Within the time frame of this study, a total of 303 patients with pHGNEC received surgical-related treatment at our center. Excluding 11 patients who were lost to follow-up and 33 patients only underwent biopsy surgery, a total of 259 patients were enrolled in this study, including 205 male patients and 54 female patients. The average age of the enrolled patients was 60.5 years (range 30–83 years), and the median follow-up time was 32 months (range 1–239 months). By the time of the last follow up, 121 patients were still alive. Among the enrolled patients, there were 146 patients with SCLC, 78 patients with LCNEC, and 35 patients with mixed type tumors (Table 1). The interquartile ranges of intraoperative blood loss for patients with pHGNEC, SCLC, and LCNEC were 200 ml, 300 ml, and 250 ml, respectively. Among the enrolled patients, 75 patients underwent preoperative bronchoscopy or puncture biopsy, with 46 cases (61.33%) diagnosed as malignant tumors, but only 34 cases (45.33%) had consistent pathological results between the biopsy and postoperative pathology. Intraoperative frozen section examination was performed in 169 patients, but only 81 cases (47.92%) were suggested to have SCLC or neuroendocrine tumors.

In the univariate analysis for DFS, several key factors emerged as significant prognostic indicators. Patients with larger maximal tumor diameters exhibited poorer outcomes (HR 1.25, 95% CI 1.13–1.37, p < 0.001). Correspondingly, higher tumor stages, both in terms of primary tumor extent (T), nodal involvement (N), distant metastasis (M), and the composite TNM staging, were consistently associated with diminished DFS. Surgical approach also played a role, with patients treated via video-assisted thoracic surgery (VATS) demonstrating markedly improved prognoses compared to thoracotomy (HR 0.53, 95% CI 0.36–0.77, p = 0.001).

Subgroup analysis of SCLC patients corroborated the detrimental impact of advanced T, N, and TNM stages. Notably, elevated preoperative carcinoembryonic antigen (CEA) levels independently predicted worse DFS (HR 2.50, 95% CI 1.31–4.79, p = 0.006). In the LCNEC subgroup, adjuvant radiotherapy was associated with relatively poorer DFS (HR 2.60, 95% CI 1.08–6.26, p = 0.033). However, peripheral tumor distribution (HR 0.39, 95% CI 0.20–0.75, p = 0.005) and VATS approach, even when converted to open thoracotomy (HR 0.22, 95% CI 0.05–0.91, p = 0.037), conferred significant prognostic benefits. Regarding the extent of surgical resection, complex or combined lobectomies were associated with worse outcomes compared to standard lobectomy or sublobar resection (HR 2.48, 95% CI 1.11–5.53, p = 0.026). Consistent with the SCLC findings, advanced T, N, and TNM stages predicted poorer DFS in the LCNEC subgroup. The Kaplan–Meier curves illustrating these DFS results are presented in Fig. 1.

Kaplan–Meier curves of DFS for patients with pHGNEC, SCLC and LCNEC: pHGNEC: A resection extent; B T stage; C TNM stage; SCLC: D pre-op CEA high; E T stage; F TNM stage; LCNEC: G resection extent; H TNM stage

Turning to overall survival (OS), increasing age (HR 1.02 per year, 95% CI 1.00–1.04, p = 0.022) and the history of other malignancies (HR 2.17, 95% CI 1.06–4.47, p = 0.035) were associated with poorer prognosis. As with DFS, larger tumor diameters significantly predicted worse OS outcomes (HR 1.25, 95% CI 1.13–1.37, p < 0.001). Surgical approach continued to play a pivotal role, with VATS associated with longer OS compared to open thoracotomy (HR 0.58, 95% CI 0.40–0.84, p = 0.004). Advancing T, N, and TNM stages corresponded with diminished OS in the overall cohort.

Within the SCLC subgroup, larger tumor size emerged as an independent adverse prognostic factor (HR 1.18 per cm, 95% CI 1.01–1.38, p = 0.032), while elevated preoperative CEA levels conferred a worse OS (HR 2.40, 95% CI 1.26–4.5, p = 0.008), consistent with the DFS findings. In the LCNEC subgroup, history of other malignancies were associated with a increase in the risk of death (HR 23.51, 95% CI 2.33–236.96, p = 0.007), and patients with endocrine comorbidities also exhibited significantly poorer OS (HR 2.50, 95% CI 1.14–5.47, p = 0.022). Mirroring the DFS results, VATS approach, even when converted to thoracotomy (HR 0.22, 95% CI 0.05–0.94, p = 0.041), conferred substantial OS benefits. Regarding tumor location, central tumors predicted improved OS compared to peripheral lesions (HR 0.44, 95% CI 0.23–0.83, p = 0.011). Both SCLC and LCNEC subgroup analyses reaffirmed the profound adverse prognostic impact of advancing T, N, and TNM stages. The Kaplan–Meier curves illustrating these OS results are presented in Fig. 2.

