ABSTRACT
Background
Pulmonary sarcomatoid carcinoma (PSC) is a rare and highly aggressive tumor. This study aimed to investigate the prognostic factors of PSC and develop a new risk classifier to predict patients' overall survival (OS).
Methods
This retrospective study consisted of 675 PSC patients registered in the Surveillance, Epidemiology, and End Results (SEER) database, and 90 patients from the First Affiliated Hospital of Xi'an Jiaotong University. The univariate and multivariate Cox regression analyses were adopted to select the best prognostic variables. A nomogram was developed to predict the OS of these patients. The nomogram was assessed using receiver operating characteristic (ROC) curves, the concordance index (C‐index), and calibration curves. Decision curve analysis (DCA) was adopted to assess its net clinical benefit. Additionally, a Lasso‐Cox regression model was utilized to explore prognostic factors in real‐world patients with PSC.
Results
The multivariate analysis revealed that sex, T, N, and M stages, surgery, radiation, chemotherapy, and liver metastasis were independent prognostic factors of OS for patients with PSC, and a nomogram was developed to predict the OS of these patients. The ROC curves and the estimated C‐index displayed that the nomogram yielded excellent performance in predicting OS. Calibration curves demonstrated satisfactory consistency between the actual and predicted OS. DCA illustrated that the nomogram would provide significant net clinical benefits. Consistent findings were discovered in the internal and external validation cohorts. Lasso‐Cox regression models revealed that surgery and immunotherapy were key protective factors for OS, while ECOG ≥ 2, underlying medical history, and distant metastasis constituted major risk factors for OS.
Conclusions
The constructed nomogram is a reliable risk classifier to achieve personalized survival probability prediction for patients with PSC. Real‐world data suggest that immunotherapy is associated with significantly prolonged survival in patients with PSC.
Keywords: antiangiogenic drugs, immune checkpoint inhibitors (ICIs), nomogram, overall survival (OS), prognostic factor, pulmonary sarcomatoid carcinoma (PSC)
1. Introduction
Pulmonary sarcomatoid carcinoma (PSC) is a rare subtype of non‐small cell lung cancer (NSCLC), which accounts for 0.1%–0.4% of all NSCLC cases [1, 2]. PSC tends to occur in older males with a smoking history [3]. PSC patients have poor outcomes; their median overall survival (OS) is approximately half that of the general NSCLC population, reflecting its highly aggressive nature and tendency for advanced presentation [2]. According to the latest definition of the World Health Organization (WHO), PSC consists of five histological subtypes—pleomorphic carcinoma, giant cell carcinoma, spindle cell carcinoma, pulmonary blastoma, and carcinosarcoma [4].
Regarding treatment approaches, surgical resection is a prognostic factor that can prolong survival time in patients with PSC. However, even after complete resection, patients have a high probability of recurrence and progression. Therefore, adjuvant therapy is important [5]. PSC is insensitive to conventional radiotherapy and chemotherapy [6]. Many cases illustrate the efficacy and safety of immunotherapy. Emerging studies indicate that both immunotherapy combined with chemotherapy and immunotherapy combined with anti‐angiogenic therapy demonstrate promising efficacy [7, 8]. A phase II clinical study showed that savolitinib exhibited encouraging efficacy and a tolerable safety profile in PSC patients with METex14 skipping mutations [9]. The efficacy of KRAS inhibitors or EGFR tyrosine kinase inhibitors in treating these unique NSCLC cases remains uncertain and warrants additional investigation [10].
Nomograms have been proven to be effective tools for identifying risk factors and predicting the prognosis of malignancies with good accuracy and convenience [5]. Herein, this study was performed to investigate the independent predictors of OS in PSC patients based on participants in the Surveillance, Epidemiology, and End Results (SEER) database, and patients from an oncology center in Northwestern China. Most importantly, we aimed to construct a reliable and personalized nomogram to predict the OS of these patients. Simultaneously, we explored the prognostic value of supplementary clinical characteristics and treatment methods in the external cohort.
2. Methods
2.1. Study Design and Participants
This was a retrospective study. Participants in this study included patients with PSC identified from the SEER database and the First Affiliated Hospital of Xi'an Jiaotong University. SEER data access was obtained via compliant registration (November 2024) under the terms of the applicable usage agreement. Patients diagnosed with PSC between 2001 and 2021 from the SEER database were included in this study. First, we included the following patients according to ICD‐O Morphology and Behavior Codes: 8022/3 (pleomorphic carcinoma), 8031/3 (giant cell carcinoma), 8032/3 (spindle cell carcinoma), 8072/3 (pulmonary blastoma), or 8980/3 (carcinosarcoma). In addition, the exclusion criteria included patients whose information on survival time, TNM stage, metastasis records, and therapeutic information was unknown; patients aged < 18 years; and patients for whom this was not their primary tumor. Finally, a total of 675 patients were enrolled from the SEER database. Additionally, 90 patients diagnosed with PSC and having complete medical records from the First Affiliated Hospital of Xi'an Jiaotong University were included in this study. This study was approved by the ethics committee of the First Affiliated Hospital of Xi'an Jiaotong University. The flow chart for patient selection, nomogram construction and validation is shown in Figure 1.
FIGURE 1.
Flow chart of the study. PSC, pulmonary sarcomatoid carcinoma; Mets, metastasis; C‐index, concordance index; ROC curve, receiver operating characteristic curve.
2.2. Cohort Establishment and Variable Selection
Patients in the SEER database were randomly divided into a training cohort and an internal validation cohort in a 7:3 ratio through the “createDataPartition” function in the R software. The training cohort was used to identify the independent predictors of OS and construct the nomogram to predict patients' survival probability. The internal validation cohort was utilized to validate its predictive ability. Additionally, ninety patients from our medical center served as an independent external validation cohort to validate the generalizability of the nomogram. Common covariates both in the SEER database and our XJTU cohort were collected, including age, sex, race, histologic type, laterality, TNM staging, distant metastases sites (bone, brain, liver, and lung), and treatment modalities (surgery, radiation, and chemotherapy). We referred to local infiltrates, lymph node involvement, and metastasis to determine the TNM staging of each patient. Age was transformed into a categorical variable by setting 65 years as the cut‐off value. Furthermore, within the XJTU cohort, this study collected data on smoking history, Body Mass Index (BMI), Eastern Cooperative Oncology Group Performance Status (ECOG‐PS), underlying medical history, baseline tumor assessments (PD‐L1 expression and genetic mutations), other laboratory examinations, and tumor treatment‐related information (treatment regimens and treatment responses). Operational definitions for certain clinical variables were also established, as illustrated in Figure 2. Underlying medical history was defined as a diagnosis of hypertension, diabetes, cardiovascular/cerebrovascular diseases, or chronic obstructive pulmonary disease. Liver dysfunction was identified by levels of ALT, AST, or total bilirubin exceeding the upper limit of normal, while kidney dysfunction was defined as an eGFR below 90 mL/min/1.73 m2. Then, the univariate and multivariate Cox regression analyses were performed to identify the independent prognostic variables of OS and thereby develop the nomogram.
FIGURE 2.
The operational definitions for the collected clinical variables were standardized and established, as illustrated.
