1st-line pembrolizumab in pts with a PD-L1 TPS ≥90% vs 50-89%, and characterize genomic and immunophenotypic differences between these PD-L1 expression groups. Methods: Pts with stage IV EGFR/ALK wild-type NSCLC and PD-L1 TPS ≥50% who received 1st-line pembrolizumab at the Dana-Farber Cancer Institute (DFCI) and Memorial Sloan Kettering Cancer Center (MSKCC), with a minimum follow-up of 3 years were included. Comprehensive tumor genomic profiling and multiplexed immunofluorescence (mIF) were performed to examine genomic and immunophenotypic correlates of very high PD-L1 expression on separate cohorts of NSCLC at the DFCI. Results: Among 396 pts, median age was 69, 53.3% were women, 90.1% had a history of tobacco use, 83.6% had a ECOG performance status of 0-1, and 28.8% had a KRAS mutation. At a median follow-up of 42.6 months, median progression-free (mPFS) and overall survival (mOS) in the entire cohort were 5.1 months, and 19.0 months, respectively. When compared to pts with a PD-L1 TPS of 50-89% (N = 252), those with PD-L1 TPS ≥90% (N = 144) had a significantly longer mPFS (6.0 vs 4.5 months, HR 0.67, p < 0.001), and longer mOS (30.2 vs 16.9 months, HR 0.66, p < 0.01). Kaplan-Meier estimates of the 3-year PFS and OS were 24.9% and 47.0% in the PD-L1 TPS ≥90% groups, and 9.4% and 27.7% in the PD-L1 TPS 50-89% group, respectively. A PD-L1 TPS ≥90% was confirmed to be an independent predictor of improved PFS (HR 0.68, p < 0.01) and OS (HR 0.67, p < 0.01) in multivariable analysis. Tumor genomic profiling from a separate cohort of 500 NSCLC samples revealed that mutations in STK11, KEAP1, FBXW7, and CTNNB1 were significantly more frequent in tumors with a PD-L1 TPS of 50-89% compared to those with a PD-L1 TPS ≥90% (q < 0.05). mIF on 91 NSCLCs identified significantly higher CD8+PD1+ T cells and PD-L1+ immune cells in tumors with PD-L1 TPS ≥90% vs 50-89% (p < 0.05). Conclusions: Pembrolizumab monotherapy continues to demonstrate a meaningful long-term survival benefit in pts with advanced NSCLC and a PD-L1 TPS ≥90%. NSCLCs with very high PD-L1 TPS may have a more favorable genomic and immunophenotypic profile. These findings have implications for treatment selection and clinical trial interpretation and design.

Immunophenotypic correlates and response to first-line pembrolizumab among elderly patients with PD-L1-high (≥ 50%) non–small cell lung cancer

9054

Background: Older age is associated with increased levels of systemic inflammation and altered immunosurveillance in cancer. Whether aging correlates with a distinct immunophenotype or impacts clinical outcomes to first-line pembrolizumab in patients with advanced non-small-cell lung cancer (NSCLC) and a PD-L1 tumor proportion score (TPS) of ≥50% is unclear. Methods: We performed a retrospective analysis of patients with NSCLC. Multiplexed immunofluorescence (mIF) for CD8+, PD-1+, PD-1+CD8+ and FOXP3+ was performed to explore tumor immunophenotype. Clinical outcomes were analyzed on a separate cohort of patients with PD-L1-high (TPS of ≥50%) NSCLC (negative for sensitizing genomic alterations in EGFR and ALK) who received treatment with first-line pembrolizumab. Variables demonstrating a univariate signal of association of p < 0.1 were included in the multivariate model. The results were compared in patients < 80 vs ≥80 years old. Results: Among 541 patients with NSCLCs profiled by mIF, the median age was 67 (28-90). When comparing patients < 80y (n = 497) to ≥80y (n = 44), there was no difference in median CD8+ T cells/mm2 (171 vs 148; p = 0.69), PD-1+ immune cells/mm2 (81.1 vs 87.2; p = 0.95), or PD-1+CD8+ T cells/mm2 (18.0 vs 13.1; p = 0.56). NSCLCs from patients ≥80y had a higher median of intratumoral-associated FOXP3+ T cells/mm2 (63.6 vs 91.1; p = 0.03). In a cohort of 271 patients with PD-L1 ≥50% who received first-line pembrolizumab, baseline clinicopathological characteristics were balanced in the < 80y (n = 225) vs ≥80y (n = 46) groups in terms of sex, tobacco use, Eastern Cooperative Oncology Group-Performance Status (ECOG-PS), histology, presence of potentially targetable driver mutations (KRAS, MET, BRAF, HER2, RET), and PD-L1 TPS distribution (50-89% vs ≥90%). Compared to patients < 80y, patients ≥80y had no difference in objective response rate (ORR 39.1% vs 28.2%; p = 0.22) or median progression-free survival (mPFS 6.0 vs 3.0 months; p = 0.16). However, patients ≥80y had a shorter median overall survival (mOS 25.7 vs 7.6 months; p = 0.02), and this result remained significant after adjusting for ECOG-PS. Among those who experienced disease progression on pembrolizumab, patients ≥80y were significantly less likely to receive any second-line systemic therapy compared to patients < 80y (55.6% vs 30.8%; p = 0.008). Conclusions: In patients with NSCLC and PD-L1 ≥50%, the ORR and mPFS to first-line pembrolizumab were similar between patients < vs ≥80 years old. OS was shorter among patients ≥80y, potentially reflecting lower use of second-line therapy in elderly patients after progression on pembrolizumab. The immunophenotypic correlates of NSCLC in older patients need further investigation.

