Next Generation Sequencing Reports Research Articles

Abstract PO3-10-07: Leveraging AI to identify factors influencing access to care and their association with overall survival- a multiracial Breast Cancer cohort

Abstract Introduction: Data from the Centers for Disease Control and Prevention (CDC) labels each US state based on its 5-Year Relative Survival of Breast Cancer (BC) patients (pts). Data is categorized into four “survival groups'', each representing a range of 5-year survival percentages. Using real-world data, this study characterized the populations in each survival group, examining pt knowledge and education about their disease as well as preferences and factors influencing their decision to explore clinical trial (CT) opportunities. Methods: Leal is an AI-based platform that employs self-reported medical profiles to match cancer pts with CTs. This study involved a diverse cohort of 14,509 BC pts who completed the questionnaire. The geographic distribution of pts across US states was analyzed, taking into account the state’s 5-year BC survival %. Four groups were defined as follows: Group A (86.4 - 89.4%), Group B (89.4 - 90.4%), Group C (90.4 - 91.5%), and Group D (92.0 - 95.0%). Parameters such as race, willingness to travel, insurance coverage, performance of genetic testing, and biomarker knowledge were considered in the analysis. Results: We found significant variations in registration rates on the Leal platform across groups. The Highest registration rate was recorded for Group D which has the highest 5-year survival rate (p< 0.0001). Group A which has the lowest 5-year relative survival had a significantly higher proportion of African Americans compared to group D (14% vs 4%, p < 0.0001). Moreover, while there was no difference in the percentage of pts undergoing Next-Generation Sequencing (NGS) in both groups, a larger proportion of pts in Group D demonstrated proficiency in interpreting the test reports (p = 0.05). Although no significant differences were found in the proportion of pts who were aware of their treatment (tx) history between the two groups, whites in both groups exhibited greater knowledge of their tx (p = 0.001). Additionally, pts in Group D showed a significantly higher willingness to travel longer distances to participate in CTs (p=0.001), and a larger percentage of them had insurance coverage. Conclusions: This study used data from CDC that categorized US states into four groups based on BC pts’ 5-Year Survival. CT participation is associated with better clinical outcomes, especially for pts with poorer prognosis. In agreement, we show a higher signup rate to the Leal platform in States with higher 5-year relative survival, possibly pointing to higher CT education and knowledge among pts in those states. This is further supported by higher proficiency in interpreting NGS reports, resulting in better-optimized CT options on the Leal platform. Group A was also characterized by a reduced willingness to travel for a CT and a larger proportion of pts without health insurance. Those are significant barriers for CT enrollment, and ultimately limit access to optimal care and may impact outcomes. Gaps in knowledge of key disease parameters such as tx history were observed for African Americans, regardless of geographic location, highlighting the need to focus on minority populations even in states where the survival rate is relatively high. In agreement, the proportion of African Americans in group A was markedly higher, correlating with current disparities in access to tx and poorer outcomes. AI-based platforms are optimized for the exact identification of barriers to access to care and for enabling relevant healthcare and support organizations to effectively address individual patient issues, consequently increasing access to CTs, particularly among different racial and ethnic populations. Citation Format: Michal Safran, Yelena Lapidot, Tzvia Bader, Avital Gaziel. Leveraging AI to identify factors influencing access to care and their association with overall survival- a multiracial Breast Cancer cohort [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO3-10-07.

Abstract 2315: AI-enabled precision oncology era: Advanced and interactive interpretation of next-gneneration sequencing (NGS) reports

Abstract Background: Next-generation sequencing (NGS) has been widely adopted in clinical practice, but there are still unmet needs among physicians regarding the interpretation and application of NGS reports. These needs include selecting multi-targeted and combination therapies, keeping up with evolving treatment options and clinical trial results, and incorporating personalized medicine based on real clinical cases. Methods: To address these challenges, we propose using large language model (LLM) to analyze and interpret clinical questions, perform semantic searches in a high-quality database, and provide customized prompts and logical outputs. Our knowledge base includes diagnostic and treatment details for 600+ tumor subtypes, 620,000+ entries, 2,000+ drug details, and 440,000+ clinical trials. The database is updated automatically using a fact engine and medical literature sources. Personalized gene interpretation is incorporated using pretraining, cosine similarity, and context learning prompts. Results: Our tool, named "SmartMTB", enables personalized interpretation of genomic testing reports for precision oncology. Users input the latest clinical features using a selection-based or question-and-answer approach. The tool identifies the treatment stage and genomic variations, provides information on mutation frequencies, sensitivity and resistance of targeted and immunotherapies, and priority selection strategies for multi-gene mutations. It matches similar clinical cases and provides real-time updates on treatment options and clinical trial literature. It also facilitates patient matching for clinical trials and considers economic factors. Conclusion: We present a groundbreaking artificial intelligence (AI) tool in cancer care that revolutionizes precision oncology. This AI system, trained on extensive datasets, swiftly and accurately analyzes and interprets complex genomic and clinical data. It enables clinicians to make informed decisions about personalized treatment strategies, significantly improving patient outcomes in cancer therapy. Citation Format: Hui Chen, Zanmei Xu, Lijuan Chen, Mingmin Wang, Peng Zhang, Fei Pang, Kai Wang. AI-enabled precision oncology era: Advanced and interactive interpretation of next-gneneration sequencing (NGS) reports [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2315.

Discordance in Recommendation Between Next-Generation Sequencing Test Reports and Molecular Tumor Boards in India.

Accurate understanding of the genomic and transcriptomic data provided by next-generation sequencing (NGS) is essential for the effective utilization of precision oncology. Molecular tumor boards (MTBs) aim to translate the complex data in NGS reports into effective clinical interventions. Often, MTB treatment recommendations differ from those in the NGS reports. In this study, we analyze the discordance between these recommendations and the rationales behind the discordances, in a non-high-income setting, with international input to evaluate the necessity of MTB in clinical practice. We collated data from MTB that were virtually hosted in Chennai, India. We included patients with malignancies who had NGS reports on solid tissue or liquid biopsies, and excluded those with incomplete data. MTB forms and NGS reports of each clinical case were analyzed and evaluated for recommendation concordance. Concordance was defined as an agreement between the first recommendation in the MTB forms and the therapeutic recommendations suggested in the NGS report. Discordance was the absence of the said agreement. The rationales for discordance were identified and documented. Seventy MTB reports were analyzed with 49 cases meeting the inclusion criteria. The recommendation discordance was 49% (24 of 49). Discordant recommendations were mainly due to low level of evidence for the drug (75% of cases). The discordance between MTB and NGS vendor recommendations highlights the clinical utility of MTB. The educational experiences provided by this initiative are an example of how virtual academic collaborations can enhance patient care and provider education across geographic borders.

