Introduction

Lung cancer is the second most common cancer among both males and females. The American Cancer Society estimated that 116,990 men and 105,510 women developed new lung and bronchus cancers in 2017, with 84,590 men and 71,280 women dying from them [1]. The estimated numbers of new cases and deaths were comparable to those in 2014 [2]. In addition, Bray et al. have reported that lung cancer was the leading cause of cancer mortality for women in 17 countries and for men in 94 countries [3]. Non-small cell lung cancer (NSCLC) and small cell lung cancer are the two major histological variants that account for 90% of lung cancers, with the major variants of NSCLC being adenocarcinoma (AD) and squamous cell carcinoma (SQCC). Lung cancer patients have a low 5-year survival rate [4] and approximately 40% of new NSCLC cases have distant metastases at diagnosis [5].

The eighth edition of the lung cancer staging system developed by the International Association for the Study of Lung Cancer (IASLC) has significant revisions regarding distant metastasis, while the lymph node (N) descriptors remained unchanged [6]. The 8th edition defines M1a as intrathoracic metastases, M1b as a single extra-thoracic metastasis, and M1c as multiple extra-thoracic metastases. For patients with distant metastasis, tumor, node, and metastasis (TNM) stage is determined only by M stage. IVA stage includes any T, any N, M1a and M1b; IVB stage includes any T, any N, and M1c. Previous studies have also demonstrated that multiorgan metastasis predict inferior survival in M1 NSCLC [7, 8]. It has been revealed that lymph node metastasis might have prognostic value in M1a NSCLC [9], and N2 disease is associated with poor survival in oligometastatic NSCLC [10], but the impact of N status on the outcome of NSCLC patients with specific organ metastasis, especially multiorgan metastases (MOM) remains to be determined. Identifying factors with prognostic value may contribute to management of disease [11], thus determining effective predictors of survival especially in late stage NSCLC is of great importance.

In the present study, we aimed to investigate the relationship between lymph node metastasis and distant organ metastasis and how N stage affects the outcome of M1 disease in NSCLC patients using surveillance, epidemiology, and end results (SEER) database.

Methods

Patients and clinicopathologic data

This retrospective study assessed (1) the association of N stage with distant metastases and (2) the prognostic value of N stage among patients with M1 NSCLC. Data for a large number of NSCLC patients were obtained from the SEER database, which was launched in 1973 by the National Cancer Institute and consists of 18 population-based cancer registries, covering approximately 28% of the total population of the United States.

The present study included NSCLC patients with M1 disease who were diagnosed between 2010 and 2013 and had complete information. We initially identified 90,817 cases of lung cancer, 20,576 cases with two or more primary sites, 6918 with incomplete information on organs metastases, 55 cases with no information on survival, 17,845 cases with small cell lung cancer, and 2810 patients with incomplete N stage and histological information. Thus, a total of 42,613 NSCLC patients were eligible analyses, with 26,457 AD cases and 8637 SQCC cases. All patients had distant metastasis at diagnosis and complete information regarding N stage. TNM stage information in the present SEER database was based on the 7th TNM staging system. N stage is defined as follows: N0 stage as no regional lymph node metastasis, N1 stage as metastasis in the ipsilateral peribronchial and/or ipsilateral hilar lymph nodes and intrapulmonary nodes, including involvement by direct extension, N2 stage as metastasis in the ipsilateral mediastinal and/or subcarinal lymph node(s), and N3 stage as metastasis in the contralateral mediastinal, contralateral hilar, ipsilateral, contralateral scalene, or supraclavicular lymph node(s). M1 is classified as two stages, M1a: presence of a malignant pleural or pericardial effusion, pleural dissemination, or pericardial disease, and metastasis in opposite lung; M1b: extrathoracic metastases. Multiorgan metastases refers to patients with two or more metastatic organs among lung, liver, brain and bone as previously described [7]. The endpoint of the present study was lung cancer specific survival, which was defined as the interval from diagnosis until death due to lung cancer or until the last follow-up.

