Skip Mediastinal Lymph Node Metastasis and Lung Cancer: A Particular N2 Subgroup With a Better Prognosis

Skip Mediastinal Lymph Node Metastasis and Lung Cancer: A Particular N2 Subgroup With a Better Prognosis Marc Riquet, MD, Jalal Assouad, MD, Patrick Bagan, MD, Christophe Foucault, MD, Françoise Le Pimpec Barthes, MD, Antoine Dujon, MD, and Claire Danel, MD Departments of Thoracic Surgery and Pathology, Surgical Center, Boisguillaume and Georges Pompidou European Hospital, Paris, France Background. Lymph node (LN) metastases from lung cancer may skip the intrapulmonary nodes directly to the mediastinum ([N1 ]N2 vs [N1 ]N2). This phenomenon is frequent. Patients with such a metastatic pattern appear to have a better prognosis following surgery. Our purpose was to further study the clinical significance and prognostic value of this particular group of (N1 )N2 patients. Methods. We retrospectively analyzed the data of 731 patients with a pn2 stage who underwent resection for non-small cell lung cancer. Patients with (N1 )N2 metastases (n 209) were compared to patients with intrapulmonary (N1 )N2 (n 522). Results. In the (N1 )N2 group, lobectomies were more frequent (54% vs 33%, p 0.00), metastases more frequently involved a single LN station (79.4% vs 56.3%, p < 0.000001), and primary tumor was more often located in the upper lobes (67.4% vs 55.6%, p 0.0066). (N1 )N2 was a factor of better prognosis (5 year survival rates 34.4% vs 18.5%, p 0.00006), which proved also significant when only a single station was involved (38.4% vs 24%, p 0.0005). These results were confirmed by multivariate analysis. Conclusions. (N1 )N2 skip metastasis is a unique subgroup of pn2 disease. Lung lymph drainage anatomy may explain the occurrence of these metastases. They form an independent prognostic factor of survival suggesting the need for further study, the results of which may lead to better knowledge of lung cancer, improved classification, and adapted adjuvant therapy. (Ann Thorac Surg 2005;79:225 33) 2005 by The Society of Thoracic Surgeons The international staging system for non-small cell lung cancer (NSCLC) [1] defines regional lymph node (LN) involvement as N1: metastasis to LN in the peribronchial or the ipsilateral hilar region or both, including direct extension; and as N2: metastasis to ipsilateral mediastinal LN (MLN) and subcarinal LN. N2 disease significantly worsens the prognosis [1, 2]. In 1971, Kirsh and colleagues [3] noted the absence of N1 disease in patients with N2 metastases. In 1986, Libshitz and colleagues [4] focused a study on this entity. The entity they called skip phenomenon was present in 29.2% (14 of 48) of N2 patients, representing 7% (14 of 200) of all patients undergoing resection for NSCLC. This condition of skip LN metastasis consisting of N2 disease without N1 involvement appeared to be more frequent in case of adenocarcinomas (10 of 14 patients). Other studies have reported this N2 subgroup to be more frequent in case of upper lobe NSCLC (43 of 62 patients) [5] and to consist most often in the involvement of only one MLN station (30 of 33 patients) [6]. This last study [6] and more recent studies [7 12] suggest that skip MLN metastases (N1 ) tend to have a better prognosis in NSCLC than when Accepted for publication June 16, 2004. Address reprint requests to Dr Riquet, Service de Chirurgie Thoracique, Hôpital Européen Georges Pompidou, 20 Rue Leblanc 75015 Paris, France; e-mail: marc.riquet@hop.egp.ap-hop-paris.fr. associated with N1 involvement. The purpose of this study was to further elucidate the clinical significance and prognostic value of skip MLN metastases by comparing 209 skip (N1 )N2 patients with 522 (N1 )N2 patients operated for NSCLC during the same period. Material and Methods Patients Between 1984 and 2001 a total of 2,863 patients underwent pulmonary surgical resection for NSCLC at Georges Pompidou European Hospital and at Boisguillaume surgical center. The surgical procedure was a potentially curative complete resection in combination with hilar and mediastinal lymphadenectomy. Patients with typical carcinoid tumor and patients who had preoperative neoadjuvant chemotherapy or radiation therapy or who had palliative surgery were excluded from the study. Pathological N classification was N0 54.1% (n 1,548), N1 20.4% (n 584), and N2 25.5% (n 731). The 731 pn2 patients form the basis of this study. Patient s ages ranged from 30 to 88 years (mean 61 years). The group included 598 males and 133 females. Skip MLN metastases (N1 )N2 were demonstrated in 209 patients (28.6%) and MLN metastases associated with N1 disease (N1 )N2 in 522 patients (71.4%). Postoperative adjuvant 2005 by The Society of Thoracic Surgeons 0003-4975/05/$30.00 Published by Elsevier Inc doi:10.1016/j.athoracsur.2004.06.081