Abstract B011: Phase 2, multicenter, open-label basket trial of nab-sirolimus for malignant solid tumors harboring pathogenic inactivating alterations in TSC1 and TSC2 (PRECISION I)

Abstract Introduction/objective: PRECISION I (NCT05103358) is an ongoing trial of nab-sirolimus in advanced cancers harboring pathogenic inactivating alterations in TSC1 or TSC2. TSC1 and TSC2 alterations have been observed in patients with various cancers, including the most common cancer of the female reproductive organs, endometrial cancer, where ~3% contain TSC1 and/or TSC2 alterations. nab-Sirolimus, approved in the USA for patients with advanced malignant perivascular epithelioid cell tumor (PEComa), is an albumin-bound mTOR inhibitor (mTORi) that inhibits the mTOR pathway via suppression of the mTORC1 complex. When TSC1 or TSC2 is inactivated via mutation or loss, the mTOR pathway may be aberrantly activated. In an exploratory analysis of AMPECT (NCT02494570), a pivotal trial of nab-sirolimus in advanced malignant PEComa, 64.3% (9/14) of patients with pathogenic inactivating TSC2 or TSC1 alterations had a confirmed response. Most treatment-related adverse events in AMPECT were grade 1/2 (none were grade ≥4) and were consistent with mTORi-class adverse events. Based on the clinical observations from AMPECT and the underlying mechanism of action of nab-sirolimus, the PRECISION I trial was designed to assess the safety and efficacy of nab-sirolimus in a tumor-agnostic study of patients with advanced cancers with TSC1 or TSC2 inactivating alterations. Patients with endometrial cancer are expected to be enrolled based on the frequency of TSC1 and TSC2 inactivating alterations. Methods: Eligible patients are ≥12 years old and mTORi-naïve, possess advanced or metastatic malignant solid tumors with TSC1 or TSC2 inactivating alterations identified using next-generation sequencing (NGS) in tumor tissue or liquid biopsy (confirmed by central review of NGS reports), and have received appropriate standard treatments, per investigator. nab-Sirolimus 100 mg/m2 is given intravenously over 30 min on days 1 and 8 of each 21-day cycle. The primary endpoint is overall response rate per independent radiographic review (IRR) using Response Evaluation Criteria in Solid Tumors v1.1. Other endpoints include duration of response, time to response, progression-free survival by IRR, overall survival, patient-reported quality of life, and safety. Trial Status: Enrollment began in March 2022 and is expedited through collaboration with leading NGS providers; ongoing study access is facilitated through a just-in-time approach to trial location activation. Citation Format: Debra L. Richardson, Gopa Iyer, Li Ding, Anita N. Schmid, Willis Navarro, David Kwiatkowski, Jordi Rodon Ahnert. Phase 2, multicenter, open-label basket trial of nab-sirolimus for malignant solid tumors harboring pathogenic inactivating alterations in TSC1 and TSC2 (PRECISION I) [abstract]. In: Proceedings of the AACR Special Conference on Endometrial Cancer: Transforming Care through Science; 2023 Nov 16-18; Boston, Massachusetts. Philadelphia (PA): AACR; Clin Cancer Res 2024;30(5_Suppl):Abstract nr B011.

Clinical characterization and therapeutic outcomes of patients (pts) with colorectal cancer (CRC) harboring somatic BRCA1/2 mutations.

134 Background: Somatic BRCA1/2 mutations in BRCA-associated cancers increase therapeutic sensitivity to platinum-based chemotherapies and PARP inhibitors, but the implications of BRCA mutations in non- BRCA-associated cancers such as CRC is unknown and hypothesized to be shaped by tumor lineage. We thus conducted a retrospective analysis of outcomes with platinum-based chemotherapy and PARP inhibition in pts with CRC, where the estimated prevalence of BRCA1/2 alterations is 3-4%. Methods: Pts with blood or tissue next generation sequencing (NGS) positive for a somatic BRCA1/2 variant and a diagnosis of CRC were identified from a single institution database. The pathogenicity of each mutation was annotated using public genomic databases (e.g. ClinVar). Pt characteristics were compared using Fisher’s exact test. Log-rank test was used for comparison of survival distribution among groups. Results: 8,258 NGS reports with a BRCA1/2 alteration from 1760 cancer pts were identified of which 189 (11%) had a diagnosis of CRC. 54/189 CRC pts (29%) had ≥1 pathogenic BRCA1/2 variant, 19 pts (10%) had ≥1 benign variant, and 134 pts (69%) had 1 ≥ variant of unknown significance (VUS); 26 pts (14%) had more than one BRCA1/2 variant. Pts with ≥1 pathogenic mutation (n=54) were more likely to have an inactivating DNA polymerase epsilon (POLE) mutation versus pts with only benign/VUS BRCA1/2 variants (n=135) (20% vs 6%, p=0.004) while prevalence of microsatellite instability (MSI-H) (19% vs 13%, p=0.37) and Ras/Raf (53% vs 53%, p=1.00) were similar between groups. Median progression free survival (PFS) among 115 CRC pts (unselected for BRCA1/2 pathogenicity) who received an oxaliplatin-based regimen for advanced disease in the 1st line (n=52) was similar to the median PFS among pts who received an irinotecan-based regimen (n=63) (7.0 mths vs 7.0 mths, p=0.48). Median PFS with an oxaliplatin-based regimen in the 1st line among 52 pts with pathogenic BRCA1/2 variants (n=13) was 5.1 mths vs 7.4 mths among pts with benign/VUS BRCA1/2 variants (n=39) (p=0.398). There were no significant differences in median PFS among pts with a pathogenic vs benign/VUS BRCA1/2 variants when the analysis was restricted to pts with BRCA1/2 variants identified in tissue (p=0.76) and to pts without co-occurring driver mutations (Ras/Raf, MSI-H, and/or POLE) (p=0.1). Overall survival in patients with advanced disease was similar between pts with a pathogenic (n=52) vs benign/VUS (n=122) BRCA variants (37.1 mths vs 45.1 mths, p=0.14). 6 pts (4 with pathogenic BRCA1/2 and 3 with VUS) received a PARP inhibitor after a median of 3 lines of treatment and the best response was disease progression in 6 pts (100%). Conclusions: In the context of CRC, somatic BRCA1/2 mutations frequently co-occur with pathogenic POLE mutations, but do not appear to confer therapeutic benefit to platinum-based chemotherapy and PARP inhibitors.

