Abstract
Background
Visceral pleural invasion (VPI) is an adverse prognostic factor in lung adenocarcinoma. Accurate preoperative estimation of VPI risk in solid tumors with pleural contact may provide useful supplementary information during preoperative assessment.
Methods
This was a retrospective study of 162 patients. All patients had surgically resected, pathologically confirmed solid lung adenocarcinoma in pleural contact between November 2018 and August 2025. Patients were classified into VPI-positive and VPI-negative groups based on postoperative pathology. The patients were randomly divided into a training cohort and a testing cohort at a ratio of 7:3 for internal validation. Multivariable logistic regression was performed to identify independent risk factors for VPI. A nomogram prediction model was developed based on the multivariable analysis. Its predictive performance was then evaluated using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA).
Results
Of the 162 patients with solid lung adenocarcinoma nodules, 58 had VPI confirmed by pathology. Multivariate logistic regression analysis identified spiculation, pleural indentation, and pleural contact length as independent risk factors for predicting VPI. A nomogram based on these three CT features showed good discriminative performance in both the training and testing cohorts, with AUCs of 0.901 (95% CI: 0.846–0.956) and 0.864 (95% CI: 0.737–0.937), respectively. Calibration curves showed the predictions matched observations well. Decision curve analysis suggested potential utility in preoperative risk assessment.
Conclusion
Among patients with solid lung adenocarcinoma nodules with pleural contact, we developed a nomogram based on routinely assessed CT semantic features to estimate the preoperative probability of VPI. The model is intended as a practical supplementary tool for preoperative VPI risk estimation.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12880-026-02458-1.
Keywords: Visceral pleural invasion (VPI), Lung adenocarcinoma, Computed tomography (CT), Nomogram
Introduction
Lung cancer is a leading cause of global cancer incidence and mortality [1, 2]. Its rates continue to rise, presenting a serious public health threat. Lung adenocarcinoma (LUAD) is the most common histological subtype of non-small cell lung cancer (NSCLC) [3]. Although treatment strategies have advanced in recent years, the overall prognosis for lung cancer remains poor. Recent population-based data indicate that the overall 5-year relative survival for lung cancer is approximately 22%, with substantial variation according to stage at diagnosis [1].
Visceral pleural invasion (VPI) is an important adverse prognostic factor in NSCLC [4–6]. It directly impacts tumor staging, treatment decisions, and outcomes. According to the 9th edition of the tumor–node–metastasis (TNM) staging system for lung cancer, when VPI is present, tumors measuring ≤ 3 cm are upstaged from T1 to T2a. This results in a shift of the clinical stage from IA to IB [7]. In current clinical practice, the extent of resection for small-sized peripheral NSCLC is increasingly informed by contemporary randomized evidence. Recent trials, including JCOG0802/WJOG4607L and CALGB/Alliance 140,503, have supported sublobar resection in selected patients, although the role of VPI in determining the adequacy of limited resection remains controversial [8, 9]. Jiang et al. [10] showed that the presence of VPI significantly worsens prognosis in patients with tumors ≤ 3 cm in diameter. Similarly, Hung et al. [11] reported that visceral pleural invasion at the PL2 level predicts higher recurrence and lower survival in NSCLC.
Chest computed tomography (CT) plays a key role in the preoperative assessment of lung adenocarcinoma [12, 13]. Previous studies have shown that several CT features—such as tumor location and size, lobulation, and pleural indentation—are associated with VPI [14, 15]. Previous CT-based studies on preoperative VPI prediction have largely been conducted in broader mixed cohorts including subsolid and solid nodules. Therefore, their findings may not be directly generalizable to solid pulmonary adenocarcinoma nodules with pleural contact, a clinically specific subgroup characterized by direct tumor–pleura apposition and potentially distinct VPI risk patterns. Compared with subsolid nodules, solid lung adenocarcinomas generally exhibit greater invasiveness and are more likely to develop VPI [18]. In addition, pleural contact length has been reported as an independent predictor of VPI [17, 19], potentially because direct tumor–pleura apposition reduces the buffering effect of intervening normal lung parenchyma and may facilitate pleural invasion.
At present, dedicated prediction models for solid lung adenocarcinoma nodules with pleural contact are still uncommon, even though this subgroup is clinically relevant and may carry a higher risk of pleural invasion. When a solid tumor abuts the pleura, the tumor–pleura interface is different from that seen in subsolid or mixed cohorts, and local extension becomes a more realistic concern; as a result, models developed in those broader populations may not translate well to pleura-abutting solid nodules. Therefore, the aim of the present study was to develop and internally validate a CT-based nomogram for preoperative estimation of VPI risk in solid pulmonary adenocarcinoma nodules with pleural contact using routinely assessed semantic CT features.
Materials and methods
Patients
This retrospective study was approved by the Institutional Review Board of the hospital, and the requirement for written informed consent was waived. For model development and internal validation, the included patients were randomly divided into a training cohort and a testing cohort at a ratio of 7:3 within the same institutional dataset. The study retrospectively analyzed the CT imaging features, clinical, and pathological data of primary pulmonary adenocarcinoma patients confirmed by surgical pathology at the hospital from November 2018 to August 2025. The inclusion criteria were as follows: (1) Pathologically confirmed diagnosis of primary pulmonary adenocarcinoma. (2) Maximum tumor diameter ≤ 3 cm. (3) Complete chest CT scan data within 1 month prior to surgery. (4) Chest CT imaging showing solid nodules with pleural contact. The exclusion criteria were as follows: (1) Patients with missing pathological or clinical data. (2) Patients with preoperative treatments such as radiotherapy or chemotherapy. (3) CT artefacts that interfered with lesion measurement. The inclusion/exclusion flow diagram is shown in Fig. 1. Based on pathological findings, patients were classified into VPI-positive and VPI-negative groups. Clinicopathological and imaging data were collected for further analysis. At our institution, the choice between lobectomy and sublobar resection was made according to routine multidisciplinary clinical judgment, taking into account tumor size and location, radiologic invasiveness, the anticipated ability to achieve an adequate surgical margin, pulmonary reserve, and patient comorbidities.