226 RIQUET ET AL Ann Thorac Surg SKIP MEDIASTINAL LN METASTASIS 2005;79:225 33 therapy was performed in 515 patients in a nonrandomized fashion according to specific management adopted by the different referring physicians: 405 patients received adjuvant radiation therapy and 110 patients received both radiation and chemotherapy or chemotherapy alone. The type of pn2 involvement was not taken into consideration in the indication for adjuvant therapy. Method (N1 )N2 patients were compared with (N1 )N2 patients in terms of the following: some characteristics (clinical N estimation, past history of cancer, type of resection performed, histology, and adjuvant therapy); the location and site of the primary tumor as well as the site and extent of MLN metastasis; survival (overall survival rates, survival rates according to the number and location of MLN stations involved, histology, and adjuvant therapy); and cause of death (systemic metastasis, local recurrence, and others). For MLN involvement the classification of Mountain and Dresler [2] was used. The N2 population was divided in single mediastinal LN (SMLN) station N2, which is N2 involving a single station either in superior mediastinal nodes (2R 4R or 3 or 4L) or aortic nodes (5 or 6) or inferior mediastinal nodes (7 or 8 or 9) and dual mediastinal LN (DMLN) station N2, which is the involvement of two or more of any combination of the above mentioned stations. The 2R and 4R LN stations were grouped together because they form the same anatomical LN chain [13]. Follow-up information was obtained either from hospital case records, or a questionnaire completed by the local chest physician or general practitioner, or from death certificates. Zero time was considered as the date of surgery and January 2003 was the closing date of the study. Univariate analysis was conducted between both populations. Actuarial survival curves were calculated by Kaplan-Meier method; statistical comparisons were made using the log-rank test. Multivariate analysis was performed using the Cox proportional hazards model. Multivariate analysis included gender, age, side of the tumor, type of surgical resection performed, histology, T, tumor size, type of N2, and adjuvant therapy. All data analyses were conducted with the two-sided test: a p value less than 0.05 was considered as statistically significant. Standard error of the mean software was used for calculation [14]. Results Characteristics Characteristics of both (N1 )N2 and (N1 )N2 populations are reported in Table 1. No statistical difference was observed concerning age, gender, past oncologic history, clinic N status, or adjuvant treatment. Pneumonectomy was performed more frequently in the case of (N1 )N2 patients (67% vs 46%, p 0.001). Postoperative complications occurred in 21% (44 of 209) and deaths in 4.3% (9 of 209) of (N1 )N2 patients and in 22.6% (118 of 521) and 6.9% (36 of 521) of (N1 )N2 patients (p 0.19). Following pneumonectomy, death occurred in 8.3% of (N1 )N2 patients and 8.3% of (N1 )N2 patients. Following lobectomy, death occurred in 0.9% of (N1 )N2 patients and in 3.6% of (N1 )N2 patients (p 0.33). Death complicated 2.6% of lobectomies (7 of 272 patients) and 8.3% of pneumonectomies (37 of 446 patients) (p 0.0019). Histology was not different between the (N1 )N2 and (N1 )N2 patient groups. However, when only SMLN stations subgroups are considered, adenocarcinomas and large cell lung cancers were more frequent in (N1 )N2 patients and squamous cell carcinomas in (N1 )N2 patients. No difference was observed among the groups concerning T status. When studying T1 and T2 together ((N1 )N2 [n 162] and (N1 )N2 [n 412]) to test visceral pleural involvement (VPI or ), there was also no difference: (N1 )N2 VPI 25.9% (42 of 112 patients) and (N1 )N2 VPI 30.6% (126 of 412 patients). (N1 )N2 was more frequent