PRECISION 1: A phase 2, multicenter, open-label basket trial of nab-sirolimus for malignant solid tumors harboring pathogenic inactivating alterations in TSC1 and TSC2.

TPS585 Background: SEER data report that US patients with advanced gastrointestinal (GI) cancers have a poor prognosis as indicated by low 5-year survival rates in patients with colorectal (15.6%), liver (3.5%), and other GI cancers, highlighting the need for more efficacious treatments. Dysregulation of the mTOR pathway contributes to tumor growth and disease progression in many cancers, including GI cancers. The tumor suppressor genes TSC1 and TSC2 are critical negative regulators of mTOR activity whose inactivation can lead to tumor cell growth. Inactivating TSC1 and TSC2 alterations have been observed in GI cancers with a combined frequency of up to 6.5% in hepatocellular carcinoma, 1.7% in colorectal adenocarcinoma and gastrointestinal stromal tumors, and 1.6% in cholangiocarcinoma. nab-Sirolimus, approved in the US for patients with advanced malignant perivascular epithelioid cell tumor (PEComa), is an albumin-bound mTOR inhibitor (mTORi) that inhibits the mTOR pathway via suppression of the mTORC1 complex. In an exploratory analysis of the pivotal AMPECT trial of nab-sirolimus in advanced malignant PEComa (NCT02494570), 8/9 (89%) and 1/5 (20%) patients with inactivating alterations in TSC2 and TSC1, respectively, had a confirmed response. Most treatment-related adverse events were grade 1/2 (none were grade ≥4) and were consistent with mTORi-class adverse events. Based on the clinical observations from AMPECT and the underlying mechanism of action of nab-sirolimus, the PRECISION 1 trial was designed to assess the safety and efficacy of nab-sirolimus in a tumor-agnostic study of patients with advanced cancers harboring TSC1 or TSC2 inactivating alterations. The trial is open to patients with GI cancers with inactivating alterations in TSC1 or TSC2. Methods: In PRECISION 1 (NCT05103358), eligible patients are ≥12 years old and mTORi-naïve, have advanced malignant solid tumors harboring TSC1 or TSC2 inactivating alterations identified using next-generation sequencing (NGS) of tumor tissue or liquid biopsy (confirmed by central review of NGS reports), and have received appropriate standard treatments, per investigator. nab-Sirolimus 100 mg/m2 is given intravenously over 30 min on days 1 and 8 of each 21-day cycle. The primary endpoint is overall response rate per independent radiographic review (IRR) using Response Evaluation Criteria in Solid Tumors v1.1. Other endpoints include duration of response, time to response, progression-free survival by IRR, overall survival, patient-reported quality of life, and safety. Enrollment began in March 2022. Collaboration with leading NGS vendors is expediting the identification of patients with qualifying TSC1 and TSC2 alterations; ongoing study access is facilitated through a just-in-time approach to trial location activation. Clinical trial information: NCT05103358 .

Next-generation sequencing reveals genetic heterogeneity and resistance mechanisms in patients with EGFR-mutated non-small cell lung cancer treated with afatinib.

Afatinib, an irreversible ErbB family inhibitor, is widely used as first-line treatment in advanced lung adenocarcinoma patients harbouring mutant epidermal growth factor receptor (EGFR). With the advancements in next-generation sequencing (NGS), comprehensive research into the clinical impact of co-occurring genetic mutations and the molecular mechanisms of acquired resistance is required for afatinib users. From January 2010 to December 2019, we enrolled patients with advanced lung adenocarcinoma with EGFR mutations using afatinib as first-line treatment, and we retrospectively collected pre- and post-afatinib treatment specimens from these patients for NGS testing. Of the 362 enrolled patients, 73 samples (68.9%) from 56 patients successfully returned complete NGS reports. In pre-afatinib treatment specimens, the most frequent co-occurring alterations were TP53, MUC16, USH2A, SNYE1, RECQL4 and FAT1; however, they were not related to progression-free survival. Small cell lung cancer transformation, EGFR p.T790M, amplification of MET, ERBB2, KRAS, EGFR, cell cycle-regulated genes and MDM2, and PTEN alterations were identified as acquired resistance mechanisms. EGFR p.T790M (p=0.0304) and APC alterations (p=0.0311) in post-afatinib specimens were significantly associated with longer overall survival, while MET amplification was significantly associated with poor overall survival (p=0.0324). The co-occurrence of TP53 alterations was significantly associated with shorter overall survival (p=0.0298). Our results show that the frequent co-occurring alterations in advanced EGFR-mutated lung adenocarcinoma did not influence the effectiveness of afatinib. EGFR p.T790M is not only the major resistance mechanism to afatinib but also related to favourable survival outcomes. MET amplification and TP53 mutations were associated with poorer overall survival.

Abstract B169: Retrospective baseline biomarker analyses in a first-in-human Phase 1 trial of the PKMYT1 inhibitor lunresertib (RP-6306) in pts with advanced solid tumors harboring CCNE1 amplification and/or deleterious alterations in FBXW7 or PPP2R1A

