Recent advances in consortium-scale genome-wide association studies (GWAS) have highlighted the involvement of common genetic variants in autism spectrum disorder (ASD), but our understanding of their etiologic roles, especially the interplay with rare variants, is incomplete. In this work, we introduce an analytical framework to quantify the transmission disequilibrium of genetically regulated gene expression from parents to offspring. We applied this framework to conduct a transcriptome-wide association study (TWAS) on 7,805 ASD proband-parent trios, and replicated our findings using 35,740 independent samples. We identified 31 associations at the transcriptome-wide significance level. In particular, we identifiedPOU3F2(p = 2.1E-7), a transcription factor mainly expressed in developmental brain. Gene targets regulated byPOU3F2showed a 2.7-fold enrichment for known ASD genes (p = 2.0E-5) and a 2.7-fold enrichment for loss-of-functionde novomutations in ASD probands (p = 7.1E-5). These results provide a novel connection between rare and common variants, whereby ASD genes affected by very rare mutations are regulated by an unlinked transcription factor affected by common genetic variations.

Background/Objectives: The tumor microenvironment (TME) plays a critical role in cancer progression by shaping immune responses and influencing patient outcomes. We hypothesized that the relative proximity of specific immune cell pairs to cancer cells within the TME could help predict their pro- or anti-tumor functions and reflect clinically relevant immune dynamics. Methods: We analyzed imaging mass cytometry (IMC) data from lung adenocarcinoma (LUAD) and triple-negative breast cancer (TNBC) cohorts. For each immune cell pair, we calculated a relative distance (RD) score, which quantifies the spatial difference in proximity to cancer cells. We assessed the prognostic and predictive significance of these RD-scores by comparing them with conventional features such as cell fractions, densities, and individual cell distances. To account for variations in cell abundance, we also derived normalized RD-scores (NRD-scores). Results: RD-scores were more strongly associated with overall patient survival than standard immunological metrics. Among all immune cell pairs, the RD-score comparing the proximity of B cells to that of intermediate monocytes showed the most significant association with improved survival. In TNBC, RD-scores also improved the distinction between responders and non-responders to immunochemotherapy and chemotherapy. Normalized RD-scores reinforced these findings by minimizing the influence of cell density and further highlighting the importance of immune cell spatial relationships. Conclusions: RD-scores offer a spatially informed biomarker that outperforms traditional metrics in predicting survival and treatment response. This approach provides a new perspective on immune cell behavior in the TME and has potential utility in guiding personalized cancer therapies and patient stratification.

Malignant pleural mesothelioma (MPM) is a rare but lethal pleural cancer with high intratumor heterogeneity (ITH). A recent study in lung adenocarcinoma has developed a clonal gene signature (ORACLE) from multiregional transcriptomic data and demonstrated high prognostic values and reproducibility. However, such a strategy has not been tested in other types of cancer with high ITH. We aimed to identify biomarkers from multi-regional data to prognostically stratify MPM patients. We generated a multiregional RNA-seq dataset for 78 tumor samples obtained from 26 MPM patients, each with one sample collected from a superior, lateral, and inferior region of the tumor. By integrating this dataset with the Cancer Genome Atlas MPM RNA-seq data, we selected 29 prognostic genes displaying high variability across different tumors but low ITH, which named PRACME (Prognostic Risk Associated Clonal Mesothelioma Expression). We evaluated PRACME in two independent MPM datasets and demonstrated its prognostic values. Patients with high signature scores are associated with poor prognosis after adjusting established clinical factors. Interestingly, the PRACME and the ORACLE signatures defined respectively from MPM and lung adenocarcinoma cross-predict prognosis between the two cancer types. Further investigation indicated that the cross-prediction ability might be explained by the high similarity between the two cancer types in their genomic regions with copy number variation, which host many clonal genes. Overall, our clonal signature PRACME provided prognostic stratification in MPM and this study emphasized the importance of multi-regional transcriptomic data for prognostic stratification based on clonal genes.

Patients with lung squamous cell carcinoma (LUSC) are often diagnosed at advanced stages, limiting opportunities for early intervention. LUSC develops through multistep progression from low‐grade lesions to high‐grade lesions, including carcinoma in situ (CIS), of which about half progress to invasive cancer while the other half regress. Although frequent mutations and copy number alterations have been documented in LUSC and observed in precursor lesions, their prognostic significance in precancers remains largely unexplored. In this study, we leveraged gene expression data from LUSC tumors in The Cancer Genome Atlas to derive transcriptional signatures corresponding to 34 key driver genomic aberrations, including mutations, amplifications, and deletions. These tumor‐derived signatures were applied to precancerous datasets to assess their ability to characterize developmental stages and predict progression risk. We found many of these signatures increased progressively across lesion stages, reflecting roles in early tumorigenesis. In particular, several signatures accurately predicted which CIS lesions would progress to invasive cancer. Furthermore, these signature scores were more strongly associated with prognosis in LUSC than the presence of genomic aberrations alone. We also examined relationships between driver‐associated signatures and the tumor immune microenvironment. Signature scores were significantly correlated with immune features such as immune cell infiltration and immune checkpoint gene expression, including CD274 (PD‐L1). These associations varied across stages, indicating dynamic immune interactions during cancer evolution. Together, our findings demonstrate that tumor‐derived driver gene expression signatures provide valuable insight into the biology and progression risk of precancerous lesions, offering potential utility for early detection and intervention in LUSC.

Patients with lung squamous cell carcinoma (LUSC) are often diagnosed at advanced stages, limiting opportunities for early intervention. LUSC develops through a multistep progression from low-grade lesions to high-grade lesions, including carcinoma in situ (CIS), of which approximately half progress to invasive cancer while the other half regress. Although frequent mutations and copy number alterations have been documented in LUSC and observed in precursor lesions, their prognostic significance in precancerous stages remains largely unexplored. In this study, we leveraged gene expression data from LUSC tumors in The Cancer Genome Atlas (TCGA) to derive transcriptional signatures corresponding to 34 key driver genomic aberrations, including mutations, amplifications, and deletions. These tumor-derived gene signatures were then applied to precancerous lesion datasets to assess their ability to characterize developmental stages and predict progression risk. We found that many of these signatures increased progressively across lesion stages, reflecting their roles in early tumorigenesis. In particular, several signatures accurately predicted which CIS lesions would progress to invasive cancer. Furthermore, these signature scores were more strongly associated with patient prognosis in LUSC than the presence of genomic aberrations alone. We also examined the relationship between driver-associated signatures and the tumor immune microenvironment. Signature scores were significantly correlated with immune features such as immune cell infiltration and immune checkpoint gene expression, including CD274 (PD-L1). Interestingly, these associations varied across lesion stages, indicating dynamic immune interactions during cancer evolution. Together, our findings demonstrate that tumor-derived driver gene expression signatures provide valuable insight into the biology and progression risk of precancerous lesions, offering potential utility for early detection and intervention strategies in LUSC. Novelty and Impact: In this study, we developed gene expression signatures using lung squamous cell carcinoma (LUSC) transcriptomic and genomic data and apply them to characterize precancerous lesions. These signatures showed high performance in predicting the progression risk of carcinoma in situ (CIS), offering a valuable tool for personalized prevention and early intervention in LUSC. These findings have the potential to improve early lung cancer detection and provide new insights into the tumor microenvironment in the context of precancer evolution of LUSC.