when tumor size exceeded 5 cm in diameter (p 0.03). Site and Extent of MLN Metastases Tumor location and MLN involvement are reported in Table 2. (N1 )N2 was observed more frequently on the right (57.9% vs 50.9%, p 0.08). (N1 )N2 was more frequent in patients with primary tumors in the upper lobes (p 0.0036). Among the (N1 )N2 patients, 79.4% exhibited metastases in a SMLN station and 20.8% in DMLN stations: the frequency of a SMLN station involvement was significantly less among (N1 )N2 patients (56%; p 0.000001). The SMLN station involvement, according to the location of the primary tumor, is reported in Table 3: 8 out of 62 (12.9%) right upper lobe tumors metastasized into the inferior mediastinum LN station 7, and 3 out of 41 (7.3%) left upper lobe tumors metastasized into the inferior mediastinum LN station 7 and 9. Inversely, 4 out of 31 (12.9%) right lower lobe tumors metastasized into the superior mediastinum LN station 4R, and 9 out of 15 (60%) left lower lobe tumors metastasized into superior mediastinum LN station, 5 and 4L. Inferior mediastinal LN station 7 involvement was present in 28.3% (47 of 166) of (N1 )N2 patients and in 39.8% (117 of 294) of (N1 )N2 patients when only a SMLN station was involved, and respectively, in 58.1% (25 of 43 patients) and 74.6% (170 of 228 patients) when dual station was involved. Survival Survival of the entire population including both (N1 )N2 and (N1 )N2 disease is reported in Table 4. Gender, histology, side of the primary tumor, and previous oncologic history were not factors of poor prognosis. Factors of significant improved prognosis were lobectomy versus pneumonectomy, SMLN station involvement, clinical N0 disease, and adjuvant therapy with chemotherapy. p(n1 )N2 disease was a factor of better prognosis as

Ann Thorac Surg RIQUET ET AL 2005;79:225 33 SKIP MEDIASTINAL LN METASTASIS 227 Table 1. Main Characteristics of (N1 )N2 and (N1 )N2 Populations Characteristics (N1 )N2 (Skip ) n 209 (N1 )N2 (Skip ) n 522 p Value Age mean 60.3 10.29 61.23 10.33 Not significant range 27 84 30 84 Sex male/female 175/34 423/99 0.39 Histology Squamous 78 (37.3%) 231 (44.2%) Adenocarcinoma 84 (40.2%) 204 (39.1%) Large cell 26 (12.4%) 39 (7.5%) 0.16 Other 8 (3.8%) 23 (4.4%) Neuroendocrine 13 (6.2%) 25 (4.8%) Histology single station Squamous 61 (36.7%) 149 (50.7%) Adenocarcinoma 69 (41.6%) 98 (33.3%) Large cell 22 (13.3%) 22 (7.5%) 0.017 Other 14 (8.4%) 25 (8.5%) Histology dual stations Squamous 17 (39.5%) 82 (36.0%) Adenocarcinoma 15 (34.9%) 106 (46.5%) 0.19 Large cell 4 (9.3%) 7 (3.0%) Other 7 (16.3%) 33 (14.5%) Past oncological history (yes/no) 30/179 (34.9%) 56/466 (27.8%) 0.17 Clinical N2 (a) 136/69 (66.3%) 373/142 (72.4%) 0.11 Pneumonectomy 96 (45.9%) 350 (67.0%) Lobectomy-segmentectomy 113 (54.1%) 172 (33.0%) 0.00000012 T factor T1 50 (23.9%) 90 (17.2%) T2 112 (53.6%) 322 (61.9%) T3 41 (19.6%) 89 (17.0%) 0.1 T4 6 (2.9%) 21 (4.0%) Size 3 cm 51 (24.4%) 173 (26.2%) 3cm S 5 cm 70 (35.5%) 216 (41.4%) 0.0079 5 cm 88 (42.1%) 169 (32.4%) Adjuvant treatment No T 69 (33.2%) 142 (27.4%) CT 12 (15.9%) 26 (14.9%) 0.22 RCT 21 51 RT 106 (50.9%) 299 (57.7%) a Data not available: n 11. CT chemotherapy; not no adjuvant treatment; RT radiation therapy; RCT radiation and chemotherapy; S size. compared to (N1 )N2 disease (5 year survival rates 34.4% vs 18.5%, p 0.00006) (Fig 1). This was also the case when only a SMLN station was involved (38.4% vs 24%, p 0.0005) (Fig 2), whereas prognosis was similar in case of DMLN station involvement (Fig 3). (N1 )N2 DMLN station involvement and (N1 )N2 SMLN station involvement proved to have similar prognostic value (24.1% vs 20.7%, p 0.29). Differences in survival between (N1 )N2 and (N1 )N2 involvement with respect to gender, histology, location, and site of the tumor, clinical N2 or surgical N2, adjuvant therapy, and past oncologic history had no prognostic significance. Differences were significant only when considering T status (T1 vs T3 T4) and type of resection (lobectomy vs pneumonectomy), and remained significant when only (N1 )N2 SMLN station involvement was considered (respectively, p 0.02 and 0.018). In case of (N1 )N2 SMLN station involvement, the stations involved (4R, 5, 4L, 9) had the similar prognostic value: 4R (n 58) 5 year survival, 38.6%; 9 (n 45), 35.5%; 5 (n 32), 36.3%; 4L (n 16), 55%.