Abstract Background: Lunresertib (RP-6306) is a first-in-class membrane-associated tyrosine- and threonine-specific Cdc2-inhibitory kinase (PKMYT1) inhibitor that disrupts the G2/M checkpoint leading to premature mitosis and catastrophic DNA damage in cells harboring synthetic lethal genomic alterations. The safety and tolerability of lunresertib alone and in combination with camonsertib (RP-3500) is being investigated in the Phase 1 MYTHIC study in patients with advanced solid tumors with either CCNE1 amplifications (amp) or deleterious alterations in FBXW7 or PPP2R1A (NCT04855656). Here, we present a comprehensive retrospective biomarker analysis aimed at understanding concordance between local vs. central and tissue- vs. plasma-based next generation sequencing (NGS). The correlation between CCNE1 copy number assessment by NGS and fluorescence in situ hybridization (FISH), as well as the relationship between CCNE1 amp and Cyclin E1 protein levels, was investigated. Methods: Patients were enrolled based on results from sponsor-approved local tumor NGS and plasma circulating tumor DNA (ctDNA) tests. To ensure molecular eligibility criteria were met, local NGS reports were centrally reviewed and functionally annotated prior to enrollment. Tumor tissue and plasma samples collected at screening were retrospectively analyzed by SNiPDx™ and Tempus xF(+), respectively. Tissue samples were further evaluated by FISH and immunohistochemistry. Results: As of June 1, 2023, 145 patients were screened with uterine (23%), colorectal (CRC, 15%), ovarian (13%), and upper GI (6%) as the most frequent cancers. Ninety-four percent (137/145) of patients screened had local NGS results available (98% tissue, 2% plasma). CCNE1 amp (40.1%) were most frequently observed in ovarian, uterine, and upper GI malignancies. FBXW7 (39.4%) and PPP2R1A (16.0%) alterations were enriched in CRC and uterine cancer, respectively. TP53 alterations were frequent across all biomarker groups (76% CCNE1, 54% FBXW7, and 88% PPP2R1A). Central tissue NGS confirmed 94% (47/50) of the enrollment alterations. Central plasma ctDNA confirmed 83.0% (39/47) of FBXW7 and PPP2R1A mutations and had high positive predictive value (100%, 17/17) but low sensitivity (42.5%; 17/40) for CCNE1 amp. CCNE1 amp call and copy number estimation by tissue NGS and FISH were highly correlated (Spearman ρ=0.65, p=0.006). Seventy-five percent of tumors with CCNE1 amp had Cyclin E1 overexpression (H-score ≥100; mean=168; range [0, 300]). Conclusion: In this study, local NGS tests were an efficient and robust strategy for directing patients into a molecularly defined Phase 1 trial. Plasma ctDNA reliably detected alterations in all 3 genes, but caution should be taken when interpreting CCNE1 amplification negative results due to a high false negative rate. FISH and genomics are appropriate methods to estimate CCNE1 amp copy number. These retrospective analyses enable the testing of the synthetic lethal hypothesis of lunresertib and CCNE1, PPP2R1A and FBXW7, assuring reliable interpretation of the clinical data. Citation Format: Elia Aguado-Fraile, Sunantha Sethuraman, Adam Petrone, Emeline Bacque, Alison Schram, Stephanie Lheureux, Elizabeth K Lee, Maria Koehler, Ian M Silverman, Victoria Rimkunas, Timothy A Yap. Retrospective baseline biomarker analyses in a first-in-human Phase 1 trial of the PKMYT1 inhibitor lunresertib (RP-6306) in pts with advanced solid tumors harboring CCNE1 amplification and/or deleterious alterations in FBXW7 or PPP2R1A [abstract]. In: Proceedings of the AACR-NCI-EORTC Virtual International Conference on Molecular Targets and Cancer Therapeutics; 2023 Oct 11-15; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2023;22(12 Suppl):Abstract nr B169.

Abstract B078: Unraveling resistance and optimizing treatment through molecular profiling in patients with NSCLC and oncogenic fusions

Abstract Introduction: In non-small cell lung cancer (NSCLC) with oncogenic fusions, resistance development complicates optimal therapeutic sequencing. This study investigates the resistance landscape in NSCLC with oncogenic fusions and evaluates the efficacy of subsequent treatment tailoring based on the molecular profile at progression. Methods: This is a single center, retrospective study, including patients with advanced NSCLC and ALK, ROS1, RET or NTRK fusions, with a molecular profile at progression after targeted therapy (TKI), between January 2015 and February 2023. Molecular profile was performed on tissue or liquid samples, with next-generation sequencing (NGS) (targeted DNA and/or RNA panels, WES/RNA sequencing). In case of TKI sequencing, more than one NGS results were available for selected patients. Alterations at time of resistance were classified into on-target, bypass alterations, and polyclonal/complex resistance (with at least two types of resistance concurrently present in the same biopsy). The primary objective was to assess the clinical utility of NGS to guide subsequent treatment at progression after TKI. Results: A total of 95 patients (61 ALK, 21 ROS1, 12 RET, 1 NTRK) had 179 samples collected. Patients were female in 51 (54%), had a median age of 53 years (IQR 41-60), never smokers in 52 (55%) cases and adenocarcinoma histology in 88 (94%) of cases. Molecular profile was performed by liquid biopsy in 83 (66%) cases, other liquids in 3 (2%) and tissue in 40 (32%) cases. There were 126 contributive samples (70%) for 69 patients. Resistance was explained by 60/126 (48%) samples for 41 patients: on-target in 27 (47%), bypass in 22 (39%), and polyclonal resistance in 8 (14%) samples. Three out of 7 samples evaluable for mutational signatures via WES showed Apolipoprotein B mRNA Editing Catalytic Polypeptide-like (APOBEC) signatures. The time until the first occurrence of resistance was 34.2 months (95%CI 26-65.3) for on-target, 23.6 months (95%CI 18.8-97.4) for by-pass and 16.5 months (95%CI 7.2-NR) for polyclonal resistance. The early occurrence of by-pass or polyclonal resistance in the first 12 months was associated with a younger age (median 37 vs 52 years, p=0.044). Among the 60 NGS results explaining resistance for 41 patients, 56 (93%) NGS results informed treatment tailoring. Drug tailoring was conducted in 16 patients for 23 NGS reports, revealing a median overall survival of 18.6 months (95% CI 11.8 – NR) for patients receiving an effective tailored treatment, versus 7.3 months (95%CI 2.9 – NR) for those without tailoring (p=0.088). Conclusion: Our findings underscore the value of using NGS to inform treatment tailoring in NSCLC patients with oncogenic fusions. Special focus should be placed on younger patients who might be prone to early development of bypass or polyclonal resistance. Citation Format: Mihaela Aldea, Vlad Pop, Arianna Marinello, Andrei Ivashkin, Damien Vasseur, Sebastien Gendarme, Marco Tagliamento, Anas Gazzah, Jordi Remon, Imen Ben-Ammar, Maxime Frelaut, Pamela Abdayem, Pernelle Lavaud, Antoine Hollebecque, Etienne Rouleau, Yohann Loriot, Luc Friboulet, Sergey Nikolaev, Antoine Italiano, Fabrice Barlesi, David Planchard, Benjamin Besse. Unraveling resistance and optimizing treatment through molecular profiling in patients with NSCLC and oncogenic fusions [abstract]. In: Proceedings of the AACR-NCI-EORTC Virtual International Conference on Molecular Targets and Cancer Therapeutics; 2023 Oct 11-15; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2023;22(12 Suppl):Abstract nr B078.