Fig. 1.
The inclusion/exclusion flow diagram. Abbreviations: CT = computed tomography. VPI = visceral pleural invasion
Pathological evaluation
All surgically resected specimens underwent routine pathological examination with hematoxylin and eosin (H&E) staining. The tumor–pleura interface on the resected specimen was routinely assessed by pathologists to determine the extent of pleural involvement. When the relationship between tumor cells and the visceral pleural elastic layer could not be confidently determined on H&E sections, additional elastic fiber staining was performed. According to the 9th edition of the TNM classification for lung cancer, PL0 was defined as no tumor invasion beyond the visceral pleural elastic layer, PL1 as tumor invasion beyond the elastic layer, and PL2 as tumor extension to the visceral pleural surface. Cases with PL1 or PL2 were classified as VPI-positive, whereas PL0 cases were classified as VPI-negative. Tumor invasion of the parietal pleura (PL3) was not included in the present study.
Imaging technique
All CT scans were performed using a SOMATOM Force scanner (Siemens Healthineers, Forchheim, Germany) during end-inspiratory breath-hold. The scanning parameters were as follows: tube voltage, 120 kV; tube current, 150–200 mA; collimation/slice thickness, 5 mm; and slice interval, 5 mm. Raw data were acquired with 5-mm collimation and additionally reconstructed into 1.25-mm thin sections (interval, 1.25 mm) for lung-window analysis. The scanning range extended from the thoracic inlet to the level of the diaphragm. After scanning, the raw images were transferred to a post-processing system. CT images were reconstructed with the following window settings: lung window (window width, 1200 HU; window level, − 600 HU) and mediastinal window (window width, 400 HU; window level, 40 HU).
Image analysis
CT images were independently reviewed by two senior radiologists, each with more than 10 years of clinical experience. Both radiologists were blinded to the patients’ pathological results. In cases of disagreement, a consensus was reached through discussion. For nodule type classification, a solid nodule was defined as a nodule in which the solid component accounted for > 80% of the entire nodule [20]. The evaluated CT semantic features were predefined based on prior literature on CT predictors of visceral pleural invasion and tumor–pleura interaction, as well as their routine availability and interpretability in clinical practice. The following CT features were evaluated: (1) nodule location (right upper/middle/lower lobe; left upper/lower lobe); (2) nodule size (the mean of the maximum long-axis diameter and the perpendicular short-axis diameter on the largest axial slice); (3) lobulation; (4) spiculation; (5) air bronchogram; (6) vacuole sign (air-containing area < 5 mm); (7) pleural indentation; (8) cavity (air-containing area ≥ 5 mm); (9) calcification; (10) vascular convergence; (11) pleural contact length. The pleural contact length was defined as the longest continuous curvilinear interface between the nodule and the visceral pleura. It was measured on lung-window images using multiplanar reconstruction (axial, coronal, and sagittal planes) on the Philips IntelliSpace Portal workstation (v6.0.5.0299), with the maximum interface length manually traced on the three-dimensional reconstructed images with MPR assistance [17, 21] (Fig. 2).
Fig. 2.
(a–c) Sagittal, axial, and coronal CT images of solid pulmonary adenocarcinoma nodules with pleural contact pathologically confirmed as VPI-positive; preoperative CT shows pleural contact lengths of 21.3 mm, 23.3 mm, and 13.2 mm. Abbreviations: VPI, visceral pleural invasion; CT, computed tomography
Statistical analysis
Statistical analysis was performed using SPSS version 30.0 and R software (version 4.4.1). The Kolmogorov–Smirnov test and the Shapiro–Wilk test were used to assess the normality of quantitative data. Data with a normal distribution were expressed as mean ± standard deviation (x ± s), and comparisons between groups were performed using the independent-sample t-test. Non-normally distributed data were expressed as median and interquartile range [M (P25, P75)], and comparisons between groups were performed using the Mann-Whitney U test. Categorical data were expressed as counts and percentages (%), and comparisons between groups were conducted using the Chi-square test. Variables showing statistically significant differences in univariate analysis were subsequently entered into the multivariable logistic regression model to identify independent predictors of VPI. To improve methodological transparency, we also evaluated model stability by considering the events-per-variable (EPV) ratio and assessed collinearity among candidate predictors using the variance inflation factor (VIF), with VIF ≤ 5 indicating no problematic multicollinearity in the final model. A nomogram model for predicting visceral pleural invasion (VPI) in lung adenocarcinoma was constructed using the “rms” package in R software. Calibration curves were applied to assess the agreement between the predicted probability of VPI and the observed incidence. In the training cohort, bootstrap resampling (1,000 repetitions) was used to obtain optimism-corrected (bias-corrected) calibration. In the testing cohort, calibration curves were estimated with bootstrap resampling (1,000 repetitions) to reduce sampling variability. Clinical decision curve analysis (DCA) was performed to evaluate the clinical utility of the nomogram.