228 RIQUET ET AL Ann Thorac Surg SKIP MEDIASTINAL LN METASTASIS 2005;79:225 33 Table 2. Tumor Location and Mediastinal Lymph Node Involvement TL (N1 )N2 (skip ) n 209 (N1 )N2 (skip ) n 522 p Value Right/left 121/88 (57.9%/42.1%) 265/257 (50.8%/49.2%) 0.081 Lobe Upper lobe 133 (69.3%) 266 (57.1%) 0.0036 Lower lobe 59 (30.7%) 200 (42.9%) Right upper lobe 69 (57.0%) 124 (46.8%) Middle lobe 9 (7.4%) 13 (4.9%) 0.082 Right lower lobe 37 (30.5%) 101 (38.1%) Right main bronchus 6 (4.9%) 27 (10.1%) Left upper lobe 64 (72.2%) 142 (55.3%) Left lower lobe 22 (25.0%) 99 (38.5%) 0.013 Left main bronchus 2 (2.3%) 16 (6.3%) Station involved (Single/dual) 166/43 (79.4%) 294/228 (56.3%) 0.0000001 TL tumor location within the lung. Cause of Death Death occurred in 78.4% (544 of 694) patients during follow-up. Deaths due to NSCLC were more frequent in case of (N1 )N2 disease, which tends to confirm the better prognosis observed in case of (N1 )N2 metastases (but this was not significant): (N1 )N2: SMLN station 62.3% (134 of 215 patients), DMLN station 70.6% (130 of 184 patients); (N1 )N2: SMLN station 54.8% (57 of 104 patients), DMLN station 60% (21 of 35 patients). Multivariate Analysis Multivariate analysis, using the Cox proportional hazard model, demonstrated a poor prognosis for the following: age (p 0.0000001), tumor size (p 0.000033), pneumonectomy (p 0.045\odds ratio [OR] 1.23\confidence interval [CI][1.00 to 1.49]), no adjuvant therapy (p 0.004\OR 1.30\CI[1.00 to 1.67]), adenocarcinoma (p 0.0074\OR 1.27\CI[1.07 to 1.51]), DMLN station (p 0.000004\OR 1.52\CI[1.27 to 1.82]), (N1 )N2 (p 0.0058\OR 1.33\CI[1.09 to 1.62]). Multivariate analysis of SNLM station involvement confirmed (N1 )N2 to be the only factor of poor prognosis (p 0.0044\OR 1.41\CI[1.11 to 1.78]), besides age and tumor size. Comment Main (N1 )N2 characteristics collected from the literature are reported in Table 5. pn2 disease is encountered in 17.6% to 26.5% of patients operated on for NSCLC with complete MLN dissection [4, 6, 8, 9, 12, 15]. Six studies [4, 6, 8, 9, 12, 15] reported 715 pn2 out of 3,415 patients (20.9%). The frequency of (N1 )N2 disease ranges from 20.2% to 38% with a mean of 32.7% (234 of 715) patients [4, 6, 8, 9, 12, 15]. Our results are within the ranges of these percentages. However, pn2 25.5% is among the highest percentage, whereas p(n1 )N2 28.6% is among the lowest. Whatever the series considered, p(n1 )N2 appears to be an important subgroup of N2 patients. The presence of intrapulmonary node disease is influenced by the rigor in which the nodes are searched for by the pathologist. Our study started twenty years ago with the aim of Table 3. Single LN Stations Involved According to Primary Tumor Location (Right 261; Left 197) Numbered According to the New International Classification [2] Right Mediastinal Lymph Node Station Left Mediastinal Lymph Node Station Lobe Involved 4R 3p 9 6(3a) 7 7 6 9 5 4L (N1 )N2 Upper lobe 51 1 2 8 2 3 2 31 7 n 166 Middle Lobe 3 3 R 102 Lower lobe 4 1 26 4 0 1 2 7 L 64 Main bronchus 4 1 3 Total 58 (34.9%) 41 (27.7%) 6 (3.6%) 34 (20.5%) 17 (10.2%) (N1 )N2 Upper lobe 65 3 10 7 2 2 41 22 n 294 Middle Lobe 5 4 R 159 Lower lobe 11 40 30 5 1 13 L 133 Main bronchus 1 20 6 1 3 Total 82 (27.9%) 74 (25.2%) 43 (14.6%) 43 (14.6%) 38 (12.9%) R right; L left.

Ann Thorac Surg RIQUET ET AL 2005;79:225 33 SKIP MEDIASTINAL LN METASTASIS 229 Table 4. N2 Patient Survival (n 694) Factors n Median 5 Year Survival p Value Gender Male 567 21 23.4 NS Female 127 20 21.3 Histology Adenocarcinoma 274 20 20.3 Squamous 292 23 25.2 NS Large cell 62 23 26.8 Other 37 21 32.0 Neuroendocrine 29 18 10.3 T T1 133 33 31.1 0.01 T2 414 19 23.3 0.0013 T3 120 18 14.3 0.0023 T4 27 18 18.7 Pneumonectomy 414 17 21.0 Lobectomysegmentectomy 280 30 25.0 0.00069 Clinical N2 a 0.02 mn2 199 28 26.9 cn2 484 19 21.8 Left 335 19 20.6 Right 359 23 25.0 Past oncologic history Yes 80 17 13.2 No 614 21 24.1 N2 single station 439 26 29.1 0.0000001 N2 dual stations 255 14 12.9 N1 N2 204 28 34.4 0.000056 N1 N2 490 18 18.5 N1 N2 1 station 161 32 38.4 0.00051 N1 N2 1 station 278 23 24.0 Adjuvant treatment b no CT 579 19 22.5 0.035 CT or RCT 110 30 24.2 a unavailable data n 11; b unavailable data n 5. Postoperative deaths are excluded (n 45). cn2 clinical N2; CT chemotherapy; mn2 microscopic N2; no CT no chemotherapy (are included: no adjuvant treatment and radiation therapy alone); NS not significant; RCT radiation and chemotherapy. Fig 1. Five year survival rates: comparison between (N1 )N2 (n 204) and (N1 )N2 (n 490) patients (all mediastinal lymph node station involved); postoperative death excluded. (1 (N1 )N2 [5 year survival rate: 34.4%, median 28 months]; 2 (N1 )N2 [5 year survival rate: 8.5%, median 18 months]). This was not different