Clinical characteristics, prognosis, and nomograms for lung invasive mucinous adenocarcinoma with distant metastasis: a population-based study.

Introduction
Lung invasive mucinous adenocarcinoma (LIMA) is a rare lung cancer subtype with poor-prognosis, accounting for 2–10% of all cases (1-4). It is recognized for its distinct histopathological features and unique molecular characteristics, which contribute to its variable prognosis (4-6). Despite its recognition, the clinicopathological features of LIMA remain poorly understood, and early detection is challenging due to its relatively low incidence (7,8). As a result, LIMA is often misdiagnosed in the early stages of the disease. Moreover, in advanced stages, LIMA tends to exhibit widespread multi-organ metastasis, which is associated with a poor prognosis (9,10). These factors contribute to the limited availability of systemic therapeutic options, as targetable biomarkers for LIMA are scarce or nonexistent.
Over recent years, there has been growing interest in the genetic and molecular profiles of LIMA, leading to significant advances in the understanding of its pathogenesis (4,11,12). However, clinical management of LIMA, particularly in patients with distant metastasis (DM), remains a subject of debate, with no standardized guidelines available (13,14). This gap in knowledge highlights the need for more precise clinical decision-making tools for LIMA patients, especially those with DM, to improve prognostic accuracy and guide treatment strategies. To date, few studies have examined the relationship between clinicopathological characteristics and the metastatic behavior of LIMA. Predictive models that assess the risk of DM and survival outcomes in LIMA patients are lacking. A reliable and accurate predictive model is urgently needed to identify high-risk patients early, facilitating timely interventions and personalized treatment strategies.
Nomograms, which are effective visual tools for individualized risk prediction, have been successfully applied in various cancers to predict prognosis and guide clinical decisions (15,16). However, no nomogram has yet been developed specifically for predicting the incidence of DM and long-term survival outcomes in LIMA patients. In this study, we aimed to construct two nomogram models: one for predicting the risk of DM and another for forecasting survival outcomes in LIMA patients with DM. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-aw-2283/rc).

Methods

Study population
Data for LIMA patients were obtained from the Surveillance, Epidemiology, and End Results (SEER) registry for the period 2004–2015 (17). The inclusion criteria were as follows: (I) patients diagnosed with LIMA based on histological codes [International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3): 8480/3, 8481/3, or 8253/3] and primary site (lung and bronchus); (II) availability of clinicopathological data, including age, sex, race, laterality, primary tumor site, pathology grade, tumor-node-metastasis (TNM) stage, and tumor size; (III) sufficient survival data; (IV) exclusion of patients diagnosed based on autopsy or death certificate only. A total of 4,439 LIMA patients were included in the analysis, of whom 830 had DM at diagnosis. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.
The study cohort was randomly divided into training (70%) and validation (30%) groups. In the training cohort, a prognostic nomogram was constructed by identifying risk factors for DM in LIMA patients. In the validation cohort, the nomogram was used to assess the performance of the identified prognostic factors. For prognosis-related analysis, DM patients were categorized based on various clinical features, and the factors influencing survival were further examined. Treatment modalities, including surgery, chemotherapy, and radiotherapy, were considered as potential prognostic factors.

Data collection
Clinical data collected included age, sex, race, laterality, primary tumor site, pathology grade, T-stage, N-stage, and tumor size. These variables were analyzed to identify risk factors for DM in LIMA patients. Additionally, survival analyses were performed to investigate the prognostic factors for LIMA patients with DM. Three therapeutic approaches (surgery, chemotherapy, and radiotherapy) were included in the analysis to assess their impact on prognosis. The primary endpoint for survival analysis was overall survival (OS), defined as the time interval from diagnosis to death or the last follow-up.

Statistical analysis
Descriptive statistics were used to summarize the baseline characteristics of the study population. Continuous variables were compared using either the independent t-test or Mann-Whitney U test, depending on the distribution of the data. Categorical variables were compared using the chi-squared test.
Univariate logistic regression analysis was first performed to identify potential risk factors for DM in LIMA patients. Variables with a P value <0.05 in the univariate analysis were subsequently included in multivariate logistic regression analysis to determine the independent risk factors for DM. Similarly, Cox proportional hazards regression was applied to identify independent prognostic factors for survival in LIMA patients with DM.
Two nomograms were developed: one to predict the risk of DM and another to forecast survival outcomes for LIMA patients with DM. The predictive accuracy of the nomograms was evaluated using calibration curves and receiver operating characteristic (ROC) analysis. Decision curve analysis (DCA) was employed to assess the clinical utility of the nomograms by examining the net benefit across a range of threshold probabilities.
Patients were stratified into high-risk and low-risk groups based on the median risk score derived from the nomogram. The performance of the nomograms was validated in the independent validation cohort.
Data analyses were performed using R software (version 4.2.2) and relevant R packages, including “table1”, “regplot”, “pROC”, “ROCR”, “rms”, “ggDCA”, “survival”, “survminer”, “survivalROC”, and “ggplot” (18,19). A two-sided P value of <0.05 was considered statistically significant.

