Comparative Study of Pulmonary Combined Large-Cell Neuroendocrine Carcinoma and Combined Small-Cell Carcinoma in Surgically Resected High-Grade Neuroendocrine Tumors of the Lung

Objectives Pulmonary large-cell neuroendocrine carcinoma (LCNEC) and small-cell lung cancer (SCLC) are both classified as pure and combined subtypes. Due to the low incidence and difficult diagnosis of combined LCNEC (C-LCNEC) and combined SCLC (C-SCLC), few studies have compared their clinical features and prognosis. Materials and Methods We compared the clinical features, mutation status of driver genes (EGFR, ALK, ROS1, KRAS, and BRAF), and prognosis between C-LCNEC and C-SCLC. Univariate and multivariate Cox regression analyses were applied for survival analysis. Results We included a total of 116 patients with C-LCNEC and 76 patients with C-SCLC in the present study. There were significant differences in distribution of smoking history, tumor location, pT stage, pN stage, pTNM stage, visceral pleural invasion (VPI), and combined components between C-LCNEC and C-SCLC (P<0.05 for all). C-SCLC was more advanced at diagnosis as compared to C-LCNEC. The incidence of EGFR mutations in C-LCNEC patients was higher than C-SCLC patients (25.7 vs. 5%, P=0.004). We found that tumor size, pN stage, peripheral CEA level, and adjuvant chemotherapy were independently prognostic factors for DFS and OS in C-LCNEC patients, while peripheral NSE level, pT stage, pN stage, VPI and adjuvant chemotherapy were independently associated with DFS and OS for C-SCLC patients (P<0.05 for all). Propensity score matching with adjustment for the confounders confirmed a more favorable DFS (P=0.032) and OS (P=0.019) in patients with C-LCNEC in comparison with C-SCLC patients upon survival analysis. Conclusions The mutation landscape of driver genes seemed to act in different way between C-SCLC and C-LCNEC, likely by which result in clinical phenotype difference as well as better outcome in C-LCNEC.


INTRODUCTION
Neuroendocrine carcinoma of the lung, which accounts for about 15-20% of primary lung cancer, is divided into four categories: small-cell lung cancer (SCLC), typical carcinoid, atypical carcinoid, and large-cell neuroendocrine carcinoma (LCNEC) (1,2). Pulmonary large-cell and small-cell neuroendocrine carcinomas are classified as high-grade neuroendocrine carcinomas (HGNEC), characterized by poor histologic differentiation, high aggressiveness, and poor prognosis (3). In 2015, the World Health Organization (WHO) classification divided SCLC into pure SCLC and combined SCLC (C-SCLC). C-SCLC is defined as a SCLC type that is mixed with other components of non-small-cell lung cancer (NSCLC), such as adenocarcinoma (AD), squamous cell carcinoma (SCC), largecell carcinoma (LCC), LCNEC, and so on. C-SCLC accounts for 2 to 28% of all SCLC (4)(5)(6). Like SCLC, LCNEC is also divided into pure LCNEC and combined LCNEC (C-LCNEC), and C-LCNEC is the LCNEC that combined with AD, SCC, and other rare types such as spindle-cell carcinoma and giant-cell carcinoma. The reported incidence of C-LCNEC ranged from 10 to 49% (7)(8)(9). Current studies on LCNEC and SCLC generally take all components as a whole instead of specified analysis with focus on pure and combined parts (10)(11)(12). Owing to the rarity and the difficulty in diagnosis of C-LCNEC and C-SCLC, it is lack of the statistical description of their clinical features, genomic landscape, prognosis, and the relevant comparisons. The diagnosis of pulmonary neuroendocrine tumors has been advanced in recent years by the development of pathological technology and the wide application of surgery in multimodal treatment of lung cancer (13,14). Therefore, genomic and clinical analysis on specific subtypes will take the advantage of increased diagnosis to improve the treatment guidance for SCLC and LCNEC. To our knowledge, no large sample studies up to now have compared C-LCNEC with C-SCLC in terms of clinical features and genomic mutation landscape. We conducted this study to fill this gap and wish the results may provide new insights into the treatment of resected C-LCNEC and C-SCLC.