when postoperative death was included: (N1 )N2 (33%, median 27); (N1 )N2 (17%, median 17) (p 0.0000001). commonly involving only one positive MLN are another example of the importance of MLN dissection [12]. Some authors [4, 15] suggest that adenocarcinomas are more likely to have (N1 )N2 metastases than squamous cell carcinomas. Most studies, including our own, show no difference in the incidence of (N1 )N2 metastasis according to histology of the primary tumor when both SMLN and DMLN involved stations are considered together. When considering (N1 )N2 metastasis in a SMLN station, we observed that adenocarcinoma was, however, significantly more frequent. As others, we have found no explanation for this possible predominance of studying NSCLC lymphatic spread and the pathologists searched the nodes with rigor and accuracy. The frequency of N2 disease encountered in the literature depends on how hard this is searched for preoperatively. When N2 is discovered before surgery, many centers use preoperative neoadjuvant therapy, thus rendering (N1 )N2 disease difficult to assess. This is the reason why we excluded such patients from the study. Furthermore, during most of the considered period, even patients with bulky N2 were considered for surgery when resectable. Finally, skip metastases are more systematically detected by MLN dissection than by sampling [12]. Apart from micrometastasis detection, skip metastases Fig 2. Five year survival rates: comparison between (N1 )N2 (n 161) and (N1 )N2 (n 278) patients (single mediastinal lymph node station involvement only); postoperative deaths excluded. (1 (N1 )N2 [5 year survival rate: 38.4%, median 32 months]; 2 (N1 )N2 [5 year survival rate: 24.1%, median 23 months]). This was not different when postoperative death was included: (N1 )N2 (37.2%, median 30), (N1 )N2 (22.7%, median 21) (p 0.0027).

230 RIQUET ET AL Ann Thorac Surg SKIP MEDIASTINAL LN METASTASIS 2005;79:225 33 Fig 3. Five year survival rates: comparison between (N1 )N2 (n 43) and (N1 )N2 (n 212) (dual mediastinal lymph node station involvement); postoperative death excluded. (1 (N1 )N2 [5 year survival rate: 20.7%, median 17 months]; 2 (N1 )N2 [5 year survival rate: 11.2%, median 14 months]). This was not different when postoperative death was included: (N1 )N2 (20.7%, median 17), (N1 )N2 (10.4%, median 13) (p 0.096). adenocarcinoma in case of (N1 )N2. We wonder whether this would not be purely coincidental. In fact, SMLN station (N1 )N2 appears of good prognosis in our series and others, whereas adenocarcinomas commonly appear of poor prognosis: both results are contradictory. The biological behavior of tumor cells participating in (N1 )N2 involvement could be another explanation [6], but this remains purely hypothetical. Yoshino and colleagues [6] state that there is no difference in the location of the cancer within the lung and the presence of (N1 )N2 or (N1 )N2 involvement. Most other authors observed a higher incidence of (N1 )N2 metastases in upper lobe primaries but not significant [4, 5, 9]. Our study (Table 2) demonstrates it to be highly significant. In a previous anatomic study [13] on 260 cadavers we reported a 23.6% incidence of segmental lymph channels draining directly into mediastinal nodes. These direct draining lymph channels were more frequent from the upper lobes (36.4% on the right and 33.6% on the left). Both our anatomic and surgical results suggest that the possible mechanisms of (N1 )N2 metastasis is the anatomic lung lymph drainage pattern. The term skip metastasis refers to the concept that a LN located along lymphatic drainage channels between primary tumor and central nodal metastasis is not itself the site of metastasis. It is preferable to use the term (N1 )N2 when anatomy has demonstrated direct lymphatic drainage from the lung to MLN bypassing N1 nodes. Other explanations provided for the presence of skip metastasis are a particular biological behavior of tumor cells [6] or the presence of possible N1 micrometastasis overlooked by routine histopathological methods, which can occur despite the rigor of the pathologist [9]. We suggest another explanation: tumors with visceral pleural involvement (VPI) present a higher frequency of single or dual station N2 involvement [16]. This may be due to desquamating tumor cells within the pleural space, which are reabsorbed by lymphatic vessels of the parietal or diaphragmatic pleura; such drainage involves MLN stations [17] and theoretically induces (N1 )N2 involvement. However, our study does not demonstrate this possible role of the VPI for (N1 )N2 metastasis (VPI : (N1 )N2 25.9%, (N1 )N2 30.6%). Yoshino and colleagues [6] and Sagawa and colleagues [10] mentioned a possible higher incidence of SMLN station