Results
Clinical characteristics and imaging features of patients
A total of 162 patients were included in this study, comprising 58 patients in the VPI-positive group and 104 patients in the VPI-negative group, with an overall VPI incidence of 35.8%. Among all 162 patients, 122 tumors (75.3%) measured ≤ 2 cm and 40 tumors (24.7%) measured > 2–3 cm; in the training cohort, 83 of 112 tumors (74.1%) were ≤ 2 cm and 29 (25.9%) were > 2–3 cm, whereas in the testing cohort, 39 of 50 tumors (78.0%) were ≤ 2 cm and 11 (22.0%) were > 2–3 cm. For internal validation, the cohort was randomly divided into a training cohort (70%, n = 112) and a testing cohort (30%, n = 50). In the training cohort, the VPI-positive group comprised 23 males and 17 females with a mean age of 62.4 ± 8.8 years, whereas the VPI-negative group included 32 males and 40 females with a mean age of 59.7 ± 11.7 years. No statistically significant differences were observed between the two groups in clinical variables, including sex, age, history of surgery, smoking history, and surgical procedure (all P > 0.05). Regarding CT imaging features, significant differences were identified between the VPI-positive and VPI-negative groups in nodule size, spiculation, pleural indentation, and pleural contact length (all P < 0.05). Compared with the VPI-negative group, patients with VPI-positive solid lung adenocarcinoma nodules had significantly larger tumor sizes (18.9 ± 4.7 mm vs. 14.5 ± 4.9 mm, P < 0.001). In addition, spiculation (70.0% vs. 26.4%, P < 0.001) and pleural indentation (55.0% vs. 34.7%, P = 0.037) were more frequently observed in the VPI-positive group. The pleural contact length was also significantly greater in the VPI-positive group [19.1 (14.4, 22.7) mm vs. 9.2 (5.6, 13.6) mm, P < 0.001]. By contrast, no significant differences were found in other CT features, including nodule location, lobulation, air bronchogram, vacuole sign, cavity, calcification, emphysema, or vascular convergence (all P > 0.05), as summarized in Tables 1 and 2.
Table 1.
Clinical characteristics of patients in training and testing cohorts
| parameters | Training cohort | Testing cohort | ||||
|---|---|---|---|---|---|---|
| VPI(+) n = 40 |
VPI(-) n = 72 |
P | VPI(+) n = 18 |
VPI(-) n = 32 |
P | |
| Gender | 0.185 | 0.264 | ||||
| Male | 23(57.5) | 32(44.4) | 13(72.2) | 18(56.2) | ||
| Female | 17(42.5) | 40(55.6) | 5(27.8) | 14(43.8) | ||
| Age(year) | 62.4 ± 8.8 | 59.7 ± 11.7 | 0.212 | 59.3 ± 8.9 | 59.7 ± 8.7 | 0.865 |
| Operation history | 0.75 | 0.368 | ||||
| Absent | 30(75) | 52(72.2) | 14(77.8) | 21(65.6) | ||
| Present | 10(25) | 20(27.8) | 4(22.2) | 11(34.4) | ||
| Smoking history | 0.9 | 0.05 | ||||
| Absent | 38(95) | 68(94.4) | 14(77.8) | 31(96.9) | ||
| Present | 2(5) | 4(5.6) | 4(22.2) | 1(3.1) | ||
| Type of surgery | 0.077 | 0.375 | ||||
| Sublobar resection | 4(10) | 17(23.6) | 6(33.3) | 7(21.9) | ||
| Lobectomy | 36(90) | 55(76.4) | 12(66.7) | 25(78.1) | ||
Table 2.
CT imaging features of patients in training and testing cohorts
| parameters | Training cohort | Testing cohort | ||||
|---|---|---|---|---|---|---|
| VPI(+) n = 40 |
VPI(-) n = 72 |
P | VPI(+) n = 18 |
VPI(-) n = 32 |
P | |
| Nodule location | 0.939 | 0.574 | ||||
| Right lower lobe | 9(22.5) | 17(23.6) | 3(16.7) | 7(21.9) | ||
| Right middle lobe | 7(17.5) | 12(16.7) | 1(5.6) | 3(9.4) | ||
| Right upper lobe | 10(25) | 17(23.6) | 5(27.8) | 9(28.1) | ||
| Left lower lobe | 9(22.5) | 13(18.1) | 2(11.1) | 7(21.9) | ||
| Left upper lobe | 5(12.5) | 13(18.1) | 7(38.9) | 6(18.8) | ||
| Nodule size(mm) | 18.9 ± 4.7 | 14.5 ± 4.9 | < 0.001 | 15.3 ± 4.9 | 17 ± 4.8 | 0.232 |
| Spiculation | < 0.001 | 0.001 | ||||
| Absent | 12(30) | 53(73.6) | 5(27.8) | 24(75) | ||
| Present | 28(70) | 19(26.4) | 13(72.2) | 8(25) | ||
| Lobulation | 0.629 | 0.768 | ||||
| Absent | 18(45) | 29(40.3) | 6(33.3) | 12(37.5) | ||
| Present | 22(55) | 43(59.7) | 12(66.7) | 20(62.5) | ||
| Air bronchogram | 0.734 | 0.977 | ||||
| Absent | 29(72.5) | 50(69.4) | 14(77.8) | 25(78.1) | ||
| Present | 11(27.5) | 22(30.6) | 4(22.2) | 7(21.9) | ||
| Vacuole sign | 0.069 | 0.684 | ||||
| Absent | 34(85) | 50(69.4) | 15(83.3) | 28(87.5) | ||
| Present | 6(15) | 22(30.6) | 3(16.7) | 4(12.5) | ||
| Pleural indentation | 0.037 | 0.022 | ||||
| Absent | 18(45) | 47(65.3) | 7(38.9) | 23(71.9) | ||
| Present | 22(55) | 25(34.7) | 11(61.1) | 9(28.1) | ||
| Cavity | 0.616 | 0.53 | ||||
| Absent | 38(95) | 70(97.2) | 18(100) | 30(93.8) | ||
| Present | 2(5) | 2(2.8) | 0 | 2(6.3) | ||
| Calcification | 1 | 1 | ||||
| Absent | 39(97.5) | 71(98.6) | 17(94.4) | 31(96.9) | ||
| Present | 1(2.5) | 1(1.4) | 1(5.6) | 1(3.1) | ||
| Emphysema | 0.491 | 0.659 | ||||
| Absent | 31(77.5) | 53(73.6) | 15(83.3) | 25(78.1) | ||
| Present | 9(22.5) | 19(26.4) | 3(16.7) | 7(21.9) | ||
| Vascular convergence | 0.225 | 0.83 | ||||
| Absent | 29(72.5) | 44(61.1) | 5(27.8) | 8(25) | ||
| Present | 11(27.5) | 28(38.9) | 13(72.2) | 24(75) | ||
| Pleural contact length (mm) |
19.1 (14.4,22.7) |
9.2 (5.6,13.6) |
< 0.001 | 16.3 ± 7.3 | 11.9 ± 6.1 | 0.026 |
VPI, visceral pleural invasion; CT, computed tomography
Interobserver agreement of CT features
The measurements of CT imaging features by two observers for the dataset showed excellent consistency for quantitative parameters (ICC ranging from 0.991 to 0.994) and strong consistency for qualitative indicators (Kappa values ranging from 0.873 to 1.000) (Supplementary Table 1).
