Defining a FISH chromosomal aneusomy panel for predicting lung cancer in the setting of CT-detected lung nodules.

Abstract

10580 Background: Early detection of lung cancer by CT is supported by the National Lung Screening Trial; but while sensitivity of CT is high, specificity is low. Biomarkers to predict the malignant nature of CT detected lung nodules would have great clinical utility. We previously found in a nested case-control study that a 4-target chromosomal aneusomy (CA) FISH panel exhibited sensitivity/specificity of 76%/88% in sputum samples of heavy smokers obtained within 18 months of lung cancer diagnosis. We hypothesized that CA-FISH may be an effective biomarker to assist with clinical decisions in the setting of CT detected lung nodules of indeterminate etiology and attempted to identify a more effective reagent. Methods: Homebrew probes encompassing genomic sequences of EGFR, NXK2-1, PIK3CA, MYC, BRF2, SOX2, PPMID, FGFR1 and the commercial reagent LSI D5S721/D5S23 (Abbott Molecular) were combined in 2-4-target FISH assays to investigate tissue copy number in early stage lung squamous cell carcinoma (SCC, N=19) and adenocarcinoma (AC, N=20). Logistic regression models were used to estimate predictive discrimination [sensitivity, specificity, area under the ROC curve (AUC)]. Results: Copy number gain was largely detected for all markers (mean range 3.39 - PPMID to 4.67 - MYC). Mean copy number of PI3CA, BRF2, SOX2 and FGFR1 were significantly higher in SCC than AC, while NKX2 and MYC were marginally higher in AC than SCC. Gene amplification was detected for all 9 markers, most frequently for SOX2 and FGFR1 (6 and 5 cases) with significant overlap between FGFR1/BRF2 and SOX2/PIK3CA. Based on the AUC results and the existence of targeted inhibitors, the probe set EGFR/FGFR1/MYC/PIK3CA was selected as a candidate for further development as an adjunct to CT screening for early detection. For this selected set,the optimal cutoff based in the linear predictor from logistic regression yields a sensitivity and specificity of 0.85. Conclusions: A highly sensitive/specific CA-FISH probe set was identified which may well complement CT screening in the early diagnosis of lung cancer. This probe set will be tested in the setting of CT detected lung nodules. (Supported by LUNGevity and NCI-Lung Cancer SPORE grants).