involvement in (N1 )N2 as compared to (N1 )N2. In our study, this is confirmed to be highly significant. Such a difference can also be explained by anatomic consideration. In our anatomic study [13], MLN are reached by two separate lymph channels in 14% of dissections. In case of (N1 )N2 dual stations, both lymph drainage may be (N1 )N2 but in some cases there may be one (N1 )N2 coexisting with one (N1 )N2 involvement (Fig 4): in such a case pn classification will be (N1 )N2 and dual (N1 )N2 involvement thus appears more frequent. (N1 )N2 involvement can concern each MLN station, whatever the cancer location within the lung. However, Okada and colleagues [15] believe that dissection of the subcarinal nodes in patients with an upper lobe tumor is not routinely necessary, especially when the nodes in both the hilum and upper mediastinum are intact, because of the rarity of (N1 )N2 subcarinal disease from upper lobes. In our study, we observed metastasis from the upper lobes into the MLN station 7 in 11 out of 103 patients (10.7%); such a pattern is also demonstrated by anatomic results in 10.4% of cases on the right [13]. Watanabe and colleagues [18] also report that the frequency of skipping metastases from the upper lobe tumors to the lower part of the mediastinum was 8.7%. We suggest that such percentages are sufficient for also supporting complete MLN dissection when operating NSCLC. The type of resection is not mentioned by authors studying skip (N1 )N2 metastases, with the exception of Gunluoglu and colleagues [8] who reported a percentage of 52.9% pneumonectomy and Misthos and colleagues [12] who reported 38%. These authors [8, 12] report that lobectomy appears to be more frequently performed than pneumonectomy in cases of (N1 )N2. This is demonstrated by our study and is easily explained by the absence of interlobar N1, rendering lobectomy easier to perform. Our percentage of pneumonectomy (61%) for N2 patients is higher than those reported in North America and Japan. However, it is similar to those of other European countries (62.5% [19], 64.6% [20], and 66% [21]. This is due to the fact that most patients were not selected and directly underwent surgery. Thus, the percentage of clinical N2 patients in our study was 70.7% (Table 1). In the study of Misthos and colleagues [12] the percentage of pneumonectomy was 38% but clinical N2 disease was not included and patients were entered in an induction chemotherapy protocol. The postoperative mortality rate we observed is due to this high number of pneumonectomies, which remains within the range of what is commonly reported (4.8% 18%) [22]. The difference in long-term survival has generally been slightly

Ann Thorac Surg RIQUET ET AL 2005;79:225 33 SKIP MEDIASTINAL LN METASTASIS 231 Table 5. Main (N1 )N2 Characteristics Collected From the Literature Characteristics Libshitz & Al 1986 [4] Tateihi & Al 1994 [5] Yoshino & Al 1996 [6] Gunluoglu & Al 2002 [8] Prenzel & Al 2003 [9] Misthos & Al 2004 [12] N2/N0 N1 (%) 48/200 (24%) 186/872 (21%) 110/619 (17.8%) 34/128 (26.5%) 45/256 (17.6%) 151/580 (26%) (N1 )N2/(N1 )N2 (%) 14/48 (29.2%) 62/186 (33%) 33/110 (30%) 11/34 (32.3%) 17/45 (38%) 44/151 (29%) Histology Adenocarcinoma 11 (78.6%) 30 (48.4%) 16 (48.5%) 6 (53%) 13 (29.6%) Squamous 0 24 (38.7%) 16 (48.5%) 9 (35%) 31 (70.4%) Large cell 2 (14.3%) 7 (11.3%) 1 (3%) 2 (12%) 0 Other 1 (7.1%) 1 (1.6%) 0 0 0 T factor T1 9 (14.5%) 9 (27.3%) 4 (23.5%) 0 T2 35 (67.3%) 20 (60.6%) 11 (64.7%) 25 (57%) T3 9 (14.5%) 4 (12.1%) 2 (11.8%) 19 (43%) T4 9 (14.5%) 0 0 0 RUL 7 26 7 5 9 LUL 3 17 10 7 7 RLL ML 2 8 9 12 2 25 LLL 2 2 4 3 3 Pneumonectomy 5 (45.5%) 19 (42.8%) Lobectomy 6 (54.5%) 25 (57.2%) Station 2R 4R 6 52 14 5 19 5 Station 7 6 Station 9 (R) 1 28 15 5 7 5 Station 9 (L) 7 Station 5/6 2 23 6 7 13 Station 4 L 1 RUL and LUL 1 and 2 7 and 4 1 and 3 2 and 1 Station 7 (and 9) 3 (and 0) 8 (and 3)?? RLL and LLL 9 and 4 5 and 2 0 and 3 Station 4 and 5 0 and 0 Clinical N2 9 (27.3%) LLL left lower lobe; LUL left upper lobe; ML middle lobe; RLL right lower lobe; RUL right upper lobe. worse for patients requiring pneumonectomy [23], which is significant in our study. The need for less pneumonectomy probably participates in the more favorable outcome observed following surgery for (N1 )N2 metastases. Fig 4. Example of dual mediastinal lymph node station involvement with coexisting (N1 )N2 and (N1 )N2 involvement. pn classification is (N1 )N2. Coexisting (N1 )N2 is underrated and (N1 )N2 involvement thus appears more frequent. In fact, (N1 )N2 is of significantly better prognosis than (N1 )N2 metastasis. Data collected from the literature are reported in Table 6: global 5 year survival rates are higher in case of (N1 )N2 [6, 7, 9 11]. N2 disease as a whole is also