Genomic correlates of acquired resistance to PD-(L)1 blockade in patients with advanced non-small cell lung cancer (NSCLC)

9021

Background: Despite improvements in survival with immune checkpoint inhibition (ICI), the majority of patients develop acquired resistance to ICI after an initial benefit. However, the mechanisms underlying acquired resistance to ICI in NSCLC are largely unknown. Methods: Patients with advanced NSCLC treated with ICI at the Dana-Farber Cancer Institute (DFCI), and whose tumors underwent genomic profiling before and after ICI, with no intervening therapies, were included. Mutations, tumor mutational burden (TMB), copy number variations (CNVs), and PD-L1 tumor proportion score (TPS) were compared between pre- and post-ICI samples. Acquired resistance was defined as the development of disease progression after an initial objective response, or stable disease ≥3 months with PD-(L)1 blockade. Results: Among 1763 patients with advanced NSCLC who received ICI, 45 had matched pre- and post-ICI tissue samples available for genomic profiling. Putative mechanisms of resistance were identified in 55% of cases (N = 25). Five patients (20%) acquired an STK11 mutation, one patient (4%) acquired a KEAP1 mutation, and another patient (4%) developed concurrent KEAP1 and SMARCA4 mutations. A patient (4%) with KRAS G12C-mutant NSCLC developed concurrent STK11 and KEAP1 mutations at resistance. In 3 cases (12%) with pre-existing STK11 or KEAP1 mutations prior to ICI administration, we identified acquired copy losses of STK11 and KEAP1, respectively, resulting in bi-allelic inactivation of these genes. Acquired beta-2-microglobulin ( B2M) mutations were detected in 3 patients (12%), one of whom developed concurrent B2M copy loss, indicating bi-allelic inactivation. Eight additional patients (32%) developed B2M gene deletions. Other acquired alterations that have been implicated in ICI resistance included CDKN2A/B loss (N = 10, 40%), including 5 with bi-allelic deletion, acquired PTEN deletions (N = 5, 20%), and MDM2 amplification (N = 2, 8%). When we examined alterations in immune checkpoint genes, we identified acquired CD274 (PD-L1) and PDCD1LG2 (PD-L2) loss in 8% of cases (N = 2), and bi-allelic deletion in one case (4%). Intervening ICI did not affect TMB (median TMB: 8.7 [pre-ICI] vs 9.1 [post-ICI] mut/Mb, P = 0.6), PD-L1 expression (median PD-L1 TPS: 3% [pre-ICI] vs 5.0% [post-ICI] mut/Mb, P = 0.5), or aneuploidy levels (as fraction of genome altered [FGA]) (median FGA: 18.4% [pre-ICI] vs 21.1% [post-ICI], P = 0.2), indicating that acquired gene level CNVs were not a reflection of increased cancer aneuploidy. In a control cohort of 30 patients with pre- and post-chemotherapy matched samples which underwent genomic profiling, no acquired mutations in STK11, KEAP1, SMARCA4, or B2M were detected. Conclusions: Mechanisms of acquired resistance to PD-(L)1 blockade are heterogenous, and new therapeutic strategies are required to delay and overcome ICI resistance in patients with NSCLC.

Differences in pathology and mutational status among colorectal cancer (CRC) patients pre-, post-, and during screening age

3596

Background: Screening protocols for CRC are broadly recommended and effective in reducing mortality. However, populations from different age groups can harbor distinct pathological and molecular profiles that can also be influenced by screening and polyp resection, especially in older ages. Methods: We retrospectively analyzed tumors from stage IV CRC patients from a central pathology laboratory in Brazil that is a reference for mutational profiling countrywide. Patients were classified into age groups as the following: pre-screening (PrSA; <45yo), screening(SA; 45-75yo) and post-screening age (PoSA; >75yo). Every tumor has been centrally reviewed by the pathologist. Groups were compared regarding clinicopathologic features and presence of RAS and BRAF mutations. Results: We included 1244 pts (164 PrSA, 919 SA and 161 PoSA). There were no difference among groups regarding sidedness(p= .68)and KRASmutations (p=.0.97). Stage IV at diagnose (p =.001), presence of signet-ring cell component (p< .001) along with poorly differentiated tumors (p= .006) were most found on young patients, while BRAFand NRASmutations where significantly more common among PosSA (table). Conclusions: PosSA and PreSA CRCs seem to present a distinct profile from SA populations, including differences in the molecular and pathologic aspects. This could impact the frequency of screening tests among different age groups. [Table: see text]