According to the European Society for Medical Oncology (ESMO), the treatment for clinical Stage I, IIA, IIB (T3N0M0), or IIIA (T4N0M0) non-small cell lung cancer (NSCLC) is the anatomic surgical resection of the involved lobe together with systematic node dissection (1, 2). The resection status after surgery has been proven to be an important predictor of prognosis and has an impact on the choice of further treatments. The residual tumor (R) classification includes: R0 (no residual tumor), R1 (microscopic residual tumor), and R2 (macroscopic residual tumor) (3). The International Association for the Study of Lung Cancer redefined the resection status into the categories: complete resection, incomplete resection, and uncertain resection. The last one included all cases without microscopic disease on resection margins but with one of the following criteria: lymphadenectomy less rigorous than systematic or lobe-specific nodal dissection, positive highest mediastinal node removed, carcinoma in situ on the bronchial margin, or positive pleural lavage cytology (4). Few previous studies focused attention on the impact of metastases in the highest mediastinal lymph node (HMLN) removed on prognosis. Moreover, the definitions of HMLN varied among these studies, leading to differences in patient selection and survival analysis. IASLC considered the highest mediastinal lymph node among those resected (3, 5).

We retrospectively reviewed patients who underwent lobectomy and systematic lymph node dissection. In patients with mediastinal nodal metastases, we investigated those who were HMLN positive. To avoid bias of the differences in the dissection of the right and left mediastinal nodal stations, because of anatomical difference (6), and according to IASLC, we defined HMLN as the lowest numerically numbered station among resected lymph node stations. These patients were defined as R-uncertain (R-u). We aimed to evaluate the prognostic value of R-u, compared with R0, in a population of patients with mediastinal node metastases.

We selected 550 patients with clinical N0 lung cancer. The baseline characteristics are presented in Table 1. The mean age was 69.70 years (SD 8.2), 343 patients (62.4%) were male and 454 (82.5%) were current or past smokers. The patient’s distribution for pT and pN classifications were as follows: pT1 296 (53.8%), pT2 174 (31.6%), pT3 58 (10.6%), and pT4 22 (4.0%); pN0 426 (77.5%), pN1 56 (10.2%), and pN2 68 (12.3%). The pN2 classification was divided into subgroups: pN2a2 39 (57.4%); pN2a1, or skip metastases, 15 (22.1%); and pN2b 14 (20.6%). The most frequent histology was adenocarcinoma (442 patients, 80.4%). VATS was performed in 385 patients (70.0%). The mean number of resected lymph nodes was 19.50 (SD 13.1), while the mean number of lymph node ratio was 22.60 (SD 17.0). The right upper lobe was the most frequently affected lobe (36.4%) while the right middle lobe was the less affected one (3.6%). Regarding the R classification in the whole population, 31 patients (5.6%) were R-u while the remaining 519 (94.4%) were R0.

The association between R classification and clinical variables in the pN2 group is displayed in Table 2. The lymph node ratio (LNR) was evaluated as the ratio between positive lymph nodes and all resected lymph nodes. The incidence of metastases in the highest mediastinal lymph node was related to the pN2 subgroups (p < 0.001). No relation was found for pT classification (p-value = 0.60), histology (p-value = 0.94), number of resected lymph nodes (p-value = 0.31), lymph node ratio (p-value = 0.18), affected lobe (p-value = 0.42), and tumor diameter (p-value = 0.62).

Table 3 shows the topographic distribution of lymph node metastasis for each affected lobe. In the right upper lobe, station 2 was the most frequent highest positive lymph node station (53.8%), while station 4 was the most frequent (100%) for the right middle lobe. In the right lower lobe, the highest positive station was found in station 7 (50.0%). Station 4 was the most common (50.0%) for the left upper lobe. In the left lower lobe, station 5 and station 7 were equally common (50.0%).

The median follow-up time was 33.9 months (SD 14.8). Postoperative survival analysis in patients with N2 disease, comparing R0 and R-u resections, is shown in Table 4. Kaplan–Meier curves are illustrated in Figure 1. The recurrence rate was 29.7% (11/37) in the R0 group, while in the R-u group it was 71.0% (22/31); 3-year DFS was 69.0% (mean time to relapse of 15.5 ± 7) and 20.0% (mean time to relapse of 9.9 ± 6.6), respectively. The mortality rate was 18.9% (7/37) in the R0 group, while in the R-u group it was 51.6% (16/31); 3-years OS was 78.0% (mean time to relapse of 23.1 ± 6.) and 40.0% (mean time to relapse of 12.7 ± 8.2), respectively.

The univariate analysis for DFS and OS, in the N2 population, included the variable listed in Table 5. The R-u variable showed a tendency to be a significant prognostic factor. The hazard ratio (HR) for DFS and OS of the R-u group was higher if compared with the R0 group [DFS: HR 4.6 (95% CI 2.2–9.6), p-value < 0.001; OS: HR 4.5 (95% CI 1.8–10.9), p-value < 0.001]. The multivariable analysis was evaluated using variables with a significant p-value at univariate analyses (Table 6). R-us remained a significant prognostic factor for DFS and OS [DFS: HR 3.2 (95% CI 1.4–7.4), p-value = 0.008; OS: HR 2.0 (95% CI 0.7–6.1), p-value < 0.001].

