For patients with metastatic disease, identification of both the primary tumor type and molecular alterations increases eligibility for targeted therapies. The 92-gene assay (CancerTYPE ID) is a validated gene expression classifier of 50 tumor types and subtypes for metastatic patients with unknown or uncertain diagnoses. Multimodal biomarker testing identifies actionable alterations to guide therapy selection. Here, a database of metastatic cases integrating molecular cancer classification with secondary biomarker analysis was analyzed to assess clinical utility in patients with lung cancer with known histology but unknown primary site of origin. MOSAIC (Molecular Synergy to Advance Individualized Cancer Care) is an IRB-approved, de-identified database of metastatic cases with unknown or uncertain diagnoses submitted for CancerTYPE ID testing and tissue type-guided multimodal biomarker testing (NeoTYPE profiles, Neogenomics). Metastatic cancers classified as adenocarcinoma, squamous cell carcinoma (SqCC), carcinoid and small/large cell carcinoma were evaluated by next-generation sequencing (NGS), fluorescent in situ hybridization (FISH) and immunohistochemical (IHC) analyses to identify primary site of tumor. Molecular diagnoses by CancerTYPE ID of 2151 patients included 271 (12.6%) with non-small cell lung carcinoma features [NSCLC; 157 lung adenocarcinomas (7.3%), 114 SqCCs (5.3%)], 10 (0.5%) lung carcinoid tumors and 71 (3.3%) with small/large cell carcinoma features. Gene fusion analysis by FISH identified 5 ALK (5.9%), 5 MET (6.0%), 1 RET (1.2%), and 1 ROS1 (1.2%) alterations in adenocarcinoma, as well as 2 MET alterations (3.4%) in SqCC. The mutational frequency of the 10 most commonly mutated genes for each lung cancer subtype are shown in Table 1. Mutations identified included genes for which targeted therapies are available, including capmatinib for MET exon 14 skipping mutations and dabrafenib for BRAF V600E mutations. Multimodal biomarker testing for pan-TRK identified 1 (1.5%) lung adenocarcinoma, 7 (15.2%) SqCC and 6 (24.0%) small/large cell carcinoma cases, which were eligible for larotrectinib and entrectinib. PD-L1 expression was seen in 95 (83.3%) lung adenocarcinoma, 58 (74.4%) SqCC, 5 (62.5%) lung carcinoid and 32 (62.8%) small/large cell carcinoma cases.Table 1Top 10 gene mutations frequency by NGSMain Type (Subtype)Top 10 Gene Mutations Detected by NGSNSCLCLung AdenocarcinomaTP53KRASKEAP1KMT2DSTK11ARID2BRCA2CHD2EPHA5KMT2C65.6%41.9%26.7%20.0%15.6%14.3%14.3%14.3%14.3%14.3%Squamous Cell Carcinoma (Lung)TP53KMT2DARID1ASMARCA4FAT1PIK3CAKEAP1LRP1BMTORNTRK371.0%32.4%23.5%23.5%22.2%20.0%17.7%16.7%16.7%16.7%OtherNeuroendocrine (lung carcinoid)*----------Neuroendocrine Small/large cell lung carcinomaTP53CTNNB1APCDICER1RANBP2RB1RUNX1SETD2SMARCA4TSC267.4%27.3%18.2%18.2%18.2%18.2%18.2%18.2%18.2%18.2%*low case number, unable to analyze; % indicated cases with detected gene abnormalities out of all cases with available test results. Open table in a new tab *low case number, unable to analyze; % indicated cases with detected gene abnormalities out of all cases with available test results. Analysis of the MOSAIC database identified a subset of patients with metastatic cancers eligible for lung cancer directed targeted therapy and immunotherapy based on tumor type, genetic alterations, and PD-L1 expression. Molecular profiling combined with cancer classification may lead to improved therapy options for patients with advanced disease of unknown primary and may provide further evidence for subsequent combination trials of targeted therapies or immunotherapies to improve patient outcomes.