Multivariate logistic regression analysis and nomogram construction
Features showing significant differences in univariate analysis (P < 0.05) were entered into the multivariate logistic regression model. Spiculation (odds ratio [OR] = 5.593, 95% confidence interval [CI]: 1.815–17.239), pleural indentation (OR = 3.770, 95% CI: 1.188–11.970), and pleural contact length (OR = 1.247, 95% CI: 1.120–1.387) remained independently associated with VPI in solid lung adenocarcinoma nodules with pleural contact (Table 3). A nomogram incorporating these predictors was subsequently developed based on the fitted regression coefficients to provide an individualized estimate of VPI risk (Fig. 3). The nomogram assigns points to each predictor according to its regression coefficient, such that variables with larger effects contribute proportionally more to the overall risk estimate. For an individual patient, the value of each CT feature is mapped to the corresponding points on the nomogram scale, and the points are summed to yield a total score. This total score is then translated into the estimated probability of VPI, enabling individualized preoperative risk prediction.
Table 3.
Logistic regression models showing variables independently associated with VPI in training cohort
| Variables | OR | 95%CI | P |
|---|---|---|---|
| Spiculation | 5.593 | 1.815–17.239 | 0.003 |
| Pleural indentation | 3.77 | 1.188–11.97 | 0.024 |
| Pleural contact length | 1.247 | 1.12–1.387 | < 0.001 |
OR, odds ratio; CI, Confidence interval
Fig. 3.
Nomogram for prediction of VPI based on training cohort. Abbreviations: VPI= visceral pleural invasion
Evaluation of model performance
The nomogram based on these three CT features demonstrated good discriminative performance in both the training and testing cohorts, with AUCs of 0.901 (95% CI, 0.846–0.956) and 0.864 (95% CI, 0.737–0.937), respectively (Fig. 4). In the training and testing cohorts, the sensitivities for predicting VPI were 90% and 83%, with corresponding specificities of 83.3% and 84.4%. Calibration performance was acceptable, with low mean absolute error (MAE = 0.02 in the training cohort and 0.056 in the testing cohort), indicating limited deviation between predicted and observed risks and only a modest decrease in performance between the two cohorts. The calibration curves showed close agreement between predicted and observed VPI probabilities across the range of risk, and this agreement was preserved after bootstrap resampling (B = 1,000), as reflected by the bias-corrected curves (Fig. 5). Decision curve analysis suggested potential utility in preoperative risk assessment (Fig. 6).
Fig. 4.
ROC curve and AUC of the predictive model. For training cohort (A), area under the curve (AUC) = 0.901 with 95% CI 0.846–0.956. For testing cohort (B), AUC = 0.864 with 95% CI 0.737–0.937. Abbreviations: ROC =receiver operating characteristic; VPI= visceral pleural invasion; AUC =area under the curve; CI =confidence interval
Fig. 5.
Calibration curves of the predictive model. A Calibration curve in the training cohort. B Calibration curve in the testing cohort
Fig. 6.
DCA of the predictive model. A DCA in the training cohort. B DCA in the testing cohort. Abbreviations: DCA=decision curve analysis
Discussion
Non-small cell lung cancer (NSCLC) remains a leading cause of cancer-related mortality worldwide, and lung adenocarcinoma (LA) is its most common subtype [22, 23]. Visceral pleural invasion (VPI) is a clinically relevant adverse feature that affects staging, treatment decisions, and prognosis in LA [21, 24–27]. Previous studies have reported that VPI occurs more frequently in solid nodules than in subsolid nodules, possibly reflecting their generally higher invasive potential and propensity to breach the pleural barrier [28, 29]. Importantly, solid nodules with pleural contact represent a specific anatomic setting characterized by direct tumor–pleura apposition with minimal intervening lung parenchyma, which may lower the effective barrier to pleural invasion; therefore, conclusions drawn from subsolid or mixed cohorts may not generalize. In this context, we developed and internally validated a CT-based nomogram using spiculation, pleural indentation, and pleural contact length to estimate VPI risk in solid pulmonary adenocarcinoma nodules with pleural contact. Recent randomized trials, including JCOG0802/WJOG4607L and CALGB/Alliance 140,503, have supported sublobar resection as an acceptable option for selected patients with small-sized peripheral NSCLC. Against this evolving surgical background, preoperative VPI risk estimation may provide supplementary information in pleura-abutting solid nodules, particularly when the adequacy of limited resection is under consideration. However, because the role of VPI in determining the extent of resection remains controversial, the proposed nomogram should be interpreted as a supplementary risk-estimation tool rather than a stand-alone determinant of surgical strategy. The incremental contribution of the present study lies primarily in its subgroup-specific clinical focus rather than methodological complexity. Specifically, we developed a CT-based nomogram for solid pulmonary adenocarcinoma nodules with pleural contact using routinely assessed semantic CT features, thereby providing a simple and clinically deployable model for this particular clinical setting.
