To our knowledge, this is the first study to characterize STRN exon 3-ALK exon 20-derived transcripts in a dedicated cohort of oncocytic tumors of the thyroid using an integrated RNA-, genomic-, and protein-based approach. Our findings indicate that STRN::ALK-positive oncocytic thyroid neoplasms comprise distinct transcript architectures associated with divergent ALK protein expression patterns. Although genomic ALK rearrangement was confirmed by FISH in all transcript-positive tumors, detectable ALK protein expression was restricted to canonical in-frame fusions, whereas noncanonical variants remained consistently immunonegative. These findings suggest that fusion architecture may influence preservation of kinase-domain organization and ALK protein expression, supporting cautious interpretation of ALK rearrangements in oncocytic thyroid neoplasms.

Importantly, both canonical and noncanonical STRN::ALK-derived transcripts were identified in tumors demonstrating genomic ALK rearrangement by FISH, indicating that genomic-level ALK disruption alone may not predict downstream protein expression. The restriction of ALK immunoreactivity to tumors harboring canonical in-frame fusions, together with the predicted preservation of kinase-domain organization in these cases, suggests that only a subset of ALK-rearranged oncocytic thyroid neoplasms generate fusion transcripts compatible with stable ALK protein expression. These observations support the existence of a hierarchical relationship across genomic, transcript, and protein levels in which the biological consequences of ALK rearrangement may depend on fusion architecture rather than rearrangement status alone.

The recurrent detection of noncanonical variants across independent PCR amplification, cloning, and sequencing experiments supports the reproducibility of these findings under the experimental conditions applied in this study. In addition, all noncanonical transcripts were consistently associated with FISH-confirmed ALK rearrangement. Together, these observations suggest that the identified variants are unlikely to represent purely stochastic technical artifacts. Nevertheless, given the use of archival FFPE material, short-amplicon RT-PCR, and cloning-based Sanger sequencing, alternative explanations, including RNA fragmentation, template switching, preferential amplification of shorter products, and recurrent sequence-dependent PCR artifacts, cannot be definitively excluded and may contribute to the observed diversity of junction structures.

Structural modeling indicated that canonical fusions preserve ALK kinase-domain architecture, whereas noncanonical variants are associated with predicted structural disruption. However, these in silico predictions should be interpreted as supportive but not definitive evidence of functional integrity. Additional biological considerations further complicate the interpretation of these variants. Out-of-frame transcripts containing premature termination codons, particularly those classified as Type II variants, may be subject to nonsense-mediated mRNA decay, potentially limiting production of stable protein products and raising the possibility that some predicted truncated proteins are not expressed in vivo. Even in the absence of preserved kinase-domain features, STRN::ALK fusion products retaining STRN-derived regions could theoretically exert biological effects independent of ALK signaling. In particular, potential ALK-independent mechanisms, including disruption of STRIPAK complex assembly and PP2A-associated signaling pathways, may be considered given the scaffolding role of STRN (striatin) in this network. Although speculative, these considerations further support interpreting noncanonical variants as rearrangements of uncertain biological significance.

Consistent with this, discordance between FISH and IHC likely reflects differences in fusion architecture rather than a single uniform ALK alteration, as also reported in other tumor types, including thyroid neoplasms [18, 22, 33, 34]. In oncocytic thyroid tumors, this is particularly relevant in the context of marked mitochondrial accumulation and genomic instability, which may promote structurally diverse rearrangements with variable biological consequences [5, 6].

Canonical STRN::ALK fusions were identified exclusively in oncocytic thyroid carcinomas within this cohort, whereas noncanonical variants were observed in both carcinomas and adenomas. Given the limited number of canonical cases, this observation should be interpreted cautiously and considered descriptive of the present cohort rather than evidence of a tumor-type-specific biological distribution. Although no significant clinicopathological differences were identified between canonical and noncanonical groups, these comparisons remain substantially underpowered and should not be interpreted as evidence of biological equivalence. Nevertheless, the restriction of canonical in-frame fusions to malignant tumors raises the possibility that structurally intact rearrangements may be more likely to generate biologically relevant ALK protein products.

