Peripheral Tumor Location Predicts a Favorable Prognosis in Patients with Resected Small Cell Lung Cancer

Background Small cell lung cancer (SCLC) is an aggressive malignancy. Surgical resection is currently only recommended for clinical stage I patients who have been carefully staged. The clinical outcomes of patients with resected SCLCs vary because the disease is highly heterogeneous, suggesting that selected patients could be considered for surgical resection depending on their clinical and/or molecular characteristics. Methods We collected data on a retrospective cohort of 119 limited-stage SCLC patients who underwent lobectomy with mediastinal lymph node dissection from March 2013 to March 2020 at Harbin Medical University Cancer Hospital. Correlations were derived using Fisher's exact test. Models of 2-year and 3-year survival were evaluated by deriving the area under receiver operating characteristic curves. Kaplan–Meier and Cox regression analyses were used to evaluate significant differences between the survival curves and hazard ratios. Results The median disease-free survival (DFS) was 35.9 months (range 0.9–105.3 months), and the median overall survival (OS) was 45.2 months (range 4.8–105.3 months). Univariate analysis showed that TNM stage was significantly correlated with DFS and OS. The 2-year disease-free rates of patients with stage I, II, and III disease were 76.4%, 50.5%, and 36.1%, respectively, and the 3-year OS rates were 75.9%, 57.7%, and 34.4%, respectively. In pN + patients, multiple (or multiple-station) lymph node involvement significantly increased recurrence and reduced survival compared with patients with single or single-station metastases. Patients with peripheral SCLCs evidenced significantly better DFS and OS than did patients with central tumors. Multivariate analysis showed that TNM stage and tumor location were independently prognostic in Chinese patients with resected limited-stage SCLC. A combination of TNM stage and tumor location was helpful for prognosis. Conclusions TNM stage and tumor location were independently prognostic in Chinese patients with resected SCLCs. Patient stratification by tumor location should inform the therapeutic strategy. The role of surgical resection for limited-stage SCLC patients must be reevaluated, as this may be appropriate for some patients.


Introduction
Lung cancer is the most common cancer worldwide. In 2020, it was the leading cause of cancer-related death in males and the second most common cause in females [1]. Lung cancer includes both nonsmall cell lung cancer (NSCLC) and small cell lung cancer (SCLC), which difer pathologically. SCLC is an aggressive malignancy characterized by rapid growth, early metastasis, and poor prognosis. Although the proportion of SCLCs among all histological lung cancer types has decreased markedly in the United States, signifcant increases in the proportions of both population-and hospital-based SCLC cases have been noted in some Chinese regions in recent years, especially among the elderly [2,3]. After appropriate treatment, about 20% of patients with limited-stage (LS) SCLC (as defned by the Veterans Administration Lung Study Group) survive. Unfortunately, for the other 80%, and for all patients with extensive-stage (ES) disease, the outcomes remain poor, despite transitory responses to initial chemotherapy and radiotherapy [4]. Although the standard front-line treatment for ES SCLC changed when two randomized phase 3 trials revealed improved survival with the addition of atezolizumab or durvalumab to platinum plus etoposide [5,6], survival remains modest compared with that of patients with other solid tumors (e.g., NSCLC). For most LS SCLC patients, concurrent chemoradiation is the standard of care; only patients with stage I to stage IIA (T1-2, N0) SCLC are recommended for surgery. Tus, fewer than 5% of SCLC patients undergo surgery, and most SCLC patients are diagnosed (unexpectedly) with SCLC during surgery [7,8]. Some evidence supports a role for surgery when LS SCLC patients have been precisely staged. It has been suggested that lobectomy for selected LS SCLC patients improved local control and survival [9][10][11][12][13]. Stage is the major prognostic factor for resected SCLC patients; however, individual survival times difer even among patients of the same stage. Apart from TNM staging, several studies have reported that certain clinical characteristics/indicators (sex, age, tumor location, adjuvant chemotherapy status, and lymph node metastatic ratio) correlated with the clinical outcomes of resected SCLC patients [9,14]. However, most studies evaluated the cohort from the Surveillance, Epidemiology, and End Results (SEER) database; clinical information was thus limited. Recent retrospective works indicated that two indices of infammation, the neutrophil-to-lymphocyte ratio (NLR) and advanced lung cancer infammation index (ALI), were prognostic in resected SCLC patients [15,16]. Tus, we here combined known and new prognostic indicators of a retrospective cohort. Te results from this study improved our understanding of how baseline patient and tumor characteristics afect the prognosis of Chinese patients with resected LS SCLCs.

