Little is known about molecular testing on tumor tissue retrieved from stained sections, for which there may be a clinical need. We retrospectively analyzed 112 sections from 56 tumor patients using either fluorescence in situ hybridization (FISH) with different probes (19 sections from 17 patients) or Sanger or targeted next generation sequencing for detection of BRAF, EGFR, KRAS, C-KIT, and TP53 mutations (93 sections from 39 patients). Tumor tissue sections had been stained by hematoxylin and eosin (H&E) (42 sections) or by immunohistochemistry for cytoplasmic or nuclear/nuclear-cytoplasmic markers (70 sections) with a peroxidase (P-IHC, with 3,3'-diaminobenzidine as chromogen) or alkaline phosphatase label (AP-IHC, with Warp Red™ as chromogen). For FISH analysis, the concordance rate between the original diagnosis and that obtained on H&E- or P-IHC-stained tissue sections (AP-IHC was not on record for this set of patients) was 95% (18 out of 19 tumor sections). Only one tumor sample, diffusely positive for MLH1, did not yield any nuclear hybridization signal. For sequencing analysis, the concordance rate was 100% on negative P-IHC and positive AP-IHC-stained sections, regardless of the subcellular localization of the reaction product. Mutations were detected in only 52% of cases expressing nuclear/nuclear-cytoplasmic markers, regardless of the sequencing technology used (p = 0.0002). In conclusion, stained sections may be a valuable resource for FISH or sequencing analysis, but on cases expressing nuclear markers sequencing results need to be interpreted cautiously.
Keywords: Direct sequencing; Fish; Immunohistochemistry; Next generation sequencing.