Statistical analyses

Patient characteristics were statistically described. Pearson’s Chi square test was used to detect differences for categorical variables. Differences in NSCLC-specific survival time were compared among patients of different N stage using Kaplan–Meier analysis and the log-rank test. Cox regression analysis was used to investigate the predictors of survival, and binary logistic regression analysis was performed to calculate the odds ratios (ORs) of experiencing various organ metastases. Differences were considered statistically significant at two-tailed p-values of less than 0.05. All analyses were performed using IBM SPSS software (version18.0; IBM Corp., Armonk, NY).

Results

Patient characteristics

Table 1 shows the clinical characteristics of total populations and subtypes, respectively. A total of 42,613 NSCLC patients (23,642 male and 18,971 female) were analyzed. The histological type AD had 26,475 cases and SQCC had 8637 cases. The median age for the overall population was 67 years [interquartile range (IQR): 59–76 years] and the median follow-up time was 4 months (IQR 1–10 months). There were 25.9% (11,057 cases) patients diagnosed at N0 stage, 8.1% (3443 cases) at N1 stage, 46.0% for N2 stage (19,609 cases), and 20.0% for N3 stage (8504 cases). The primary tumors were more common on the right side (54.9% overall, with 28.4% on the right upper lobe) and less common on the left side (39.1% overall) (Table 2).

Table 1 Characteristics of overall and NSCLC subtypes with distant metastasis (SEER database 2010–2013, n = 42,613)
Full size table
Table 2 The location of NSCLC and its common subtypes (SEER database 2010–2013, n = 42,613)
Full size table

Association between N stage and M stage in M1 NSCLC patients

Table 3 shows that in NSCLC patients, the proportion of N0 in M1a was significantly higher than that in M1b, 34.0% and 22.7% respectively (P < 0.001). Interestingly, the proportions of patients with N1, N2 and N3 were all lower in M1a NSCLC patients than that in M1b (P < 0.001). In the subgroup analysis, similar results were obtained in AD patients, with N0 rate as high as 36.8% in M1a disease. While in SQCC patients, N0 rate was also higher in M1a group although the difference between M1a and M1b was relatively smaller compared with AD patients, and the N2 proportion between M1a and M1b group failed to reach statistical difference.

Table 3 Association between N stage and M stage among M1 NSCLC and its subtypes (SEER database 2010–2013, n = 42,613)
Full size table

Association between N stage and MOM

The proportion of patients diagnosed at stage N0, N1, N2, and N3 who developed MOM was 18.7%, 25.0%, 29.3%, and 33.6%, respectively (Table 4). Compared with the N0 group, the odds of having MOM were higher for patients diagnosed at stage N1 (OR 1.45 [1.33–1.59], P < 0.001), N2 (OR 1.80 [1.70–1.90], P < 0.001), and N3 (OR 2.19 [2.06–2.34], P < 0.001). After adjusting for age, sex, and tumor size, the ORs remained significant in N1 group (OR 1.30 [1.17–1.44]), N2 group (OR 1.70 [1.59–1.81]), and N3 group (OR 1.96 [1.81–2.11]) (all P < 0.001). Similar results were observed in the AD and SQCC subgroups (Table 4).

Table 4 Association between N stage and MOM (SEER database years of 2010–2013)
Full size table

Association between N stage and disease-specific survival

The effects of N stage on disease-specific survival among overall NSCLC patients and two histological subtypes are shown in Figs. 1, 2 and 3. Survival decreased with increasing N stage (P < 0.001), with the best survival observed in the N0 group, intermediate survival observed in N1 group, and similarly poor survival observed in both N2 and N3 groups (N3 vs. N2, P = 0.511) (Fig. 1). Similar results were observed when conducting analyses in AD and SQCC subgroups (Fig. 2).