2.3. Statistical Analysis
All categorical variables were presented as frequencies and percentages, and for continuous variables, normality (Shapiro–Wilk test) and homogeneity of variance (F‐test) were evaluated to determine the appropriate statistical methods. Continuous variables exhibiting a normal distribution and homogeneous variance were compared using the independent samples t‐test, whereas the Wilcoxon rank‐sum test was employed otherwise. Differences between categorical variables were assessed using the Chi‐square test or Fisher's exact test. In this study, continuous variables were transformed into categorical variables, with cutoff values determined based on clinical consensus or the maximum Youden index (for variables without established thresholds). Kaplan–Meier survival curves were generated to compare the survival difference between different groups, with a log‐rank test being adopted to test the statistical significance. The univariate and multivariate Cox regression analyses were applied to identify independent prognostic factors of OS for patients with PSC. Variables with a p value < 0.1 in the univariate Cox regression analysis were then incorporated into the multivariate Cox regression analysis. The hazard ratio (HR) and 95% confidence interval (CI) of each variable were estimated to investigate their associations with the OS of patients with PSC. The significant variables in the multivariable Cox regression analysis were selected for nomogram construction, using the “rms” and “regplot” packages in R software to visualize the nomogram. In addition, the concordance index (C‐index), receiver operating characteristic (ROC) curve, and calibration curve were used to assess the discrimination and calibration abilities of the nomogram in each cohort. Furthermore, decision curve analysis (DCA) was adopted to compare the net clinical benefits of the nomogram and the traditional TNM staging system when used to guide clinical decision‐making. In the external XJTU validation cohort, a 10‐fold cross‐validated Lasso‐Cox regression analysis was applied. Utilizing the least absolute shrinkage and selection operator (Lasso) alongside L1 regularization, the regression coefficients of irrelevant variables were shrunk to zero. This approach was employed to screen for prognostic factors in the XJTU cohort while effectively addressing the issue of multicollinearity associated with small sample sizes. A positive Lasso regression coefficient indicates that an increase in the variable leads to an elevated risk of a fatal outcome, designating the variable as a risk factor; conversely, a negative coefficient implies that an increase in the variable corresponds to a reduced risk of death, identifying it as a protective factor. The statistical difference was considered significant when p < 0.05. All tests were two‐sided. All statistical analyses and visualizations were performed using R software (version 4.4.2).
3. Results
3.1. Clinical Characteristics of the Participants
The mean age of patients in the entire SEER database was 68.0 years. The cohort consisted of 392 males and 283 females. Pleomorphic carcinoma was the most prevalent histologic type, accounting for 43.9% of cases. Overall, 47.6% of PSC patients were diagnosed with TNM stage IV. Regarding the detailed therapeutic regimens, 40.3% of patients received surgery, 45.8% received chemotherapy, and 37.2% received radiation. Additionally, we observed that the bone was the most common distant metastatic organ, accounting for 17.8% of patients, followed by the brain (13.0%) and lung (12.3%). Patients in the SEER database were randomly divided into the training and internal validation cohorts in a ratio of 7:3, with no significant differences in most baseline characteristics between these cohorts, except for brain metastasis (Table 1; all other p values > 0.05). In the XJTU cohort, a total of 90 patients with PSC were identified from January 2013 to April 2025. Most clinicopathological parameters were comparable between the SEER database and our XJTU cohort. Table 1 summarizes in detail the clinical characteristics of patients with PSC enrolled in this study.
TABLE 1.
Demographic and clinical characteristics of patients with PSC.
| Characteristics | Whole population (n = 675) | Training cohort (n = 472) | Internal validation cohort (n = 203) | P1 | XJTU Cohort (n = 90) | P2 | |
|---|---|---|---|---|---|---|---|
| Sex (n, %) | Female | 283 (41.9) | 198 (41.9) | 85 (41.9) | 1 | 13 (14.4) | < 0.001 |
| Male | 392 (58.1) | 274 (58.1) | 118 (58.1) | 77 (85.6) | |||
| Age (n, %) | < 65 years old | 249 (36.9) | 180 (38.1) | 69 (34.0) | 0.339 | 45 (50.0) | 0.046 |
| ≥ 65 years old | 426 (63.1) | 292 (61.9) | 134 (66.0) | 45 (50.0) | |||
| Race (n, %) | White | 537 (79.6) | 370 (78.4) | 167 (82.2) | 0.292 | 0 (0) | < 0.001 |
| Black | 84 (12.4) | 65 (13.8) | 19 (9.4) | 0 (0) | |||
| Others | 54 (8.0) | 37 (7.8) | 17 (8.4) | 90 (100) | |||
| Histologic type (n, %) | pleomorphic carcinoma | 296 (43.9) | 213 (45.1) | 83 (40.9) | 0.076 | 20 (22.2) | < 0.001 |
| giant cell carcinoma | 88 (13.0) | 67 (14.2) | 21 (10.3) | 0 (0) | |||
| spindle cell carcinoma | 180 (26.7) | 125 (26.5) | 55 (27.1) | 23 (25.6) | |||
| carcinosarcoma | 105 (15.6) | 64 (13.6) | 41 (20.2) | 30 (33.3) | |||
| pulmonary blastoma | 6 (0.8) | 3 (0.6) | 3 (1.5) | 2 (2.2) | |||
| Unknown | 0 (0) | 0 (0) | 0 (0) | 15 (16.7) | |||
| Laterality (n, %) | Bilateral | 2 (0.3) | 2 (0.4) | 0 (0) | 0.352 | 1 (1.1) | 0.462 |
| Left | 291 (43.1) | 210 (44.5) | 81 (39.9) | 42 (46.7) | |||
| Right | 382 (56.6) | 260 (55.1) | 122 (60.1) | 47 (52.2) | |||
| T stage (n, %) | T1 | 77 (11.4) | 53 (11.2) | 24 (11.8) | 0.812 | 9 (10.0) | 0.803 |
| T2 | 143 (21.2) | 101 (21.4) | 42 (20.7) | 23 (25.5) | |||
| T3 | 139 (20.6) | 93 (19.7) | 46 (22.7) | 15 (16.7) | |||
| T4 | 316 (46.8) | 225 (47.7) | 91 (44.8) | 43 (47.8) | |||
| N stage (n, %) | N0 | 332 (49.2) | 226 (47.9) | 106 (52.2) | 0.681 | 37 (41.2) | 0.012 |
| N1 | 80 (11.9) | 55 (11.7) | 25 (12.3) | 11 (12.2) | |||
| N2 | 202 (29.9) | 146 (30.9) | 56 (27.6) | 22 (24.4) | |||
| N3 | 61 (9.0) | 45 (9.5) | 16 (7.9) | 20 (22.2) | |||
| M stage (n, %) | M0 | 354 (52.4) | 254 (53.8) | 100 (49.3) | 0.313 | 44 (48.9) | 0.421 |
| M1 | 321 (47.6) | 218 (46.2) | 103 (50.7) | 46 (51.1) | |||
| TNM stage (n, %) | I | 95 (14.0) | 68 (14.4) | 27 (13.3) | 0.577 | 10 (11.1) | 0.632 |
| II‐III | 259 (38.4) | 186 (39.4) | 73 (36.0) | 34 (37.8) | |||
| IV | 321 (47.6) | 218 (46.2) | 103 (50.7) | 46 (51.1) | |||
| Surgery (n, %) | None | 403 (59.7) | 277 (58.7) | 126 (62.1) | 0.442 | 55 (61.1) | 0.726 |
| Yes | 272 (40.3) | 195 (41.3) | 77 (37.9) | 35 (38.9) | |||
| Radiation (n, %) | None | 424 (62.8) | 294 (62.3) | 130 (64.0) | 0.728 | 69 (76.7) | 0.008 |
| Yes | 251 (37.2) | 178 (37.7) | 73 (36.0) | 21 (23.3) | |||
| Chemotherapy (n, %) | None | 366 (54.2) | 244 (51.7) | 122 (60.1) | 0.053 | 29 (32.2) | < 0.001 |
| Yes | 309 (45.8) | 228 (48.3) | 81 (39.9) | 61 (67.8) | |||
| Bone metastasis (n, %) | None | 555 (82.2) | 393 (83.3) | 162 (79.8) | 0.323 | 73 (81.1) | 0.647 |
| Yes | 120 (17.8) | 79 (16.7) | 41 (20.2) | 17 (18.9) | |||
| Brain metastasis (n, %) | None | 587 (87.0) | 402 (85.2) | 185 (91.1) | 0.035 | 80 (88.9) | 0.413 |
| Yes | 88 (13.0) | 70 (14.8) | 18 (8.9) | 10 (11.1) | |||
| Liver metastasis (n, %) | None | 632 (93.6) | 446 (94.5) | 186 (91.6) | 0.171 | 88 (97.8) | 0.288 |
| Yes | 43 (6.4) | 26 (5.5) | 17 (8.4) | 2 (2.2) | |||
| Lung metastasis (n, %) | None | 592 (87.7) | 413 (87.5) | 179 (88.2) | 0.898 | 72 (80.0) | 0.066 |
| Yes | 83 (12.3) | 59 (12.5) | 24 (11.8) | 18 (20.0) | |||
Note: P1, p value between training cohort and internal validation cohort; P2, p value between training cohort and XJTU cohort.
