Paraffin-based molecular diagnosis of lung cancer reproduces morphologic and molecular subtypes of lung cancer

Abstract

7579 Background: Molecular profiling of lung cancer by gene expression analyses has documented potential to guide therapy. However, quality ‘fresh’ tissue for RNA analyses is generally unavailable in clinical practice. We introduce a (q)RT-PCR assay and analytic method for profiling lung cancers from clinically obtained formalin-fixed paraffin-embedded tissues. Methods: Approximately 1,000 DNA microarrays were analyzed to select genes distinguishing the major histological variants (adenocarcinoma, small cell carcinoma, etc) and previously described molecular subtypes of lung cancer. Based on gene expression, a classifier of 62 genes was constructed to assign each sample to its morphologic tumor type, as well as to risk stratify by molecular subtypes. A real-time qRT-PCR assay was developed to evaluate the expression of the 62 genes from paraffin-embedded samples. This assay was used to profile RNA extracted from a cohort of surgically treated lung cancer patients. Samples were procured as fresh frozen and formalin-fixed, paraffin-embedded and were archived between 1–15 years. Results: Fifty-eight of 62 genes passed performance criteria and were used for analyses. Normalized gene expression was used for sample classification. The cohort contained a broad spectrum of tumors in proportions consistent with clinical practice. Gene amplification was successful in 139 of 142 (98%) lung cancer samples. Matched frozen-paraffin and replicate paraffin samples had mean correlations of approximately 80%. Linear discriminant analysis of gene expression data agreed with morphologic classification by light microscopy in >99% of cases. More importantly, the method successfully re-identified molecular subtypes of lung cancer for the first time through the use of a paraffin-based assay. Clinical outcomes previously associated with molecular tumor subtypes, including differential survival and metastatic patterns, were again seen in our cohort. Conclusions: We describe for the first time a clinically meaningful and robust molecular diagnosis of a clinical cohort of lung cancer patients which is complementary to morphologic cancer diagnosis. This assay is easily implemented using specimens routinely collected in current patient care. No significant financial relationships to disclose.