Volume Doubling Time Emerges as Decisive Imaging Biomarker in Lung Cancer Screening Debate

A new editorial published in the journal Cancer Imaging argues that the way radiologists measure lung nodules on CT scans may be one of the most consequential — and contested — decisions in modern lung cancer screening. Writing in an editorial published on 21 October 2026, Matthijs Oudkerk examines a paper by Milena Petranovic and colleagues on ACR Lung-RADS v 2022, which appears in the same issue of the journal. That paper refines the methodological CT guidelines for the early detection of pulmonary nodules with an increased lung cancer probability, with a particular focus on the first CT screening round. Its central concern is whether the simple two-dimensional diameter measurement of a nodule is precise enough for effective lung cancer screening implementation, compared with the more sophisticated automated three-dimensional CT nodule segmentation now available across much of the world.

The question is not merely technical. According to the editorial, Petranovic and colleagues state that in terms of the final mortality outcome of CT lung cancer screening, there is no evidence that automated 3D nodule volume segmentation is superior in reducing mortality. They therefore argue that new studies should prove the necessity of 3D volume implementation in lung cancer screening — notwithstanding the fact that European guidelines all prescribe 3D nodule volume segmentation as a prerequisite for screening implementation. Oudkerk describes this divergence between the American and European positions as an important diversity in guidelines that requires further explanation, and he traces its roots back through two decades of screening research on both sides of the Atlantic.

The conceptual foundation for using a CT tumor biomarker to track nodule growth rate was published by Yankelevitz and colleagues from an American institute in 2000. Yet this CT biomarker was never implemented in the randomized controlled National Lung Screening Trial in the United States. The NLST was performed across 33 imaging centers without standardization of CT data acquisition and image analysis. Because isotropic 3D CT nodule volume measurement was unavailable at that time, nodule diameter measurement was understandably used as a surrogate for nodule size assessment. The turning point came with the introduction of 16-MDCT systems at the RSNA in 2002, which enabled isotropic 3D CT data acquisition and, through the first release of computer-aided detection for lung nodule segmentation, true 3D nodule volume assessment.

Those 16-MDCT systems were implemented across four imaging centers using identical CT systems and data acquisition and analysis protocols for the execution of the NELSON randomized controlled lung cancer screening trial. Concurrently, the NELSON study designed an “indeterminate” nodule category to facilitate the introduction of nodule volume growth rate as a new imaging biomarker. By repeating the CT examination after three months, the Volume Doubling Time — VDT — could be introduced as a biomarker to select actionable lung nodules and to rule out suspicious nodules with slow or absent growth rates. In this way, the two landmark trials came to reflect what the editorial calls the current paradigm split: US guidelines advocating diameter nodule measurement without implementing nodule growth rate as an imaging biomarker, versus European guidelines based on volumetry for nodule detection and growth rate evaluation.

The stakes of that split are quantified in the mortality figures the editorial assembles. The NLST 6.5-year mortality reduction, based on the National Cancer Institute analysis by Pinsky, was 8 percent for men and 27 percent for women, dropping to an overall mean mortality reduction of 8 percent after 12.3 years of follow-up. Conversely, the 10-year mortality reduction in the NELSON study, based on volumetry and growth rate assessment, was 24 percent for men and 39 percent for women, with reductions of up to 60 percent after eight years. These two trials, Oudkerk writes, provide the best possible comparison between prospective randomized evaluations of a nodule diameter methodology and a volumetry-based approach using nodule growth rate as a biomarker.

Developing a new randomized controlled trial to compare both methodologies head-to-head, with mortality as a primary endpoint, is described as neither feasible nor affordable. The costs of the existing trials were already high: nearly 250 million US dollars for the NLST and over 10 million euros for the NELSON. This is especially true, the editorial notes, because automated 3D nodule volume segmentation by computer-aided detection — further enhanced by AI optimization beyond Hounsfield Unit thresholding — is already widely implemented across European imaging departments. Given that trajectory, it is likely only a matter of time before volumetry becomes the dominant approach in the United States as well, particularly as an absolute prerequisite for Volume Doubling Time measurement.

