Surgical Outcomes of Synchronous Multiple Primary Non-Small Cell Lung Cancers

The prognostic indicators for synchronous multiple primary non-small cell lung cancer (NSCLC) vary across reports. In present study, the prognostic factors for the patients with synchronous multiple primary NSCLC were analyzed in a large cohort. A total of 285 patients with synchronous multiple primary NSCLC who underwent radical surgical resection and with complete follow-up information were included in this study. The Kaplan-Meier method were used for survival analysis, Cox proportional hazards regression models were used for risk factors evaluation. Among them, 94 (33.0%) patients had bilateral tumors and 51 (17.9%) had multiple (≥3) tumors. The 5-year disease-free survival (DFS) and overall survival (OS) rate was 58.7% and 77.6%, respectively. Univariate analysis identified parameters conferring shorter OS including male gender, symptomatic disease, negative family history, large maximal tumor size, not all adenocarcinomas, advanced highest T stage, and lymph node involvement. Multivariate analysis showed that male gender (p = 0.020), symptomatic disease (p = 0.017), and lymph node involvement (p < 0.001) were independent adverse prognosticators. For patients with multiple adenocarcinomas, the 5-year DFS and OS rate was 59.6% and 82.4%, respectively. The subtypes other than lepidic predominant (p < 0.001) and lymph node involvement (p = 0.002) were the independent unfavorable prognosticators. In conclusion, we identified independent prognosticators which will provide the valuable clues for postoperative management of patients with synchronous multiple primary NSCLC.

In a large cohort of patients, we analyzed surgical outcomes of synchronous multiple primary non-small cell lung cancer (NSCLC) to investigate the prognostic values of various clinical parameters for long-term survival. The present study, to our knowledge, is the largest investigation on clinical outcome of patients treated with surgery for synchronous multiple primary NSCLC.

Materials and Methods
Information collection. The medical records of patients who underwent complete pulmonary resection for lung cancer from January 2010 to December 2014 at the Department of Thoracic surgery, Cancer Hospital of Chinese Academy of Medical Sciences were reviewed. The demographic characteristics were recorded for further analysis, including: age, gender, major complaint (symptomatic disease was defined as persistent symptoms such as dry cough prior to diagnosis; asymptomatic disease was defined as lung cancers identified by either health check-up and screening or incidental discovery without any symptoms), smoking (never smokers were defined as consumption of < 100 cigarettes during their lifetimes; light smokers, consumption of < 20 pack-years; moderate smokers, consumption of 20-40 pack-years; and heavy smokers, consumption of > 40 pack-years), family history of cancer (in first degree relatives), preoperative serum biomarker profiling (carcinoembryonie antigen [CEA], cancer antigen 125 [CA125], cytokeratin 19 fragments [CYFRA , squamous cell carcinoma antigen [SCCA] and neuron specific enolase [NSE]), type of surgical resection, as well as the numbers, location (laterality and lobe), size (maximum diameter), and histological type of tumors, the highest T and N stage of each patient, and postoperative adjuvant chemotherapy. Informed consents were signed by all patients. This study was approved by the Institutional Review Board of Cancer Hospital of Chinese Academy of Medical Sciences and conducted according to the guidelines approved by the ethics committee. Patient selection. The synchronous preoperative computed tomography (CT), positron emission tomography (PET) and the intra-and postoperative histopathologic diagnoses were used to verify the existence of more than one malignant tumors. Patients were classified with SMPLCs if they met the modified criteria of Martini and Melamed for the diagnosis 8 : (1) Tumors with different histopathologic characteristics (e.g., adenocarcinoma vs. squamous cell carcinoma); (2) Tumors with differently predominant histologic subtypes (e.g., ratio of acinar, bronchoalveolar, and papillary percentage for adenocarcinomas); (3) Tumors with similar histology a. arising from separate foci (e.g., in the case of squamous cell carcinoma, the presence of in situ carcinoma in both tumors); b. without regional or mediastinal lymph node metastasis; c. without distant metastasis. Patients with small cell lung cancer, atypical adenomatous hyperplasia or preoperative neoadjuvant therapy were excluded.
