Introduction
Abstract
Fibroblast growth factors (FGF) and receptors (FGFR) are involved in various cellular processes, including tissue development, angiogenesis and tissue regeneration. The FGFR1, FGFR2 and FGFR3 genes can be abnormally activated by chromosomal translocations, mutations or amplifications, which lead to the deregulation of intracellular signaling and may promote the development of cancer. These alterations also constitute predictive biomarkers allowing to select the patients that may benefit from FGFR inhibitors.
Objective
The aim of the Master work was to study the accuracy of break-apart FISH analysis for the detection of FGFR rearrangements in tumor tissues, by correlating the hybridization patterns observed by FISH with the presence and type of fusion transcript detected by RNA-based NGS fusion panels in various types of cancer.
Methodology
From the Lausanne Institute of Pathology database, we retrieved 36 samples with both FISH and NGS data available for FGFR1, FGFR2 and/or FGFR3 rearrangements. Of these, 13 underwent a double FGFR2/FGFR3 analysis, giving us a database containing 49 paired results. The results were analyzed in detail, including the types of cancer, studied genes, observed FISH patterns and the fusion partners detected by NGS. FISH patterns were classified into classical, atypical, equivocal and no rearrangement, and correlated with fusion transcript results by NGS, which we considered the gold standard. Performance parameters were calculated for each single FISH probe and for FGFR FISH assays in general.
Results
The cohort comprised 16 urothelial carcinomas, 5 cholangiocarcinomas and 15 other cancers, including 8 cases with FGFR gene fusions detected by NGS. FGFR1 FISH (n=10) frequently showed atypical rearrangements (9/10), all negative by NGS. FGFR2 FISH (n=22) revealed 4 classical and 3 atypical rearrangements of whom 6 had fusions confirmed by NGS, including 3 cholangiocarcinomas. FGFR3 FISH (n=17) were frequently equivocal (8/17) or discordant with NGS (4/17).
Overall, FISH had low specificity for FGFR1 rearrangements, high sensitivity and acceptable specificity for FGFR2, and was frequently inconclusive for FGFR3. NGS was necessary to accurately detect FGFR fusions in most scenarios, except for cholangiocarcinomas with a classical FGFR2 rearrangement, where FISH was also reliable.
Conclusion
RNA-based NGS remains the gold standard for FGFR fusion detection in cancer tissues, while break- apart FISH is less reliable, and applicable only in specific scenarios. These data have implications for biomarker testing in cancer patients.