The research, led by Xinyu Li and Guangming Lu of Nanjing Medical University together with colleagues at Jinling Hospital, The Affiliated Changsha Central Hospital, and The First People’s Hospital of Chenzhou, is a two-center retrospective study built on patients who underwent complete resection of pT3–4N0–2M0 non-small cell lung cancer. The development cohort, drawn from a center treating consecutive patients between January 2015 and December 2021, comprised 135 patients, of whom 30 experienced a documented local recurrence as their first event. An entirely separate external center contributed 31 patients treated between January 2018 and December 2021, with 5 local-recurrence events, providing an independent test of the model’s generalizability. The team analyzed three complementary data types for each patient: preoperative contrast-enhanced computed tomography images, postoperative hematoxylin-and-eosin-stained whole-slide histopathology images, and routine clinicopathological variables such as tumor characteristics recorded after surgery.

What distinguishes the work from many prior artificial intelligence studies in oncology is its rigorous handling of a statistical subtlety that is often ignored: competing risks. In this patient population, some individuals die or develop distant metastases before a local recurrence is ever observed, which means those events preclude the outcome of interest. Standard survival models such as ordinary Cox regression can produce distorted risk estimates in this setting. The researchers instead used penalized Fine–Gray models, a framework specifically designed for competing-risk data, in which distant-first recurrence and death before local recurrence were treated as competing events. The primary outcome was defined precisely as the time from surgery to the first documented local recurrence, and the models were evaluated using time-dependent area under the curve values that account for these competing risks, along with Brier scores and calibration assessments at the two-year mark.

The imaging arm of the pipeline relied on radiomics, the high-throughput extraction of quantitative features from medical images. The team segmented both the intratumoral region—the tumor itself—and a peritumoral ring extending three millimeters beyond the tumor boundary on preoperative contrast-enhanced CT scans. This choice reflects a growing recognition in oncologic imaging that the tissue immediately surrounding a tumor, with its infiltrating immune cells, stromal changes, and early invasion, carries prognostic information that the tumor core alone does not. Radiomic features, quantifying properties such as texture heterogeneity, intensity distributions, and spatial patterns within each volume of interest, were filtered and selected using methods including the least absolute shrinkage and selection operator to prevent overfitting, and feature stability was assessed through intraclass correlation coefficients consistent with the Image Biomarker Standardization Initiative.

On the pathology side, the researchers trained a convolutional neural network on whole-slide images using a weakly supervised strategy. Rather than requiring pathologists to laboriously annotate which microscopic regions harbor prognostically important features—an expensive and inconsistent process—the network learned from slide-level labels alone, aggregating information across thousands of image patches to produce what the authors call pathomics features: numerical descriptors of the tissue’s cellular and architectural landscape. This pathology deep-learning model, referred to as Path-DL, was trained separately using a fixed 7:3 patient-level split and, critically, was not retrained within the cross-validation folds, a design decision that guards against the subtle information leakage that has undermined many published machine-learning studies in medicine. Gradient-weighted class activation mapping, or Grad-CAM, provided a way to visualize which regions of the slides the network attended to, offering pathologists a window into the model’s reasoning.

The final multimodal model fused the clinicopathological, intratumoral radiomics, peritumoral radiomics, and pathomics components at the score level, allowing each modality to contribute its own estimate of risk that could then be combined. The results, obtained through repeated nested five-fold cross-validation in the development cohort, tell a clear story about where predictive power resides. The clinicopathological model alone achieved a two-year area under the curve of just 0.492—essentially no better than a coin flip—underscoring how little conventional variables reveal about local recurrence risk in this population. Intratumoral and peritumoral radiomics performed meaningfully better, with AUCs of 0.689 and 0.694 respectively. The pathology deep-learning model reached 0.804, and the full multimodal fusion model topped the field at 0.827, with a 95 percent confidence interval of 0.689 to 0.936.

External validation, the true test of any predictive model, painted a more cautious but still encouraging picture. In the 31-patient external cohort, the fusion model achieved a two-year AUC of 0.683 (95 percent CI, 0.411–0.917), ahead of the clinicopathological model’s 0.310 and modestly above the radiomics models, which scored 0.605 and 0.616. The wide confidence intervals, an unavoidable consequence of the small external sample and its five events, temper any claim of proven clinical utility, but the direction of the results is consistent with the internal findings: the multimodal approach carries information that standard clinical assessment lacks. Formal paired comparisons quantified the incremental value of fusion over the pathology model alone. The paired difference in AUC was 0.023 (95 percent CI, −0.097 to 0.062) internally and 0.064 (−0.270 to 0.413) externally, while differences in Brier scores, which capture both discrimination and calibration, were 0.001 in both settings with confidence intervals straddling zero. In other words, the fusion model was not statistically superior to the pathology deep-learning model on its own—a finding the authors present transparently rather than overselling.