Abstract LB_B12: Phase 2, multicenter, open-label basket trial of nab-Sirolimus for malignant solid tumors harboring pathogenic inactivating alterations in TSC1 and TSC2 (PRECISION I)

Abstract BACKGROUND. nab-Sirolimus, approved in the US for patients with advanced malignant PEComa, is a novel albumin-bound mTOR inhibitor (mTORi) that inhibits the mTOR pathway via suppression of the mTORC1 complex. When TSC1 or TSC2 is inactivated via mutation or loss, the mTOR pathway may be aberrantly activated. Clinically, in an exploratory analysis of the pivotal AMPECT trial of nab-sirolimus in advanced malignant PEComa (NCT02494570), 8/9 (89%) and 1/5 (20%) patients with inactivating alterations in TSC2 and TSC1, respectively, had confirmed response (Wagner, J Clin Oncol, 2021). TSC1 and TSC2 alterations have been observed in patients with various cancers. Most treatment-related adverse events in AMPECT were grade 1/2 (none were grade ≥4) and were consistent with mTORi-class adverse events. Based on the clinical observations from AMPECT and the underlying mechanism of action of nab-sirolimus, the PRECISION I trial (NCT05103358) was designed to assess the safety and efficacy of nab-sirolimus in a tumor-agnostic study of advanced cancers with TSC1 or TSC2 inactivating alterations. METHODS. Eligible patients are ≥12 years old and mTORi-naïve, possess advanced malignant solid tumors with TSC1 or TSC2 inactivating alterations identified using next-generation sequencing (NGS) in tumor tissue or liquid biopsy (confirmed by central review of NGS reports), and have received appropriate standard treatments, per investigator. nab-Sirolimus 100 mg/m2 will be given  intravenously over 30 min on Days 1 and 8 of each 21-day cycle. The primary endpoint is overall response rate per independent radiographic review (IRR) using RECIST v1.1. Other endpoints include duration of response, time to response, progression-free survival by IRR, overall survival, patient-reported quality of life, and safety. Enrollment began March 2022. Collaboration with leading NGS vendors will expedite the identification of patients with qualifying TSC1 and TSC2 mutations; ongoing study access is facilitated through a just-in-time approach to trial location activation. Based on the prevalence of TSC1 and TSC2 inactivating alterations, the most frequent tumor types expected are bladder, hepatobiliary, endometrial, soft tissue sarcoma, ovarian, and esophagogastric. Citation Format: Candace Haddox, Gopa Iyer, Michael J. Demeure, Li Ding, Anita N. Schmid, Willis H. Navarro, David J. Kwiatkowski, Jordi Rodon Ahnert. Phase 2, multicenter, open-label basket trial of nab-Sirolimus for malignant solid tumors harboring pathogenic inactivating alterations in TSC1 and TSC2 (PRECISION I) [abstract]. In: Proceedings of the AACR-NCI-EORTC Virtual International Conference on Molecular Targets and Cancer Therapeutics; 2023 Oct 11-15; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2023;22(12 Suppl):Abstract nr LB_B12.

Confirmed Pathogenic Germline Variants in Cancer Predisposition Genes Incidentally Detected in Somatic Genomic Profiling of Multiple Myeloma Patients

Introduction: We previously showed that ~10% of patients with multiple myeloma (MM) carry pathogenic or likely-pathogenic (P/LP) germline variants (PGVs) in known cancer predisposition genes, and that moderate/high-penetrance PGVs in DNA repair genes may be associated with MM risk and favorable response to alkylating chemotherapy. Increasingly applied in MM, genomic profiling by next-generation sequencing (NGS) can identify prognostically-relevant and therapeutically-actionable somatic mutations, but it may also incidentally unveil PGVs, which can have significant implications for patients and their families. We here test the diagnostic accuracy of a novel referral tool designed to flag likely PGVs warranting confirmatory germline testing among mutations identified in somatic NGS studies of 1161 MM patients from our institution. Methods: We retrospectively reviewed 1715 commercial NGS reports of 1161 MM patients treated at the Icahn School of Medicine at Mount Sinai (ISMMS) between 2015 & 2022 (targeted 448-gene somatic profiling panels performed by a single genomic testing company for clinical purposes on bone marrow aspirate samples). In collaboration with certified genetic counselors at our institution, we applied evidence-based criteria to develop a tool to flag P/LP variants of likely germline origin warranting a referral for confirmatory germline testing (“Likely PGVs”). The resulting referral tool flags variants detected in 35 well-established cancer predisposition genes at a variant allele frequency (VAF) of >10%, as well as 24 cancer predisposition founder variants at any VAF. Next, we tested the positive predictive value (PPV), negative predictive value (NPV), sensitivity and specificity of our approach for the detection of true PGVs by analyzing paired germline whole-exome sequencing (WES) in 382/1161 (33%) MM patients in the study cohort. Germline WES data were generated for research purposes using a clinically-validated pipeline for PGV detection and annotation. Results: 385 P/LP variants were reported in somatic NGS studies of 382 MM patients who had paired somatic/germline sequencing data. Of these, 38/385 (9.9%) were found to be true PGVs in confirmatory germline testing. True PGVs detected in more than one gene included nine APC, seven CHEK2, five BRCA1, three BRCA2, three ATM and three MUTYH variants. The referral tool flagged 41/385 (10.6%) of the cohort's somatic variants as likely PGVs, identifying 32/38 (84.2%) true PGVs. The six PGVs missed by the referral tool were detected in the following genes: one APC, one CHEK2, one MITF, one MUTYH, one NF1 and one SHDA variant. In contrast, the nine variants that were mislabeled as likely PGVs but were in fact somatic were: two ATM, two BRCA2, two TSC1, one CHEK2, one BRIP1, and one RET. With a variant allele threshold of >10%, the tool had a PPV of 78.0% and a NPV of 98.3% for the detection of true PGVs. The sensitivity and specificity were 84.2% and 97.5%, respectively. Among the 692/1161 MM patients who did not have confirmatory germline testing (67%), the proportion of variants flagged as likely PGVs was very similar to that of the paired somatic/germline group (77/696, 11.1%), and variants were detected in similar genes. Assuming a PPV of 78%, 60 of these variants could represent true PGVs. Conclusion: A novel tool designed to flag likely PGVs among mutations identified in somatic genomic profiling of MM patients had high sensitivity and specificity for the detection of true PGVs. 9.9% of all MM patients with paired somatic/germline testing were found to have a true PGV, validating the results of our prior study and once again highlighting the need to implement screening strategies in high-risk subgroups of MM patients, such as those with strong family history of cancer or prior personal history of cancer.