Preoperative evaluation and Surgical approach. The routine preoperative staging workup for lung cancer patients in our hospital includes chest radiography, computed tomography (CT) of the thorax, abdominal ultrasonography, bronchoscopy, magnetic resonance image (MRI) of the brain, whole body bone scan, and cardiopulmonary function test. Endobronchial ultrasonography-guided transbronchial needle aspiration or, positron emission tomography (PET) CT scan was used in patients who had suspected mediastinal lymph nodes. A curative operation for pulmonary resection and lymphadenectomy via thoracotomy or video-assisted thoracic surgery (VATS) were performed.
Pathology examination. For adenocarcinomas, the tumors were reviewed by a specialized thoracic pathologist (Dr. Feng), and further classified according to the International Association for the Study of Lung Cancer (IASLC) classification 9 and predominantly histological subtype of largest tumor was recorded for each patient.

Follow-up.
Postoperative surveillance was scheduled regularly in the outpatient clinic with chest CT, serum tumor markers measurement (CEA, CA125, CYFRA 21-1, SCCA and NSE), and abdominal ultrasonography. Bone scanning and MRI of brain were performed annually. The follow-up frequency was every three months for the first 2 years, every six months for the third year, and once per year for subsequent years.
Statistical analysis. Disease-free survival (DFS) was calculated from the date of the initial surgery to the date of recurrence or distant metastasis; and overall survival (OS) was calculated from the date of the initial surgery to the date of cancer-related death or last follow-up for censored patients. One patient was lost to follow-up. The latest follow-up date was August 12, 2015. Patients who died from cardiopulmonary complications and non-cancer-related reasons were not included in the survival analysis. Continuous variables were summarized as mean ± standard deviation (SD). Actuarial survival was estimated using Kaplan-Meier method, and log-rank tests were used for univariate analysis. Multivariate analyses were performed using the Cox proportional hazard models to determine prognostic parameters associated with survival. All data analyses were performed using SPSS software version 20.0, and a P value of less than 0.05 was considered as statistically significant.

Results
Demographics of SMPLC patients. During the study period, 290 patients met the modified criteria of Martini and Melamed and were classified with SMPLC. 5 were excluded from data analysis, including 1 patient who died during the perioperative period, 1 who died 45 days postoperatively of respiratory failure after the second operation (0.69%), 2 patients who died from non-cancer-related reasons (1 from acute myocardial infarction and 1 from cerebrovascular accident), and 1 patient who was lost to follow-up. A total of 285 patients were studied in detail to determine prognostic factors (Fig. 1). In this study, long-term survival was defined as 5-years' cancer-specific survival. Patient demographic characteristics were summarized in Table 1. The median age was 60 years old (range 39-78), and 55.8% of the patients were female. Synchronous multiple primary NSCLCs were detected in a higher proportion of asymptomatic patients (56.1%) and non-smokers (62.8%). A total of 102 patients (35.8%) had a family history of cancer (first-degree relatives: parents, children and sibs). The preoperative (first surgery) serum carcinoembryonic antigen (CEA) level was elevated (≥ 5.0 ng/ mL) in 48 patients (21.9%). Table 2. Among the 94 (33.0%) patients with bilateral tumors, a single-stage bilateral operation was performed in 1 patient using VATS, while two-stage bilateral operations were performed in 93 patients; consisting of 48 VATS, 27 thoracotomies and 18 combined surgeries. A total of 87 patients (30.5%) underwent a standard surgical resection called a multi-lobectomy, including 6 pneumonectomies, 37 bilobectomies and 44 lobectomies, for synchronous cancers. Lobectomy combined with sublobar resection (segmentectomy and wedge resection) was performed in 139 patients (48.8%), and 59 patients (20.7%) underwent sublobar resections. The resection margins in all cases were negative.