Patients and Methods
We retrospectively studied a cohort of 119 LS SCLC patients who underwent lobectomy with mediastinal lymph node dissection from March 2013 to March 2020 at Harbin Medical University Cancer Hospital. We enrolled all patients with SCLC only (as confrmed histopathological analysis). Tis study was approved by the Medical Ethics Committee of Harbin Medical University Cancer Hospital (Harbin, China). Te need for informed patient consent was waived because of the retrospective nature of the study. Te primary endpoints were disease-free survival (DFS) and overall survival (OS). DFS was defned as the time from surgery to disease recurrence or death, and OS was defned as the time from surgery to death from any cause, or to May 2022 for patients who remained alive. Clinicopathological and treatment characteristics, including sex, age (<65 or ≥65 years), smoking history (never-smokers versus current smokers), pathological TNM stage, tumor location, number of lymph node metastases, the neoadjuvant or/and adjuvant therapy regimen (chemotherapy or radiotherapy), and preoperative hematological parameters (e.g., the neutrophil, lymphocyte, and platelet counts), were abstracted from the electronic medical record system. Current smokers included those with a smoking index ˃400 packs per year throughout life and those who had quit for ˂5 years; the remaining patients were defned as never-smokers. Staging after surgery employed the 8 th edition of the American Joint Committee on Cancer criteria [17]. Primary tumor location was defned as central or peripheral upon bronchoscopy performed by experienced specialists. Central tumors involved the segmental or more proximal bronchi. Tumors with primary sites distal to the subsegmental bronchi and tumors that were not evident via bronchoscopy were defned as peripheral SCLCs. Te NLR was calculated by dividing the absolute neutrophil count by the absolute lymphocyte count, and the PLR was the platelet count divided by the absolute lymphocyte count.
All statistical analyses were performed with the aid of GraphPad Prism version 8 (Prism Software Inc., San Diego, CA, USA) and SPSS version 26 (SPSS Software Inc., Chicago, IL, USA). Numerical variables were presented as the mean ± standard deviation, otherwise presented as the median and range. One-way ANOVA analyses were used for the analysis of the NLR and PLR in patients with diferent stages. Models of the NLR and PLR at 2-year DFS and 3-year OS were evaluated using the area under the receiver operator characteristic curves (AUCs). Correlations were derived using the Pearson chi-squared test or the Fisher exact test to compare categorical variables. Kaplan-Meier analysis and the log-rank Mantel-Cox test were used to compare survival curves and hazard ratios (HRs). To identify independent prognostic factors, a multivariate survival analysis with HRs proceeded using Cox's regression model of factors that were signifcant on univariate analysis. All P values were twosided, and P < 0.05 was considered to indicate statistical signifcance.

Prognostic Utilities of Clinical Characteristics and Clinicopathological Features for Resected LS SCLC Patients.
Female sex, younger age, and stage I disease were previously shown to be associated with a better prognosis in LS SCLC patients [14]. Tus, we frst evaluated the prognostic signifcance of baseline clinical characteristics ( Table 2). Univariate analysis showed that TNM stage was signifcantly correlated with both DFS (P < 0.0001) and OS (P < 0.0001). Te 2-year disease-free rates for patients with stage I, II, and III disease were 76.4%, 50.5%, and 36.1%, respectively, and the 3-year fgures were 75.9%, 57.7%, and 34.4%, respectively (Figure 1(a)). Sex, age, or smoking status was not signifcantly prognostic, although females and never-smokers exhibited longer survival times. In addition, several pathological tumor characteristics, including the T and N stages, were prognostic. Two patients initially classifed as T2 stage were pathologically confrmed to be T3 stage and were excluded for analyses of the correlation between T stage and prognosis due to a small sample size. Te results showed that there was no statistical signifcance for the correlation between T stage and prognosis; however, it seems that patients with T2 stage trended toward a shorter DFS (23.6 months vs. NR, P = 0.1188) and OS (32.2 months vs. NR, P = 0.0581) than T1 stage (Figure 1(b)). Univariate analysis of pathological N stages in predicting prognosis revealed that N stages were signifcantly associated with DFS (P = 0.0004) and OS (P = 0.0001) in resected SCLC, which was consistent with previous studies [9,14] (Figure 1(c)).
Resected NSCLC patients with lymph node metastases varied in terms of prognosis, but multiple lymph node station metastases correlated with an inferior prognosis [18]. It was previously shown that the sites and numbers of metastatic lymph nodes correlated with the prognosis of resected N2 SCLC patients [19]. Tus, we analyzed the association between metastatic lymph node status and DFS and OS in pathological stage N+ patients. Tose with several metastatic lymph nodes had recurrence sooner than did patients with single metastatic lymph nodes (13.5 vs. 37.0 months, P � 0.0663), and the survival time of the former patients was signifcantly shorter (25.4 months vs. NR, P � 0.0131) (Figure 2(a)). LS SCLC patients with multiple-station lymph node metastases exhibited a shorter time to recurrence (11.5 vs. 37.0 months, P � 0.0084) and poorer survival than did patients with single-station metastatic lymph nodes (20.7 months vs. NR, P � 0.0013) (Figure 2(b)). Further analysis stratifed by N stage revealed that this was true only for patients with N2-stage disease ( Figure 3). Such patients with metastases in multiple lymph nodes of several stations exhibited much earlier relapses and death as compared with those with single afected lymph nodes.