Abbreviation: PSC, Pulmonary Sarcomatoid Carcinoma.
3.2. Univariate and Multivariate Cox Regression Analysis
We performed a univariate Cox regression analysis in the training cohort to identify potential prognostic factors of OS in patients with PSC. The results showed that age, sex, race, histologic type, laterality, T stage, N stage, M stage, TNM stage, bone metastasis, brain metastasis, liver metastasis, lung metastasis, surgery, radiation, and chemotherapy were associated with the OS of these patients (p‐value < 0.1). Next, we incorporated the above variables into a multivariate Cox regression analysis to minimize the impact of confounders. Furthermore, considering the correlation between the M stage and the TNM stage system, we only selected the former for the regression equation. Ultimately, the results indicated that sex, T stage, N stage, M stage, surgery, radiation, chemotherapy, and liver metastasis were the independent prognostic factors of OS for patients with PSC (Table 2). Among them, male sex (HR: 1.26, 95% CI: 1.00–1.59), T4 (HR: 2.54, 95% CI: 1.64–3.95), N2 (HR: 1.52, 95% CI: 1.13–2.03), N3 (HR: 2.24, 95% CI: 1.50–3.35), M1 (HR: 1.93, 95% CI: 1.35–2.76), and liver metastasis (HR: 1.87, 95% CI: 1.18–2.94) were significantly associated with poor OS in patients with PSC. Conversely, patients who received surgery (HR: 0.34, 95% CI: 0.24–0.50), radiation (HR: 0.74, 95% CI: 0.57–0.96), and chemotherapy (HR: 0.31, 95% CI: 0.24–0.41) had longer OS (Table 2).
TABLE 2.
Univariate and multivariate Cox analyses for the prediction of OS of patients with PSC.
| Characteristics | Univariate analysis | Multivariate analysis | |||||
|---|---|---|---|---|---|---|---|
| HR | 95% CI | p | HR | 95% CI | p | ||
| Sex | Female | Reference | Reference | ||||
| Male | 1.33 | 1.06–1.67 | 0.012 | 1.26 | 1.00–1.59 | 0.053 | |
| Age | 65 years old | Reference | Reference | ||||
| ≥ 65 years old | 1.25 | 1.00–1.58 | 0.052 | 1.13 | 0.87–1.47 | 0.343 | |
| Race | Black | Reference | Reference | ||||
| White | 0.71 | 0.52–0.97 | 0.030 | 0.91 | 0.65–1.27 | 0.568 | |
| Others | 0.83 | 0.52–1.34 | 0.454 | 1.03 | 0.62–1.70 | 0.910 | |
| Histologic type | Pleomorphic carcinoma | Reference | Reference | ||||
| Giant cell carcinoma | 2.27 | 1.66–3.11 | < 0.001 | 1.08 | 0.76–1.53 | 0.676 | |
| Spindle cell carcinoma | 1.98 | 1.51–2.59 | < 0.001 | 1.10 | 0.80–1.49 | 0.561 | |
| Carcinosarcoma | 1.23 | 0.87–1.74 | 0.233 | 1.15 | 0.80–1.66 | 0.437 | |
| Pulmonary blastoma | 0.82 | 0.20–3.34 | 0.787 | 0.47 | 0.11–2.03 | 0.313 | |
| Laterality | Bilateral | Reference | Reference | ||||
| Left | 0.10 | 0.02–0.40 | 0.001 | 0.32 | 0.07–1.37 | 0.125 | |
| Right | 0.09 | 0.02–0.37 | 0.001 | 0.31 | 0.07–1.34 | 0.117 | |
| T stage | T1 | Reference | Reference | ||||
| T2 | 1.00 | 0.63–1.60 | 0.997 | 1.04 | 0.64–1.68 | 0.884 | |
| T3 | 1.59 | 1.01–2.50 | 0.045 | 1.40 | 0.87–2.28 | 0.169 | |
| T4 | 2.57 | 1.71–3.86 | < 0.001 | 2.54 | 1.64–3.95 | < 0.001 | |
| N stage | N0 | Reference | Reference | ||||
| N1 | 1.31 | 0.91–1.88 | 0.140 | 1.29 | 0.89–1.88 | 0.178 | |
| N2 | 2.27 | 1.76–2.93 | < 0.001 | 1.52 | 1.13–2.03 | 0.005 | |
| N3 | 3.90 | 2.74–5.57 | < 0.001 | 2.24 | 1.50–3.35 | < 0.001 | |
| M stage | M0 | Reference | Reference | ||||
| M1 | 3.79 | 3.00–4.78 | < 0.001 | 1.93 | 1.35–2.76 | < 0.001 | |
| TNM stage | I | Reference | — | — | — | ||
| II‐III | 2.91 | 1.79–4.73 | < 0.001 | — | — | — | |
| IV | 8.72 | 5.40–14.08 | < 0.001 | — | — | — | |
| Surgery | None | Reference | Reference | ||||
| Yes | 0.22 | 0.17–0.29 | < 0.001 | 0.34 | 0.24–0.50 | < 0.001 | |
| Radiation | None | Reference | Reference | ||||
| Yes | 1.22 | 0.98–1.53 | 0.07 | 0.74 | 0.57–0.96 | 0.026 | |
| Chemotherapy | None | Reference | Reference | ||||
| Yes | 0.56 | 0.45–0.69 | < 0.001 | 0.31 | 0.24–0.41 | < 0.001 | |
| Bone metastasis | None | Reference | Reference | ||||
| Yes | 2.67 | 2.04–3.51 | < 0.001 | 0.88 | 0.64–1.21 | 0.441 | |
| Brain metastasis | None | Reference | Reference | ||||
| Yes | 2.28 | 1.71–3.03 | < 0.001 | 1.32 | 0.93–1.87 | 0.124 | |
| Liver metastasis | None | Reference | Reference | ||||
| Yes | 3.42 | 2.28–5.15 | < 0.001 | 1.87 | 1.18–2.94 | 0.007 | |
| Lung metastasis | None | Reference | Reference | ||||
| Yes | 2.49 | 1.85–3.36 | < 0.001 | 0.81 | 0.56–1.16 | 0.242 | |
Abbreviations: CI, confidence interval; HR, hazard ratio; OS, overall survival; PSC, Pulmonary Sarcomatoid Carcinoma.