For the 233 patients with multiple lung adenocarcinoma, the 5-year DFS and OS rate was 59.6% and 82.4%, respectively. The univariate analysis showed that symptomatic disease (HR = 2.16, p = 0.011), larger maximal tumor size (p < 0.001), advanced pT (p < 0.001), lymph node involvement (p < 0.001), adjuvant chemotherapy (HR = 2.28, p = 0.008) as well as the subtype other than lepidic predominant (p < 0.001) were significantly associated with the shorter PFS (see Supplementary Fig. S3). The negative family history of cancer (HR = 3.73, p = 0.025), larger maximal tumor size (p = 0.027), advanced pT (p < 0.001), lymph node involvement (p = 0.005) and the subtype other than lepidic predominant (p < 0.001) significantly influenced the OS of patients with multiple lung ADC (see Supplementary Fig. S4). Adjusting for the above factors associated with PFS or OS, the multivariate analysis illustrated that only lymph node involvement (p < 0.001) and the subtype other than lepidic predominant (p = 0.002) were the independent prognosticators for the worse DFS and OS for patients with multiple lung ADC (Tables 5 and 6).

Discussion
In a retrospective series of 285 patients who underwent surgery for synchronous multiple primary NSCLC, the 5-year DFS and OS rate was 58.7% and 77.6%, respectively, which are comparable to previous reports [10][11][12][13][14][15] . Multivariate analysis identified symptomatic disease and lymph node involvement as the unfavorable prognostic variables for DFS, while symptomatic disease, lymph node involvement as well as male gender were the independent poor prognosticators for OS in patients with NSCLC. For the 233 patients with multiple lung ADC, lymph node involvement and the subtype other than lepidic predominant were independent predictors for both DFS and OS.
Among the patients who underwent adjuvant chemotherapy treatment, 60 had lymph nodes metastases, while only 8 patients had positive lymph nodes in the non-treated group. The poor DFS observed in the adjuvant chemotherapy group may have been substantially influenced by lymph node involvement rather than treatment. Moreover, univariate analysis showed no difference in OS between the two groups, which are consistent with previous reports suggesting no benefit of postoperative adjuvant treatment 8,13,16,17 . Two groups reported that adjuvant treatment might be favorable in multivariate analyses. However, their conclusion was weakened due to a selection bias 18 or lack of statistical significant difference in survival 19 . These discrepancies may be attributable to differences in the selection criteria used in different studies, and thus stratified analyses are guaranteed to identify the influential factors. We further analyzed whether adjuvant chemotherapy provided a benefit in the subgroups with or without lymph node involvement. In the negative subgroup, the presence of adjuvant chemotherapy did not significantly affect the 5-year OS rate (85.6% vs 85.0%, p = 0.752). However, in the positive subgroup, patients who received postoperative adjuvant treatment had a much better 5-year OS rate (72.1% vs 0.0%, p = 0.012). Therefore, SMPLC patients with positive lymph node involvement may benefit from postoperative adjuvant chemotherapy treatment.