characterized by higher 5 year survival rates when SMLN station is involved compared to DMLN station involvement [6, 7, 10, 11]. We also found the better prognostic value of single station N2 involvement to be highly significant (Table 4). It has been suggested that the improved (N1 )N2 prognosis be explained by the lesser frequency of DMLN station involvement [6]. This is also supported by observing that 5 year survival rates appear not significantly higher when (N1 )N2 SMLN station involvement was compared with (N1 )N2 SMLN station involvement (p 0.09 [6] and 0.08 [10]). In a previous study [24] including lesser patients ((N1 )N2 153, (N1 )N2 408) we also observed the difference in survival between (N1 )N2 and (N1 )N2 SMLN station as not significant: we believed that (N1 )N2 global poor prognosis was explained by the higher frequency of DMLN station involvement in that group [24]. Our current study significantly demon-

232 RIQUET ET AL Ann Thorac Surg SKIP MEDIASTINAL LN METASTASIS 2005;79:225 33 Table 6. Five Year Survival Rates (%) Collected From the Literature (N1 )N2 Tateishi and colleagues [5] n 62 22% Yoshino and colleagues [6] n 33 35% (N1 )N2 n 77 13% Total N2 n 186 24% (N1 )N2 SMLN Station n 26 31% n 30 38% NS p 0.096 Yano and colleagues [7] n 111 18% Sagawa and colleagues [10] n 76 37% n 102 22% n 178 28% n 43 56.6% (N1 )N2 SMLN Station n 49 15% n 51 33% Gawryckowski and colleagues [11] n 23 26.1% p 0.02 p 0.08 n 41 0% p 0.0015 Prenzel and colleagues [9] n 17 41% n 28 14% n 15 48% n 24 13% p 0.018 p 0.015 strates that the prognosis of(n1 )N2 is better even in case of SMLN station involvement, which Prenzel and colleagues [9] also recently stressed on a smaller number of patients (n 45, with only 6 DMLN stations). We have no explanation for this better prognosis. Prenzel and colleagues [9] found a smaller number of tumor infiltrating MLN per patient in case of (N1 )N2 disease and suggest that this could explain the difference. The fact that (N1 )N2 SMLN station involvement shares the same poor prognosis as (N1 )N2 DMLN station involvement in our study supports this point of view. In a previous study, we demonstrated that survival rates were similar between extralobar N1 involvement and N2 involvement when N2 involvement was limited to a SMLN station [25]. With respect to anatomy, patients who have N2 disease and N1 nodes may have lymphatic channels that bypass N1 nodes and such patients have a better prognosis than patients who first seeded N1 nodes and then seeded N2 nodes. Since only the first nodes filtering the lymph channels from the primary tumor are seeded, SMLN station (N1 )N2 involvement may represent a particular N1 subgroup of patients. In conclusion, (N1 )N2 involvement is a frequent pstaging discovery. (N1 l)n2 demonstrates a better prognosis than other N2 subgroups. Further studies are needed to better understand this particular (N1 )N2 involvement. These results should lead to a better understanding of lung cancer, improved classification, and more appropriate adjuvant therapy. References 1. Mountain CF. Revisions in the international system for staging lung cancer. Chest 1997;111:1710 7. 2. Mountain CF, Dresler CM. Regional lymph node classification for lung cancer staging. Chest 1997;111:1718 23. 3. Kirsh MM, Kahn DR, Cago O, et al. Treatment of bronchogenic carcinoma with mediastinal metastases. Ann Thorac Surg 1971;12:11 7. 4. Libshitz HI, McKenna RJ, Mountain CF. Patterns of mediastinal metastases in bronchogenic carcinoma. Chest 1986;90: 229 32. 5. Tateishi M, Fukuyama Y, Hamatake M, Kohdono S, Ishida T, Sugimachi K. Skip mediastinal lymph node metastasis in non-small cell lung cancer. J Surg Oncol 1994;57:139 42. 6. Yoshino I, Yokoyama H, Yano T, et al. Skip metastasis to the mediastinal lymph nodes in non-small cell lung cancer. Ann Thorac Surg 1996;62:1021 5. 7. Yano T, Fukuyama Y, Yokoyama H, et al. Long-term survivors with pn2 non-small cell lung cancer after complete resection with a systematic mediastinal node dissection. Eur J Cardiothorac Surg 1998;14:152 5. 8. Gunluoglu Z, Solah O, Metin M, Gurses A. The prognostic significance of skip mediastinal lymphatic metastasis in resected non-small cell lung cancer. Eur J Cardiothorac Surg (letter) 2002;21:595. 9. Prenzel KL, Monig SP, Sinning JM, et al. Role of skip metastasis to mediastinal lymph nodes in non-small cell lung cancer. J Surg Oncol 2003;82:256 60. 10. Sagawa M, Sakurada A, Fujimura S, et al. Five year survivors with resected pn2 non-small cell lung carcinoma. Cancer 1999;85:864 8. 11. Gawrychowski J, Gabriel A, Lackowska B. Heterogeneity of stage IIIA non-small cell lung cancer (NSCLC) and evaluation of late results of surgical treatment. Eur J Surg Oncol 2003;29:178 84. 12. Misthos P, Sepsas E, Athanassiadi K, Kakaris S, Skottis I. Skip metastases: analysis of their clinical significance and prognosis in the IIIA stage of non-small cell lung cancer. Eur J Cardiothorac Surg 2004;25:502 8. 13. Riquet M, Hidden G, Debesse B. Direct lymphatic drainage of lung segments to mediastinal nodes. J Thorac Cardiovasc Surg 1989;97:625 32. 14. Kwiatkowski F. SEM (Statistiques, Epidémiologie, Médecine): un outil de gestion informatique et statistique