The implications for postoperative management of locally advanced lung cancer are nonetheless significant. Patients with resected pT3–4 non-small cell lung cancer currently receive relatively uniform surveillance and adjuvant treatment decisions, guided largely by stage and nodal status. A validated tool that stratifies local recurrence risk from data that already exist in the medical record—preoperative CT scans and the pathology slides prepared after surgery—could, if confirmed in larger prospective studies, allow clinicians to intensify follow-up imaging, consider adjuvant radiotherapy, or enroll high-risk patients in clinical trials while sparing lower-risk individuals unnecessary intervention. The inclusion of competing-risk methodology also means the model’s outputs are expressed as cumulative incidence of actual local recurrence, the quantity that matters in the clinic, rather than an inflated hazard-based surrogate.

The study’s transparency about its own limitations is notable and aligns with reporting standards such as the Checklist for Artificial Intelligence in Medical Imaging. Sample size is the most obvious constraint: 135 development patients with 30 events is modest by machine-learning standards, and 31 external patients cannot establish generalizability with statistical confidence. The retrospective design introduces the usual risks of selection bias, although the use of consecutive patients at both centers mitigates this. The models were locked before external evaluation, and the study was approved by the Institutional Review Board of Jinling Hospital with informed consent waived owing to the retrospective design, conducted in accordance with the Declaration of Helsinki. Shapley additive explanations, or SHAP, were used to interpret feature contributions, and decision curve analysis was employed to assess the clinical net benefit of the models across a range of risk thresholds—techniques that move the work beyond raw accuracy metrics toward genuine clinical decision support.

The research was supported by the Science and Technology Innovation 2030-Major Projects (grant 2020AAA0109500) and the General Program of the National Natural Science Foundation of China (grant 82371958), with technical support from the Deepwise multimodal research platform. The authors, including Yu Zong, Changsheng Zhou, Yang Cao, Zhen Zhou, Jianrui Li, Zhiyuan Sun, Xiaoqing Cheng, and Liying Wang, declare no competing interests. The article is published open access under a Creative Commons Attribution 4.0 license, with the accepted manuscript shared early under a citable permanent DOI ahead of the final version of record.

For the field of AI-assisted oncology, the study is a case study in methodological discipline: nested cross-validation for internal evaluation, a frozen upstream network to prevent data leakage, competing-risk statistics matched to the clinical question, external validation with locked models, and honest reporting of null findings in paired comparisons. It suggests that the microscopic world captured on a postoperative slide, read by a weakly supervised neural network, may hold more prognostic signal for local recurrence than anything clinicians currently measure—and that fusing it with imaging-based tumor and microenvironment features pushes performance modestly further. The next step, which the authors’ careful framing implicitly calls for, is prospective validation in larger, multi-institutional cohorts before such a model can guide real-world decisions for the thousands of patients who each year face the uncertainty of life after surgery for locally advanced lung cancer.

Lung adenocarcinoma, a subtype of non-small cell lung cancer, remains a formidable clinical challenge due to its tendency for postoperative recurrence even after surgical resection, which is currently the mainstay curative approach for early-stage disease. Predicting which patients are at higher risk for recurrence has remained elusive with conventional imaging biomarkers. Standard PET/CT parameters such as maximum standardized uptake value (SUVmax) commonly reflect tumor metabolism but fall short in predicting postoperative outcomes robustly. The emerging concept explored in this study is the spatial relationship of the metabolic “hot spot” — the point of highest radiotracer uptake — relative to key tumor anatomical landmarks.

The research team retrospectively analyzed data from 164 patients with surgically treated, pathologically confirmed stage IA–IIIA lung adenocarcinoma. All participants had undergone preoperative ^18F-Fluorodeoxyglucose PET/CT imaging, a powerful tool that maps glucose metabolism within tumors. Beyond conventional metabolic metrics, the researchers introduced and meticulously quantified two novel parameters: the normalized distance from the maximum uptake point (hot spot) to the tumor centroid, termed NHOCmax, and the normalized distance from the hot spot to the tumor perimeter, termed NHOPmax. These metrics effectively capture where within the tumor the metabolic peak is situated, normalized for tumor size, providing unique insights into tumor heterogeneity and aggressiveness.

Remarkably, the study found that NHOPmax, the distance from the highest glucose-avid point to the tumor’s outer edge, was the most potent predictor of postoperative recurrence and disease-free survival (DFS). It achieved an area under the curve (AUC) of 0.682 with an impressive sensitivity of 78.8%, outperforming traditional PET parameters in prognostic ability. This finding suggests that tumors with metabolic hot spots located closer to the perimeter rather than the center may confer a different biological behavior and risk profile, possibly reflecting invasive tumor fronts or areas of active proliferation.