In the latest revision of T descriptors in the forthcoming 8 th TNM classification system for lung cancer, additional pulmonary tumor nodules in the same lobe or in different ipsilateral lobes continue will be categorized as T3 or T4, respectively 2 . However, in our present study, the 5-year OS rate in patients with tumors located in the same lobe was 71.8%, which is much better than the reported 52-56% of T3 stage 2 . Yu et al. compared survival between SMPLCs and matched-stage solitary primary lung cancers after surgical treatment, indicating that SMPLCs had an excellent and comparable surgical outcome, which was somewhat discordant with outcomes in T4 or M1 stage patients in the current TNM classification system 8 . In our study, the 5-year OS rate of patients with bilateral tumors was 82.7%, which indicated that it was inappropriate to classify the synchronous multiple primary lung cancers as T3, T4 or M1 stage in the current TNM system. In the present study, the location of simultaneous tumors was not a prognostic indicator of survival, regardless of the laterality or the distribution of the tumors within the lobe. Similar results have been reported previously, showing no difference in survival according to tumor distribution 8,14,16,17,[19][20][21][22] . Some studies have argued that having bilateral tumors seems a favorable prognostic indicator 6,18 . However, in the study reported by Trousse et al. 18 , more patients with unilateral advanced stage diseases were included in the analysis. Tanvetyanon et al. 6 proposed that bilateral tumors had better outcomes mainly based on the literature 23 , but in the referenced study, the authors admitted that they selected more lymph node negative subset in bilateral synchronous tumors, which contributed to the higher 5-year survival rate. Conversely, Ishikawa et al. claimed that bilateral tumors predicted poor outcomes 13 , because additional unfavorable factors such larger size of the second tumor and inadequate resection were included. Therefore, with regard to synchronous multiple primary NSCLCs, tumor distribution might not be significantly related to long term survival if selection bias could be strictly controlled.
Tanvetyanon and colleagues observed that adenocarcinoma was independently associated with better outcomes (p < 0.001) 24 , which is in agreement with our results. The 5-year DFS and OS rate for multiple lung ADCs or predominant ADCs in our study was 60.4% and 82.6%, respectively. Lymph node metastases, indicating a more aggressive disease, were confirmed in approximately one-third of the patients with predominant non-ADCs. Regardless of the diagnostic approaches (low-dose CT or chest radiography), lung cancer screening identified the majority of adenocarcinomas, and many cases were detected at the early stages 25 , allowing these patients to benefit from radical surgical resection. It is therefore reasonable that patients without symptomatic disease had better survival than their counterparts.
To the best of our knowledge, we have reported the largest series of surgical outcomes for synchronous multiple primary NSCLCs (Table 7). We found that the independent prognosticators included male gender, symptomatic disease, and lymph node involvement for patients with SMPLC, and lymph node involvement and subtype other than lepidic predominant for patients with multiple lung ADC, which were also identified in previous identical studies 13,16,18,20,26,27 . Although the Martini and Melamed clinicopathological criteria are widely accepted as a basis for modification and enrichment, remarkable variation is still observed in different reports. The explanations may include the arbitrary nature of modified criteria, selection bias, lack of comprehensive histological assessment or multidisciplinary team revision and an absence of genetic molecular analyses.
In our study, 35.8% of the patients had a family history of cancer in first degree relatives (parents, children and sibs). They had better OS, which has not been described in previous studies. Notwithstanding the fact that the relationship between family history and increased risk of lung cancer is widely accepted, especially in non-smokers 28 , strong evidence and convincing mechanisms for this association remain unidentified. A previous study from our institution reported that patients with a positive family history of lung cancer had a better prognosis (p = 0.015) 29 . Further investigations of synchronous multiple primary NSCLCs may be useful for identifying the genetic factors that predispose individuals to lung cancer.
The strength of our study includes the largest number of enrolled patients with synchronous multiple primary NSCLCs. Modified criteria in the present series enabled a more comprehensive data analysis. However, there are still some limitations in our study. First, it was inevitable that inherent selection bias would be present due to the nature of a single-institution, retrospective study. Second, we did not select patients who underwent incomplete or no surgery, which may have underpowered the ability to identify predictors associated with prognosis. Third, the difficulty we experienced in summarizing the precise TNM stage instead of the pT stage or pN stage in each patient led to an inability to determine the 5-year DFS and OS rate for each stage, which impeded a comparison of stage-matched counterparts. In summary, for the patients with synchronous multiple primary NSCLC, male gender, symptomatic disease and lymph node involvement were the independent unfavorable prognosticators. In addition, lymph node involvement and subtype other than lepidic predominant were the independent unfavorable prognosticators for patients with multiple lung ADC. Radical surgical resection is feasible and effective treatment for patients with synchronous multiple primary NSCLC, and the identified prognosticators will provide valuable clues for the postoperative management of this population.