Impact of Primary Tumor Location on Prognosis.
SCLC typically presents as a large hilar mass accompanied by bulky mediastinal lymphadenopathy. Most patients who were (unexpectedly) diagnosed with pathological SCLC presented with solitary peripheral nodules before surgery. Two prior studies reported that peripheral SCLCs more commonly expressed TTF-1 than did central tumors and were associated with a poorer prognosis [20,21]. However, the clinical signifcance of tumor origin in resected LS SCLC patients remains unclear. Out of all our patients, 60 presented with peripheral tumors (53.3% stage I (32/60) and 46.7% stages II and III (28/60)) and 59 with central tumors (39.0% stage I (23/59) and 61.0% stages II and III (36/59)). Te TNM stages of our peripheral and central SCLC patients did not difer (P � 0.142). TTF-1 immunoreactivity was apparent in 87 tumors, and peripheral tumors expressed TTF-1 more frequently (91.7%, 44/48) than did central SCLCs (71.8%, 28/39) (P � 0.0215). Unlike previous studies of unresected SCLC patients, we found that patients with peripheral SCLCs exhibited signifcantly longer DFS (P � 0.0021) and OS (P � 0.0035) than did those with central tumors (Figure 4). Te median DFS and OS were 17.7 and 32.2 months, respectively, in those with central tumors compared with "not attained" in patients with peripheral tumors. Te 3-year DFS and the 5-year OS were 32.2% and 22.6%, respectively, in those with central tumors, but 63.3% and 44.1%, respectively, in patients with peripheral tumors.

Te Impacts of Infammation-Based Scores on DFS and OS.
Previous studies have indicated that several hematological variables, including the lymphocyte count and NLR (both of which refect immune system and infammation status), correlated with LS SCLC prognosis [22]. A study of Caucasian patients with stage I or II resected SCLC showed that the preoperative NLR (but not the PLR) was associated with longer OS [16]. We frst sought to assess the correlations between the NLR and PLR and TNM stage.       Considering TNM stage and tumor location as independent predictive factors, we next integrated these two factors to predict prognosis. Survival analysis showed that patients with stage I or II peripheral tumors and stage I central tumors enjoyed better prognoses than those with stage III peripheral tumors and stage II or III central tumors ( Figure 6). Tus, integration of TNM stage and tumor location was meaningfully prognostic in terms of both DFS and OS.

