Characteristics and Prognostic Value of Lymphatic and Blood Vascular Microinvasion in Lung Cancer

Transcription

1 Characteristics and Prognostic Value of Lymphatic and Blood Vascular Microinvasion in Lung Cancer Alex Arame, MD, Pierre Mordant, MD, Aurélie Cazes, MD, Christophe Foucault, MD, Antoine Dujon, MD, Françoise Le Pimpec Barthes, MD, PhD, and Marc Riquet, MD, PhD Department of Thoracic Surgery and Histology, Georges Pompidou European Hospital, Assistance Publique-Hôpitaux de Paris, Faculté de Médecine Paris Descartes, Université Paris, and Department of Thoracic surgery, Cedar surgery center, Bois Guillaume, France Background. The prognostic value of vascular microinvasion (VMI) in non-small cell lung cancer (NSCLC) has been a matter of discussion in recent decades. The last T N M classification does not take VMI into account, but many points remain questionable. Methods. A retrospective study was performed of patients undergoing operations for NSCLC during a 20-year period. Lymphatic VMI (LVMI) was classified as group (G) 1, blood VMI (BVMI) as G2, LVMI and BVMI as G3, and no VMI as G4. The demographic, pathologic, T N M characteristics, and long-term survival of each group were analyzed. Results. A total of 3,868 patients (G1, 334; G2, 642; G3, 172; G4, 2,720), mean age years, underwent different types of resection, with complete lymphadenectomy in 88.5%. Adenocarcinomas were more frequent in G1 and G3, and squamous cell carcinomas in G2. In G2, more N1 tumors needed more extensive resections. G1 was equally distributed regardless of tumor size, but G2 prevalence increased with augmenting size. Nodules in the same lobe were significantly more frequent in LVMI than in BVMI. After exclusion of patients with R1 and R2 resections, multivariate analysis confirmed that LVMI and BVMI were independent prognostic factors as well as age, sex, type of resection, T extension, and N involvement. Conclusions. VMI is generally associated with a poorer prognosis. LVMI is less frequent than BVMI but has lower survival rates. The benefit of adjuvant therapy in VMI patients needs to be evaluated. (Ann Thorac Surg 2012;94:1673 9) 2012 by The Society of Thoracic Surgeons Contrary to non-small cell lung cancer (NSCLC) invading the visceral pleura categorized as T2 or to NSCLC with separate nodule in the same lobe or in a different ipsilateral lobe categorized as T3 and T4, respectively, vascular microinvasion (VMI) was not considered in the international staging system of 2009 [1]. However, VMI represents an NSCLC with a poorer outcome, and attention had already been drawn to this factor of prognosis at the end of the 1950s [2, 3]. Because this potentially important prognostic factor has not been extensively studied, we sought to revisit its frequency and prognostic significance. The goals of this study were to analyze the clinical presentation of lymphatic (LVMI) or blood (BVMI) vessel microinvasion in the overall population of patients undergoing surgical resection in a curative attempt and then to analyze related prognosis in the subgroup of patients undergoing complete surgical resection. Accepted for publication July 23, Address correspondence to Dr Riquet, Georges Pompidou European Hospital, Departments of General Thoracic Surgery and Pathology, rue Leblanc, Paris Cedex 15, France; marc.riquet@ egp.aphp.fr. Patients and Methods The study was approved by the Thoracic Surgery Society Ethic Committee (CERC-SFCTCV). Need for informed patient consent was waived. The clinical records of patients who underwent operations for NSCLC during a 20-year period ending in December 2009 in Georges Pompidou European Hospital (Paris) and Cedar Surgery Centre (Bois Guillaume) were reviewed. The preoperative workup included chest