Clinical Challenge Small lung adenocarcinomas that appear as ground-glass opacities (GGOs) or part-solid nodules on CT are increasingly detected by lung cancer screening. Accurately predicting which nodules harbor invasive disease and have metastatic potential determines whether patients need segmentectomy vs. lobectomy and mediastinal lymph node dissection.
Current Standard - CTR The consolidation tumor ratio (CTR) - the proportion of a nodule's diameter that appears solid on 2D CT - is widely used to assess malignancy. However, CTR is measured in only one dimension and may not accurately reflect the actual solid component volume in complex or irregular nodules.
Novel Metric - SVR This study evaluated the solid volume ratio (SVR), which uses 3D volumetric CT analysis to measure the proportion of the total nodule volume that is solid. Three-dimensional measurement is hypothesized to more accurately capture the true extent of solid tumor, which correlates with invasiveness.
Study Population 343 patients who underwent surgical resection for cT1 lung adenocarcinoma between January 2015 and December 2023 were analyzed using AI-based Infervision software for 3D CT segmentation and SVR calculation.
Infervision AI Platform The study used Infervision AI software to automatically segment lung nodules in 3D from CT scans, measuring both the total nodule volume and the solid component volume. This AI-based approach provides reproducible, observer-independent volumetric measurements.
SVR Calculation SVR was defined as the volume of the solid component divided by the total nodule volume, expressed as a percentage. Unlike CTR, which depends on a single axial measurement, SVR integrates the full 3D shape of both the solid and GGO components.
Outcome Variables Two primary outcomes were assessed: pathological grading (low vs. high malignancy, using the 2021 WHO classification distinguishing minimally invasive and well-differentiated adenocarcinoma from invasive and poorly differentiated forms) and lymph node metastasis (LNM) status confirmed by surgical pathology.
Statistical Analysis AUC comparisons between SVR and CTR were performed using receiver operating characteristic analysis with DeLong test for statistical significance. Optimal cutoff values were determined by Youden index to maximize the sum of sensitivity and specificity.
Pathological Grading AUC SVR achieved an AUC of 0.777 for predicting high-grade malignancy, compared to 0.761 for CTR. While this difference may appear modest, SVR's volumetric approach reduces the impact of nodule shape irregularities that cause 2D CTR measurements to diverge from true solid fraction.
SVR Threshold for High Malignancy An SVR cutoff of greater than 5% predicted high malignancy with a sensitivity of 97.2% and a negative predictive value (NPV) of 96%. This means that nodules with SVR at or below 5% are very unlikely to be high-grade cancers, providing a clinically useful threshold for conservative management.
Practical Impact The high NPV of 96% means that a patient with SVR less than or equal to 5% can be managed with minimally invasive surgery or watchful waiting with high confidence, avoiding the morbidity of unnecessarily extensive resection.
LNM Prediction AUC SVR showed a substantially superior AUC of 0.873 for predicting lymph node metastasis compared to CTR's AUC of 0.804. This larger advantage suggests that 3D solid volume quantification is particularly important for identifying the subset of tumors with metastatic potential.
SVR Threshold for LNM An SVR cutoff of greater than 47.1% predicted lymph node metastasis with a sensitivity of 97.3% and an NPV of 99.5%. The extremely high NPV means that patients with SVR below this threshold have a very low probability of harboring lymph node disease, potentially sparing them from systematic lymph node dissection.
Clinical Surgical Implications The ability to predict LNM preoperatively with 99.5% NPV using SVR could guide decisions about the extent of lymph node dissection during surgery - patients below the threshold might be candidates for sentinel node biopsy rather than full mediastinal dissection, reducing operative risk and recovery time.
Advantage Over CTR for LNM The larger AUC gap for LNM prediction (0.873 vs. 0.804) compared to malignancy grading (0.777 vs. 0.761) suggests that 3D solid volume is especially informative for the most clinically consequential prediction - whether cancer has spread to lymph nodes.
Geometric Limitations of CTR CTR measures the diameter of the solid component in a single 2D cross-section. For nodules with irregular shapes, asymmetric solid components, or non-spherical geometry, a single diameter ratio poorly represents the actual proportion of solid tissue, leading to measurement error.
SVR Captures True Solid Burden By measuring volumes in 3D, SVR integrates the solid component across all spatial dimensions, providing a more accurate reflection of actual invasive tumor burden. This is especially important for part-solid nodules where the solid component may be eccentrically distributed.
AI Enabling Scalable 3D Analysis Manual 3D volumetric measurement of complex nodules would be impractically time-consuming. AI-based automated segmentation with platforms like Infervision makes SVR calculation scalable for routine clinical use, enabling this superior metric to be practically deployed.
Single-Center Retrospective Design The study was conducted at a single institution, limiting generalizability. Multicenter prospective validation across different CT scanner manufacturers, reconstruction protocols, and patient demographics is needed before SVR thresholds can be adopted broadly.
AI Software Dependence The SVR measurements were generated using proprietary Infervision software. Validation with other AI segmentation platforms would be needed to confirm that the identified SVR thresholds are software-independent and reflect true biological properties rather than algorithmic artifacts.
Integration into Lung-RADS The current Lung-RADS reporting system uses CTR as the primary metric. Future guideline updates could consider incorporating SVR, particularly for part-solid nodules where the distinction between 2D and 3D assessment is most consequential.
Longitudinal SVR Tracking Tracking SVR changes over time in nodules under surveillance could identify the rate of solid component growth as a dynamic predictor of malignant transformation, adding a temporal dimension to the static single-time-point measurements used in this study.