Spiculation has been proposed to reflect asymmetric infiltrative tumor growth along adjacent bronchi, vessels, and lymphatic channels, accompanied by stromal proliferation and neovascularization; on CT, it typically appears as fine radiating or corona-like strands extending from the nodule margin [30, 31]. Clinically, the presence of spiculation is often associated with a more aggressive phenotype and poorer prognosis [32]. A prior meta-analysis further demonstrated a significant association between spiculation and VPI (OR = 2.581, 95% CI: 1.640–4.062), supporting spiculation as an important semantic CT feature for preoperative VPI risk stratification [33]. Chen et al. [19] reported that spiculation was associated with an increased risk of VPI in pleura-contacting lung adenocarcinoma (OR = 2.70), and that combining spiculation with a pleural contact length > 10 mm improved sensitivity to 90%, with an AUC of 0.73. In our cohort, spiculation was present in 42.0% of nodules overall and was substantially more frequent in the VPI-positive group (70.7%). Moreover, multivariable logistic regression identified spiculation as an independent predictor of VPI in solid pulmonary adenocarcinoma nodules with pleural contact, consistent with the observations reported by Chen et al. However, the independent contribution of spiculation to VPI prediction has not been consistently reported across studies. For instance, Wang et al. investigated CT features associated with VPI in subpleural clinical stage IA lung adenocarcinoma and found that spiculation was significantly more common in the VPI-positive group than in the VPI-negative group (47.5% vs. 12.1%); nevertheless, spiculation was not retained as an independent risk factor in their multivariable model [34]. Such discrepancies may be attributable to differences in study populations and tumor–pleura interfaces. Notably, our study specifically focused on solid nodules with pleural contact, whereas other cohorts may have included lesions without pleural contact or with only indirect pleural proximity. Taken together, we speculate that the predictive value of spiculation may be more pronounced in tumors with pleural contact, potentially because these lesions exhibit stronger local invasiveness and are more likely to breach the visceral pleural barrier; this may partly explain why spiculation emerged as an independent predictor in our cohort.
The pleural indentation sign is believed to arise from intratumoral fibrosis, scar formation, and connective-tissue remodeling that generate traction on adjacent structures. When this traction becomes sufficiently strong, it can pull the visceral pleura inward toward the lesion and deform the pleural surface. In many cases, the adjacent pleura shows no overt thickening or adhesion, suggesting a localized pulling effect rather than diffuse pleural disease. On CT, the sign appears as focal inward retraction or wrinkling of the visceral pleura next to the nodule [35]. Previous studies [14, 34, 36, 37] reported pleural indentation as an independent predictor of VPI, possibly because it reflects tumor–pleura interaction and early pleural involvement. In our cohort of solid pulmonary adenocarcinoma nodules with pleural contact, pleural indentation was present in 55.0% of VPI-positive cases versus 34.7% of VPI-negative cases. After adjustment in multivariable logistic regression, the between-group difference remained statistically significant, supporting pleural indentation as an independent predictor of VPI. This relationship may relate to the biology of predominantly solid tumors—higher cellularity, a limited lepidic (ground-glass) component, and faster proliferation—which could promote local extension toward the pleura and increase the likelihood of both pleural indentation and VPI.
In the multivariate logistic regression analysis, pleural contact length (OR = 1.247, 95% CI: 1.120–1.387, P < 0.001) was identified as an independent predictor of VPI in solid pulmonary adenocarcinoma nodules with pleural contact. Previous studies have also shown that pleural contact length is independently associated with VPI. Chen et al. [19] reported that a contact length > 10 mm (P = 0.001) served as an independent predictor of VPI in solid pulmonary adenocarcinoma nodules. Deng et al. [38] found that, in peripheral lung adenocarcinoma with pleural contact, both the total pleural contact length (15.4 mm vs. 12.2 mm, P = 0.001) and the solid-component contact length (11.7 mm vs. 4.2 mm, P < 0.001) were significantly greater in the VPI-positive group, and the solid-component contact length remained an independent risk factor for VPI (P < 0.001). These findings are broadly consistent with our results. However, our findings differ from those of Heidinger et al. [39], who found that although there was a difference in contact length between the VPI-positive and VPI-negative groups, it was not an independent predictor of VPI. This discrepancy may be attributed to differences in measurement methods: Heidinger et al. measured linear contact length, while our study measured curved contact length, which may more accurately reflect the extent of contact between the nodule and the pleura. Overall, although pleural contact length cannot replace biopsy, it serves as a non-invasive indicator suggesting the presence of VPI and plays a crucial role in staging diagnosis, treatment planning, and prognostic evaluation.
Previous studies have shown that lung adenocarcinoma nodules with VPI exhibit more aggressive imaging features, which is consistent with our findings [16, 40, 41]. In this study, tumors in VPI-positive solid pulmonary adenocarcinoma patients were larger [18.9 ± 4.7 mm vs. 14.5 ± 4.9 mm, P < 0.001], had longer pleural contact lengths [19.1 (14.4, 22.7) mm vs. 9.2 (5.6, 13.6) mm, P < 0.001], and had higher frequencies of spiculation (70% vs. 26.4%, P < 0.001) and pleural indentation (55% vs. 34.7%, P = 0.037). Previous studies have reported tumor size as an independent risk factor for predicting VPI [16]. However, in our multivariable logistic regression analysis, tumor size was not identified as an independent risk factor for VPI, which may be related to differences in case selection and measurement methods. These imaging features have been associated with higher malignancy potential and poorer prognosis [42–44].
