301 Sex-Based Variation in Diagnostic IHC Marker Expression Across Normal Human Tissues: A GTEx-Based Analysis

Abstract Introduction/Objective Immunohistochemical (IHC) markers are foundational tools in pathology for identifying the tissue of origin, particularly in poorly differentiated, metastatic, or ambiguous tumors. These markers guide diagnostic decision-making and therapeutic direction. While many are presumed to exhibit tissue-specific expression, less attention has been given to how biological variables such as sex might influence baseline marker expression in non-diseased tissues. This gap poses diagnostic risks, particularly in hormone-responsive or sex-specific cancers, where subtle expression differences may impact interpretation. Although prior research has explored tissue-based protein expression using immunostaining (e.g., via the Human Protein Atlas), these approaches are limited by antibody performance, categorical scoring, and technical variability. The Genotype-Tissue Expression (GTEx) Project provides an unparalleled resource for examining transcript-level expression in over 50 normal human tissue types with matched donor metadata, including sex. In this study, we leverage GTEx v8 RNA-seq data to examine sex-based differences in the expression of five widely used diagnostic IHC markers GATA3, CDX2, TTF1 (NKX2-1), PAX8, and S100B. We hypothesize that certain markers exhibit significant sex-specific expression patterns that could influence clinical interpretation and propose that incorporating sex-aware transcriptomic references may improve diagnostic accuracy in surgical pathology. Methods/Case Report We performed a secondary analysis of publicly available transcriptomic data from the GTEx Project (v8), which includes RNA-seq expression profiles (in transcripts per million, TPM) across 54 non-diseased human tissue types. Five diagnostic immunohistochemical (IHC) markers—GATA3, CDX2, TTF1 (NKX2-1), PAX8, and S100B were selected based on their routine use in pathology for determining lineage or tissue of origin. GTEx sample-level and subject-level metadata were used to stratify expression data by tissue type and donor sex. Median TPM values were calculated for each marker across all tissues, and sex-based comparisons were made using the non-parametric Mann-Whitney U test. To assess the magnitude and direction of sex-based variation, we computed log2(female/male) fold-changes for each tissue-marker combination. Statistical significance was defined as p < 0.05, with adjustment for multiple comparisons where applicable. We visualized expression patterns using heatmaps, tissue-stratified boxplots, and barplots of sex-based fold changes. Tissue-specific patterns were further examined for markers with significant sex differences. All data analyses were conducted using Python (v3.11) with the pandas, seaborn, and matplotlib libraries. This approach enabled the development of a sex-stratified, transcript-level atlas for commonly used IHC markers in non-diseased human tissues. Results Significant sex-based expression differences were identified for several diagnostic IHC markers. PAX8 demonstrated the strongest sex-biased expression, with significantly higher transcript levels observed in female kidney and thyroid tissues (p < 1e-15), consistent with its known regulation in hormonally responsive tissues. S100B also showed marked sex-based variation, with increased expression in male brain and nerve tissues (p < 1e-8). TTF1 (NKX2-1), although more broadly expressed, exhibited modest but statistically significant differences between sexes in the lung and thyroid (p < 1e-5). In contrast, GATA3 and CDX2 showed no significant differences in transcript expression by sex across all tissues analyzed. Heatmaps and barplots of log2(female/male) fold changes highlighted tissue-specific sex effects for PAX8 and S100B, suggesting that these markers may be influenced by hormonal or genetic sex-related factors. Boxplots confirmed non-overlapping expression distributions in key tissues for these markers. The results suggest that diagnostic interpretation of these IHC markers—particularly when used as lineage indicators in cancer diagnosis may benefit from contextual sex-aware references. This study is among the first to systematically explore sex-specific variation in diagnostic marker expression using transcriptomic data, providing a foundation for more personalized pathology and machine learning-based diagnostics. Conclusion Our study reveals that biological sex significantly influences the transcript-level expression of certain diagnostic IHC markers across normal human tissues. Markers such as PAX8 and S100B show consistent and significant sex-biased expression patterns, which may affect IHC interpretation in hormone-responsive organs like the kidney, thyroid, and brain. These findings underscore the importance of integrating sex-aware data into diagnostic pathology frameworks, especially in the era of precision medicine. By leveraging GTEx v8 one of the most comprehensive transcriptomic resources from non-diseased human donors we provide a reproducible, quantitative, and antibody-independent reference for baseline IHC marker expression. This work supports more accurate diagnostic panel design and helps mitigate potential misinterpretation arising from biologic variability. Furthermore, the dataset and visual outputs generated here can serve as valuable resources for AI and digital pathology models that incorporate molecular features. Future directions include expanding the marker panel, validating findings with protein-level data, and assessing diagnostic misclassification rates in clinical cohorts. Overall, our study bridges a key gap in the field and demonstrates the relevance of sex-stratified reference data for improving interpretive accuracy, diagnostic equity, and educational resources in pathology and laboratory medicine.