https://orcid.org/0000-0003-1400-1437
Introduction
This research aimed to identify which subtype of pulmonary large-cell neuroendocrine carcinoma (LCNEC) could benefit from adjuvant chemotherapy (aCT). Unlike other histological types of lung cancer, LCNEC accounts for only 1–3 per cent of all lung cancers with poor prognosis1,2. Neuroendocrine lung tumours consist of typical and atypical carcinoids, which are low-grade tumours, whereas LCNECs and small-cell lung carcinomas (SCLCs) are high-grade tumours. Owing to its rarity, the standard regimen for LCNEC remains controversial, as some physicians favour an SCLC-type regimen, whereas others prefer non-SCLC-type treatment3. Therefore, efforts are ongoing to establish gene expression, specifically for RB1 gene4, activating mutations, and signalling pathways to identify patients with LCNEC5.
No effective biomarkers have been reported to date for predicting the superior outcome of aCT in patients with LCNEC4,5. The immune status of tumour cells (TCs) and tumour-infiltrating lymphocytes (TILs) is critical to chemotherapy effectiveness6. Programmed death ligand 1 (PD-L1) expression may predict response to aCT in patients with LCNEC and was assessed here.
Methods
The study was performed in accordance with the Declaration of Helsinki. All data were analysed anonymously. The Institutional Review Board of National Cancer Centre/Cancer Hospital, Chinese Academy of Medical Sciences, approved this study (2022060809432402).
Patients
Patients with LCNEC who received surgical treatment (2010–2019) at the National Cancer Centre in China were reviewed retrospectively. The diagnosis of LCNEC was confirmed by two independent reviewers based on the neuroendocrine lung tumour classification in resection specimens7. Patients with preoperative clinical stage below IIB proceeded directly to surgery. Among those with preoperative clinical stage IIIA disease (including cN2), surgery was undertaken in suitable patients after multidisciplinary team discussion. Patients who met the following criteria were excluded: no formalin-fixed paraffin-embedded (FFPE) specimens available; lost to follow-up; follow-up less than 12 months; postoperative treatment uncertainty; died within 3 months after surgery; or underwent other treatment modalities. Patients with complete pathological tissue samples and detailed clinicopathological and prognostic information (overall survival (OS) time) were selected.
Immunohistochemistry
FFPE tissues were subjected to immunohistochemical (IHC) staining for PD-L1 (22C3; DAKO, Carpinteria, CA 93013, USA). Clinical data are shown in Table S1. PD-L1 expression in the samples was calculated using the HALO digital pathology software platform. The classifier block in HALO v3.0.311.314 analytics software was applied to distinguish TCs from TILs (Fig. S1). An Indica Labs—Multiplex IHC v2.2.0 block was used to count the number of positive cells, total cell number, and, finally, to calculate the percentages of positively stained cells.
Statistical analysis
Associations between PD-L1 expression and clinical factors were evaluated using χ2 tests. Cox regression models and Kaplan–Meier survival curves were used to compare outcomes between patient groups. OS was calculated from the date of surgery to the date of death or end of follow-up. All statistical tests were two-sided, and P < 0.050 was deemed statistically significant. The best cut-off values for continuous variables were calculated using X-tile software version X86 (Yale University, USA) and the R survminer package (R Foundation for Statistical Computing, Vienna, Austria). Microsoft® Excel version 2016 (Microsoft, Redmond, WA, USA), SPSS® version 25.0 (IBM, Armonk, NY, USA) and R Studio version 3.6.1 software packages were used for data analysis.
Results and discussion
Of 173 patients, 76 with complete pathological tissue samples and detailed clinicopathological and prognostic information (OS time) were identified. After quality control, 76 patients with and without aCT (surgery + chemotherapy versus surgery alone), comprising 47 and 29 patients respectively, were included in the evaluation of PD-L1 expression (Fig. S2).
Because both TCs and TILs expressed PD-L1, the positive cell rate of PD-L1 in TCs and TILs was calculated by the HALO platform. Patients were divided into groups with high PD-L1 in TCs (19 patients; over 11.2 per cent of cells), low tumour PD-L1 (28 patients; less than 11.2 per cent), high PD-L1 in TILs (18 patients; over 26.9 per cent), and low PD-L1 in TILs (29 patients; less than 26.9 per cent) (Figs S3 and S4).