Results

Clinical characteristics of the study population
After applying the predefined inclusion and exclusion criteria, a total of 4,439 LIMA patients were included in the study. The baseline clinical characteristics of the study population are summarized in Table 1. Chi-squared tests revealed no significant differences in any of the variables between the training and validation cohorts. The majority of LIMA patients were elderly and female. The primary site of LIMA tumors was predominantly located in the lower lobe of the lung. Tumor size was most commonly in the 20–40 mm range (42.4%). The majority of patients had low-grade tumors (Grade I–II), and a significant proportion of patients had tumors with lower levels of infiltration (T1–T2) and no lymph node metastasis (N0). A total of 830 patients had DM (M1), with 586 in the training cohort and 244 in the validation cohort.

Risk variables of DM in LIMA patients
To identify risk factors associated with DM in LIMA patients, univariate and multivariate logistic regression analyses were performed. The univariate analysis identified six factors significantly associated with DM: age, primary tumor site, grade, tumor size, T-stage, and N-stage (Table 2). Subsequently, multivariate logistic regression revealed that four factors—primary site, grade, T-stage, and N-stage—were independently associated with the presence of DM in LIMA patients.

Diagnostic nomogram of DM in LIMA patients
Based on the independent risk factors identified, a diagnostic nomogram was constructed to predict the likelihood of DM in LIMA patients (Figure 1A). The nomogram demonstrated excellent diagnostic performance, with a ROC curve area under the curve (AUC) of 0.836 in the training cohort (Figure 1B). DCA confirmed the clinical reliability of this nomogram (Figure 1C), and calibration curves showed good agreement between observed and predicted outcomes (Figure 1D).
To validate the nomogram, it was tested in the independent validation cohort, yielding an AUC of 0.846 (Figure 1E). The DCA and calibration curves for the validation cohort further validated the excellent diagnostic performance of the nomogram (Figure 1F,1G).
For comparison, the ROC curves for each independent clinical variable were plotted, showing that the AUC for the nomogram was superior to that of any individual variable in both the training and validation cohorts (Figure 2A,2B).

Prognostic factors for LIMA patients with DM
In the 830 LIMA patients with DM, various clinical and treatment-related factors were analyzed to identify prognostic variables. The demographic and clinical characteristics are detailed in Table 3. The majority of patients were older adults (74.5%), with the white race being the most prevalent (83.3%). Most patients had right-sided tumors (60.5%). No significant differences were found between the training and validation cohorts based on Chi-squared tests.
Table 4 presents the results of Cox regression analyses for prognostic factors. These analyses revealed that age, sex, race, laterality, grade, tumor size, T-stage, N-stage, and treatment modalities (surgery, chemotherapy, and radiotherapy) were significantly associated with prognosis. However, only grade, tumor size, T-stage, and chemotherapy were identified as independent prognostic factors for survival in LIMA patients with DM.

Prognostic nomogram for LIMA patients with DM
Using the independent prognostic variables, a prognostic nomogram was developed to predict the survival probability of LIMA patients with DM (Figure 3). Calibration curves for 12-, 24-, and 36-month survival demonstrated good concordance between observed and predicted survival outcomes in the training cohort (Figure 4A-4C). DCA further confirmed the clinical utility and efficiency of this nomogram in the training cohort (Figure 4D-4F). Consistent results were observed in the validation cohort, with calibration curves (Figure 5A-5C) and DCA (Figure 5D-5F) both verifying the model’s robust performance.
Patients were stratified into low- and high-risk groups based on the median survival time in the training cohort. Kaplan-Meier (K-M) survival analysis showed that low-risk patients had significantly better OS than high-risk patients (Figure 6A,6B). The ROC curves for the nomogram in the training cohort for 12-, 24-, and 36-month survival yielded AUCs of 0.751, 0.754, and 0.784, respectively (Figure 6C), and in the validation cohort, the AUCs were 0.766, 0.783, and 0.781 (Figure 6D). These findings further confirm the strong discriminative ability of the nomogram for predicting the prognosis of LIMA patients with DM.
Additionally, a comparison of the prognostic nomogram with individual independent prognostic factors revealed that the nomogram provided greater predictive value for survival at 12, 24, and 36 months (Figure 7A-7F).