The editorial also revisits how measurement thresholds have evolved. Tumor growth evaluation by a radiologist’s “eyeballing” has been a cornerstone of oncological radiology for decades, and categorization into progression, regression, or stable disease remains valid today. In the NLST, all manually measured baseline nodules larger than 4 mm were classified as suspicious and followed through diagnostic work-up or a repeat CT to determine nodule growth, without standardized nodule volume quantification. That approach led to extremely high false-positive rates of 24 percent. In 2014, the NELSON consortium published additional baseline outcome data demonstrating that the distribution between benign and malignant nodules shifted toward an increased lung cancer probability for screening participants with nodules larger than 100 cubic millimeters — equal to a 5.76 mm diameter if the nodule were perfectly spherical — compared to participants with no nodules at all. In the same year, Lung-RADS 1.0 guidelines raised the action threshold from 4 mm to 6 mm, representing an extrapolated volume of 113 cubic millimeters, before a nodule required further diagnostic work-up, a change based on a retrospective analysis of the NLST data by Pinsky and colleagues. Very recently, in 2025, the 100 cubic millimeter — or 6 mm — threshold was prospectively confirmed by the SUMMIT study.

The regulatory landscape has continued to diverge. In 2017, the European position statement on lung cancer screening recommended a 100 cubic millimeter volume threshold, with 80 cubic millimeters as a correction factor for less sophisticated segmentation software, alongside growth rate as an imaging biomarker — specifically, a volume doubling time of fewer than 400 days for actionable nodules. To this day, Lung-RADS 2.0 from 2022 continues to recommend mean diameter as the primary measure of size, despite published research evidence outside the US demonstrating that diameter-based growth measurements lead to incorrect tumor classification in the majority of cases. Recent prospective data from large-scale screening cohorts in 2025 and 2026 have, according to the editorial, heavily reinforced this discrepancy. Large trial evaluations demonstrated that a volumetric rule-out threshold of under 100 cubic millimeters achieves an identical negative predictive value to long-axis diameter thresholds but encompasses significantly more participants. This approach greatly reduces the number of unnecessary intermediate interval scans and safely lowers false-positive referral rates.

The most recent European guidance has moved decisively toward volume. The 2025 and 2026 European Society of Thoracic Imaging guidelines have formalized an aggressiveness-guided nodule management protocol, establishing volumetry as the absolute preferred measure for solid nodules and reserving manual diameter measurements strictly as a fallback. The editorial catalogs the shortcomings of manual measurement in stark terms: low reproducibility with inferior inter- and intra-reader agreement, low sensitivity for growth, highly inconsistent measurement of irregular nodule margins, and high false-positive rates together make the sole application of manual diameter measurements — even more so for growth rate assessment — unacceptable in terms of modern screening reliability.

Against this backdrop, the editorial points to automation as the practical bridge to volumetry-based screening. Nodule volume segmentation based on AI implementation in lung cancer screening enables longitudinal pulmonary nodule matching by which manual correction is needed in only 1.5 percent of participants, providing standardized fully automated nodule volume doubling time quantification and calculation. This development, Oudkerk concludes, will substantially impact workload and costs in lung cancer screening and enhance safe and reliable implementation of screening programs. Whether American guidelines will follow the European lead remains an open question, but the editorial makes clear that the debate over diameter versus volume is now backed by two decades of randomized trial evidence, prospective cohort data from 2025 and 2026, and a maturing AI infrastructure that makes volumetric assessment practical at scale.

Subject of Research: The use of CT-based tumor volume doubling time as an imaging biomarker for early detection of lung cancer in screening programs

Article Title: Tumor growth rate as a new imaging biomarker in early detection of lung cancer

Article References: Oudkerk, M. (2026). Tumor growth rate as a new imaging biomarker in early detection of lung cancer. Cancer Imaging, 26(1), Article 146. https://doi.org/10.1186/s40644-026-01138-1

Image Credits: AI Generated

DOI: 10.1186/s40644-026-01138-1

Keywords: lung cancer screening, volume doubling time, CT volumetry, pulmonary nodules, NELSON trial, NLST, Lung-RADS, imaging biomarker, AI segmentation, false positives, ESTI guidelines, Cancer Imaging