The Global Cancer Report 2020 highlights lung cancer as the leading cause of cancer-related mortality worldwide, with adenocarcinoma recognized as the predominant histological subtype (1). In 2021, the World Health Organization (WHO) updated the classification of lung adenocarcinomas, dividing them into in-situ, microinvasive, and invasive categories, based on the extent of their invasive progression (2).

The widespread adoption of early lung cancer screening programs has led to increased detection of stage IA lung adenocarcinomas, which typically present as lung nodules. Surgical excision is the main treatment modality for these patients (3). For early-stage IA lung adenocarcinoma, sublobectomy is now the treatment of choice (4–6). Nonetheless, in some instances, recurrences and metastases occur, with STAS significantly contributing to these outcomes (7).

STAS was introduced by the WHO in 2015 as a distinct pattern of invasion in invasive lung adenocarcinoma (IAC). This pattern is marked by the presence of pathological micropapillary clusters, solid nests, or single cells that extend beyond the tumor margin, infiltrating the adjacent lung parenchyma (8). Research indicates that STAS correlates with a worse prognosis, prompting several scholars to advocate for lobectomy in cases of IA lung adenocarcinoma showing STAS positivity (9–12).

Limitations exist with intraoperative frozen sections for STAS prediction, making preoperative clinical and imaging evaluations more predictive (13). Historical data suggest variability in the predictive reliability of clinical and imaging features for STAS (8, 14–16). Studies by Onozato et al. (15) and Shiono et al. (14) suggest that smoking increases the risk of STAS, although Uruga et al. (16) observed no such correlation. Similarly, Warth et al. (17) identified male gender as a risk factor for STAS, a finding not corroborated by Kadota (8).

Other research indicates that a tumor diameter exceeding 2 cm does not reliably predict STAS (18, 19). Yin et al. (20) observed that pulmonary nodules with partly solid features tend to develop STAS, while Toyokawa et al. (19) determined that a solid component over 50% is predictive of STAS. Key CT indicators such as solid nodules, spiculation, vacuoles, well-defined borders, lobulation, and pleural traction are frequently observed in tumors positive for STAS (20, 21). However, the focus on individual features has limited the overall predictive accuracy. Combining multiple diagnostic features has shown to enhance performance, with Ding et al. (22) achieving an AUC of 0.724 using a combination of CTR, tumor size, vacuoles, and spiculation, while Gao et al. (23) reached an AUC of 0.808 by integrating age, nodule type, and SUVmax.

Despite some improvements in diagnostic efficacy, the limitations of small sample sizes and variability in features persist, underscoring the need for further studies involving more comprehensive datasets. To address this gap, we propose a large-scale investigation to thoroughly assess the link between clinical and imaging features and STAS in stage IA lung adenocarcinoma. Our objectives include identifying independent predictors, developing and validating a predictive nomogram for STAS, and aiding in the selection of appropriate treatment strategies for patients with stage IA lung adenocarcinoma.

Lung cancer continues to be a leading factor in cancer-related deaths, with adenocarcinoma identified as the most common subtype (28). The evolution of screening technologies has significantly improved the identification of early-stage lung cancer, positioning surgical intervention as the most effective treatment option for these instances. In light of the findings from the JCOG studies (29–32), sublobectomy has emerged as a prevalent surgical method. Nevertheless, the occurrence of STAS in patients is intricately associated with heightened recurrence rates and reduced survival rates. Lobectomy generally results in more favorable outcomes for individuals diagnosed with STAS-positive stage IA lung adenocarcinoma (11, 33). Therefore, precise preoperative assessment of STAS is crucial for guiding the selection of surgical interventions (8, 10). This study introduced a nomogram that integrates preoperative clinical and CT imaging data to estimate the likelihood of STAS, thereby facilitating surgical planning for early-stage lung adenocarcinoma.

Previous findings by Onozato et al. (15) and Shiono et al. (14) indicate that smoking increases the likelihood of developing STAS, while Warth et al. (17) observed a higher propensity for STAS among male patients, a result that contrasts with ours. In our analysis, although gender and smoking history were significant in univariate analysis, they did not emerge as independent predictors of STAS, aligning with other research (8, 16). This variation could be attributed to differences in the characteristics of the study populations, measurement protocols, and analytical models used in different studies.