Discussion
SCLC is aggressive, and surgical resection is currently recommended (only) for clinical stage I patients who have been carefully staged. In practice, most SCLC patients who undergo surgical resection are diagnosed intraoperatively or postoperatively. Te largest published retrospective analysis of the SEER database found that patients with stage I or II SCLC undergoing resection had better survival than did    [23]. Similar retrospective studies, including patients with LS SCLC, also suggested that the role of surgery as an LS SCLC treatment required reevaluation [24,25]. Te clinical outcomes of resected SCLC patients of the same stage vary, as the disease is heterogeneous. Tis suggests that certain patients might beneft from surgical resection. However, the clinical and/or molecular characteristics that might identify such patients are unknown. To the best of our knowledge, this is the frst attempt to employ tumor location to predict the DFS and OS of resected LS SCLC patients. We found that both tumor location and TNM stage were independently prognostic. Central and peripheral SCLCs difer in terms of genome (in) stability, driver gene somatic copy numbers, and mutational signatures, suggesting that the biological characteristics difer [26]. Previous studies regarding the signifcance of SCLC tumor location have yielded  inconsistent results, and another study showed that peripheral SCLCs expressed TTF-1 more frequently than did central SCLCs and were associated with a poorer prognosis [20]. We agree (only) in terms of the TTF-1 expression frequency. We found the exact opposite in terms of prognosis. Te reason for this diference could be that the study populations difered. Nevertheless, a similar study enrolling a larger population found that peripheral SCLC was associated with longer OS than that of central SCLC [27], which was consistent with our study. Notably, our focus was on resected LS SCLC patients. Te populations in the studies cited above were dominated by ES SCLC patients receiving systematic therapy.
Our survival analyses showed that stage II peripheral SCLC patients after surgery experienced a prognosis similar to that of central SCLC patients of stage I, for whom surgical resection is currently recommended, suggesting that selected patients with N1 SCLC might beneft from surgery. Given the inferior DFS and OS of peripheral SCLC patients of stage III and central SCLC patients of stages II and III, our study supports the idea that such patients should receive the systematic therapy of the current guidelines. Previous studies of NSCLC patients who underwent surgery reported that the numbers and sites of involved lymph nodes correlated signifcantly with prognosis, especially for N2-stage patients [28][29][30]. We describe here, for the frst time, how lymph node involvement afects the prognosis of pN1-and N2-stage resected SCLC patients. Patients with multiple involved lymph node numbers/stations experienced a median OS similar to that of patients receiving chemoradiation, indicating the importance of preoperative staging. Stratifed analysis by N stage revealed that lymph node status was prognostic principally in patients of pN2 status, further emphasizing the high-level heterogeneity of pN2-stage resected SCLC, consistent with the fndings of a prior study of N2-stage resected SCLC patients [31]. When we compared survival with the N stage, we found that N1-stage patient survival was similar to that of N0-stage patients and signifcantly longer than that of N2stage patients, suggesting that some patients with N1-stage SCLC (as revealed by preoperative imaging) should be considered for surgical resection. Similarly, a previous study of stage I and stage II SCLC patients in the SEER database supported possible surgical resection to treat stage II SCLC [32].
Tis study has some limitations. Tis was a single-center retrospective study, and we were thus unable to consider all relevant factors. In addition, selection bias may have been in play, contributing to diferences between our fndings and those of others. A recent study reported that the preoperative NLRs and PLRs predicted the survival of patients with LS SCLC; however, we did not fnd this [33]. However, the  baseline characteristics of patients difered between the two studies. We enrolled more patients with stage I (46.2%) than did the cited work (22.3%). Our median NLR and PLR values were 1.711 and 100.0, respectively, which are lower than those in the cited study. Zeng et al. [34] enrolled some patients with histologically combined SCLC, while we enrolled only pure SCLC patients. However, on univariate analysis, we found that females and never-smokers exhibited longer survivals, as have previous studies. Our small sample size and short follow-up period may limit the statistical power of the analyses of correlations between sex and smoking and prognosis.

Conclusion
We found that TNM stage and tumor location were independently prognostic in Chinese patients with resected LS SCLC. Tis aids the evaluation of how surgery afects the prognoses of diferent groups and suggests that patient stratifcation by tumor location would optimize therapies. We found that the survival of pN1-and pN0-stage patients was similar and signifcantly longer than that of pN2 patients. In addition, the lymph node involvement status of pN2-stage resected SCLC patients was meaningful in terms of DFS and OS prognoses. Future prospective studies should reevaluate the utility of surgical resection in selected LS SCLC patients with lymph node metastases.

Data Availability
Te data used to support the fndings of this study are available from the corresponding authors upon request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.

Authors' Contributions
Y. Gao collected and analyzed data, provided scientifc input, and wrote the manuscript. Y. Dong, Y. Zhou, and G. Chen collected and analyzed data and wrote the manuscript. X. Hong provided intellectual and scientifc input and wrote the manuscript. Q. Zhang conceived the study, designed the experiments, analyzed data, and wrote the manuscript.