3.3. Nomogram Development and Validation
Subsequently, we developed the nomogram based on the eight independent prognostic factors of OS in patients with PSC to achieve personalized survival probability prediction. As illustrated in Figure 3, clinicians could easily predict the 6‐, 12‐, and 24‐month OS probability of each patient according to the constructed nomogram. The ROC curves revealed that the nomogram had a remarkable discrimination ability in predicting the 6‐month (AUC: 0.898), 12‐month (AUC: 0.860), and 24‐month (AUC: 0.854) OS of PSC in the training cohort (Figure 4A). Similar results were also observed in the internal and external validation cohorts (Figure 4B,C). Additionally, the estimated C‐index demonstrated that the constructed nomogram had excellent discrimination power in predicting the OS of these patients compared with the classic TNM staging system in the training cohort (Figure 4D). Consistent findings were noted in the internal and external validation cohorts (Figure 4E,F). Furthermore, we also generated calibration curves to evaluate the calibration ability of the nomogram, which demonstrated high consistency between the actual and predicted 12‐month OS in all the cohorts (Figure 4G–I). To conclude, the constructed nomogram had a promising performance in predicting the survival probability of patients with PSC.
FIGURE 3.
Nomogram constructed to predict the 6‐, 12‐, and 24‐month OS of patients with PSC according to the eight independent prognostic factors identified in the multivariate Cox regression analysis. As an example, the red dots represent a hypothetical male patient with T2N0M0 disease and no liver metastasis, who underwent surgery but did not receive postoperative chemotherapy or radiation. According to the nomogram, the total points for this patient are 401, and the estimated probabilities of the OS time being less than 6, 12, and 24 months are 19.2%, 32.1%, and 42.1%, respectively.
FIGURE 4.
Assessment of the discrimination, calibration, and risk stratification abilities of the constructed nomogram. (A–C) Receiver operating characteristic (ROC) curves for predicting the 6‐, 12‐, and 24‐month OS of patients with PSC in the training cohort (n = 472) (A), internal validation cohort (n = 203) (B), and XJTU cohort (n = 90) (C). (D–F) Comparison of the C‐index between the constructed nomogram (Model A) and the classic TNM staging system (Model B) for predicting the 6‐, 12‐, and 24‐month OS in the training cohort (n = 472) (D), internal validation cohort (n = 203) (E), and XJTU cohort (n = 90) (F). (G–I) Calibration curves for predicting the 12‐month OS of patients in the training cohort (n = 472) (G), internal validation cohort (n = 203) (H), and XJTU cohort (n = 90) (I). (J–L) Kaplan–Meier survival curves displaying the risk stratification ability of the nomogram in the training cohort (n = 472) (J), internal validation cohort (n = 203) (K), and XJTU cohort (n = 90) (L). Log‐rank tests were used to calculate the p‐values to assess the statistical significance of survival differences between the high‐risk and low‐risk groups. Abbreviations: ROC, receiver operating characteristic; C‐index, concordance index; PSC, pulmonary sarcomatoid carcinoma; OS, overall survival.
3.4. Risk Stratification Ability Assessment of the Nomogram
Ultimately, all patients with PSC were divided into low‐risk and high‐risk groups based on the median of the total points derived from the nomogram to assess the risk stratification ability of the nomogram in each cohort. Kaplan–Meier survival curves demonstrated that the OS of patients with PSC in the high‐risk group was significantly worse than that of patients in the low‐risk group in the training cohort (Figure 4J), internal validation cohort (Figure 4K), and our XJTU cohort (Figure 4L). These results supported that the constructed nomogram had excellent risk stratification ability.
3.5. Clinical Utility Evaluation of the Nomogram
We applied DCA to compare the net clinical benefit for patients with PSC when the nomogram and the TNM staging system were adopted to guide clinical practice. We observed that the nomogram could provide greater net clinical benefit for these patients compared with the TNM staging system in predicting the 6‐month, 12‐month, and 24‐month OS at specific risk thresholds in the training cohort (Figure 5A–C), internal validation cohort (Figure 5D–F), and external validation cohort (Figure 5G–I).
FIGURE 5.
Decision curve analysis comparing the net clinical benefit of the constructed nomogram versus the classic TNM staging system. The net clinical benefits for predicting 6‐, 12‐, and 24‐month OS are displayed for the training cohort (n = 472) (A–C), the internal validation cohort (n = 203) (D–F), and the external XJTU cohort (n = 90) (G–I). Abbreviations: OS, overall survival.
3.6. Real‐World Clinical Characteristics and Prognostic Factors in Patients With PSC
Considering potential omissions of clinically relevant variables in the SEER database, we systematically supplemented the clinical data of PSC patients from the First Affiliated Hospital of Xi'an Jiaotong University (XJTU). This included patient demographics, relevant laboratory parameters, and treatment‐related information to establish a research dataset better aligned with real‐world clinical settings. As demonstrated in Table 3, the XJTU cohort (n = 90) predominantly exhibited a favorable performance status (ECOG score 0–1; n = 76, 84.4%), adequate nutritional status (n = 56, 62.2%), and an absence of underlying diseases (n = 55, 61.1%). Moreover, 31.1% (n = 28) of the patients had a non‐smoking history. Treatment modalities included chemotherapy (n = 61, 67.8%), surgical intervention (n = 35, 38.9%), radiotherapy (n = 21, 23.3%), immunotherapy (n = 21, 23.3%), and targeted therapy (n = 19, 21.1%). Additionally, all patients were stratified according to the latest AJCC 9th edition staging criteria for N and M classifications. Given the limited sample size of the XJTU cohort, Lasso‐Cox regression analysis was employed to mitigate multicollinearity effects. One patient with an overall survival of 0 months was excluded to minimize potential bias. The optimal coefficients for each prognostic factor were determined by minimizing the partial likelihood deviance, with coefficient variation curves generated across logarithmic λ‐value sequences (Figure 6A). Through 10‐fold cross‐validated Lasso regression, surgical intervention and immunotherapy were identified as independent protective factors for OS, while an ECOG score ≥ 2, underlying medical history and distant metastasis constituted independent risk factors for OS (Figure 6B). We constructed the model based on these five variables. To validate the clinical relevance of variable selection, Kaplan–Meier survival curves and risk score distributions were plotted post‐risk stratification. The Kaplan–Meier survival curves demonstrated that the OS of patients with PSC in the high‐risk group was significantly worse than that of patients in the low‐risk group (Figure 6C). In Figure 6D, we present the integrated visualization comprising the risk score scatter plot, survival time scatter plot, and expression heatmap. These results collectively indicate the good predictive ability of our model for risk assessment. Specifically, our findings demonstrate prolonged survival in patients receiving surgery or immunotherapy, whereas those with an ECOG score ≥ 2, underlying medical history and M‐stage metastasis exhibited inferior prognosis.