Relevance of next generation sequencing (NGS) data re-analysis in the diagnosis of monogenic diseases leading to organ failure

BackgroundIn 2018, our center started a program to offer genetic diagnosis to patients with kidney and liver monogenic rare conditions, potentially eligible for organ transplantation. We exploited a clinical exome sequencing approach, followed by analyses of in silico gene panels tailored to clinical suspicions, obtaining detection rates in line with what reported in literature. However, a percentage of patients remains without a definitive genetic diagnosis. This work aims to evaluate the utility of NGS data re-analysis for those patients with an inconclusive or negative genetic test at the time of first analysis considering that (i) the advance of alignment and variant calling processes progressively improve the detection rate, limiting false positives and false negatives; (ii) gene panels are periodically updated and (iii) variant annotation may change over time.Methods114 patients, recruited between 2018 and 2020, with an inconclusive or negative NGS report at the time of first analysis, were included in the study. Re-alignment and variant calling of previously generated sequencing raw data were performed using the GenomSys Variant Analyzer software.Results21 previously not reported potentially causative variants were identified in 20 patients. In most cases (n = 19), causal variants were retrieved out of the re-classification from likely benign to variants of unknown significance (VUS). In one case, the variant was included because of inclusion in the analysis of a newly disease-associated gene, not present in the original gene panel, and in another one due to the improved data alignment process. Whenever possible, variants were validated with Sanger sequencing and family segregation studies. As of now, 16 out of 20 patients have been analyzed and variants confirmed in 8 patients. Specifically, in two pediatric patients, causative variants were de novo mutations while in the others, the variant was present also in other affected relatives. In the remaining patients, variants were present also in non-affected parents, raising questions on their re-classification.ConclusionsOverall, these data indicate that periodic and systematic re-analysis of negative or inconclusive NGS data reports can lead to new variant identification or reclassification in a small but significant proportion of cases, with benefits for patients’ management.

The Clinical Characteristics, Diagnostic Methods, Treatment, and Outcomes of Listeria monocytogenes Meningoencephalitis: A Case Series Study from China.

Listeria monocytogenes meningoencephalitis (LMM) is a rare but dangerous infection of the nervous system. This study intends to summarize and discuss the clinical features, diagnostic methods, treatment, and outcomes. LMM patients treated in the First Affiliated Hospital of Zhengzhou University from June 2013 to December 2022 were retrospectively studied. The clinical data, magnetic resonance imaging (MRI), biochemical and hematological parameters, pathogen tests, treatments, and outcomes information were collected and summarized. Ten LMM patients were enrolled in this study. Three were female and seven were male, with a mean age of 54.5 years old. Seven (70%) of the patients had chronic conditions that either affected the liver (1 case), heart (1 case), or immune system (5 cases). The main clinical manifestations were fever (100%), headache, (60%), and disturbance of consciousness (60%). Out of the nine patients taken MRI, enhanced magnetic resonance imaging was observed with cerebral parenchyma and meningeal enhancement in 3 patients, whereas six patients showed non-specific abnormal signals and brain edema. Seven (70%) patients were positive for L. monocytogenes by bacterial culture and five patients were positive by further next-generation sequencing (NGS) test in CSF. All patients were treated with antibiotics based on antibiotic sensitivity tests or experience, with seven (70%) improved but three (30%) died within one month. LMM patients have a high mortality rate. Considering the time and reliability of NGS reports, it would be better to identify L. monocytogenes infection in the CSF than bacterial culture. While, after diagnosis, giving timely and appropriate antibiotics would still be a challenge to achieve good outcomes.

Relevance and accuracy of ChatGPT-generated NGS reports with treatment recommendations for oncogene-driven NSCLC.

1555 Background: Next-generation sequencing (NGS) is a routine clinical practice in advanced NSCLC. NGS reports are information-dense and clinical interpretation remains a challenge. ChatGPT is a large language model (LLM) AI chatbot that can generate text in response to user-generated prompts. We sought to assess the clinical relevance and accuracy of ChatGPT-generated NGS reports with first-line (1L) treatment recommendations for NSCLC patients with targetable driver oncogenes. Methods: Eight driver oncogenes with FDA-approved targeted treatment for 1L stage IV NSCLC were identified in the latest NCCN Clinical Practice Guidelines available to the AI model (version 5, September 2021). The prompt, “Create a next-generation sequencing report with a list of first-line treatment options for a patient with stage IV non-small cell lung cancer with an [oncogenic driver].” was run in a separate “new chat” 10 times for each driver oncogene (n=80). Each ChatGPT output was recorded and scored. The Relevance Score (RS) awarded 1 point for every NCCN preferred option and 0.5 points for each “other recommended” treatment listed in the AI-generated output, divided by the maximum possible score for the driver oncogene. Spurious recommendations were awarded 0 points. The Accuracy Score (AS) represents reported treatment options listed in NCCN over the total number of treatments in a report. Percentage of reports listing an NCCN-preferred 1L therapy, a clinical trial as an option, and character and word count were also captured. Results: The average length of the AI-generated NGS reports was 117 words (range: 44 – 232). The median number of treatments recommended was 5 (range: 3 – 8). An oncogenic driver-specific preferred 1L treatment was included in 55 reports (68.8%), and a recommendation to explore clinical trials was listed in 43 reports (53.8%). The RS for the total sample was 0.59 (95% CI: 0.52 – 0.65), and the AS was 46.0% (95% CI: 40.2% – 51.8%). Conclusions: ChatGPT can rapidly generate concise NGS reports with treatment options for NSCLC with driver oncogenes. Recommendation relevance was moderate, and accuracy was limited with high variability across oncogenes. Overall, ChatGPT recommendations were promising given the complexity of the task with no prompting or training provided to the AI. As LLM AI platforms mature, they may generate more relevant and accurate NGS reports, offering a potentially valuable tool for NGS report annotation for clinicians, and increased accessibility for patients.[Table: see text]

Importance of clinical decision making in precision oncology: Discordance in recommendation between next generation sequencing test vendor and molecular tumour board in India.