Patients
We retrospectively analyzed 1,250 patients with high-grade neuroendocrine carcinoma who underwent pulmonary resection in our organization. All surgically resected specimens were independently reviewed by two professional pathologists according to the 2015 WHO criteria for pulmonary neuroendocrine carcinoma (1). The initial diagnosis and classification of the C-LCNEC and C-SCLC were based on the neuroendocrine tumor morphology, which was further confirmed by immunohistochemistry. The tumor staging was based on the 8th edition TNM staging system proposed by the International Association for the Study of Lung Cancer (IASLC) (15). The treatment response was evaluated according to the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1. Besides NSCLC components, LCNEC might also be mixed with SCLC, but these tumors are classified as C-SCLC and will be excluded from our study. Other exclusion reasons included uncertain diagnosis, palliative surgery, and incomplete medical record and follow-up data.
As presented by the study flow chart (Figure 1), we finally included 192 cases of surgically resected combined high-grade neuroendocrine carcinoma in total. Specially, there were 116 patients with C-LCNEC and 76 patients with C-SCLC. The studies involving human participants were reviewed and approved by Shanghai Chest Hospital and affiliation of ethics committee. A written informed consent was signed by included patients before they donated tumor tissues for the purpose of scientific research.

Molecular Analysis
The patient's genomic DNA and RNA were extracted from formalin-fixed, paraffin-embedded tumor tissues and were subjected to the analysis of genetic alterations. We evaluated EGFR mutations by amplification refractory mutation system (ARMS) method, ALK expression by immunohistochemistry (IHC), ROS1 fusion by in situ hybridization (FISH), and KRAS/BRAF mutations by polymerase chain reaction (PCR) amplification method.

Follow-Up
Patients were diagnosed between 2008 and 2018, with median follow-up of 76 months and 55.7% deaths at the end of follow-up. Survival information was obtained through inpatient and outpatient records or telephone. Routine examinations such as chest computed tomography (CT) scans, brain magnetic resonance imaging or CT, tumor markers, and the neck and abdominal ultrasound were performed every 3 months for the first 2 years after surgery, followed by every 6 months examinations until 2-5 years. After 5 years, the patients were examined once a year. Patients were censored who were lost to follow-up or did not achieve the endpoint event. Disease-free survival (DFS) was defined as the time from surgery to disease recurrence or last follow-up if censored. Overall survival (OS) was the period between the date of surgery and death, but was extended to the end of follow-up in the presence of censored situation.

Statistical Analysis
We compared the differences in clinicopathological characteristics between C-LCNEC and C-SCLC patients. Chisquare test or Fisher's exact test was used for the comparisons in presence of discrete variables. As a non-parametric analysis method, Mann-Whitney U test was adopted for comparisons of continuous variables that were distributed non-normally. Student t test was used for normal distribution variables. Survival analysis was performed by Cox proportional hazards regression model, and the survival difference was visualized by Kaplan-Meier curve and log-rank test.
To reduce the selection bias as much as possible, propensity score matching (PSM) method was applied to sample C-LCNEC and C-SCLC patients at a 1:1 ratio via nearest-neighbor method without replacement (16). Propensity scores for included patients were calculated by a multiple logistic regression with adjustment for clinical features including gender, age, smoking history, primary site, laterality, tumor location, tumor size, pT stage, pN stage, VPI, combined components, CEA, CYFRA21-1, SCCA, NSE, adjuvant chemotherapy, and postoperative adjuvant radiotherapy (PORT). Finally, A total of 75 pairs were successfully matched. We used the SPSS software of version 26.0 (IBM Corporation, Armonk, NY, USA) for statistical analysis, and two-tailed P<0.05 was considered statistically significant.

Patient Population
The patients' clinicopathologic characteristics are shown in Table 1. We included 116 patients pathologically diagnosed as C-LCNEC, and other 76 patients with C-SCLC. C-SCLC patients had a higher proportion of smoking than C-LCNEC patients (P=0.027) and tended to be central-type carcinoma (P=0.015). The pathological staging of C-SCLC was more advanced compared to C-LCNEC, in terms of pT, pN, and pTNM stage (P=0.012, 0.020, and 0.003; respectively). C-LCNEC appeared to have a higher incidence of VPI than C-SCLC (P=0.002). Among 116 C-LCNEC patients, the most common were LCNEC combined with adenocarcinoma (LCNEC/AD, 70.7%, n=82), and then LCNEC combined with squamous cell carcinoma (LCNEC/SCC, 29.3%, n=34). As for C-SCLC, the percentage of SCLC plus adenocarcinoma and squamous cell carcinoma was 50%, respectively. The proportion of combined with AD in C-LCNEC was higher than that in C-SCLC (P=0.013).To adjust the confounding factors between the two groups, we sampled the two patient population with the PSM method. We found similar results upon PSM adjustment by such comparisons mentioned above ( Table 1).