In this study, the nomogram showed good predictive performance in both the training and testing cohorts, with AUCs of 0.901 (95% CI: 0.846–0.956) and 0.864 (95% CI: 0.737–0.937), respectively. Previously published VPI prediction models can be broadly categorized into CT semantic-feature-based models, radiomics-based models, and combined models. These studies differ not only in model performance, but also in cohort composition, model inputs, validation strategy, interpretability, and intended clinical use. For example, Huang et al. [37] developed a multicenter nomogram combining CT features and radiomics features for subpleural invasive adenocarcinoma ≤ 2 cm, whereas Wang et al. [45] and Zhao et al. [46] proposed radiomics-based or combined models in broader early-stage lung adenocarcinoma cohorts. By comparison, the present study specifically focused on solid pulmonary adenocarcinoma nodules with pleural contact, a clinically specific subgroup characterized by direct tumor–pleura apposition and potentially distinct VPI risk patterns. In addition, unlike radiomics-based approaches that rely on lesion segmentation and may be sensitive to image acquisition and reconstruction heterogeneity, our model was constructed using routinely assessed CT semantic features and pleural contact length, which may offer greater interpretability and facilitate implementation in routine preoperative assessment. However, we do not claim superiority over previously published models based solely on numerical performance, because no direct head-to-head comparison was performed within the same cohort. Rather than competing on technical complexity, we aimed to provide a pragmatic tool for supplementary preoperative VPI risk estimation in routine clinical assessment. In current thoracic surgical practice, the choice of operative procedure is determined by multiple factors, including tumor size, radiologic invasiveness, expected surgical margin, pulmonary reserve, and patient comorbidities. Therefore, the proposed nomogram is not intended to serve as a stand-alone determinant of treatment strategy. Rather, it may provide supplementary preoperative information in selected borderline clinical scenarios involving solid nodules with pleural contact. Importantly, because the present study did not evaluate the relationship between predicted VPI risk and actual surgical procedures or postoperative outcomes, its potential impact on clinical management remains to be established in future studies. Accordingly, the relative value of different VPI prediction models should be interpreted in light of their target populations and intended clinical settings. More complex radiomics-based models may provide additional quantitative information, whereas semantic-feature-based models such as ours may offer better interpretability and easier integration into routine workflow. Future studies with direct head-to-head comparison in external cohorts would help better define the complementary roles of these approaches.
This study has several limitations. It was a retrospective single-center study with a relatively limited sample size, which may introduce selection bias and restrict generalizability. Although the use of a single institutional dataset may provide relative consistency in CT acquisition, surgical workflow, and pathological assessment, it may also limit the transportability of the model to other institutions with different patient-selection pathways, imaging protocols, reconstruction parameters, and clinical practice patterns. The current 70%/30% split-sample approach constitutes internal validation rather than true external validation, and the relatively limited number of VPI-positive cases may affect model stability and restrict generalizability. Therefore, the present findings should be interpreted cautiously. In addition, because the current model was developed and internally validated within the same institutional setting, its broader applicability to external populations remains uncertain. We also did not perform subgroup analysis according to the anatomical location of pleural contact (e.g., costal, mediastinal, diaphragmatic, or fissural pleura), and such anatomic heterogeneity may influence the risk of VPI. Although the final model was kept parsimonious and no problematic multicollinearity was identified, the relatively limited number of VPI events warrants cautious interpretation of model stability. Further multicenter studies with external validation in independent cohorts are needed to further assess the reproducibility, robustness, and clinical applicability of the model across institutions with different patient populations, imaging protocols, and clinical workflows.
Conclusion
In summary, we identified spiculation, pleural indentation, and pleural contact length as independent predictors of VPI in solid lung adenocarcinoma nodules with pleural contact and developed a CT-based nomogram for preoperative risk estimation. The model demonstrated good discrimination and acceptable calibration under internal validation conditions, and decision curve analysis suggested potential utility in preoperative risk assessment, thereby supporting its value as a supplementary tool for estimating VPI risk before surgery. Future multicenter studies with external validation are needed to further clarify its clinical integration, robustness, and broader applicability.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
Not applicable.
Abbreviations
- CT
Computed Tomography
- VPI
Visceral Pleural Invasion
- ROC
Receiver Operating Characteristic Curve
- NSCLC
Non-Small Cell lung Cancer
- OR
Odds Ratio
- AUC
Area Under The Curve
- DCA
Decision Curve Analysis
- TNM
Tumor–Node–Metastasis
Author contributions
H.Z. and D.Z. were the guarantors of the integrity of the entire study. R.X., H.Z., and J.T. contributed to the study conception and design. R.X., M.D., W.Y., and J.T. performed the literature research. R.X., Q.Y., and J.Z. contributed to the clinical studies and data collection. R.X. and M.D. performed data analysis. Q.Y., J.Z., and J.T. conducted the statistical analysis. R.X., D.Z., and W.Y. drafted the manuscript. H.Z. critically revised and edited the manuscript. All authors read and approved the final manuscript.
Funding
This work was supported by the Health Research Program of Anhui [AHWJ2023BAb20009]; and the Key Natural Science Research Project of the Anhui Educational Committee [2024AH050796]; and the Key Project of the Clinical Research Cultivation Program of the Second Affiliated Hospital of Anhui Medical University (2020LCZD12).
Data availability
The data underlying this article are not publicly available due to patient privacy and institutional restrictions, but will be shared on reasonable request to the corresponding author.
Declarations
Ethics approval and consent to participate
This retrospective study was approved by the Ethics Committee of the Second Affiliated Hospital of Anhui Medical University (Approval No. SL-YX2025-089) and was conducted in accordance with the Declaration of Helsinki and its later amendments. The requirement for informed consent was waived due to the retrospective nature of the study and the use of anonymized data.
Consent for publication
Not applicable.
Author information
Rui Xu- Master of Medicine candidate, Jianing Tang- Master of Medicine candidate, Qianyao Yuan- Master of Medicine candidate, Mengjiao Ding- Master of Medicine candidate, Jingjing Zhang- Master of Medicine candidate, Wenjun Yao-Doctor of Philosophy, Dai Zhang-Doctor of Philosophy, Hong Zhao -Doctor of Philosophy.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rui Xu and Jianing Tang contributed equally to this work and share first authorship.
Contributor Information
Dai Zhang, Email: zhangdai_casia@126.com.