Ann Thorac Surg RIQUET ET AL 2005;79:225 33 SKIP MEDIASTINAL LN METASTASIS 233 adapté á la recherche en cancérologie. Bull Cancer 2000;87:715 21. 15. Okada M, Tsubota N, Yoshimura M, Miyamoto Y. Proposal for reasonable mediastinal lymphadenectomy in bronchogenic carcinomas: role of subcarinal nodes in selective dissection. J Thorac Cardiovasc Surg 1998;116:949 53. 16. Manac h D, Riquet M, Medioni J, Le Pimpec Barthes F, Dujon A, Danel C. Visceral pleura invasion by non-small cell lung cancer: an underrated bad prognostic factor. Ann Thorac Surg 2001;71:1088 93. 17. Okiemy G, Foucault C, Avisse C, Hidden G, Riquet M. Lymphatic drainage of the diaphragmatic pleura to the peritracheobronchial nodes. Surg Radiol Anat 2003;25:32 5. 18. Watanabe Y, Shimizu J, Tsubota M, Iwa T. Mediastinal spread of metastatic lymph nodes in bronchogenic carcinoma: mediastinal nodal metastases in lung cancer. Chest 1990;97:1059 65. 19. Van Klaveren RJ, Festen J, Otten HJ AM, Cox AL, de Graaf R, Lacquet LK. Prognosis of unsuspected but completely resectable N2 non-small cell lung cancer. Ann Thorac Surg 1993;56:300 4. 20. Conill C, Astudillo J, Verger E. Prognostic significance of metastases to mediastinal lymph node levels in resected non small cell lung carcinoma. Cancer 1993;15:1199 202. 21. Regnard JF, Magdeleinat P, Azoulay D, et al. Results of resection for bronchogenic carcinoma with mediastinal lymph node metastases in selected patients. Eur J Cardiothorac Surg 1991;5:583 7. 22. Klemperer J, Ginsberg RJ. Morbidity and mortality after pneumonectomy. Chest Surg Clin N Am 1999;9:675 93. 23. Daly BDT. Late results. Chest Surg Clin N Am 1999;9:675 93. 24. Riquet M, Dujon A, Le Pimpec Barthes F, et al. Cancer bronchiques N2 sans métastase aux ganglions intrapulmonaires. Ann Chir Thorac Cardiovasc 2003;7: 95(abstract). 25. Riquet M, Manac h D, Le Pimpec Barthes F, Dujon A, Chehab A. Prognostic significance of surgical pathologic N1 diseases in non-small cell carcinoma of the lung. Ann Thorac Surg 1999;67:1572 6. The Thoracic Surgery Foundation for Research and Education The Thoracic Surgery Foundation for Research and Education (TSFRE) extends best wishes to all its colleagues and friends for 2005. We wish to extend our thanks not only for your support in the past year, but for the past many years. Your support of TSFRE has made a significant difference in the growth and accomplishments of the organization. Your past commitments to TSFRE have furthered many invaluable research and education activities that have made powerful differences in the practice of thoracic surgery. The transition period between the old year and the new year is also a time of reflection for many of us. We urge you to take this period of time to give some thought to the legacy you will be leaving behind, particularly within thoracic surgery. TSFRE would ask that you consider, in acknowledgment of what cardiothoracic surgery has done for you, the ways in which you might leave your legacy on our specialty. The great progress that cardiothoracic surgery has witnessed in the past decade is the result of extraordinary research and education activities. You can ensure that these vital programs continue well into the future, continuing to transform cardiothoracic surgery to help our patients. To do so, please consider the making of a planned gift to TSFRE. The most common planned gift is a charitable bequest to TSFRE through your will after you have properly ensured that your family and other loved ones will be financially secure. Under certain circumstances, a donation from your estate might give TSFRE $10,000 while representing only about $2,500 that would go to your family should the payment of estate, income and capital gains taxes be required. Real Estate as a planned gift can be a very practical mechanism, even if the real estate property has decreased in value. A donor may even give real estate and yet retain the use of the property. Sometimes a real estate gift can provide an ongoing source of income to the donor or to another designated individual. Through your generosity, you continue to allow our specialty to evolve. We urge you to continue to do so to your utmost capability. 2005 by The Society of Thoracic Surgeons Ann Thorac Surg 2005;79:233 0003-4975/05/$30.00 Published by Elsevier Inc