Associations between the clinical data for patients with LCNEC and PD-L1 expression are shown in Table S1. The association between OS and PD-L1 levels in patients with LCNEC who received aCT was evaluated by Kaplan–Meier analysis. Patients with high tumour PD-L1 levels had significantly shorter OS (P < 0.001) (Fig. 1a), but PD-L1 in TILs had no relationship with OS (P = 0.92) (Fig. 1d). High tumour PD-L1 levels showed strong predictive ability for OS (3-year OS: area under the curve (AUC) 0.816; 5-year OS: AUC 0.783) (Fig. 1b). Additionally, the receiver operating characteristic (ROC) curve of PD-L1 in TCs for 5-year OS was compared with that for the lung cancer staging system. The data demonstrated that the predictive ability of TC PD-L1 for OS outperformed that of the tumour staging system (Fig. 1c). However, PD-L1 in TILs had no value as prognostic predictor (Fig. 1d–f). TC PD-L1 was a predictor of OS independently of other factors, including sex, age, size, and pTNM stage (Table S2).
Impact of programmed death ligand 1 in tumour cells and tumour-infiltrating lymphocytes on overall survival and predictive ability in patients with large-cell neuroendocrine carcinoma who underwent adjuvant chemotherapy and surgery a Overall survival (OS) according to programmed death ligand 1 (PD-L1) expression level in tumour cells (TCs); b receiver operating characteristic (ROC) curves for ability of PD-L1 in TCs to predict OS; c ROC curves for ability of PD-L1 in TCs versus lung staging system to predict 5-year OS; d OS according to PD-L1 expression level in tumour-infiltrating lymphocytes (TILs); e ROC curves for ability of PD-L1 in TILs to predict OS; and f ROC curves for ability of PD-L1 in TILs versus lung staging system to predict 5-year OS. AUC, area under curve. aP < 0.001, bP = 0.92 (log rank test).
As LCNEC is rare, there is a lack of extensive literature and RCTs definitively describing an optimal treatment approach3–8. Although discordant results have been reported, chemotherapy appears to provide an additional benefit in early-stage LCNEC9. Among the 47 patients with LCNEC who had aCT in the present cohort, 26 received a SCLC-type regimen and 21 a Non-Small Cell Lung Cancer-type regimen. There was no survival difference between the two chemotherapy regimens (Fig. S5).
The surgery group had better OS than the surgery + aCT group (Fig. 2a). However, multivariate Cox regression analysis demonstrated that aCT was not an independent prognostic factor (Table S3). The groups showed no difference in PD-L1 expression (Fig. S6). In the group with high PD-L1 levels in TCs, patients who received aCT had significantly shorter OS than those who underwent surgical treatment alone (P = 0.002) (Fig. 2b). There was a significant survival difference between the surgery + aCT versus surgery alone groups only in the TC PD-L1-positive subgroup (HR 4.39, 95 per cent c.i. 1.06 to 18.12; P = 0.041) (Tables S4–S8). In the TC PD-L1-positive cohort, the surgery + aCT group had a dramatically worse prognosis among patients with early-stage disease, poor differentiation, or smokers (Fig. S7).
Overall survival curves for patients with large-cell neuroendocrine carcinoma according to treatment modality in whole cohort and subgroups based on expression level of programmed death ligand 1 in tumour cells and tumour-infiltrating lymphocytes a Whole cohort, b patients with high programmed death ligand 1 (PD-L1) expression in tumour cells (TCs), c patients with low PD-L1 expression in TCs, d patients with high PD-L1 expression in tumour infiltrating lymphocytes (TILs) and e patients with low PD-L1 expression in TILs. aCT, adjuvant chemotherapy. aP = 0.017, bP = 0.002, cP =0.78, dP =0.006, eP = 0.34 (log rank test).
Among patients with low PD-L1 expression, OS rates in those who received aCT were similar to those in patients who underwent surgery alone (Figs 2c and 2e). Although patients with high PD-L1 levels in TILs who recieved aCT had shortened OS (Fig. 2d), the 50 per cent OS time was much shorter in the cohort with high PD-L1 levels in TCs. These results showed that patients with LCNEC and high PD-L1 levels in TCs did not benefit from aCT.
Funding
This work was supported by the National Natural Science Foundation of China (82002610, 82002432); Beijing Hope Run Special Fund of Cancer Foundation of China (LC2019B18, LC2021L01, LC2021B12); Beijing Municipal Science & Technology Commission (Z211100002921058); Beijing Natural Science Foundation (J20010); CAMS Innovation Fund for Medical Sciences (CIFMS) (2020-I2M-C&T-B-071); and China Scholarship Council (CSC) (202106210312).
Acknowledgements
Yun Che, Zhiwen Luo, and Yanan Cao contributed equally to this work.
Disclosure
The authors declare no conflict of interest.
Supplementary material
Supplementary material is available at BJS online.
Data availability
Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information.
References