Hong Zhao, Email: 178331090@qq.com.
References
- 1.Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229–63. [DOI] [PubMed] [Google Scholar]
- 2.Qi J, Li M, Wang L, et al. National and subnational trends in cancer burden in China, 2005-20: an analysis of national mortality surveillance data. Lancet Public Health. 2023;8(12):e943–55. [DOI] [PubMed] [Google Scholar]
- 3.Luo G, Zhang Y, Rumgay H, et al. Estimated worldwide variation and trends in incidence of lung cancer by histological subtype in 2022 and over time: a population-based study. Lancet Respir Med. 2025;13(4):348–63. [DOI] [PubMed] [Google Scholar]
- 4.Ruan Z, Zhuo X, Xu C. Diagnosis, treatment, and prognosis of stage IB non-small cell lung cancer with visceral pleural invasion. Front Oncol. 2023;13:1310471. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Yu Y, Huang R, Wang P, et al. Sublobectomy versus lobectomy for long-term survival outcomes of early-stage non-small cell lung cancer with a tumor size ≤ 2 cm accompanied by visceral pleural invasion: a SEER population-based study. J Thorac Dis. 2020;12(3):592–604. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Yang X, Sun F, Chen L, et al. Prognostic value of visceral pleural invasion in non-small cell lung cancer: A propensity score matching study based on the SEER registry. J Surg Oncol. 2017;116(3):398–406. [DOI] [PubMed] [Google Scholar]
- 7.Zhang B, Fang WT, Zhong H. [Introduction to the 9(th) edition of TNM classification for lung cancer]. Zhonghua Zhong Liu Za Zhi. 2024;46(3):206–10. [DOI] [PubMed] [Google Scholar]
- 8.Huang W, Deng HY, Lin MY, et al. Treatment Modality for Stage IB Peripheral Non-Small Cell Lung Cancer With Visceral Pleural Invasion and ≤ 3 cm in Size. Front Oncol. 2022;12:830470. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Altorki N, Wang X, Kozono D, et al. Lobar or Sublobar Resection for Peripheral Stage IA Non-Small-Cell Lung Cancer. N Engl J Med. 2023;388(6):489–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Jiang L, Liang W, Shen J, et al. The impact of visceral pleural invasion in node-negative non-small cell lung cancer: a systematic review and meta-analysis. Chest. 2015;148(4):903–11. [DOI] [PubMed] [Google Scholar]
- 11.Hung JJ, Jeng WJ, Hsu WH, Chou TY, Lin SF, Wu YC. Prognostic significance of the extent of visceral pleural invasion in completely resected node-negative non-small cell lung cancer. Chest. 2012;142(1):141–50. [DOI] [PubMed] [Google Scholar]
- 12.Dai X, Li H, Qiu H, et al. Investigating the correlation between the cytological grading system and computed tomography features in early-stage lung adenocarcinoma. Oncology. 2025:1–13. [DOI] [PubMed]
- 13.Zou PL, Ma CH, Li X, Luo TY, Lv FJ, Li Q. Early Lung Adenocarcinoma Manifesting as Irregular Subsolid Nodules: Clinical and CT Characteristics. Acad Radiol. 2025;32(4):2320–9. [DOI] [PubMed] [Google Scholar]
- 14.Qi LP, Li XT, Yang Y, et al. Multivariate Analysis of Pleural Invasion of Peripheral Non-Small Cell Lung Cancer-Based Computed Tomography Features. J Comput Assist Tomogr. 2016;40(5):757–62. [DOI] [PubMed] [Google Scholar]
- 15.Wei SH, Zhang JM, Shi B, Gao F, Zhang ZX, Qian LT. The value of CT radiomics features to predict visceral pleural invasion in ≤ 3 cm peripheral type early non-small cell lung cancer. J Xray Sci Technol. 2022;30(6):1115–26. [DOI] [PubMed] [Google Scholar]
- 16.Luo Q, Li H, Liu X, et al. Development and validation of a nomogram for predicting visceral pleural invasion in patients with early-stage non-small cell lung cancer. Transl Lung Cancer Res. 2024;13(12):3352–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Ahn SY, Park CM, Jeon YK, et al. Predictive CT Features of Visceral Pleural Invasion by T1-Sized Peripheral Pulmonary Adenocarcinomas Manifesting as Subsolid Nodules. AJR Am J Roentgenol. 2017;209(3):561–6. [DOI] [PubMed] [Google Scholar]
- 18.Fu F, Zhang Y, Wen Z, et al. Distinct Prognostic Factors in Patients with Stage I Non-Small Cell Lung Cancer with Radiologic Part-Solid or Solid Lesions. J Thorac Oncol. 2019;14(12):2133–42. [DOI] [PubMed] [Google Scholar]
- 19.Chen Y, Huang Q, Zhong H, Li A, Lin Z, Guo X. Correlations between iodine uptake, invasive CT features and pleural invasion in adenocarcinomas with pleural contact. Sci Rep. 2023;13(1):16191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Henschke CI, Yip R, Smith JP, et al. CT screening for lung cancer: part-solid nodules in baseline and annual repeat rounds. AJR Am J Roentgenol. 2016;207(6):1176–84. [DOI] [PubMed] [Google Scholar]
- 21.Seok Y, Jeong JY, Lee E. Extent of visceral pleural invasion and the prognosis of surgically resected node-negative non-small cell lung cancer. Thorac Cancer. 2017;8(3):197–202. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer J Clin. 2024;74(1):12–49. [DOI] [PubMed] [Google Scholar]
- 23.Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin. 2021;71(3):209–49. [DOI] [PubMed] [Google Scholar]
- 24.Kim BG, Choi J, Lee SK, et al. Impact of adjuvant chemotherapy on patients with stage IB non-small cell lung cancer with visceral pleural invasion. J Thorac Dis. 2024;16(2):875–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Pathak R, Goldberg SB, Canavan M, et al. Association of Survival With Adjuvant Chemotherapy Among Patients With Early-Stage Non-Small Cell Lung Cancer With vs Without High-Risk Clinicopathologic Features. JAMA Oncol. 2020;6(11):1741–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Mathey-Andrews C, Abruzzo AR, Venkateswaran S, et al. Segmentectomy vs Lobectomy for Early Non-Small Cell Lung Cancer With Visceral Pleural Invasion. Ann Thorac Surg. 2024;117(5):1007–14. [DOI] [PubMed] [Google Scholar]
