One-year mortality risk prediction model for patients with interstitial lung disease and lung cancer.

BACKGROUND

The association of interstitial lung disease (ILD) with lung cancer (LC) has generated increased research interest in recent years. We aimed to characterize the clinical features and prognostic factors of patients with ILD and LC and to develop a 1-year mortality risk prediction model for these patients.

METHODS

The retrospective study enrolled patients with ILD and LC admitted to Nanjing Drum Tower Hospital from 2017 to 2022. The demographic data, histological type and staging of LC, high-resolution computed tomography (HRCT) patterns of ILD, laboratory examinations, and therapeutic and follow-up information were collected. The primary endpoint for the prediction model was all-cause 1-year mortality. Logistic regression analysis was used to identify risk predictors and further establish a nomogram to predict 1-year mortality. Area under the curve (AUC), calibration curves, and decision curves were used to assess the utility of the nomogram.

RESULTS

A total of 206 patients with concurrent ILD and LC were included. Adenocarcinoma was the most common pathological subtype (94/206, 45.6%), followed by squamous cell carcinoma (55/206, 26.7%) and small-cell lung cancer (SCLC) (42/206, 20.4%). Moreover, 43.7% (90/206) of tumors were located inside ILD lesions. Among the patients with non-small cell lung cancer (NSCLC), 90 were diagnosed with advanced-stage disease (> stage IIIA) while 28 patients with SCLC were at the extensive phase. The most common HRCT pattern of ILD was usual interstitial pneumonia (UIP) (102/206, 49.5%). The all-cause 1-year mortality rate was 41.3%. The prediction model incorporated age, sex, neutrophil count, and lactate dehydrogenase (LDH) and albumin (Alb) levels. The AUC values in training and internal validation sets were 0.775 and 0.716 respectively. Calibration curves indicated strong consistency, and decision curves confirmed the clinical net benefit achievable at different risk thresholds.

CONCLUSIONS

We developed a 1-year mortality risk prediction model for patients with concurrent ILD and LC to identify those with high risk of death and facilitate precise management. Future multicenter studies are needed for further external validation.