roentgenogram, bronchoscopy, computed tomography (CT) scan of the chest, spirometry, lung perfusion scan, and a thorough search for distant metastases, including in recent years, positronemission tomography imaging. Mediastinoscopy was performed to exclude N3 disease and to confirm N2 involvement in patients included in various neoadjuvant treatment protocols, depending on the demand of different referring centers. The staging system was the International Staging System for NSCLC of 2009 [1]. The classification of Mountain and Dresler [4] was used for mediastinal lymph nodes involvement. R2 was gross tumor left behind, R1 marginal microscopic invasion, and R0 complete resection. N3 disease and distant metastases precluded surgical intervention. The study excluded patients who underwent an exploratory thoracotomy by The Society of Thoracic Surgeons /$36.00 Published by Elsevier Inc

2 1674 ARAME ET AL Ann Thorac Surg PROGNOSTIC VALUE OF VMI IN NSCLC 2012;94: Table 1. Surgical Management Variable a Group 1 Group 2 Group 3 Group 4 (LVMI) (BVMI) (Both) (Neither) Total n 334 (9) N 642 (17) n 172 (4) n 2,720 (70) n 3,868 (100) No. (%) No. (%) No. (%) No. (%) No. (%) p Value Induction treatment 70 (21) (10) 111 (17) (15) 36 (21) (5) 510 (19) (70) 727 (19) (100) 0.48 Resection Segmentectomy/wedge 37 (10) (9) 50 (8) (12) 22 (13) (5) 315 (12) (74) 424 (11) (100) Lobectomy/sleeve 191 (57) (9) 358 (56) (16) 84 (49) (4) 1,590 (59) (71) 2,223 (58) (100) Bilobectomy 13 (4) (9) 29 (5) (19) 12 (7) (8) 96 (4) (64) 150 (4) (100) Pneumonectomy 88 (26) (9) 199 (31) (20) 52 (30) (5) 679 (25) (67) 1,018 (26) (100) Completion pneumonectomy 5 (2) (9) 6 (1) (11) 2 (1) (4) 40 (2) (76) 53 (1) (100) Complete lymphadenectomy 301 (90) 561 (87) 150 (87) 2,411 (89) 3,423 (89) 0.58 Postoperative Complication 83 (24.9) 198 (31) 61 (36) 593 (22) 935 (24) Death 16 (4.8) 38 (5.9) 8 (4.7) 130 (4.8) 192 (5) 0.69 a Percentage (in parenthesis) by group is shown in regular type and percentage (in parenthesis) in italic is by type of treatment. BVMI blood vascular microinvasion; LVMI lymphatic vascular microinvasion. We more particularly analyzed the pathologic and prognostic characteristics of these patients, focusing on the VMI. Sections of the primary tumor were fixed in 10% formalin and embedded in paraffin. Slides from each block were stained conventionally with hematoxylin and eosin. Elastin stains were used to evaluate involvement of the visceral pleura. Vascular invasion was determined by the presence of intravascular tumor cells in blood or lymphatic vessels. Distinction between LVMI and BVMI was made according to the aspect of the vessel wall, the opening of the vessel, and its localization. In difficult cases, elastin and podoplanin stains were used to allow clear visualization of blood and lymphatic vessels, respectively. Other pathologic studies, including immunostains, were performed at the discretion of the pathologist. LVMI was categorized as group 1, BVMI as group 2, both LVMI and BVMI as group 3, and no VMI as group 4. Follow-up information was obtained from the hospital patient records, from a questionnaire completed by the chest physician or general practitioner, or from death certificates. The main outcome was the overall survival, defined as the interval between the date of operation and the date of death or the last follow-up visit for censored patients. Mean follow-up duration was months. Actuarial survival curves were estimated by the Kaplan- Meier method. Statistical comparisons between survival distributions were made using the log-rank test. Multivariate analysis was performed using the Cox proportional hazards model for overall survival analysis. Univariate analysis used the outcome variables of sex, age, type of resection, histologic assessment, and type of