The apparent rarity and heterogeneity of these alterations, together with differences in cohort composition, may also account for the absence of ALK rearrangements in two large genomic profiling studies of oncocytic thyroid carcinomas by Gopal et al. and Ganly et al. [5, 6], despite isolated reports of ALK-rearranged oncocytic thyroid tumors confirmed by FISH and immunohistochemistry [26, 27]. In the present cohort, although STRN::ALK-derived transcripts were detected in 9 of 56 tumors (16%) using this targeted assay, only 2 of 56 cases (3.6%) harbored canonical in-frame fusions associated with detectable ALK protein expression. These observations suggest that biologically relevant canonical STRN::ALK alterations likely represent a substantially smaller subset of oncocytic thyroid neoplasms than overall transcript detection rates alone might imply, while structurally atypical transcripts may contribute to apparent variability across studies. Differences in cohort composition, assay design, and sensitivity for detection of structurally altered transcripts may also contribute to these discrepancies. Together, these observations suggest a potential role for integrated molecular and protein-based assessment in the interpretation of ALK status and stratification of cases for future investigation of ALK-targeted therapies.

This study has limitations inherent to its design. The targeted RT-PCR approach was restricted to a single STRN exon 3-ALK exon 20 junction configuration and therefore did not capture the full spectrum of ALK fusion diversity. Accordingly, ALK fusions involving alternative breakpoints within STRN or ALK, as well as rearrangements involving non-STRN fusion partners, may have been missed.

In addition, RNA degradation in archival FFPE material may have contributed to false-negative RT-PCR results despite the potential presence of underlying genomic ALK rearrangements. The possibility of RT-PCR-negative/FISH-positive discordant cases cannot be excluded, as FISH analysis was restricted to RT-PCR-positive tumors, introducing a degree of ascertainment bias. Consequently, the reported detection rate reflects assay performance rather than the true prevalence of ALK rearrangements in oncocytic thyroid tumors.

FFPE-derived RNA is inherently fragmented, increasing susceptibility to artifacts in short-amplicon assays. Although sequence inspection did not identify consistent microhomology, direct repeats, or other features suggestive of polymerase slippage, sequence-dependent artifacts cannot be excluded. Moreover, the analyzed products derive from RT-PCR-amplified, cloned, and sequenced fragments and therefore represent assay-derived transcript segments rather than full-length fusion transcripts. Thus, the data represent transcript fragments generated under assay conditions rather than complete fusion structures.

Cloning-based Sanger sequencing provides limited resolution of transcript complexity and is subject to amplification bias, which may preferentially enrich specific amplicons while underrepresenting low-abundance variants. Consequently, functional interpretation based on immunohistochemistry and in silico structural modeling should be considered preliminary, as these methods do not provide direct experimental validation of biological activity. Orthogonal approaches, including long-read RNA sequencing, comprehensive genomic profiling, and functional assays, are required to define transcript architecture and biological relevance. Notably, the recurrent detection of structurally distinct STRN::ALK-derived transcripts across independent cases supports the robustness of the findings and argues against isolated technical artifacts.

Finally, although all cases underwent centralized expert histopathological review according to WHO 2022 criteria, formal inter-observer agreement analysis was not performed, and the modest sample size limited clinicopathological correlations. Future studies incorporating larger, independent cohorts with integrated RNA-based and functional validation will be essential to confirm and extend these findings, particularly regarding the biological significance of noncanonical variants.

In summary, STRN::ALK rearrangements in oncocytic thyroid tumors appear to comprise two categories with distinct structural and predicted functional characteristics: rare canonical in-frame fusions associated with detectable ALK protein expression and predicted preservation of kinase-domain features, and more frequent noncanonical variants that are structurally altered and of uncertain biological relevance. Collectively, these findings suggest that detection of STRN::ALK-derived transcripts alone may not uniformly reflect biologically relevant fusion protein expression, supporting the importance of integrative molecular interpretation when evaluating STRN::ALK-positive oncocytic thyroid tumors.