- 27.Van Schil PE, Asamura H, Nishimura KK, et al. The International Association for the Study of Lung Cancer Lung Cancer Staging Project: Proposals for the Revisions of the T-Descriptors in the Forthcoming Ninth Edition of the TNM Classification for Lung Cancer. J Thorac Oncol. 2024;19(5):749–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Yip R, Ma T, Flores RM, et al. Survival with Parenchymal and Pleural Invasion of Non-Small Cell Lung Cancers Less than 30 mm. J Thorac Oncol. 2019;14(5):890–902. [DOI] [PubMed] [Google Scholar]
- 29.Sun Q, Li P, Zhang J, et al. CT Predictors of Visceral Pleural Invasion in Patients with Non-Small Cell Lung Cancers 30 mm or Smaller. Radiology. 2024;310(1):e231611. [DOI] [PubMed] [Google Scholar]
- 30.Choromańska A, Macura KJ. Evaluation of solitary pulmonary nodule detected during computed tomography examination. Pol J Radiol. 2012;77(2):22–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Bankier AA, MacMahon H, Colby T, et al. Fleischner Society: Glossary of Terms for Thoracic Imaging. Radiology. 2024;310(2):e232558. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Cui SL, Qi LL, Liu JN, et al. A prediction model based on computed tomography characteristics for identifying malignant from benign sub-centimeter solid pulmonary nodules. J Thorac Dis. 2024;16(7):4238–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Long Y, Li Y, Lin L, et al. Predicting visceral pleural invasion in part-solid and solid nodules using CT features: a systematic review, meta-analysis, and independent cohort validation. J Comput Assist Tomogr. 2025. [DOI] [PubMed]
- 34.Wang Y, Lyu D, Zhou T, Tu W, Fan L, Liu S. Multivariate analysis based on the maximum standard unit value of (18)F-fluorodeoxyglucose positron emission tomography/computed tomography and computed tomography features for preoperative predicting of visceral pleural invasion in patients with subpleural clinical stage IA peripheral lung adenocarcinoma. Diagn Interv Radiol. 2023;29(2):379–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Meng Y, Gao J, Wu C, et al. The prognosis of different types of pleural tags based on radiologic-pathologic comparison. BMC Cancer. 2022;22(1):919. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Deng HY, Li G, Luo J, Alai G, Zhuo ZG, Lin YD. Novel biologic factors correlated to visceral pleural invasion in early-stage non-small cell lung cancer less than 3 cm. J Thorac Dis. 2018;10(4):2357–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Huang R, Zhao C, Yang J, et al. Nomogram based on radiomics and CT features for predicting visceral pleural invasion of invasive adenocarcinoma ≤ 2 cm: A multicenter study. Eur J Radiol. 2025;190:112227. [DOI] [PubMed] [Google Scholar]
- 38.Lyu D, Wang Y, Tu W, et al. Prediction of visceral pleural invasion of clinical stage IA lung adenocarcinoma based on computed tomography features. Transl Cancer Res. 2025;14(3):1596–608. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Heidinger BH, Schwarz-Nemec U, Anderson KR, et al. Visceral Pleural Invasion in Pulmonary Adenocarcinoma: Differences in CT Patterns between Solid and Subsolid Cancers. Radiol Cardiothorac Imaging. 2019;1(3):e190071. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Shang X, Hu B, Gao F, Ren W. Correlation between computed tomography imaging and pathological stages and subtypes in early lung adenocarcinoma. J Cancer Res Ther. 2020;16(7):1569–74. [DOI] [PubMed] [Google Scholar]
- 41.Li L, Yang Q, Luo D, Wang X, Liu Z, Huang R. Baseline computed tomography imaging findings could assist in early diagnosis of visceral pleural invasion for newly discovered early subpleural non-small cell lung cancer: T1 or T2. J Thorac Dis. 2024;16(9):5779–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Liu J, Li Y, Long Y, et al. Predicting High-risk Lung Adenocarcinoma in Solid and Part-solid Nodules on Low-dose CT: A Multicenter Study. Acad Radiol. 2025;32(5):2966–76. [DOI] [PubMed] [Google Scholar]
- 43.Zuo YQ, Gao D, Cui JJ, et al. Peritumoral and intratumoral radiomics for predicting visceral pleural invasion in lung adenocarcinoma based on preoperative computed tomography (CT). Clin Radiol. 2025;80:106729. [DOI] [PubMed] [Google Scholar]
- 44.Yang S, Yang L, Teng L, et al. Visceral pleural invasion by pulmonary adenocarcinoma ≤ 3 cm: the pathological correlation with pleural signs on computed tomography. J Thorac Dis. 2018;10(7):3992–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Wang Y, Lyu D, Hu S, et al. Nomogram using intratumoral and peritumoral radiomics for the preoperative prediction of visceral pleural invasion in clinical stage IA lung adenocarcinoma. J Cardiothorac Surg. 2024;19(1):307. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Zhao F, Zhao Y, Ye Z, Yan Q, Sun H, Zhou G. Integrating radiomics features and CT semantic characteristics for predicting visceral pleural invasion in clinical stage Ia peripheral lung adenocarcinoma. Discov Oncol. 2025;16(1):780. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data underlying this article are not publicly available due to patient privacy and institutional restrictions, but will be shared on reasonable request to the corresponding author.