NCI Funds BU-Led Project to Turn Lung Pre-Cancer Atlas Into Nodule Biomarkers

Bioengineer.org reported on October 11, 2026 that the National Cancer Institute's Human Tumor Atlas Network has awarded a five-year, $3.4 million U01 grant to a Boston University-led team developing biomarkers for lung ground glass nodules ^[1]. The project is led by Joshua D. Campbell, PhD, of the Chobanian & Avedisian School of Medicine, who also directs the university's Bioinformatics Program ^[1]. Funded under Award Number U01CA313003, the work aims to help clinicians judge whether a faint spot on a CT scan is early cancer or something benign ^[1].

The clinical problem

Ground glass nodules are hazy areas on CT that may reflect early adenocarcinoma, precancerous change, inflammation or old scarring ^[1]. The story notes that radiologists find them more often as more people undergo CT screening ^[1]. An indolent nodule may prompt an invasive surgical resection, while a progressing one may be watched until the window for curative early intervention closes ^[1]. The team describes its goal as a multimodal strategy that "reduces false-negative rates, ensuring earlier and more targeted intervention while minimizing overtreatment" ^[1].

How the project is built

The award is the Phase II follow-on to the Lung Pre-Cancer Atlas, a cellular map of early lung cancer development that Campbell and colleagues helped build under Avrum Spira's leadership during the network's first phase ^[1].

• Biopsy: interventional pulmonologists at Roswell Park Comprehensive Cancer Center, led by co-principal investigator Nathaniel Ivanick, MD, and at Boston Medical Center, led by co-investigator Ehab Billatos, MD, will collect part-solid nodule biopsies; the team will also evaluate robotic-assisted bronchoscopy, which uses steerable, computer-navigated instruments to reach the peripheral lung ^[1].
• Pathology: co-principal investigator Daniel Merrick, MD, a University of Colorado pathologist, will review every sample to establish the histological reference against which biomarkers are calibrated ^[1].
• Imaging: William Hsu, PhD, at UCLA, will lead automated extraction of radiomic features, meaning quantitative measurements from CT scans of the nodule, surrounding tissue and whole lung; the team will determine which features best predict whether a biopsy yields a usable diagnosis ^[1].
• Molecular: co-principal investigator Sarah Mazzilli, PhD, director of the BU Spatial Biology Core, will direct spatial transcriptomics, which maps gene activity to each histological region of a sample instead of averaging across it ^[1].

The key biopsy metric is diagnostic yield, the proportion of biopsies that return a definitive result of cancer, precancer or benign tissue ^[1]. According to the story, a substantial fraction of current biopsies of these nodules are inconclusive ^[1]. The spatial profiles will be combined with CT features and AI-based risk scores in models intended to diagnose cancer even in biopsies pathologists judged inconclusive, and to predict which nodules are likely to progress toward invasive disease ^[1].

Why it matters to research administrators

The project spans pulmonology, pathology, cancer biology, imaging science and data science across Boston University, Roswell Park, the University of Colorado Anschutz and UCLA ^[1]. Specimens collected across participating centers will feed what the story describes as one of the largest cohorts of part-solid nodule biopsies assembled to date, with a single expert pathologist reviewing every sample ^[1].

Mazzilli said every profile the team generates will become part of a public atlas, so that researchers everywhere can apply their own methods to the data ^[1]. The story frames this commitment around the molecular profiles and does not say whether CT data or pathology reads will also be shared ^[1].

What to watch

• Diagnostic yield findings for robotic-assisted bronchoscopy on part-solid nodules, which the team plans to evaluate systematically ^[1].
• Whether CT-based forecasting of inconclusive biopsies holds up; the story says it could help physicians choose which nodules to sample and which tool to use before any molecular test is run ^[1].
• Releases of spatially resolved molecular profiles into the public atlas ^[1].
• Whether the consortium produces a template for turning large molecular atlases into clinical decision tools, which the story presents as contingent on the models performing as intended ^[1].

The story reports no enrollment targets, interim milestones or model performance figures, so any assessment of the biomarkers depends on results the team has yet to publish ^[1].