Scholarlink

Background <p>Genetic aberrations are among the critical driving factors of lung cancer. Importantly, the impact of genetic variations on proteomic dysregulations with the goal of characterizing potential diagnostic biomarkers at the population-level requires additional investigation. Modeling such proteogenomic interactions is crucial in understanding early-stage biological disruptions to inform biomarker discovery, successful clinical trials, and developing effective therapeutics.</p> Methods <p>We investigated two complementary aspects of lung cancer risk. First, we performed a genome-wide association study of lung cancer using population-scale datasets, then examined whether lung cancer risk-associated variants influence plasma protein levels using the UK Biobank Pharma Proteomics Project data. Second, we identified plasma proteomic dysregulations in presymptomatic and symptomatic patients with the objective of pinpointing diagnostic biomarkers through leveraging machine learning methods.</p> Results <p>Using the identified proteins, machine learning models achieved median cross-validated AUCs of 0.85–0.88 (0–4 years before diagnosis [YBD]), 0.81–0.84 (5–9 YBD), and 0.80–0.86 (0–9 YBD). Performing survival analyses within the 5–9 YBD group, elevated levels of eight proteins, such as CALCB, PLAUR, and CD74, were found to significantly associate with lower survival. We identified 22 disease-associated proteins, of which 14 have been previously implicated in lung cancer, including CEACAM5, CXCL17, GDF15, WFDC2 along with 8 novel proteins. These proteins were enriched in pathways related to cytokine signaling, interleukin regulation, neutrophil degranulation, and lung fibrosis.</p> Conclusions <p>While these findings do not establish mechanistic causality, they highlight proteomic alterations reflecting systemic changes preceding the diagnosis. Our study contributes to understanding genome–proteome relationships in lung cancer and identifies circulating proteins warranting further investigation as potential early biomarkers for screening and risk stratification.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning-based proteogenomic data modeling identifies circulating plasma biomarkers for early detection of lung cancer

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月13日

查看全文
  • Marcela A. Johnson,
  • Shirley Nieves-Rodriguez,
  • Liping Hou,
  • Bevan Emma Huang,
  • Assieh Saadatpour,
  • Abolfazl Doostparast Torshizi

摘要

Background

Genetic aberrations are among the critical driving factors of lung cancer. Importantly, the impact of genetic variations on proteomic dysregulations with the goal of characterizing potential diagnostic biomarkers at the population-level requires additional investigation. Modeling such proteogenomic interactions is crucial in understanding early-stage biological disruptions to inform biomarker discovery, successful clinical trials, and developing effective therapeutics.

Methods

We investigated two complementary aspects of lung cancer risk. First, we performed a genome-wide association study of lung cancer using population-scale datasets, then examined whether lung cancer risk-associated variants influence plasma protein levels using the UK Biobank Pharma Proteomics Project data. Second, we identified plasma proteomic dysregulations in presymptomatic and symptomatic patients with the objective of pinpointing diagnostic biomarkers through leveraging machine learning methods.

Results

Using the identified proteins, machine learning models achieved median cross-validated AUCs of 0.85–0.88 (0–4 years before diagnosis [YBD]), 0.81–0.84 (5–9 YBD), and 0.80–0.86 (0–9 YBD). Performing survival analyses within the 5–9 YBD group, elevated levels of eight proteins, such as CALCB, PLAUR, and CD74, were found to significantly associate with lower survival. We identified 22 disease-associated proteins, of which 14 have been previously implicated in lung cancer, including CEACAM5, CXCL17, GDF15, WFDC2 along with 8 novel proteins. These proteins were enriched in pathways related to cytokine signaling, interleukin regulation, neutrophil degranulation, and lung fibrosis.

Conclusions

While these findings do not establish mechanistic causality, they highlight proteomic alterations reflecting systemic changes preceding the diagnosis. Our study contributes to understanding genome–proteome relationships in lung cancer and identifies circulating proteins warranting further investigation as potential early biomarkers for screening and risk stratification.