Weekly Respiratory Research Analysis
This week’s respiratory literature highlights rapid advances in host-directed antivirals, AI-enabled diagnostics, and multimodal precision tools that bridge omics and clinical text. A high-impact preclinical study identifies HGS as a druggable host factor for pan-coronavirus assembly and a repurposing candidate with in vivo activity. Several clinically oriented papers show substantial gains from integrating molecular biomarkers with large language models for lower respiratory infection diagnosis
Summary
This week’s respiratory literature highlights rapid advances in host-directed antivirals, AI-enabled diagnostics, and multimodal precision tools that bridge omics and clinical text. A high-impact preclinical study identifies HGS as a druggable host factor for pan-coronavirus assembly and a repurposing candidate with in vivo activity. Several clinically oriented papers show substantial gains from integrating molecular biomarkers with large language models for lower respiratory infection diagnosis, while pathology image-based pathomics is maturing as a treatment-selection biomarker in lung cancer.
Selected Articles
1. Targeting the host factor HGS-viral membrane protein interaction in coronavirus infection.
A genome-wide CRISPRi screen identified HGS as a conserved host factor that binds the coronavirus M protein to enable ERGIC trafficking and virion assembly. M-derived peptides and a repurposed compound, riboflavin tetrabutyrate (RTB), disrupt HGS–M binding, retain M in the ER, and block virion formation, showing broad anti-coronavirus activity in vitro and in vivo.
Impact: Identifies a conserved, druggable host–virus interface and a repurposing candidate with in vivo efficacy — a pivotal step toward host-directed broad-spectrum antivirals that may be less susceptible to viral resistance.
Clinical Implications: If safety and PK/PD are favorable, RTB or HGS-targeting peptides could progress to early-phase clinical trials as broad-spectrum coronavirus therapeutics; host-targeting strategies could complement direct-acting antivirals.
Key Findings
- Genome-wide CRISPRi screen identifies HGS as essential for virion assembly via direct M protein binding.
- M-derived peptides and riboflavin tetrabutyrate disrupt HGS–M interaction, causing M retention in ER and blocking assembly.
- Agents show broad anti–pan-coronavirus activity in vitro and in vivo.
2. Integrating a host biomarker with a large language model for diagnosis of lower respiratory tract infection.
A multimodal classifier combining the pulmonary transcriptomic biomarker FABP4 with GPT-4 analysis of EMR free-text substantially outperformed either modality alone and physician admission diagnoses for LRTI in critically ill adults (AUC ~0.93; external validation AUC ~0.98). The approach demonstrates reproducible gains from integrating molecular host signals with LLM-based clinical text interpretation.
Impact: Demonstrates practical, externally validated improvement in a high-stakes diagnostic problem by fusing host transcriptomics with LLM-derived clinical reasoning — a template for rapid translation of multimodal diagnostics in critical care.
Clinical Implications: Could be deployed as a decision-support tool to improve discrimination of infectious vs non-infectious respiratory failure, guiding antimicrobial use and targeted testing; multicenter prospective impact studies are needed before clinical rollout.
Key Findings
- Combined FABP4 + GPT-4 classifier AUC 0.93 (cohort) and AUC 0.98 (independent validation), accuracy superior to clinicians.
- Multimodal integration outperforms FABP4-only and LLM-only models.
- External validation supports reproducibility in critically ill adults.
3. A pathomics model for predicting response to chemo-immunotherapy in lung squamous cell carcinoma: A multicenter study.
A pathomics model built from whole-slide images and transcriptomics predicted a T cell–inflamed gene-expression profile and stratified lung squamous cell carcinoma patients for first-line chemo-immunotherapy benefit. In a prospective multicenter trial cohort, high pathomics score showed significant interaction with CIT and large PFS/OS improvements; results were replicated in independent cohorts and associated with immune-hot microenvironments.
Impact: Operationalizes a scalable, tissue-based biomarker with prospective multicenter validation to guide first-line chemo-immunotherapy selection in LUSC — potentially reducing unnecessary CIT exposure and optimizing outcomes where molecular assays are limited.
Clinical Implications: Pathomics scoring could be integrated into pathology workflows to select LUSC patients for first-line chemo-immunotherapy; prospective implementation and cost-effectiveness studies are the next step.
Key Findings
- Pathomics score predicted T cell–inflamed GEP (TCGA AUC training 0.80, validation 0.71).
- Significant pathomics score × treatment interaction in a prospective multicenter trial with large PFS/OS benefit for high-score patients receiving CIT (PFS HR 0.31; OS HR 0.30).
- Results replicated in two independent cohorts and linked to an immune-hot tumor microenvironment.