e15050 Background: Molecular tumour boards (MTB) facilitate discussions among medical professionals, scientists, bioinformaticians and geneticists for clinical application of precision oncology, via interpretation of various genomic sequencing studies, most commonly Next Generation Sequencing (NGS). NGS test vendor recommendation may differ from the MTB recommendations. We aim to study the discordance rate between these recommendations. Methods: MTB discussions between April 1, 2021 and December 31, 2021, were retrospectively evaluated. MTBs which assessed NGS reports with or without significant genetic mutations but with drug recommendations were included. NGS testing was done with multiple vendors. Patients from all over India, Nepal, Bangladesh, Dubai and Abu Dhabi were discussed. We excluded MTBs which assessed NGS data without any drug recommendations. We compared therapeutic recommendations provided in the NGS report with the first-choice recommendation in the MTB discussion. Recommendation discordance was defined as disagreement between the NGS recommendation and the first MTB recommendation that either instructed its implementation either immediately or on disease progression. Results: 70 MTB data was studied. 49 MTBs assessing NGS reports of 49 unique patients (median age 54 years; range, 32-82) were included in the study. 13 patients had lung cancer, 10 colorectal, 6 breast, 7 hepatopancreatobiliary, 4 carcinoma of unknown primary and 9 others. 17 did not provide any treatment recommendations and hence were excluded. 4 were discussions based on NGS reports previously discussed in an MTB that was already included in the study, hence were excluded. Recommendation discordance rate was 49% (24 of 49). Conclusions: To our knowledge, this is the first study that assessed discordance rate in recommendations between NGS report and MTB from a low- and middle-income country setting. Accurate clinical translation of complex genomic data is at the forefront of personalized cancer care.

Characteristics and predictive modeling of HER2 mutation in patients with non-small cell lung cancer.

e13559 Background: Recent advances in targeted therapies have increased the survival and quality of life for non-small cell lung cancer (NSCLC) patients. Understanding clinical characteristics and predicting HER2 status may facilitate efficient use of next-generation sequencing (NGS) testing and promote precision medicine/targeted therapies. Methods: NSCLC patients ≤65 years old, diagnosed between 01/2011 and 06/2021 who have NGS test results were identified in from two National Cancer Institute designated cancer centers in the Northeastern region of the United States. Demographics, behavioral, and clinical characteristics were extracted from electronic health record (EHR) data and EHR-linked cancer registry. Prevalence prediction modelling of HER2/ERBB2 mutation was performed using machine learning (ML) methods including logistic regression (LR), random forest (RF), naïve Bayes (NB), extreme gradient boosting (XGBoost) and multi-layer perceptron (MLP). 10- fold cross validation and k-nearest neighbor method were used to impute missing values. Under-sampling, oversampling and feature selection methods were tried. Results: Among 761 NSCLC patients with NGS reports available, 353 (46%) were ≤65 years old. Among 353 (mean age 57, 58% female, 88% non-Hispanic, 69% white, 57% had metastasis, 58% prescribed chemotherapy) 25 (7%) had a HER2 mutation. Based on univariate significance tests with LR on the complete data: cerebrovascular disease, BMI between 27.5-29.9, KRAS, EGFR and MET were significant with odds ratios (ORs) ranging from 0.2-3.9. In our prediction models, BMI, cerebrovascular disease, weight loss, hepatic metastases, diabetes, congestive heart failure, and KRAS, EGFR, MET and RET alteration status were significant predictors in multivariate analyses, suggesting the potential usefulness of these features in predicting HER2 mutation among patients with NSCLC. Area under the ROC curve (AUC) varied across different models from 70.9% (RF), 70% (XGB and MLP), 69.5% (LR) to 68.2% (Naïve Bayes). The best performing model with RF yielding an AUC ~71% can provide relative high specificity and acceptable sensitivity depending on the cut off. Conclusions: Our prediction models yielded AUC around 70% with the best performing model’s AUC being 71% in a subgroup aged ≤65. After further validation, our prediction model may be used to identify patients appropriate for repeat NGS testing who did not have HER2 testing in first molecular profiling. Future work is needed to predict status in repeat testing with previously test negative patients who relapse or have progression of disease. Random forest (RF) Model performance across different cutoff for the predicted probability. [Table: see text]

Comprehensive multigene profiling impact on clinical decisions in patients with advanced cancers: A multicenter, retrospective analysis.

e15163 Background: Comprehensive molecular profiling (CMP) via next generation sequencing (NGS) is currently applied in clinical practice and has enabled the detection of biomarkers of response to targeted treatment. Given the lack of the data about optimal range of analyzed alterations, the clinical usefulness of CMP has not been established. To address the challenges in incorporating genomics in treatment of oncology patients, we conducted a retrospective study. Methods: Preliminary analysis of NGS reports and clinical data of patients with advanced solid tumors was performed. CMP was conducted in 261 cancer patients (non-small cell lung cancer - 30,6%, colorectal cancer - 24,8%, pancreatic cancer - 7,4%, breast cancer- 6,2%; 59,3% were female, median number of prior systemic treatment lines was 2 (0-12)). In the present study, tumor molecular profile analysis was performed using 160 - 400 gene NGS panels, containing the majority of clinically significant genes for cancer treatment selection. We assessed the percentage of patients who were recommended to receive molecularly-matched therapy (MMT) after CMP, immunotherapy (IT) distribution of recommended therapies according to ESCAT tiers, median progression-free survival (PFS) – ratio ( = PFS after CMP / PFS before CMP, PFS2/PFS1), median PFS2. Results: MMT after CMP was recommended for 55 (21,1%) patients. ESCAT tiers could be assessed in 48 MMT patients (87%). For 31% of patients, administered therapy was considered as ESCAT I tier recommendation (data on mPFS2/1, mPFS2 not available). ESCAT tiers II, III, IV and IT administration were considered in 31, 8, 8 and 22% of patients respectively (detailed data is presented in Table 1). Most commonly altered genes were TP53 (53%), RAS (29%), APC (18%), CDKN2A/B (16%), PIK3CA(14%), NF1 (10%), BRCA1/2 (9%). Positive predictive factors for MMT after CMP were MSI-high status (HR 0,13, 95% CI 0,03 - 0,67), presence of EGFR alteration (HR 0,32, 95% CI 0,12 - 0,86). Presence of RAS mutation was considered a negative predictive factor (HR 2,78, 1,07 - 7,26). Conclusions: Available NGS panels offer a wide range of analyzed alterations that are not feasible in clinical practice. Most of the alterations that can cause impact on treatment plan can be detected with other methods than NGS. Further studies are needed to determine optimal range of alterations depending on tumor site. [Table: see text]

Real world early experience of precision neoadjuvant systemic therapy in patients with resectable NSCLC: Single institution experiences and perspectives.

