RESUMEN
BACKGROUND: Clinical guidelines highly recommended the detection of potentially targetable genetic aberrations such as anaplastic lymphoma kinase (ALK) rearrangements in patients with non-small cell lung cancer (NSCLC). Few methods, such as the ALK break apart FISH assay and IHC for ALK protein, are approved for routine diagnostics. However, some challenges exist in selecting the most reliable, robust and cost-effective algorithm, especially for large-scale screening of NSCLC patients. MATERIALS AND METHODS: A total of 4002 FFPE samples from Russian patients with NSCLC were tested for ALK rearrangement using two algorithms: FISH testing only (2334 samples) and IHC screening supported by FISH in positive or equivocal cases (1546 samples). Cross validation of the methods was performed blindly in 122 specimens. All discrepant IHC/FISH cases were examined for unbalanced 5'/3'-end ALK expression and variant-specific RT PCR. RESULTS: The sensitivity and specificity of IHC compared to FISH was 100% and 99%, respectively, therefore initial IHC screening was strongly supported. The prevalence of ALK rearrangements was estimated to be 7.8% and 6.6% for the FISH and IHC/FISH groups, respectively, with significant correlations for female gender, non-smoking status and age below 60. The use of FISH after IHC screening revealed 10 additional positive cases among equivocal samples (13.4%). Seven IHC/FISH cases (0.5% of the total group) characterized as discordant were reevaluated, and four were reclassified as truly discrepant. The PCR-based investigation revealed chimeric transcripts in IHC-/FISH+ and IHC+/FISH borderline samples, while no transcript was found in two IHC+/FISH- cases. CONCLUSION: These results reveal the utility of the two-step testing algorithm for the evaluation of potentially complicated cases with preanalytic defects, providing additional information for IHC equivocal cases without a significant increase in cost. The evaluation of discrepant cases appears to be necessary to better understand ALK biology and should be included in the routine testing algorithm.