Prognostic comparison between DM and non-metastatic LIMA patients
Using the independent prognostic variables, a prognostic nomogram was developed to predict the survival probability of LIMA patients with non-metastatic (Figure 8). It can be observed that the presence of DM is a significant adverse prognostic factor for LIMA. The survival curve of the DM group was significantly lower than that of the non-metastatic group, and the difference was highly statistically significant (Figure 9).

Discussion
LIMA is a rare subtype of lung adenocarcinoma, characterized by atypical clinical manifestations. It is often misdiagnosed as tuberculosis or pneumonia in its early stages (20-22). LIMA is also highly aggressive and metastatic, with a strong propensity for DM in advanced stages, which significantly contributes to its poor prognosis (9,10,23). Therefore, identifying risk factors and prognostic indicators for DM in LIMA patients is critical for improving clinical management and outcomes.
Currently, while several studies have explored the molecular markers associated with DM in LIMA patients (14,24,25), most have focused on genetic and molecular biomarkers rather than clinical factors. For example, HNF4a, MUC6, CSK2, and GATA6 have been proposed as potential biomarkers linked to DM in LIMA (4,26-28). However, to our knowledge, no predictive models have been established for early detection of DM in LIMA based on these markers. This study represents the first large-scale, systematic effort to develop nomogram models that evaluate both the risk of DM and the prognosis of LIMA patients with DM. The findings from various validation methods demonstrate that these nomograms exhibit robust predictive ability, offering valuable tools for clinical risk stratification and prognosis prediction.
In our study, we found that the prevalence of DM among LIMA patients was 18.7%. Several significant risk factors for DM were identified, including primary tumor site, grade, T-stage, and N-stage. Notably, age was not found to be significantly associated with the presence of DM, which differs from some previous studies that reported an age-related risk (29). Our results also confirm the previously recognized association between the TNM stages and the occurrence of DM. Importantly, the diagnostic nomogram developed in this study demonstrated superior predictive power when compared to each individual clinical variable, emphasizing the value of a comprehensive, multi-variable predictive model. However, it is worth noting that due to the limitations of the SEER database, smoking status could not be included in our analysis, although it is an important factor that may influence the development of DM in LIMA.
The prognosis of LIMA patients, particularly those with DM, remains poorly understood, and survival data for this specific cohort are sparse (11,25,30,31). Discrepancies in survival outcomes across studies suggest that the prognosis of LIMA patients has yet to reach a clinical consensus. Our study fills this gap by developing a predictive nomogram for survival outcomes in LIMA patients with DM. The nomogram outperformed individual predictors, highlighting its potential to guide clinical decision-making and tailor treatment strategies. Our findings provide a novel approach for predicting patient prognosis and enhancing personalized care.
Surgical resection remains an important treatment option for LIMA patients, as it has been shown to extend survival (32-35). However, many LIMA patients are diagnosed with metastatic disease at an advanced stage and are not candidates for surgery. In these cases, chemotherapy is often the next treatment option. Our study reinforces the importance of surgery and chemotherapy in improving survival in LIMA patients with DM, in line with previous reports (10,29). Interestingly, radiotherapy did not significantly impact survival outcomes, suggesting that its role in treatment may be limited and warrants further investigation. Our study also confirmed that older age and male sex are associated with worse prognosis, which aligns with established findings in lung cancer research. Similarly, tumor size was a significant predictor of survival, consistent with previous studies in other lung cancer subtypes (34). Additionally, higher T-stage, N-stage, and tumor grade were all associated with worse survival outcomes, highlighting their importance in clinical prognosis assessment.
Despite the strengths of our study, there are several limitations that must be acknowledged. First, the relatively small sample size of LIMA patients with DM (n=830) may introduce selection bias, potentially limiting the generalizability of the results. Second, the majority of the cohort was composed of white patients, which may affect the external validity of our findings in other racial and ethnic groups. Further studies in more diverse populations are needed to confirm these results. Third, the nomogram models were developed using data from the SEER database, which is based on retrospective information. Prospective validation through clinical trials and real-world data is necessary to confirm their applicability in clinical practice. Finally, while immunotherapy and targeted therapies are promising treatment modalities for various cancers, their potential role in the treatment of LIMA was not explored in this study due to the lack of relevant data in the SEER database. Future research incorporating these therapies would be valuable.

Conclusions
In conclusion, this study identifies key clinical risk factors for DM in LIMA patients and prognostic factors for those with DM using logistic and Cox regression analyses. Based on these findings, we developed two nomograms that can assist clinicians in risk stratification and personalized treatment planning. These tools have the potential to improve clinical outcomes by guiding early detection and therapeutic decisions in LIMA patients with DM.

Supplementary
The article’s supplementary files as