Spatial heterogeneity challenges limited sampling of claudin 18 expression in cholangiocarcinoma
Abstract
Abstract Background Claudin 18.2 (CLDN18.2) has emerged as a therapeutic target in gastrointestinal cancers, and CLDN18-directed strategies are entering trials in cholangiocarcinoma. Because biomarker classification may depend on the proportion of tumor cells with moderate-to-strong staining, spatial heterogeneity could affect assessment from limited tissue. We aimed to characterize intratumoral heterogeneity of CLDN18 immunohistochemical expression in cholangiocarcinoma. Methods We analysed 85 whole-section blocks from 38 patients in a selected, CLDN18-enriched cohort derived from a previously published cholangiocarcinoma series. CLDN18 expression was quantified as the percentage of tumor cells with moderate-to-strong (2+/3+) membranous staining and classified at a ≥ 75% threshold. Spatial expression patterns were annotated, and patient-level whole-section classifications were compared with prior tissue-microarray results. Results At patient level, 11 of 38 cases were uniformly above the ≥ 75% threshold across evaluated blocks, 4 were heterogeneous, and 23 were uniformly below threshold. Among patients with at least two assessable blocks, interblock heterogeneity was observed in 4 of 29 cases (13.8%). Overall, 29 of 85 blocks (34.1%) met the ≥ 75% threshold. Tissue-microarray and whole-section classifications agreed in 30 of 38 patients (78.9%; κ = 0.57). Independent reassessment showed high agreement for block-level threshold classification (92.9%; κ = 0.85), while spatial-pattern assignment showed substantial agreement (81.2%; κ = 0.67). Conclusions In this selected cohort enriched for tumors with prior partial or strong CLDN18 expression, staining was consistent across sampled blocks in most cases, while 4 of 29 multiblock cases crossed the ≥ 75% threshold between blocks. A single TMA core, used here as a model of limited tissue sampling, did not always reflect the broader whole-section classification. These findings identify sampling heterogeneity as a practical consideration for CLDN18 assessment in CCA and suggest that spatial pattern may provide useful context alongside percentage staining.