The univariate analysis was also carried out in the R-u group (Table 7). We consider lymph node macro-metastases when the metastatic part is bigger than 2 mm. Variables that showed a tendency to be significant prognostic factors were pT [OS: HR 1.9 (95% CI 1.1–3.6), p-value = 0.02], pN2 subgroups [DFS: HR 1.7 (95% CI 0.9–2.9), p-value = 0.04; OS: HR 1.9 (95% CI 1.0–3.9), p-value = 0.03], number of resected lymph nodes [DFS: HR 0.9 (95% CI 0.9–1.0), p-value = 0.03], lymph node ratio [DFS: HR 1.0 (95% CI 1.0–1.0), p-value = 0.04], number of the positive lymph node in the higher station [DFS: HR 0.4 (95% CI 0.2–1.1), p-value = 0.03], macro-metastases [DFS: HR 30.5 (95% CI 3.9–240.0), p-value < 0.001; OS: HR 3.9 (95% CI 1.1–14.0), p-value = 0.01], and tumor diameter at CT scan [DFS: HR 1.0 (95% CI 1.0–1.1), p-value = 0.02]. In multivariable analysis, evaluated using the variables that had a significant p-value at univariate analyses, variables that confirmed to be significant prognostic factors were pT [OS: HR 1.9 (95% CI 0.9–3.7), p-value = 0.04] and macro-metastases (DFS: HR 28.8 (95% CI 3.5–239.5), p-value = 0.002; OS: HR 3.5 (95% CI 0.9–12.6), p-value = 0.05] (Table 8).

It is known that the diffusion of metastatic cells through the lymphatic pathway generally follows a specific pattern: intrapulmonary nodes, hilar nodes (N1), mediastinal nodes (N2) in the caudal–cranial direction, final to the supraclavicular (N3) nodes and distant organs (12). As previous studies evaluated, the pN2 groups are heterogeneous with regard to prognosis. The difference could come from the number of involved lymph nodes and stations and the specific patterns of lymphatic spread (13, 14).

At the time of surgery, the highest mediastinal lymph node resected represents the margin of cancer dissemination. Therefore, a metastasis in this station could be considered as a positive margin. The rationale lies in the possibility of cranial lymph node involvement or distant micro-metastases. Thereby, the involvement of cervical nodes or a more extensive mediastinal involvement may be underestimated. The finding of metastasis in the highest mediastinal lymph node dissected might be an important parameter to differentiate this subgroup of patients with a poor prognosis.

Previous studies investigated the difference in prognosis between complete and uncertain resection. In the study by Zheng et al., R0 and R-u 5-year survival rate and median survival time were 29% and 36.48 months vs. 13% and 24.43 months, respectively (p < 0.0001) (6). Osarogiagbon et al. found similar results: mean OS in R0 was 62 months, while in R-u it was 32 months (p-value < 0.0001) (15). In the literature, few studies paid attention to the prognostic impact of the highest mediastinal lymph node metastases and, as previously said, the definition of HMLN varied among studies. Gagliasso et al. used the IASLC definition of HMLN, so they found a better 5-year survival rate compared to less rigorous lymph node evaluation and carcinoma in situ in the bronchial margin (28.8% vs. 44.2% and 40.0%, respectively) (16).

Ren et al., following the IASLC definition, considered being HMLN positive as an independent risk factor for DFS and OS: among patients with N2 metastases, those with positive HMLN had significantly worse survival compared to R0 (DFS: 36 vs. 44 months, p < 0.001; OS: 50 vs. 59 months, p < 0.001) (17).

Sakao et al. defined HMLN as nodes lying above a horizontal line at the upper rim of the left innominate vein. They found that, in patients with advanced N2 disease, patients with positive HMLN had a 5-year survival rate of 21.0% compared to 52.0% of negative HMLN. Furthermore, patients with negative HMLN, even if they have multilevel N2 status, positive cN status, or T2–3 tumor status, had a better prognosis (18).

Two studies found no prognostic difference regarding survival between complete and incomplete resection. Both studies used a stricter definition of HMLN: for the right side they considered 2R, and for the left side 4l, 5, or 6. In these studies, the metastasis on HMLN did not show survival differences in completely resected N2 NSCLC (19, 20).

The results of our study support the idea that the presence of metastases in the highest mediastinal lymph node among those resected is a negative prognostic factor for DFS and OS. R-u had a higher recurrence and mortality rate compared to R0.

This study has some limitations. First, the retrospective nature of the study and the small cohort of patients may have affected the validity of the study. The category of R-u is wide; we focused only on the metastases on the highest mediastinal lymph nodes. Other studies should be done to compare the prognostic value of these subgroups and to evaluate the necessity of a specific adjuvant therapy. Future prospective for developing a preoperative evaluation of the lymphatic pathway and the sentinel node might be useful (21). So far, we suggest a systematic lymph node dissection for a better staging of the mediastinal lymph node status.

The group of pN2 is not homogeneous from a prognostic point of view. R classification proposed by IASLC showed a significant improvement in survival discrimination. R-u delineates the crossing area between complete resection (R0) and incomplete resection (R1 and R2). The R-u for the presence of metastasis in the HMLN removed seems to be an independent prognostic factor for mortality and recurrence. The definition of HMLN varied among studies, whereby standardization of the definition was needed. The finding of metastasis in the HMLN represents the margin of cancer dissemination at the time of surgery, so it could imply metastasis into the N3 node or distant metastasis.