Kaplan–Meier curves of OS for patients with pHGNEC, SCLC and LCNEC: pHGNEC: A resection extent; B history of other malignancies; C T stage; SCLC: D TNM stage; SCLC: E pre-op CEA high; F T stage; G TNM stage; LCNEC: H resection extent; I history of other malignancies; J TNM stage

In the multivariate analysis for DFS, patients who underwent VATS surgery exhibited substantially better outcomes compared to those treated via thoracotomy(HR 0.50, 95% CI 0.33–0.75, p = 0.001). Advancing T stage was associated with diminished DFS. Patients with T2 (HR 1.84, 95% CI 1.10–3.06, p = 0.020) and T4 (HR 3.76, 95% CI 1.53–9.24, p = 0.004) lesions demonstrated worse prognosis respectively. Similarly, advancing composite TNM stage emerged as a potent adverse prognostic factor (Table 2).

Subgroup analysis of SCLC patients corroborated these findings. Advancing T stage continued to predict poorer DFS. Similarly, advancing TNM stage was associated with substantially diminished DFS in the SCLC subgroup. Furthermore, elevated preoperative CEA levels emerged as an independent adverse prognostic factor (HR 2.71, 95% CI 1.36–5.39, p = 0.004) (Table 3). In the LCNEC subgroup, VATS approach was associated with improved DFS compared to thoracotomy (HR 0.35, 95% CI 0.14–0.89, p = 0.027). This benefit was maintained even when conversion to open thoracotomy was required (HR 0.19, 95% CI 0.04–0.99, p = 0.049). Advancing to stage IV disease emerged as a profound adverse prognostic factor (HR 13.66, 95% CI 2.23–83.72, p = 0.005) (Table 4).

Turning to the multivariate analysis of OS, increasing age at diagnosis emerged as an independent risk factor (HR 1.03, 95% CI 1.01–1.06, p = 0.002). As with DFS, undergoing VATS surgery was a powerful predictor of improved OS (HR 0.49, 95% CI 0.33–0.74, p = 0.001). Consistent with the DFS findings, advancing T stage corresponded with poorer OS, as patients with T2 (HR 1.64, 95% CI 1.01–2.65, p = 0.044) and T4 (HR 3.20, 95% CI 1.37–7.51, p = 0.007) lesions exhibited poorer prognosis1. Similarly, advancing TNM stage emerged as a potent predictor for unfavorable prognosis (Table 2).

Within the SCLC subgroup, advancing T stage continued to predict poorer OS, and advancing TNM stage also emerged as a profound predictor. Additionally, elevated preoperative CEA levels independently predicted poorer OS (HR 2.39, 95% CI 1.21–4.72, p = 0.012) (Table 3). In the LCNEC subgroup, the presence of history of other malignancies emerged as an extraordinary adverse prognostic factor (HR 180.32, 95% CI 11.60–2802.78, p < 0.001). However, undergoing VATS approach conferred a significant OS benefit (HR 0.41, 95% CI 0.17–0.99, p = 0.048). Advancing to stage III and IV disease also emerged as profound predictors (Table 4). Notably, intraoperative blood loss emerged as an adverse prognostic factor in univariate OS analysis for the overall cohort and both subgroups. However, as the analysis results display only four decimal places, the precise hazard ratios could not be reported. In the multivariate OS analysis for the overall cohort, each milliliter of intraoperative blood loss was associated with a marginal 0.03% increased risk of death (multivariable HR 1.0003, 95% CI 1.0001–1.0005, p = 0.014).

Nomogram model that included the important predictors in the Cox analysis was established to predict the prognosis of pHGNEC, including SCLC and LCNEC (Figs. 3 and 4). Internal verification also showed that the nomogram could accurately predict the C-index of DFS for included pHGNEC, SCLC and LCNEC, which was 0.757, 0.756 and 0.793. Internal verification showed that the nomogram could accurately predict the C-index of OS for included pHGNEC, SCLC and LCNEC, which was 0.748, 0.741 and 0.787. The calibration curve showed that there was good concordance between the predicted and observed values of 1-year and 3-year OS and DFS internal validation cohorts (Figs. 5 and 6).

Nomogram model predicting the 1-, 3- and 5-year DFS in included patients. The nomogram is used by summing all points identified on the scale for each variable. The total points projected on the bottom scales indicate the probabilities of 1-, 3- and 5-year survival: A pHGNEC; B SCLC; C LCNEC

Nomogram model predicting the 1-, 3- and 5-year OS in included patients. The nomogram is used by summing all points identified on the scale for each variable. The total points projected on the bottom scales indicate the probabilities of 1-, 3- and 5-year survival: A pHGNEC; B SCLC; C. LCNEC

The calibration curves for predicting patient DFS at 1-, 3- and 5-year in the internal verification: A pHGNEC; B SCLC; C LCNEC. The DFS predicted by the nomogram model is plotted on the x-axis, and the actual DFS is plotted on the y-axis

The calibration curves for predicting patient OS at 1-, 3- and 5-year in the internal verification: A pHGNEC; B SCLC; C LCNEC. The OS predicted by the nomogram model is plotted on the x-axis, and the actual OS is plotted on the y-axis