e20594 Background: The incidence of early-stage NSCLC is increasing with lung cancer screening. However, up to 55% of NSCLC patients (pts) who undergo curative surgery die of recurrence. Neoadjuvant chemoimmunotherapy increases pathological clinical response and reduces tumor recurrences in resectable NSCLC pts without known EGFR/ALK alterations in March 2022. The uptake of this practice is low and there is an unmet need in incorporating effective targeted therapies for driver oncogenes in pts with early-stage NSCLC. As precision neoadjuvant therapy holds the great promise in improving surgical and pathological outcomes and cure rate, the objective of this study was to develop a clinical workflow to timely coordinate the care between oncology specialties and biomarker testing in pts with early-stage NSCLC. Methods: This study included early-stage NSCLC pts with no biomarker testing (Cohort A, 15) and biomarker testing (Cohort B, 19). Driver genomic alterations with targeted therapy in the metastatic setting were identified by next generation sequencing (NGS). Descriptive analysis was used for the type and frequency of each mutation and unpaired t-test was used for time interval analysis. Results: In Cohort A, 10/15 (67%) pts received neoadjuvant chemotherapy (4) or chemoimmunotherapy (6) before surgery, 3 proceeded to surgery, and 2 received concurrent chemoradiation for unresectable tumor. Of 19 pts in Cohort B, 2 did not have biomarker testing due to small tumor size. Of the 17 pts who had NGS testing, 7 (41%) had a targetable oncogene: EGFR l858R (2), EML4-ALK (1), MET amplification (1), RET fusion (1), and EGFR E20in (2). Two received matched targeted therapies on clinical trials. Median (range) dates for NGS testing were 12 (6-44) and for NGS report from pathologic diagnosis were 48 (17-93). Barriers for getting a matched neoadjuvant therapy included no matched targeted therapy (2), tumor progression (1), unresectable tumor (1), and patient preference (1). Of the pts that underwent surgery, there was no statistically significant difference in median time (range) to diagnosis between Cohort A [N = 12, 33 (7-100) days] and Cohort B [N = 8, 34 (9-82) days] (p = 0.74). The median time (range) to surgery from pathologic diagnosis was shorter in Cohort B [78 (29-137)] than Cohort A [166 (56-201)] (p = 0.0002). The difference was due to more pts in the Cohort A receiving neoadjuvant therapy than the biomarker cohort [10/12 (83%) vs 2/7 (25%)]. Conclusions: In this analysis, time to pathological diagnosis were similar between Cohorts A and B. Time for NGS testing was also comparable for standard of care in metastatic disease. However, time to NGS results from pathological diagnosis was longer than anticipated in Cohort B, which is an area of improvement in our workflow in prospective clinical evaluation for precision neoadjuvant treatment of early-stage NSCLC.

Evaluating the impact of a molecular tumor board in a community oncology center: Results from the Avera Cancer Institute's experience.

e13500 Background: The use of molecular tumor boards (MTB) has been widely documented at large academic medical centers, but there is limited experience and data available in smaller settings like community oncology centers. The Avera Cancer Institute has had a molecular tumor board in place since 2014, but the composition and methodology have changed over time. Our 10 years of hands-on experience and involvement in studies like I-PREDICT showed that in addition to molecular matching, considering factors like tumor type, previous treatment responses, comorbidities, medication coverage, and treatment availability provide crucial insights for determining optimal therapy for patients. Methods: We analyzed next-generation sequencing (NGS) testing results from August 2019 to December 2022. At the time of NGS testing result, providers were offered to have the case reviewed and presented at MTB. Our multidisciplinary team consists of members with backgrounds in bioinformatics, medical oncology, pathology, translational science, trial coordination, and pharmacy. Recommendations from the MTB discussions were provided in three categories: standard of care, clinical trials, and off-label options. Results: From August 2019 to December 2022, 1,514 qualified or resulted NGS reports were returned and considered for analysis. 733 (48%) were presented at the molecular tumor board and from this group, 629 patients were evaluable. Oncologists chose a recommended therapy 66% of the time. Of the remaining 34%, the most common reason for not utilizing recommendations was that patients were unable to receive further lines of systemic therapy. A subset analysis of 202 cases (June 2021 – August 2022) who either received a recommended therapy (n = 88, 43.6%, group 1) or not (n = 114, 56.4%, group 2) showed a survival benefit. Median overall survival was not reached in group 1, as compared with 14 months in group 2 (HR = 0.58; 95% CI, 0.37 – 0.92; p< 0.02). The median PFS was 8.2 months in group 1 versus 5 months in group 2 (HR, 0.67; 95% CI, 0.48 - 0.92; p< 0.017). Conclusions: The results of our study demonstrate the positive impact of a molecular tumor board review in helping oncologists determine the best treatment options for their patients. Our findings show that recommendations made by the board are well received and frequently adopted by providers. While providers generally prefer standard of care options or clinical trials, they are open to considering all available options to make informed therapy decisions.

A Real-World Experience from a Single Center (LPCE, Nice, France) Highlights the Urgent Need to Abandon Immunohistochemistry for ROS1 Rearrangement Screening of Advanced Non-Squamous Non-Small Cell Lung Cancer.

The detection of ROS1 rearrangements in metastatic non-squamous non-small cell lung carcinoma (NS-NSCLC) permits administration of efficient targeted therapy. Detection is based on a testing algorithm associated with ROS1 immunohistochemistry (IHC) screening followed by ROS1 FISH and/or next generation sequencing (NGS) to confirm positivity. However, (i) ROS1 rearrangements are rare (1-2% of NS-NSCLC), (ii) the specificity of ROS1 IHC is not optimal, and (iii) ROS1 FISH is not widely available, making this algorithm challenging to interpret time-consuming. We evaluated RNA NGS, which was used as reflex testing for ROS1 rearrangements in NS-NSCLC with the aim of replacing ROS1 IHC as a screening method. ROS1 IHC and RNA NGS were prospectively performed in 810 NS-NSCLC. Positive results were analyzed by ROS1 FISH. ROS1 IHC was positive in 36/810 (4.4%) cases that showed variable staining intensity while NGS detected ROS1 rearrangements in 16/810 (1.9%) cases. ROS1 FISH was positive in 15/810 (1.8%) of ROS1 IHC positive cases and in all positive ROS1 NGS cases. Obtaining both ROS1 IHC and ROS1 FISH reports took an average of 6 days, while obtaining ROS1 IHC and RNA NGS reports took an average of 3 days. These results showed that systematic screening for the ROS1 status using IHC must be replaced by NGS reflex testing.