TABLE 3.
Demographic and clinical characteristics of patients with PSC from the First Affiliated Hospital of Xi'an Jiaotong University (XJTU).
| Characteristics | XJTU Cohort (n = 90) | Characteristics | XJTU Cohort (n = 90) | ||
|---|---|---|---|---|---|
| ECOG (n, %) | 0–1 | 76 (84.4) | Targeted therapy (n, %) | None | 71 (78.9) |
| ≥ 2 | 14 (15.6) | Yes | 19 (21.1) | ||
| BMI (n, %) | < 18.5 | 5 (5.6) | Surgery (n, %) | None | 55 (61.1) |
| 18.5–23.9 | 56 (62.2) | Yes | 35 (38.9) | ||
| > 24 | 29 (32.2) | Chemotherapy (n, %) | None | 29 (32.2) | |
| Underlying medical history (n, %) | None | 55 (61.1) | Yes | 61 (67.8) | |
| Yes | 35 (38.9) | N stage (n, %) | N0 | 37 (41.1) | |
| Smoking history (n, %) | None | 28 (31.1) | N1 | 11 (12.2) | |
| Yes | 62 (68.9) | N2a | 6 (6.7) | ||
| Liver dysfunction (n, %) | None | 53 (58.9) | N2b | 16 (17.8) | |
| Yes | 37 (41.1) | N3 | 20 (22.2) | ||
| Kidney dysfunction (n, %) | None | 78 (86.7) | M stage (n, %) | M0 | 44 (49.0) |
| Yes | 12 (13.3) | M1a | 12 (13.3) | ||
| Radiation (n, %) | None | 69 (76.7) | M1b | 11 (12.2) | |
| Yes | 21 (23.3) | M1c1 | 11 (12.2) | ||
| Immunotherapy (n, %) | None | 69 (76.7) | M1c2 | 12 (13.3) | |
| Yes | 21 (23.3) | ||||
FIGURE 6.
Predictor selection and validation using a Lasso regression model. (A) Coefficient path plot of the Lasso regression model. Each line represents the coefficient of a predictor variable, starting from the full model (left) and shrinking to zero as log(λ) increases (right). (B) Ten‐fold cross‐validation plot for optimal λ selection. The left vertical dotted line indicates λmin, and the right vertical dotted line shows λ1 − SE. The predictors in this study were selected using the λmin criterion. (C) Kaplan–Meier survival curves demonstrating the risk stratification ability of the Lasso‐Cox model. (D) Integrated visualization comprising the risk score scatter plot (top), survival time and status scatter plot (middle), and a clinical feature heatmap (bottom) to indicate the predictive ability of the LASSO‐Cox model for risk assessment. Variable encodings in the heatmap are as follows: ECOG (0: 0–1; 1: ≥ 2); Underlying medical history (0: None; 1: Yes); Immunotherapy (0: None; 1: Yes); Surgery (0: None; 1: Yes); and M stage (0: M0; 1: M1a; 2: M1b; 3: M1c1; 4: M1c2). Abbreviations: Lasso, least absolute shrinkage and selection operator; SE, standard error; ECOG, Eastern Cooperative Oncology Group.
A real‐world study previously demonstrated improved survival in patients with advanced pulmonary sarcomatoid carcinoma treated with immunotherapy, with or without chemotherapy, suggesting it as the optimal first‐line treatment currently available [11]. Accordingly, we collected data on PD‐L1 expression and the treatment patterns of patients receiving immune checkpoint inhibitors (ICIs). Furthermore, we compared the survival of patients treated with ICIs versus those who did not receive them. Tumor cell PD‐L1 expression was assessed in 35 patients: 17 (48.6%) patients had a PD‐L1 TPS of 50% or higher, and 4 (11.4%) patients had a TPS of less than 1%. A subgroup analysis was performed specifically for patients undergoing immunotherapy. Among the 21 patients who received ICIs, pembrolizumab (33.3%) and sintilimab (28.6%) were the most commonly used agents. Most of these patients (85.7%) received ICI‐based therapy as their first‐line treatment. Regarding combination strategies, 15 patients (71.4%) received ICIs combined with chemotherapy. Detailed information on immunotherapy for PSC is summarized in Tables 3 and 4. Subsequently, we analyzed the difference in OS between the subgroups with and without ICI‐based therapy. The results showed that ICI‐based therapy significantly prolonged OS compared to the group without such treatment (Figure 7A). Similarly, we observed consistent findings in the subgroup of patients with advanced‐stage disease (unresectable stage III and stage IV) (Figure 7B).
TABLE 4.
Characteristics of immunotherapy.
| Characteristics | N (%) |
|---|---|
| Line of ICIs therapy | |
| 1 | 18 (85.7) |
| 2 | 1 (4.8) |
| ≥ 3 | 2 (9.5) |
| Type of ICIs | |
| Pembrolizumab | 7 (33.3) |
| Sintilimab | 6 (28.6) |
| Tislelizumab | 3 (14.3) |
| Toripalimab | 2 (9.5) |
| Sugemalimab | 2 (9.5) |
| Camrelizumab | 1 (4.8) |
| Strategies of ICIs | |
| ICI alone | 1 (4.8) |
| ICI + Chemotherapy | 15 (71.4) |
| ICI + targeted therapy | 2 (9.5) |
| ICI + Chemotherapy + targeted therapy | 3 (14.3) |
FIGURE 7.
Kaplan–Meier survival analysis stratified by immunotherapy. (A) Comparison of OS between patients receiving immunotherapy (n = 21) and those not receiving immunotherapy (n = 68). (B) Comparison of OS between patients with advanced pulmonary sarcomatoid carcinoma receiving immunotherapy (n = 12) and those not receiving immunotherapy (n = 40). Advanced disease was defined as unresectable stage III and stage IV. Abbreviations: OS, overall survival.
4. Discussion
Given the low incidence and high clinical and molecular heterogeneity of Pulmonary Sarcomatoid Carcinoma (PSC), previous research has largely been constrained by small‐sample case reports or retrospective analyses. While public databases like SEER enable large‐scale exploration, they often lack granular clinical details such as specific treatment regimens, immunotherapy applications, and comprehensive performance status. This data dimensionality gap hinders the development of precise, multidimensional prognostic tools. To bridge this gap, our study integrated 675 patients from the SEER database with a well‐characterized real‐world cohort of 90 patients from the XJTU center. We constructed a robust nomogram based on the SEER cohort and, crucially, employed a Lasso‐Cox regression algorithm in the XJTU cohort to identify modern prognostic factors.