Fig. 1
figure 1

Kaplan–Meier analysis of disease specific survival according to N stage in non-small cell lung cancer

Full size image
Fig. 2
figure 2

Kaplan–Meier analysis of disease specific survival according to N stage in a adenocarcinoma and b squamous carcinoma

Full size image
Fig. 3
figure 3

Kaplan–Meier analysis of disease specific survival according to N stage in a M1a and b M1b non-small cell lung cancer

Full size image

Further analysis showed that in M1a patients, N0 stage was associated with better outcome compared with higher N stage, with 1- and 3-year survival rates as high as 42.3% and 19.3%, respectively (P < 0.001) (Table 5) (Fig. 3a). Similar results were obtained in M1b disease (Table 5) (Fig. 3b). M1a patients had better survival than M1b patients, but the difference between N3M1a and N0M1b was not significant (P = 0.574) (Table 5). In the subgroup analysis for the M1b group, better survival was observed in N0 stage compared with N2 or N3 stage if patients had bone metastasis only, brain metastasis only, or MOM (P < 0.001, P < 0.001, and P = 0.002, respectively), but there was no significant difference in survival among each N stage patients with liver metastasis only (Fig. 4).

Table 5 Kaplan–Meier analysis of the disease specific survival in M1 NSCLC patients (SEER database years of 2010–2013)
Full size table
Fig. 4
figure 4

Kaplan–Meier analysis of disease specific survival according to N stage in M1b non-small cell lung cancer with a bone metastasis, b brain metastasis, c liver metastasis, and d multiorgan metastases

Full size image

Cox regression analysis was performed to investigate the effect of N stage on disease-specific mortality risk of M1 NSCLC patients. Compared with patients diagnosed at N0 stage, patients diagnosed at N1 stage had higher hazard for crude mortality (HR 1.12 [1.07–1.17], P < 0.001) and for age, sex, M stage, and tumor size adjusted mortality (HR 1.13 [1.07–1.19], P < 0.001). For crude mortality, higher hazards were also observed for patients diagnosed at N2 stage (HR 1.23 [1.20–1.27], P < 0.001) and N3 stage (HR 1.23 [1.19–1.27], P < 0.001). Furthermore, for age, sex, M stage, and tumor size adjusted mortality, N2 group (HR 1.26 [1.22–1.30], P < 0.001) and N3 group (HR 1.29 [1.23–1.34], P < 0.001) had higher hazards compared to N0 group. Similar results were observed for the AD and SQCC subgroups (Table 6).

Table 6 Association between N stage and cancer-specific mortality among M1 NSCLC and its subtypes (SEER database 2010–2013, n = 42,613)
Full size table

Discussion

In the newly developed 8th edition American Joint Committee on Cancer (AJCC) TNM staging manual for NSCLC, T descriptor and M descriptor were both revised, while N descriptors remained unchanged. Lymph node involvement is an important demarcation criterion for the staging of M0 patients, whereas stage IV is classified as IVA and IVB only based on the M stage, without concern of N status. It is well accepted that lymph node metastasis is a key predictor of outcome in M0 NSCLC [12], but few studies have focused on whether N status plays an important role in M1 patients [9, 10]. This study evaluated the relationships between lymph node metastasis and distant metastasis in NSCLC patients and further identified the prognostic value of N stage in M1 patients.

Our primary finding is that patients with M1a (intra-thoracic metastasis) NSCLC trends to have lower N stage compared with M1b (distant metastasis) patients, as N1, N2, and N3 rates in M1b were all significantly higher than those in M1a (Table 3). We observed that nearly 34% of M1a and 22.7% of M1b patients had no lymph node involvement in our study, which could be explained by that blood vessel dissemination is responsible for distant metastasis of cancer [13]. Nevertheless, we observed that lymph node metastasis is associated with higher risk of MOM in NSCLC patients. It has been a subject of considerable debate whether cancer cells in lymph nodes can seed distant organs. It is possible that primary tumors may remain noninvasive until somehow triggered to metastasize via both lymphatics and blood vessels simultaneously [14]. As lymph nodes are often the first site of metastasis, it is also possible that lymphatics serve as part of pathway contributing to the subsequent dissemination to distant organs [15]. Experimental evidence has demonstrated that inhibiting lymphatic dissemination could be a promising method for preventing distant metastases [16,17,18]. The exact mechanism of how lymphatic spread leads to systematic metastasis remains unclear, most likely through the left lymphatic duct (thoracic duct) or the right lymphatic duct and subsequently into the subclavian veins [14]. Recent studies revealed that lymph node metastases can disseminate by invading lymph node blood vessels rather than by transiting through efferent lymphatic vessels in a mouse model, but whether this form of tumor cell spreading occurs in cancer patients remains to be determined [19, 20]. Our findings also shed light on the important role of lymphatic route in distant metastasis.