Genetic Alterations
The mutation data of driver genes such as EGFR, ALK, ROS1, KRAS, and BRAF were available in 110 patients, including 70 C-LCNEC and 40 C-SCLC patients. As shown in Figure 2, we only found the alterations of ALK and EGFR genes in these patients. The incidence of EGFR mutations in C-LCNEC patients was found to be higher than C-SCLC patients (25.7 vs. 5%, P=0.004). Specially, there were 18 patients with EGFR mutations of C-LCNEC, including 10 patients with 19 exon deletions, 7 patients with 21 exon L858R mutation, and 1 patient with exon 20 insertion. For C-SCLC patients, we only identified two patients with EGFR 19 exon deletions. The incidence of ALK rearrangement were relatively lower, with only four patients TABLE 1 | Clinical demographics before and after propensity score matching for patients with resected C-LCNEC and C-SCLC.

Survival Comparison Between C-LCNEC and C-SCLC
We enrolled the factors that might affect DFS and OS in C-LCNEC and C-SCLC patients into Cox regression models for survival analysis (Tables 2, 3). We initially made univariate Cox analysis and found tumor size, pN stage, CEA, and adjuvant chemotherapy were the main DFS modulator for C-LCNEC with statistical significance (P<0.05 for all,  Table 3). These main survival effectors in univariate Cox regression models were further included in multivariate analysis. As results, we found that tumor size, pN stage, peripheral CEA level, and adjuvant chemotherapy were independently associated with DFS and OS in C-LCNEC patients ( Figure 3). For C-SCLC patients, NSE, pT stage, pN stage, VPI, and adjuvant chemotherapy were independently predictors for DFS and OS (Figure 3). We next compared the outcomes of patients between C-LCNEC and C-SCLC and found that there was a longer OS in patients with C-LCNEC (P=0.006, Figure 4B). But we only observed an insignificant trend towards favorable DFS in patients with C-LCNEC (P=0.059, Figure 4A). PSM method adopted to reduce the confounders between the groups resulted in better DFS and OS of C-LCNEC patients than C-SCLC patients, highlighting the indeed difference in prognosis between the two lung cancer subtypes (for DFS, P=0.032, and for OS, P=0.019; Figures 4C, D).