N involvement. All data analyses were conducted with the two-sided test, and a p value of less than 0.05 was considered as statistically significant. Data were analyzed using SEM statistical software (Anticancer Centre Jean Perrin, Clermont-Ferrand, France) [5]. Results Overall Population The population consisted of 3,868 patients, with 3,039 men (78 %) and 3,444 smokers (89%). Mean age was years. VMI was observed in 1,148 patients (30%), with LVMI (group 1) in 334 (9%) and BVMI (group 2) in 642 (17%), and both LVMI and BVMI (group 3) in 172 (4%). Main surgical characteristics are given in Table 1. Interestingly, the frequency of LVMI or BVMI was not influenced by induction therapy. Frequency of LVMI was similar whatever the type of surgical resection, but frequency of BVMI was not: BVMI was more frequent with more extensive resection. Postoperative complications were more frequent in patients with BVMI (group 2) or both LVMI and BVMI (group 3). Postoperative mortality of the entire population was 4.9% (192 of 3,868), and did not differ from one group to another (p 0.69). Tumor extension characteristics are summarized in Table 2. In groups 1 and 3 (LVMI presence), nodal involvement was present in 67% of patients (329 of 506), a significantly higher rate than the 48% in group 2 (309 of 642) and the 36% in group 4 (986 of 2,720; p 10 6 ). N1 involvement was more frequent in BVMI and N2 involvement in LVMI. Other pathologic characteristics are summarized in Table 3. The frequency of adenocarcinomas was higher in groups 1 and 3 (LVMI presence), whereas squamous cell carcinomas (SCC) were more frequent in group 2 (BVMI). Tumors with pleural invasion were more frequent in groups 1, 2, and 3 than in group 4 (p 10 6 ). The frequency of LVMI was the same regardless of tumor size, but the frequency of BVMI significantly increased with increasing

3 Ann Thorac Surg ARAME ET AL 2012;94: PROGNOSTIC VALUE OF VMI IN NSCLC 1675 Table 2. Tumor Extension Variable a Group 1 Group 2 Group 3 Group 4 (LVMI) (BVMI) (Both) (Neither) Total n 334 (9%) n 642 (17%) n 172 (4%) n 2,720 (70%) n 3,868 (100%) No. (%) No. (%) No. (%) No. (%) No. (%) p Value Tumor T1 69 (21) (6) 134 (21) (12) 25 (14) (2) 922 (34) (80) 1,150 (30) (100) T2 211 (63) (10) 364 (57) (18) 104 (61) (5) 1,362 (50) (67) 2,041 (53) (100) T3 46 (14) (8) 108 (17) (19) 41 (24) (7) 385 (14) (66) 580 (15) (100) T4 8 (2) (8) 36 (6) (37) 2 (1) (2) 51 (2) (53) 97 (3) (100) Nodes N0 127 (38) (6) 333 (52) (15) 50 (29) (2) 1,734 (64) (77) 2,244 (58) (100) N1 60 (18) (8) 159 (25) (22) 39 (23) (6) 448 (17) (64) 706 (18) (100) N2 147 (44)(16) 150 (23) (16) 83 (48) (9) 538 (20) (59) 918 (24) (100) N2 1 station 80 (24) 99 (15) 45 (26) 371 (14) 595 (15) N2 2 stations 67 (20) 51 (8) 38 (22) 167 (6) 323 (9) Stages Stage I 88 (26) (6) 228 (36) (15) 33 (19) (2) 1,216 (45) (78) 1,565 (41) (100) Stage II 58 (17) (6) 168 (26) (18) 29 (17) (3) 699 (28) (73) 954 (25) (100) Stage III 165 (49) (14) 208 (32) (18) 94 (55) (8) 684 (18) (60) 1,151 (30) (100) Stage IV 23 (7) (12) 38 (6) (19) 16 (9) (8) 121 (5) (61) 198 (5) (100) a Percentage in parenthesis and regular type is within a certain group (column). Percentage in parenthesis and italic is within a certain tumor, node, or stage extension (line). BVMI blood vascular microinvasion; LVMI lymphatic vascular microinvasion. size, from 13% to 22% (p ). Finding another nodule in the same lobe was more frequent in group 1. Pleural lavage cytology was studied in the last 1,128 T1 and T2 patients. Positive cytology was more frequent when LVMI was present (p ). Survival analysis of the overall population according to VMI is depicted in Figure 1. Overall survival rates were 43.1% at 5 years and 26.3% at 10 years. The overall median survival was 45 months. Survival rates at 