The multivariate Cox regression analysis identified that sex, T stage, N stage, M stage, surgery, radiation, chemotherapy, and liver metastasis were independent predictors of OS in patients with PSC. This is in general agreement with previously published studies [5, 12]. Furthermore, a core innovation of this study is the deep dimensional extension performed in the XJTU cohort using the Lasso‐Cox regression algorithm. By integrating variables missing in public databases—such as ECOG status, the presence of underlying medical history, and smoking history—we identified that poor ECOG scores, underlying medical history, and distant metastasis significantly increase mortality risk, while surgery and immunotherapy serve as key protective factors.
While the SEER‐based nomogram focuses on anatomical and basic treatment variables, the XJTU cohort reveals that a poor ECOG score (≥ 2) and the underlying medical history significantly increase mortality risk. This finding is highly relevant to the typical PSC “patient profile”—elderly males with extensive smoking histories who often present with cardiovascular or respiratory conditions like COPD.
Regarding demographic factors, although the p‐value for gender (p = 0.053) narrowly missed statistical significance in our multivariate analysis, we retained it due to its strong clinical relevance. Male gender is historically associated with poorer clinical outcomes in PSC [3, 13]. Interestingly, while smoking drives genetic instability and higher mutation burdens (such as TP53 alterations) associated with poor prognosis in NSCLC [14, 15, 16, 17], smoking status did not demonstrate a significant adverse effect in our LASSO‐Cox analysis. This may be attributed to the fact that smoking history was only recorded as a dichotomous variable (yes/no) in this cohort, without incorporating a smoking index for quantitative assessment, thereby failing to capture the dose–response relationship between smoking and prognosis. Furthermore, studies highlight that women with advanced lung cancer may derive a significantly larger benefit from chemo‐immunotherapy combinations than men [18]. This phenomenon could relate to sex hormones, as studies highlight androgen‐mediated promotion of CD8+ T cell dysfunction in cancer [19]. These may explain why males have a poorer prognosis. Research on sex differences in PSC patients is currently limited, and the influence of sex on prognostic outcomes remains to be further elucidated.
In terms of tumor biological aggressiveness, our study confirms that distant metastasis—particularly liver metastasis—is a potent independent risk factor. Since liver metastasis typically indicates disease progression to stage IV, it often portends an extremely poor survival prognosis for these patients. However, our multivariate regression analysis did not reveal any statistical significance for lung, brain, or bone metastases. A review of the relevant literature indicates a correlation between liver metastasis and the efficacy of ICIs; the prevailing view suggests that the unique immune‐tolerant microenvironment of the liver suppresses systemic immune responses, leading to a poor prognosis [7]. Concurrently, other studies have also confirmed that liver metastasis is an independent adverse prognostic factor for shortened progression‐free survival (PFS) [8]. These organ‐specific prognostic differences suggest that for PSC patients with liver metastasis, immunotherapy as a monotherapy may struggle to effectively overcome the disease's highly malignant biological behavior, highlighting the need to explore alternative combination treatment strategies in the future. Furthermore, our Lasso‐Cox analysis found distant metastasis (M stage) to be a far more decisive risk factor than regional lymph node involvement (N stage). This reflects the profound aggressiveness of PSC, where early hematogenous dissemination often eclipses the impact of local nodal spread. Consequently, while the 8th and newly published 9th editions of the TNM staging system provide robust baseline stratifications [5], relying solely on anatomical staging is insufficient. Future staging refinements for PSC should prioritize the extent of metastatic spread (e.g., single vs. multiple organ systems) to optimize clinical decision‐making.
Crucially, our analysis highlights a significant evolution in treatment efficacy. In the historical SEER cohort (spanning 2001–2021), conventional chemotherapy and radiation emerged as independent protective factors. Several studies have shown that neoadjuvant or adjuvant chemotherapy can provide significant survival benefits to postoperative patients with PSC [20, 21]. However, in the modern XJTU cohort, these traditional modalities were not selected by the Lasso‐Cox model, reflecting the high primary resistance of PSC to standard cytotoxic therapies [20, 22]. Instead, surgery (for early disease) [23, 24] and immunotherapy emerged as key protective factors. Since the late 2010s, the recognition of high PD‐L1 expression [25, 26] and high tumor mutation burden (TMB) [27] in PSC has paved the way for immune checkpoint inhibitors (ICIs) [6]. Our real‐world data confirms that ICI‐based therapy significantly prolongs OS, even in advanced or metastatic stages [7, 28]. Among our patients, combination strategies—such as ICIs with chemotherapy [29, 30] or anti‐angiogenic drugs like Anlotinib [31, 32, 33] —demonstrated robust efficacy. Emerging evidence also suggests that neoadjuvant immunotherapy combined with chemotherapy can induce pathologic complete responses in potentially resectable PSC, representing a promising frontier [34, 35]. Given the high heterogeneity of PSC, future models must integrate biomarkers like PD‐L1, TMB, and TILs to pinpoint populations that will derive the maximum benefit from these precision therapies [36, 37]. Currently, targeted therapies and immunotherapies are increasingly used in the field of PSC, which heralds the beginning of the precision therapy model for treating PSC; however, larger prospective studies are warranted to further validate these findings in PSC patients.
The clinical utility of our integrated approach is demonstrated by the nomogram's superior risk stratification and net clinical benefit compared to the traditional TNM system. By identifying high‐risk individuals through a combination of anatomical, physiological, and therapeutic indicators, clinicians can tailor follow‐up intensities—implementing more frequent radiological monitoring for high‐risk patients while avoiding over‐treatment in low‐risk groups.
Despite the advantages of our study, inevitable limitations remain. First, selection and informative biases are inherent to the retrospective design. Second, while the SEER database provided a large sample size, it lacks granular data on detailed chemotherapy regimens, radiation dosages, and specific surgical types. Third, although targeted therapies and immunotherapies provide significant survival benefits, their impact could not be fully explored within the SEER dataset. Finally, the XJTU cohort has a limited sample size, and confounders such as PD‐L1 expression could not be entirely excluded. Well‐designed, multicenter, prospective studies are urgently needed to validate these findings and deepen our understanding of this field.
5. Conclusion
We identified key independent predictors of OS in PSC patients and constructed a nomogram that effectively predicts survival probabilities, offering significant net clinical benefits. While the SEER cohort highlighted traditional prognostic factors, our real‐world XJTU cohort established surgery, immunotherapy, ECOG status, underlying medical history, and distant metastasis as modern prognostic determinants. Large‐scale prospective studies remain essential to validate our findings and refine precision treatment models for PSC.
Author Contributions
Xiaoyu Zhang: software, visualization. Yimeng Li: methodology, conceptualization, software. Huan Gao: visualization. Zhiping Ruan: writing – review and editing. Xuan Liang: writing – review and editing. Shuwen Cui: conceptualization, methodology, software, data curation, validation, investigation, visualization, writing – review and editing, writing – original draft. Yuanyuan Fan: methodology, investigation. Xiao Fu: conceptualization, writing – review and editing. Kaidiriye Setiwalidi: validation, visualization. Xiaoqiang Zheng: writing – review and editing. Chunli Li: writing – review and editing. Yingming Zhang: data curation, software. Tao Tian: conceptualization, writing – review and editing. Yu Yao: conceptualization, writing – review and editing.