Further statistical scrutiny demonstrated that NHOPmax was largely independent of other metabolic parameters like SUVmax, total lesion glycolysis (TLG), and metabolic tumor volume (MTV), indicating it conveys distinctive prognostic information. In both univariate and multivariate logistic regression analyses, NHOPmax showed a robust inverse association with postoperative recurrence risk, symbolizing that higher NHOPmax values — meaning the hot spot is positioned further from the perimeter — corresponded to superior patient outcomes.

Survival analysis added compelling weight to these observations, establishing NHOPmax as an independent predictor of disease-free survival. Patients with NHOPmax values exceeding the threshold of 0.43 experienced significantly longer DFS, underscoring the clinical utility of this novel imaging biomarker in stratifying recurrence risk. Integrating NHOPmax into postsurgical follow-up protocols could enable clinicians to tailor adjuvant therapies more precisely and optimize patient counseling.

The introduction of spatial PET parameters like NHOPmax transcends the traditional reliance on metabolic intensity alone. This paradigm shift emphasizes tumor microenvironment organization and heterogeneity as critical facets influencing cancer progression. By quantifying the positional metabolic gradients within tumors, clinicians could gain refined insights into tumor biology and behavior, potentially applicable beyond lung adenocarcinoma to other solid tumors.

Such an imaging biomarker dovetails seamlessly with the growing field of radiomics, where complex image features are computationally extracted and leveraged for clinical predictions. NHOPmax exemplifies a clinically actionable radiomic feature distilled from widely accessible PET/CT scans, enhancing translational value. Future research integrating NHOPmax with molecular and genomic tumor profiles could unlock synergistic prognostic models, propelling the era of precision oncology forward.

This study’s findings are especially poignant in the context of stage IA–IIIA lung adenocarcinoma, where surgical resection yields curative potential but recurrence risk remains a pressing concern. Current prognostic tools, including tumor-node-metastasis (TNM) staging, lack granularity in identifying which resected patients harbor micrometastatic disease or aggressive tumor phenotypes. NHOPmax adds a layer of nuanced, noninvasive risk stratification that could redefine postoperative monitoring intensity and therapeutic decision-making.

Moreover, the ease of calculating NHOPmax from routine ^18F-FDG PET/CT scans elevates its clinical feasibility. Since PET/CT imaging is standard for lung cancer staging, implementing NHOPmax quantification would require minimal alterations to imaging protocols, facilitating seamless adoption. This methodology also circumvents the need for invasive tissue sampling or complex molecular assays, democratizing risk assessment in diverse clinical settings.

While the current study is retrospective and single-institutional, it paves the way for prospective multicenter trials validating NHOPmax’s prognostic prowess. Evaluating its predictive capacity in conjunction with novel systemic therapies such as immunotherapy or targeted agents could further elucidate its role in evolving lung cancer treatment landscapes. Additionally, refining computational algorithms for automated NHOPmax measurement may enhance reproducibility and expedite clinical workflows.

In essence, this research pioneers a new dimension in oncologic imaging biomarkers by leveraging the spatial metabolic architecture of tumors. NHOPmax emerges not just as a statistical predictor, but as a window into the biological complexity underpinning tumor aggressiveness and recurrence. By translating this insight into clinical practice, oncologists may soon wield a powerful tool to preempt postoperative relapse and personalize patient care.

Together, these advancements highlight the transformative potential of enhancing PET imaging metrics beyond conventional parameters. The nuanced evaluation of glucose metabolism topography within lung adenocarcinoma introduces a critical step forward in precision diagnostics, prognostics, and therapeutics. As medicine gravitates towards individualized approaches, such innovative imaging biomarkers will undoubtedly play a pivotal role in shaping future lung cancer management strategies.

The implications extend beyond recurrence prediction: NHOPmax and similar spatial biomarkers might serve as early surrogate endpoints in clinical trials or as markers to select patients for intensified adjuvant therapies. They could also stimulate biologic investigations into the mechanisms driving differential metabolic distribution, unveiling novel targets to thwart invasion and metastasis.

In conclusion, the study’s identification of NHOPmax from ^18F-FDG PET/CT scans as a robust, independent predictor of postoperative recurrence in lung adenocarcinoma represents a major stride in oncologic imaging and prognosis. Its incorporation into clinical workflows promises to refine patient stratification, inform treatment decisions, and ultimately improve survival outcomes in this challenging malignancy. As the oncology community embraces increasingly sophisticated imaging analytics, such breakthroughs underscore the synergistic power of technology and clinical science in confronting cancer’s complexities.

Subject of Research: Predictive PET imaging biomarkers for postoperative recurrence in lung adenocarcinoma

Article Title: Novel PET imaging biomarkers as predictors of postoperative recurrence in lung adenocarcinoma

Article References:
Zheng, C., Miao, J., Xu, L. et al. Novel PET imaging biomarkers as predictors of postoperative recurrence in lung adenocarcinoma. BMC Cancer 25, 874 (2025). https://doi.org/10.1186/s12885-025-14263-0

Image Credits: Scienmag.com

DOI: https://doi.org/10.1186/s12885-025-14263-0