Serum tumor markers serve as crucial non-invasive tools for detecting cancer, contributing to early lung cancer screening and monitoring for postoperative progression and metastasis (34). Shimomura et al. (35) reported elevated preoperative CEA levels in STAS-positive patients, a finding echoed by other research (36, 37). Our analysis also identifies high CEA levels as an independent risk factor for STAS. However, Qin et al. (38) did not observe a correlation between CEA levels and STAS, possibly due to the inclusion of patients at various stages of lung cancer.

The role of maximum tumor diameter as a predictor of STAS has been debated. While Toyokawa et al. (39) and others (18) suggest that diameters larger than 2 cm are unreliable predictors, conflicting evidence exists (19, 40). Yin et al. (20) conducted a meta-analysis and found no predictive value for diameters exceeding 2 cm; however, their analysis might have been impacted by treating nodule diameter as a binary variable. In a contrasting study, Qin et al. (38) treated maximum diameter as a continuous variable in a cohort of 503 patients and found it to be an independent predictor of STAS, which supports our conclusions. Our research corroborates that maximum tumor diameter is an independent indicator of STAS in stage IA lung adenocarcinoma, particularly when the diameter surpasses 1.45 cm.

The aggressiveness of lung cancer is closely linked to the proportion of solid tumor components observed on CT scans; a higher solid component indicates a more significant pathological invasion and an elevated probability of STAS. Research by multiple authors (14, 38) has shown a positive correlation between the core-to-total ratio (CTR) and the likelihood of STAS. Qi et al. (41) highlighted CTR as the most accurate CT characteristic for forecasting STAS in lung adenocarcinomas measuring ≤2 cm, whereas Jia et al. (42) demonstrated that an increase in solid components substantially raises the risk of STAS. While several studies (14, 43) have indicated that solid nodules frequently exhibit STAS in imaging, Yin et al. (20) proposed that part-solid nodules might present a greater risk of STAS, potentially reflecting the small number of purely solid nodules in their sample. In their analysis of 327 patients, Toyokawa et al. (19) concluded that a solid component percentage exceeding 50% was predictive of STAS, a finding that contrasts with that of Margerie-Mellon et al. (40), possibly due to variances in nodule size and the heterogeneity of ground-glass components. Our research corroborates that nodule type acts as an independent predictor of STAS, with an increase in solid components markedly heightening the risk.

Qualitative CT features also significantly influence STAS prediction. Yin et al. (20) explored common CT qualitative features linked with STAS, while Gu et al. (21) delved into additional characteristics. Gu’s findings suggest that tumors positive for STAS more frequently exhibited traits such as solid nodules, spiculation, vacuoles, well-defined borders, lobulation, pleural traction, and vascular clustering signs compared to STAS-negative adenocarcinomas. Our findings further substantiate pleural traction signs and vascular convergence as independent risk factors for STAS, with the observed discrepancies potentially arising from our focused analysis on stage IA lung adenocarcinoma patients.

The nomogram prediction model surpasses traditional correlation analyses by synthesizing multiple pertinent features, thereby enhancing both the precision and efficiency of predictions (44, 45). In recent times, radiomics-based nomograms have been devised to improve diagnostic accuracy for STAS. Nonetheless, the practical implementation of these tools is somewhat restricted by the sophisticated technical requirements associated with radiomics (46, 47). Contrastingly, other investigations (23, 37) have employed more straightforward clinical imaging data to estimate STAS risk; however, these studies typically involve smaller cohorts and do not encompass a thorough analysis of both clinical and imaging variables. Variability in study populations further complicates the detailed examination specifically tailored to patients with stage IA lung adenocarcinoma. In our study, we conducted an extensive evaluation of the interplay between clinical and imaging attributes and STAS across a large sample set. We successfully developed and validated a prediction model for STAS in stage IA lung adenocarcinoma, which produced favorable outcomes. When a patient is assessed as high-risk for STAS, the recommended clinical approach includes a lobectomy accompanied by extensive lymph node dissection to optimize prognostic results.

While the findings are promising, it is essential to address several limitations that warrant further examination. The retrospective design of this study presents a challenge in addressing selection bias. The research focuses solely on patients from our institution, classifying it as a single-center retrospective study with a limited sample size. While showing steady performance across different time frames, the lack of multicenter and prospective data restricts the broader applicability and validation of these results. Furthermore, the imaging characteristics were quantified through manual methods, which may introduce a potential for bias.

By amalgamating clinical and CT imaging features, we crafted a nomogram prediction model capable of forecasting the occurrence of STAS in patients with stage IA lung adenocarcinoma. This model exhibits robust predictive strength for STAS, significantly aiding clinical management and informing decision-making processes concerning this condition.