Lymph node status is a powerful prognostic predictor as many studies have demonstrated that lymphatic metastasis is an independent prognostic factor for NSCLC [21, 22]. Lymphatic metastasis after surgical resection of NSCLC has been shown to be the most important determinant of long-term outcome [21], and even lymphatic micrometastasis could predict poor prognosis [23]. In the present study, earlier N stage was associated with better disease-specific survival, based on the result of Kaplan–Meier analysis that N0 patients had the best survival while N2 and N3 was associated with worse survival. Analyses of the current N descriptors at clinical and pathological staging in the IASLC database showed that they clearly separate lung cancers of significantly different prognosis [6]. In this cohort of patients, the 3-year survival rates were significantly lower than AJCC guidelines estimates due to the research design that our study only included patients with M1 disease at diagnosis. Despite the difference and limitation, our findings did highlight a similar trend to AJCC guidelines’ estimates, that the survival was best for patients diagnosed at N0 stage and poorest at N2 or N3 stage. Similar trends were noticed in the sub-group analysis of M1a and M1b disease. In addition, our results showed that although M1a disease was associated with better survival compared with M1b, N3M1a had equal survival as N0M1b. To exclude the impact of other prognostic factors, Cox regression analysis was employed to confirm the effect of N status in NSCLC and its two pathological subtypes AD and SQCC, showing that increasing N status was associated with higher risk of mortality after adjusted for age, sex, tumor size and M stage. These findings provide evidence for the prognostic value of lymph node metastasis in M1 NSCLC, and may be relevant for the next version of TNM staging system.

Interestingly, in the subgroup analysis we found that the effect of lymph node metastasis was significantly attenuated in liver metastasis only group, which might be related to the strong negative effect of liver metastasis as previous studies have reported [8, 24, 25]. The exact mechanism why liver metastasis is associated with poorer survival is still not clear. More research will be necessary to study the molecular mechanisms of liver metastasis in lung cancer in order to devise effective treatment strategies. In the bone metastasis only group, brain metastasis only group, and MOM group, the best survival was observed in N0 patients, and N2–3 was associated with worse survival, while the difference between N0 and N1 group did not reach statistical significance. Although bone metastasis, brain metastasis, and MOM were associated with poor survival, our results showed that lymph node metastasis still had prognostic value and should be considered in the management of disease.

This study has several limitations. Firstly, metastases were identified based on the 7th edition of the AJCC guidelines, which did not provide information on the number of metastasis in the metastatic organ. Secondly, metastasis information was only available for the four common metastatic organs (lung, bone, brain, and liver) starting from 2010. This limited the data and population that could be included and omitted the influence of patients with metastases to other organs. In addition, the SEER database did not provide details regarding the methods used to determine N stage, and information such as comorbidities, treatment details, and gene mutations were also not available, which prevented us from further stratification analysis. Therefore, more research will be needed to confirm our findings.

Conclusion

In conclusion, these results reveal lymph node metastasis is associated with higher risk of MOM in M1 NSCLC. Our study also provides evidence that lymph node status has important prognostic value in patients with M1 NSCLC and may help guide the next TNM staging system. Knowledge of these findings may be helpful when making precise prognostic evaluation and treatment decisions for NSCLC patients.