5 years were 28.7% in group 1, 38.5% in group 2, 19.2% in group Table 3. Pathologic Characteristics Variable a Group 1 Group 2 Group 3 Group 4 (LVMI) (BVMI) (Both) (Neither) Total n 334 (9%) n 642 (17%) n 172 (4%) n 2,720 (70%) n 3,868 (100%) No. (%) No. (%) No. (%) No. (%) No. (%) p Value Histology Adenocarcinoma 186 (56) (10) 279 (44) (16) 96 (56) (5) 1,251 (46) (69) 1,812 (47) (100) Squamous cell 103 (31) (6) 287 (45) (18) 58 (34) (4) 1,152 (42) (72) 1,600 (41) (100) Large cell 31 (9) (9) 49 (8) (15) 14 (8) (6) 224 (8) (70) 318 (6) (100) Adenosquamous 13 (4) (15) 18 (3) (21) 2 (1) (2) 52 (2) (61) 85 (1) (100) Miscellaneous 1 (1) (2) 9 (1) (17) 2 (1) (4) 41 (2) (77) 53 (11) (100) Visceral pleura invasion 110 (33) (11) 220 (34) (22) 58 (34) (6) 614 (23) (61) 1,002 (26) (100) 10 6 Tumor size, cm (47) (9) 225 (35) (13) 61 (36) (3) 1,302 (48) (75) 1,745 (45) (100) (34) (8) 259 (40) (19) 77 (45) (6) 896 (33) (67) 1,345 (35) (100) (12) (8) 96 (15) (19) 19 (11) (4) 336 (12) (69) 490 (13) (100) 7 25 (8) (9) 62 (10) (21) 15 (9) (5) 186 (7) (65) 288 (7) (100) Nodule same lobe 36 (11) (20) 30 (5) (17) 13 (8) (7) 102 (4) (56) 181 (5) (100) Nodule other lobe 18 (5) (10) 26 (4) (14) 13 (8) (7) 127 (5) (69) 184 (5) (100) 0.18 a Percentage in parenthesis and regular type is within a certain group (column). Percentage in parenthesis and italic is within a certain histology, tumor size, or presence of a second nodule (line). BVMI blood vascular microinvasion; LVMI lymphatic vascular microinvasion.

4 1676 ARAME ET AL Ann Thorac Surg PROGNOSTIC VALUE OF VMI IN NSCLC 2012;94: Fig 1. Overall survival of the four groups: 1, lymphatic vascular microinfiltration; 2, blood vascular microinfiltration; 3, lymphatic and blood vascular microinfiltration; 4, neither type of vascular microinfiltration. 3, and 47.5% in group 4 (p 10 6 ), suggesting vascular invasion could be of prognostic value in NSCLC. Prognostic Analysis in R0 Patients To further analyze the prognostic value of VMI, we excluded patients with R1 and R2 resections. The characteristics and causes of death of R0 patients are summarized in Table 4. There was no difference between groups in occurrence of local recurrence or distant metastases, or both. The survival curves of R0 patients according to nodal extension are shown in Figure 2. Survival for patients with R0 N0 tumors was poorer in group 3 than in group 1, poorer in group 1 than in group 2 and poorer in groups 1 and 3 than in group 4 (Fig 2A). Comparing R0 T1 N0 with R0 T2 N0, survival at 5 years among the groups was different in case of T1 N0: group 1 (n 35), 48%; group 2 (n 86), 59.8%; group 3 (n 12), 25%; and group 4(n 708), 66% (p ). However, survival at 5 years was not different in T2 N0: group 1 (n 80), 51.7%; group 2 (n 197), 51.4%; group 3 (n 30), 38.4%; and group 4 (n 811), 53.5% (p 0.18). In R0 N1 and R0 N2 patients, survival was not different among groups 1, 2, and 3 but was significantly better in group 4 than in the three others (Fig 2B and C). The 5-year survival rates in these patients, with or without adjuvant treatment, were not different between groups, and even when considering only the patients in stage I. Median, 5-year and 10-year survival were 43 months, 42%, 28% in patients with induction treatment (n 845), 32 months, 39%, 24% in patients with adjuvant treatment (n 298), 33 months, 35%, 16% in patients with both (n 280), and 68 months, 53%, 33% in patients with neither (n 1902). Multivariate analysis confirmed that LVMI and BVMI were independent prognostic factors as well as N involvement (N2 and N1), age, T, sex, and type of resection (p 10 6, Table 5). Comment VMI was first reported in 1957 [2], but since then, most reports lack homogeneity. BVMI and LVMI may be studied as two different entities in the same