A tailored histology-driven molecular profiling algorithm proposal for salivary gland cancers.

Salivary gland cancers (SGCs) are rare, heterogeneous malignancies with limited systemic therapies. This study aimed to define a histology-driven molecular profiling (MP) algorithm for SGCs to optimize testing strategies and support therapeutic decision-making. We retrospectively analyzed all SGC patients who underwent MP (2016-2023) for gene rearrangements and pathogenic/likely pathogenic DNA variants (P-LPV). We considered adenoid cystic carcinomas (AdCCs) versus non-AdCCs, the latter as low aggression (LA) or high aggression (HA). MP included DNA/RNA next-generation sequencing (NGS). Immunohistochemistry (IHC) ± fluorescence in situ hybridization (FISH) were used for human epidermal growth factor receptor 2 (HER2) status, and IHC for androgen receptor (AR) status. FISH, NGS and/or IHC were used for neurotrophic tyrosine receptor kinase (NTRK) detection. Among a total of 253 SGCs, AdCCs (132/253, 52%) had MYB/MYBL1 fusions in 47% of cases, all lacking P-LPV. Among HA non-AdCCs (84/253, 33%), P-LPV were found in 55% of cases, including TP53 (34%), PIK3CA (18%), HRAS (14%) and PTEN (5%), in the absence of gene fusions. We found P-LPV in 50% of LA non-AdCCs (37/253, 15%), with ETV6-NTRK3 fusion in one secretory carcinoma. HER2 analysis identified HER2 positivity in 23% of cases (all HA non-AdCC), with 36% HER2-low and 41% HER2-negative. We found less P-LPV in HER2-positive (17.6%) versus HER2-low (67%) or HER2-negative (64%), regardless of AR status. MP-guided therapy was delivered in 4% of cases. Based on our results, we propose a stepwise, histology-driven MP algorithm for SGCs. In AdCCs, we recommend initial FISH for MYB/MYBL1 fusions: fusion-positive cases typically lack other P-LPV, sparing further MP. In fusion-negative cases, DNA NGS can be considered. In HA non-AdCCs, HER2 and AR should be tested upfront, proceeding with DNA NGS in HER2-low/negative tumors. In LA non-AdCCs, we recommend ETV6-NTRK3 FISH followed by DNA/RNA NGS in fusion-negative cases. Our data support the clinical implementation of a tailored, histology-driven MP algorithm, potentially optimizing the genomic testing resources in SGCs.