When morphology suggests a fusion: A pathology-centered escalation framework from histology and immunohistochemistry...

Gene fusions have moved from ancillary molecular findings to defining events for an expanding group of neoplasms, with consequences for tumor classification, diagnostic certainty, and, in selected settings, targeted therapy. Yet the practical problem facing pathologists is not simply whether fusion testing is available, but when morphology should trigger it, which assay should be selected, and when conventional testing should be bypassed in favor of RNA-based sequencing. This structured narrative review synthesizes established morphologic, immunohistochemical, and molecular principles into a cross-cutting, tumor-type-agnostic, pathology-centered decision framework. We first summarize morphologic red flags that raise the pre-test probability of a fusion-driven neoplasm, including monomorphic spindle-cell, primitive round-cell, epithelioid/clear-cell, and secretory or salivary-like patterns, as well as unusual age-site combinations and morphology-immunophenotype discordance. We then position immunohistochemistry (IHC) as a triage layer rather than a molecular surrogate in all cases, emphasizing marker specificity, clone- and platform-dependent performance, and the distinction between screening and near-diagnostic surrogate markers. The relative roles of fluorescence in situ hybridization (FISH), reverse-transcription polymerase chain reaction (RT-PCR), DNA-based next-generation sequencing (NGS), targeted RNA-based NGS (RNA-NGS), and whole-transcriptome approaches are compared from a diagnostic decision-making perspective. Particular attention is given to the potential role of direct-to-RNA testing in tissue-limited or fusion-diverse cases, pre-analytical limitations of formalin-fixed paraffin-embedded (FFPE) material, and the interpretation of discordant, novel, or multiple fusion calls. Finally, we propose an expert-derived, hypothesis-generating closed-loop escalation framework in which molecular data resolve morphologic questions rather than replace morphologic reasoning. The framework has not been prospectively validated; its decision thresholds and potential effects on diagnostic yield, tissue use, cost, and time to diagnosis require formal evaluation.