DISCUSSION
Pulmonary neuroendocrine carcinoma is relatively aggressive and its diagnosis linked to worse survival in comparison with other lung cancer subtypes. Because the role of surgery has not been fully recognized before, chemotherapy combined with radiotherapy has been the major treatment for pulmonary  (20). As a subtype of NSCLC, surgery was the primary treatment for localized (I-IIIA) LCNEC. Survival benefits from    (21)(22)(23). Because patients could benefit more from surgery than first-line chemoradiotherapy in early-stage patients (24,25), the number of patients received surgery has gradually increased, which may provide more available tissue for pathologists to accurately identify the combined components. To our knowledge, this study is the first, with relatively large sample size, to describe the clinicopathological features and prognosis of C-LCNEC and C-SCLC patients who had undergone surgery, and make relevant comparison between them. According to a series of previous studies, it was reportedly to be comparable regarding clinicopathological characteristics and prognosis between LCNEC and SCLC (26,27). However, Varlotto JM et al. analyzed the data from Surveillance, Epidemiology, and End Results Program (SEER) database and found that there were significant differences between LCNEC and SCLC in tumor size, lymph node metastasis, and differentiation degree (10). However, there was lack of evidence for clinical feature comparisons between C-LCNEC and C-SCLC. In our study we found that C-SCLC patients tended to have a history of smoking than C-LCNEC patients. C-LCNEC were mainly located at periphery lung, while C-SCLC were the tumors mainly in central location. In addition, the pathological staging of C-SCLC was more advanced compared to C-LCNEC, but C-LCNEC was characterized with VPI phenotype and combined AD components with manifestation of higher proportion than C-SCLC. These differences revealed may provide a potential tool for us to distinguish C-LCNEC from C-SCLC.
As expected, the clinical factors that influence the prognosis of C-SCLC and C-LCNEC were also revealed to be not identical. We found that tumor size, pN stage, peripheral CEA level, and adjuvant chemotherapy were independently prognostic factors for DFS and OS of C-LCNEC patients. Our study was first to analyze the prognostic factors of C-LCNEC in the presence of relatively large sample size and may make better understanding of prognosis of such a low-incidence carcinoma type. In line with previous studies, for C-SCLC patients, peripheral NSE level, pT stage, pN stage, VPI, and adjuvant chemotherapy were independently survival predictors for DFS and OS (4,28).
Because of low incidence, survival comparison has not been previously reported between C-SCLC and C-LCNEC cancers. The previous focus was mainly on generalized SCLC and LCNEC, and demonstrated a better prognosis of LCNEC than SCLC (11,12,29). But other studies found no difference in prognosis between the two groups (26,30,31). In our study, we compared the outcomes of C-LCNEC with C-SCLC and revealed a prolonged OS of patients with C-LCNEC than those with C-SCLC, but the DFS was comparable. After PSM sampling, the prolonged OS and DFS were both statistically significant for the comparison of C-LCNEC versus C-SCLC, likely due to the higher neuroendocrine component of C-SCLC in nature.
It has been reported that the presence of driver mutations was very rare in LCNEC and SCLC but usually occurred in combined subtypes (32)(33)(34). Yokomizo et al. found EGFR mutations in three of 15 (20%) patients with C-SCLC (35). In another study, NGS was performed in 10 C-LCNEC patients, and five of them had driver gene alteration (32). Natasha et al. performed targeted next-generation sequencing testing on 45 LCNEC patients and classified LCNEC into two major types: SCLC-like (n=18; 40%), characterized byTP53+RB1 co-mutation/loss; NSCLC-like (n=25; 68%), characterized by the lack of coaltered TP53+RB1 (36). This study demonstrated the biological heterogeneity of LCNEC and provided reference for the classification and management of LCNEC patients.  In our study, EGFR mutations were presented in 18 patients (25.7%) and 4 patients (5.7%), for C-LCNEC and C-SCLC, respectively. ALK rearrangement was only observed in patients with C-LCNEC, and we did not detect alterations in other driver genes for both carcinoma types. A total of 10 patients received TKI treatment after distant metastasis and obtained good survival benefits, with an ORR rate of 60%, which suggests that genetic testing for patients with combined HGNEC is feasible and beneficial, especially for C-LCNEC patients.
Collectively, these results suggest that patients with C-LCNEC and C-SCLC may have different genetic background, which is a potential reason for the different clinical features and prognosis between them.
This article has several limitations. First of all, although we adopted the PSM method to reduce the confounders, the retrospective design of the present study may inevitably introduce bias. Second, although our sample size was relatively larger compared with previous studies, the limited sample size when grouped into C-LCNEC and C-SCLC may reduce the statistical power. Lastly, only a small number of patients in this study underwent driver gene testing, and lack of comprehensive genetic testing made it unable to capture a full spectrum of genomic profiles of HGNEC. Therefore, larger and more comprehensive studies were needed to validate and refine our findings.
In conclusion, the driver mutation landscape between C-LCNEC and C-SCLC acted in a different way, which is a potential cause for different distribution of clinicopathological features and prolonged outcome of C-LCNEC patients. Comparisons of the genetic and clinical dimensions between the two lowincidence carcinoma types may provide potential tools for clinical treatment decision.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Shanghai Chest Hospital and affiliation of ethics committee. A written informed consent was signed by included patients before they donated tumor tissues for the purpose of scientific research.

AUTHOR CONTRIBUTIONS
YW, FQ, and YC both have substantial contributions to the conception or design of the work, the collection and analysis of data, and the writing and editing of the article. The rest authors have given substantial contributions to the work by providing editing and writing assistance. All authors contributed to the article and approved the submitted version.