Funding
This work was supported by Shaanxi Administration of Traditional Chinese Medicine No (SZY‐KJCYC‐2025‐ZY‐018). The Foundations of the First Affiliated Hospital of Xi'an Jiaotong University Nos (2024‐QN‐43), Nos (2024‐MS‐5).
Ethics Statement
This study was approved by the ethics committee of the First Affiliated Hospital of Xi'an Jiaotong University (XJTU1AF2025LSYY‐420). We conducted this study following the requirements of the Declaration of Helsinki. This study is solely a retrospective investigation that does not involve the collection of biological specimens, does not interfere with patient treatment, does not impact patients' personal interests, and will not cause any adverse reactions. Therefore, the requirement for written informed consent was waived by the ethics committee.
Conflicts of Interest
The authors declare no conflicts of interest.
Acknowledgements
We thank the Surveillance, Epidemiology, and End Results (SEER) database and Biobank of the First Affiliated Hospital of Xi'an Jiaotong University for providing clinical data.
Contributor Information
Yu Yao, Email: 13572101611@163.com.
Xiao Fu, Email: fuxiao@xjtufh.edu.cn.
Data Availability Statement
The data that support the findings of this study are available from National Cancer Institute. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from the author(s) with the permission of National Cancer Institute.
References
- 1. Yendamuri S., Caty L., Pine M., et al., “Outcomes of Sarcomatoid Carcinoma of the Lung: A Surveillance, Epidemiology, and End Results Database Analysis,” Surgery 152, no. 3 (2012): 397–402, 10.1016/j.surg.2012.05.007. [DOI] [PubMed] [Google Scholar]
- 2. Steuer C. E., Behera M., Liu Y., et al., “Pulmonary Sarcomatoid Carcinoma: An Analysis of the National Cancer Data Base,” Clinical Lung Cancer 18, no. 3 (2017): 286–292, 10.1016/j.cllc.2016.11.016. [DOI] [PubMed] [Google Scholar]
- 3. Wei Y., Wang L., Jin Z., et al., “Biological Characteristics and Clinical Treatment of Pulmonary Sarcomatoid Carcinoma: A Narrative Review,” Translational Lung Cancer Research 13, no. 3 (2024): 635–653, 10.21037/tlcr-24-127. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Wang T., Wang Z., Zhou J., et al., “Effect of Histological Subtype on the Efficacy of Perioperative Chemotherapy in Pulmonary Sarcomatoid Carcinoma: A Retrospective Study Based on SEER Population,” Translational Lung Cancer Research 13, no. 4 (2024): 749–762, 10.21037/tlcr-24-41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Yi X., Zhu B., Zhang J., Tang G., Luo H., and Zhou X., “Novel Model for Cancer‐Specific Survival of Patients With Pulmonary Sarcomatoid Carcinoma: A Population‐Based Analysis and External Validation,” Asian Journal of Surgery 47, no. 1 (2024): 184–194, 10.1016/j.asjsur.2023.07.002. [DOI] [PubMed] [Google Scholar]
- 6. Jiao Y., Liu M., Luo N., Guo H., and Li J., “Successful Treatment of Advanced Pulmonary Sarcomatoid Carcinoma With the PD‐1 Inhibitor Toripalimab: A Case Report,” Oral Oncology 112 (2021): 104992, 10.1016/j.oraloncology.2020.104992. [DOI] [PubMed] [Google Scholar]
- 7. Hazama D., Nakahama K., Kodama H., et al., “Effectiveness and Safety of Immune Checkpoint Inhibitors Alone or in Combination With Chemotherapy in Pulmonary Sarcomatoid Carcinoma,” JTO Clinical and Research Reports 5, no. 1 (2024): 100613, 10.1016/j.jtocrr.2023.100613. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Xie M., Chu T., Dong X., et al., “Heterogeneity in Advanced Pulmonary Sarcomatoid Carcinoma and Its Efficacy to Immune Checkpoint Inhibitors,” European Journal of Cancer 209 (2024): 114260, 10.1016/j.ejca.2024.114260. [DOI] [PubMed] [Google Scholar]
- 9. Lu S., Fang J., Li X., et al., “Once‐Daily Savolitinib in Chinese Patients With Pulmonary Sarcomatoid Carcinomas and Other Non‐Small‐Cell Lung Cancers Harbouring MET Exon 14 Skipping Alterations: A Multicentre, Single‐Arm, Open‐Label, Phase 2 Study,” Lancet Respiratory Medicine 9, no. 10 (2021): 1154–1164, 10.1016/s2213-2600(21)00084-9. [DOI] [PubMed] [Google Scholar]
- 10. Du Z., Qin Y., Lv Y., et al., “Clinical Characteristics and Survival Outcomes in Patients With Pulmonary Sarcomatoid Carcinoma: A Multicenter Retrospective Study,” Clinical & Translational Oncology 27, no. 7 (2025): 2993–3000, 10.1007/s12094-024-03823-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Amrane K., Cabarrou B., Justeau G., et al., “Efficacy of First‐Line Treatments for Advanced Pulmonary Sarcomatoid Carcinomas: A Real‐World Analysis,” ESMO Open 10, no. 8 (2025): 105343, 10.1016/j.esmoop.2025.105343. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Liang X., Cheng Y., Yuan Z., et al., “Clinical, Pathological and Treatment Factors Associated With the Survival of Patients With Pulmonary Sarcomatoid Carcinoma,” Oncology Letters 19, no. 6 (2020): 4031–4039, 10.3892/ol.2020.11472. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Ullah A., Ahmed A., Yasinzai A. Q. K., et al., “Demographics and Clinicopathologic Profile of Pulmonary Sarcomatoid Carcinoma With Survival Analysis and Genomic Landscape,” Cancers (Basel) 15, no. 9 (2023): 2469, 10.3390/cancers15092469. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Jiang X., Liu Y., Chen C., et al., “The Value of Biomarkers in Patients With Sarcomatoid Carcinoma of the Lung: Molecular Analysis of 33 Cases,” Clinical Lung Cancer 13, no. 4 (2012): 288–296, 10.1016/j.cllc.2011.11.004. [DOI] [PubMed] [Google Scholar]
- 15. Xu S., Wang Y., Ren F., et al., “Impact of Genetic Alterations on Outcomes of Patients With Stage I Nonsmall Cell Lung Cancer: An Analysis of the Cancer Genome Atlas Data,” Cancer Medicine 9, no. 20 (2020): 7686–7694, 10.1002/cam4.3403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Zhang Z., Xue J., Yang Y., et al., “Influence of TP53 Mutation on Efficacy and Survival in Advanced EGFR‐Mutant Non‐Small Cell Lung Cancer Patients Treated With Third‐Generation EGFR Tyrosine Kinase Inhibitors,” MedComm 5, no. 6 (2024): e586, 10.1002/mco2.586. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Vieira T., Antoine M., Ruppert A. M., et al., “Blood Vessel Invasion Is a Major Feature and a Factor of Poor Prognosis in Sarcomatoid Carcinoma of the Lung,” Lung Cancer 85, no. 2 (2014): 276–281, 10.1016/j.lungcan.2014.06.004. [DOI] [PubMed] [Google Scholar]
