Clinicopathological characteristics and prognosis of pulmonary spindle cell carcinoma: a population-based retrospective study using SEER data

Background: Pulmonary spindle cell carcinoma (PSCC) is a rare pulmonary sarcomatoid carcinoma subtype accounting for 0.17–0.4% of lung malignancies. Due to its rarity, its clinicopathological features and prognostic factors remain poorly characterized. This retrospective study evaluated the clinical characteristics and survival-associated factors of PSCC.

Methods: Patients with PSCC and other non-small cell lung cancers (NSCLCs) diagnosed between 2004 and 2014 were identified from the Surveillance, Epidemiology, and End Results (SEER) database. In response to disease-definition concerns, the primary reanalysis was restricted to International Classification of Diseases for Oncology, Third Edition (ICD-O-3) code 8032/3. Cancer-specific survival (CSS) and overall survival (OS) were analyzed using Kaplan-Meier and Cox proportional hazards regression models. Tumor-node-metastasis (TNM) stage and other clinicopathological variables were evaluated, and an internally derived CSS prognostic model was rebuilt.

Results: The strict PSCC cohort included 250 ICD-O-3 8032/3 patients and was compared with 220,703 other NSCLC patients. Strict PSCC patients had a median age of 72 years, 53.6% were male, 28.4% had stage III disease, and 47.6% had stage IV disease. Compared with other NSCLCs, strict PSCCs were larger and less differentiated. Median CSS was 5 months, with 1-, 3-, 5-, and 10-year CSS rates of 29.2%, 17.6%, 16.3%, and 14.7%, respectively. Median OS was 4 months, with 1-, 3-, 5-, and 10-year OS rates of 27.1%, 14.8%, 13.2%, and 8.5%, respectively. In multivariable CSS analysis, female sex, lower T/N/M stages, surgery, and chemotherapy were associated with longer CSS. The rebuilt CSS model had a C-index of 0.757.

Conclusions: Strictly defined PSCC exhibits aggressive clinicopathological features and poor CSS and OS. Treatment-related findings should be interpreted as associations rather than causal evidence of treatment benefit because treatment allocation was not randomized and key confounders were unavailable. The internally derived prognostic model may provide exploratory prognostic information but requires external validation.