Predictive model for sarcopenia in patients with non-small cell lung cancer and malignant pleural effusion

Abstract

Background: Sarcopenia in patients with non-small cell lung cancer (NSCLC) is often indicative of a more aggressive disease course and a poorer prognosis. Nevertheless, there have been limited studies that specifically examined clinical parameters to predict sarcopenia in individuals with malignant pleural effusion (MPE). Our objective is to investigate the potential correlations between commonly utilized clinical variables and reduced muscle mass in NSCLC patients who also have MPE.

Methods: This retrospective study examined the clinicopathological data and imaging characteristics of NSCLC patients admitted to the hospital with MPE. The Least Absolute Shrinkage and Selection Operator (LASSO) algorithm was employed to select the most appropriate variables for model creation, effectively reducing the chance of overfitting. Logistic regression analysis was conducted to pinpoint the independent factors predicting sarcopenia in NSCLC patients with MPE. Subsequently, a nomogram was formulated to estimate the sarcopenia risk for individual patient. The efficacy of this nomogram was assessed through various metrics, including the receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis (DCA).

Results: A total of 139 patients, with an average age of 66 years and a majority being male (56.8%), were included in this study. Multivariate logistic regression analysis revealed that age, body mass index (BMI), albumin (Alb), and cytokeratin-19-fragment (CY21-1) were all independent predictors of sarcopenia in NSCLC patients with MPE. A nomogram was developed to facilitate personalized prediction of sarcopenia for individual patient. The ROC curve demonstrated that the nomogram model incorporating these predictive factors achieved an area under the curve (AUC) of 0.889, indicating its discriminatory power in predicting sarcopenia. The calibration curve demonstrated a strong concordance between the actual and the anticipated sarcopenia risk. DCA further confirmed that the nomogram showed good clinical applicability and net benefits in sarcopenia prediction.

Conclusions: Certain commonly used clinical characteristics were found to be associated with decreased skeletal muscle mass. Specifically, age, BMI, Alb, and CY21-1 levels emerged as predictive indicators for sarcopenia among NSCLC patients with MPE. These indicators have the potential to serve as effective alternatives to traditional computed tomography (CT) evaluation in assessing sarcopenia.

Keywords: Malignant pleural effusion; Nomogram; Non-small cell lung cancer; Predictive indicators; Sarcopenia.

Fig. 1

A representative single-slice CT image of the third lumbar vertebrae level used to calculate the SMA. Specialized computer software was employed to measure the SMA of the manually outlined areas (indicated by the green line). SMA, skeletal muscle cross-sectional area

Fig. 2

Study flow diagram

Fig. 3

LASSO algorithm to discern potential predictors of sarcopenia. The left panel illustrates the variable selection phase in the context of the LASSO penalty. The horizontal axis denotes the log-transformed penalized parameter, lambda, while the vertical axis represents the coefficients of individual variable, which progressively diminish towards zero as lambda increases. Ultimately, variables retaining nonzero coefficients are chosen for subsequent analysis. The right panel displays the results of a 10-fold CV for the LASSO model. LASSO, Least Absolute Shrinkage and Selection Operator; CV, cross-validation

Fig. 4

Correlation analysis between selected indicators and LSMI in lung cancer patients with MPE. The results indicated a negative correlation between the LSMI and both age (A) and CY21-1 levels (D). In contrast, there existed a positive correlation between the LSMI and both BMI (B) and serum Alb levels (C). LSMI, lumbar skeletal muscle index; MPE, malignant pleural effusion; BMI, bodymass index; Alb, albumin; CY21-1, cytokeratin-19-fragment

Fig. 5

A nomogram developed for the diagnostic assessment of sarcopenia BMI, body mass index; Alb, albumin; CY21-1, cytokeratin-19-fragment

Fig. 6

ROC curve to illustrate the nomogram’s discriminative power BMI, body mass index; Alb, albumin; CY21-1, cytokeratin-19-fragment; ROC, receiver operating characteristic

Fig. 7

The calibration curve of the nomogram for predicting sarcopenia

Fig. 8

Decision curve analysis of the nomogram for predicting sarcopenia. BMI, body mass index; Alb, albumin; CY21-1, cytokeratin-19-fragment