- 18. Conforti F., Pala L., Bagnardi V., et al., “Sex‐Based Heterogeneity in Response to Lung Cancer Immunotherapy: A Systematic Review and Meta‐Analysis,” Journal of the National Cancer Institute 111, no. 8 (2019): 772–781, 10.1093/jnci/djz094. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Kwon H., Schafer J. M., Song N. J., et al., “Androgen Conspires With the CD8+ T Cell Exhaustion Program and Contributes to Sex Bias in Cancer,” Science Immunology 7, no. 73 (2022): eabq2630, 10.1126/sciimmunol.abq2630. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Abdallah H. M., Martinez‐Meehan D., Lutfi W., et al., “Adjuvant Chemotherapy for Pulmonary Sarcomatoid Carcinoma: A Retrospective Analysis of the National Cancer Database,” Journal of Thoracic and Cardiovascular Surgery 163, no. 5 (2022): 1669–1681, 10.1016/j.jtcvs.2021.01.081. [DOI] [PubMed] [Google Scholar]
- 21. Sun L., Dai J., Chen Y., et al., “Pulmonary Sarcomatoid Carcinoma: Experience From SEER Database and Shanghai Pulmonary Hospital,” Annals of Thoracic Surgery 110, no. 2 (2020): 406–413, 10.1016/j.athoracsur.2020.02.071. [DOI] [PubMed] [Google Scholar]
- 22. Chaft J. E., Sima C. S., Ginsberg M. S., et al., “Clinical Outcomes With Perioperative Chemotherapy in Sarcomatoid Carcinomas of the Lung,” Journal of Thoracic Oncology 7, no. 9 (2012): 1400–1405, 10.1097/JTO.0b013e3182614856. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Lin Y., Yang H., Cai Q., et al., “Characteristics and Prognostic Analysis of 69 Patients With Pulmonary Sarcomatoid Carcinoma,” American Journal of Clinical Oncology 39, no. 3 (2016): 215–222, 10.1097/coc.0000000000000101. [DOI] [PubMed] [Google Scholar]
- 24. Ouziane I., Boutayeb S., Mrabti H., Lalya I., Rimani M., and Errihani H., “Sarcomatoid Carcinoma of the Lung: A Model of Resistance of Chemotherapy,” North American Journal of Medical Sciences 6, no. 7 (2014): 342–345, 10.4103/1947-2714.136920. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Kim S., Kim M. Y., Koh J., et al., “Programmed Death‐1 Ligand 1 and 2 Are Highly Expressed in Pleomorphic Carcinomas of the Lung: Comparison of Sarcomatous and Carcinomatous Areas,” European Journal of Cancer 51, no. 17 (2015): 2698–2707, 10.1016/j.ejca.2015.08.013. [DOI] [PubMed] [Google Scholar]
- 26. Velcheti V., Rimm D. L., and Schalper K. A., “Sarcomatoid Lung Carcinomas Show High Levels of Programmed Death Ligand‐1 (PD‐L1),” Journal of Thoracic Oncology 8, no. 6 (2013): 803–805, 10.1097/JTO.0b013e318292be18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Schrock A. B., Li S. D., Frampton G. M., et al., “Pulmonary Sarcomatoid Carcinomas Commonly Harbor Either Potentially Targetable Genomic Alterations or High Tumor Mutational Burden as Observed by Comprehensive Genomic Profiling,” Journal of Thoracic Oncology 12, no. 6 (2017): 932–942, 10.1016/j.jtho.2017.03.005. [DOI] [PubMed] [Google Scholar]
- 28. Wu M., Zhou N., Guan M., Wang Y., and Wang Y., “Efficacy of Immune Checkpoint Inhibitors in Pulmonary Sarcomatoid Carcinoma and Predictive Potential of Mutated TP53,” Lung Cancer 199 (2025): 108068, 10.1016/j.lungcan.2024.108068. [DOI] [PubMed] [Google Scholar]
- 29. Birsen G., Gounant V., Girard N., et al., “Efficacy of First‐Line Immune Checkpoint Inhibitors in Pulmonary Sarcomatoid Carcinoma,” Translational Lung Cancer Research 13, no. 9 (2024): 2212–2221, 10.21037/tlcr-24-263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Du H., Song X., and Wu F., “Evaluation of the Effectiveness and Safety Profile of First‐Line Immune Checkpoint Inhibitors Combined With Chemotherapy in Pulmonary Sarcomatoid Carcinoma,” Lung Cancer (Auckland) 16 (2025): 73–83, 10.2147/lctt.S494990. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Bao Z., Yu X., Zheng K., Zhai K., Cui H., and Xu M., “Case Report: Sintilimab Combined With Anlotinib as Neoadjuvant Chemotherapy for Metastatic Bone Tumor Resection in Patients With PSC,” Frontiers in Immunology 15 (2024): 1372279, 10.3389/fimmu.2024.1372279. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Li R., Di X., Li Y., Li H., and Liu C., “Dramatic Response of Advanced Pulmonary Sarcomatoid Carcinoma to Tislelizumab Combined With Anlotinib: A Case Report,” Frontiers in Oncology 15 (2025): 1531700, 10.3389/fonc.2025.1531700. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Wu S., Wu S., Liao X., et al., “Pembrolizumab Combined With Anlotinib Improves Therapeutic Efficacy in Pulmonary Sarcomatoid Carcinoma With TMB‐H and PD‐L1 Expression: A Case Report and Literature Review,” Frontiers in Immunology 14 (2023): 1274937, 10.3389/fimmu.2023.1274937. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Guo Y., Liu X., Tang H., et al., “Case Report: Pathological Complete Response Induced by Immunochemotherapy in a Case of Pulmonary Sarcomatoid Carcinoma Staged IIIA‐N2,” Frontiers in Immunology 15 (2024): 1374270, 10.3389/fimmu.2024.1374270. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Guo X., Wang J., Li D., Wang B., Zhu H., and Guo H., “Case Report: The Outcomes of Neoadjuvant Immunotherapy Combined With Chemotherapy in Pulmonary Sarcomatoid Carcinoma: Case Series and Literature Review,” Frontiers in Immunology 15 (2024): 1467755, 10.3389/fimmu.2024.1467755. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Ding K., Peng Z., and Xu Y., “Role of Immunotherapy in Pulmonary Sarcomatoid Carcinoma: Review of Current Approaches and Related Biomarkers,” Therapeutic Advances in Medical Oncology 16 (2024): 17588359241249041, 10.1177/17588359241249041. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Li H., Chen R., Guo W., Xiang S., Huang J., and Chen H., “Immunotherapy as a Neoadjuvant Preoperative Treatment for Locally Advanced Pulmonary Sarcomatoid Carcinoma: A Case Report of Clinical Efficacy,” Frontiers in Oncology 15 (2025): 1483766, 10.3389/fonc.2025.1483766. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are available from National Cancer Institute. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from the author(s) with the permission of National Cancer Institute.