report [6 12], both grouped without distinction [13 17], or as BVMI only [2, 3, 13 15, 18 20]; this is the most frequent type of study, including in a recent meta-analysis [21]. As a counterpart, we found only one study analyzing LVMI alone [22]. VMI is rarely studied in three different groups occulting the existence of coexisting BVMI and LVMI, despite its frequency ranging from 4.5% in our series to 8.5% in others [23, 24]. Indeed, the frequency of VMI is generally appreciated with great variability in the literature. BVMI may range from 6.2% to 77% [21] and LVMI Table 4. Characteristics and Causes of Death in R0 Patients Variable Group 1 (LVMI) Group 2 Group 3 Group 1 3 Group 4 n 258 (BVMI) (Both) (Any VMI) (No VMI) No. (%) n 540 n 129 n 927 n 2,407 Medical TTT Induction 14 (5) 38 (7) 11 (9) 63 (7) 235 (10) Adjuvant 89 (34) 167 (31) 54 (42) 310 (33) 535 (22) Both 36 (14) 44 (8) 11 (9) 91 (10) 189 (8) Cause of death Lung cancer 107 (41) 160 (30) 56 (43) 323 (35) 600 (25) Other cancer 11 (4) 28 (5) 5 (4) 44 (5) 104 (4) Other cause 34 (13) 70 (13) 19 (15) 123 (13) 315 (13) Unknown 44 (17) 99 (18) 26 (20) 169 (18) 452 (19) Total deaths 196 (76) 357 (66) 106 (82) 659 (71) 1,471 (61) BVMI blood vascular microinvasion; LVMI lymphatic vascular microinvasion; TTT treatment; VMI vascular microinvasion.

5 Ann Thorac Surg ARAME ET AL 2012;94: PROGNOSTIC VALUE OF VMI IN NSCLC 1677 Fig 2. Survival in R0 patients. (A) Overall median survival in case of R0 N0 by groups: 1, 55 months; 2, 73 months; 3, 41 months; 4, 84 months (global p value ; between group 1 and 2, p 0.055; group 2 and 3, p ; group 1 and 4, p ; group 2 and 4, p 0.29). (B) Overall median survival in case of R0 N by groups: 1, 36 months; 2, 31 months; 3, 21 months; 4, 50 months (global p ; between group 1 and 2, p 0.72; group 2 and 3, p 0.64; group 1 and group 4, p ; group 2 and 4, p (C) Overall median survival in case of R0 N2 by group: 1, 23 months; 2, 17 months; 3, 17 months; and 4, 25 months (global p ; between group 1 and 2, p 0.85, group 2 and 3, p 0.07, group 1 and 4, p 0.012; group 2 and 4, p 0.029). (Group 1 lymphatic vascular microinfiltration; group 2 blood vascular microinfiltration; group 3 lymphatic and blood vascular microinfiltration; group 4 neither.) has been reported between 15% [22] and 36% [8]. All together, VMI was present in 26% of patients in our series, as observed by others regrouping LVMI and BVMI [13, 14, 16]. The difference in frequency observed from one series to another might be explained by different methods of pathologic detection. Detection of LVMI is more difficult than BVMI [24]. Wang and colleagues [21] suggested unification and precise definition of methods for evaluating vascular or lymphatic microinvasion. Such definition is mandatory to solve this problem in the perspective of further research concerning this topic. In the literature, VMI may be studied, depending on the histologic assessment of the tumor, as only adenocarcinomas and SCC [6], adenocarcinomas alone [22], or SCC and all other histologies [7]. In fact, VMI may be present whatever the histologic evaluation, but its frequency may vary according to the histologic assessment: Table 5. Multivariate Analysis in R0 Patients Variables HR (95% CI) p Value N1 vs N0 (ref) 1.42 ( ) 10 6 N2 vs N0 (ref) 2.03 ( ) 10 6 Age 62 vs 61 years (ref) 1.48 ( ) 10 6 T2 vs T1 (ref) 1.31 ( ) 10 6 T3 T4 vs T1 (ref) 1.72 ( ) 10 6 Male vs female (ref) 1.31 ( ) Group 1 vs group 4 (ref) 1.28 ( ) Group 2 vs group 4 (ref) 1.17 ( ) Induction treatment 0.53 Adjuvant treatment 0.53 CI confidence interval; group 1 lymphatic vascular microinvasion; group 2 blood lymphatic vascular microinvasion; group 4 no lymphatic or blood vascular microinvasion; HR hazard ratio. we observed more adenocarcinomas in LVMI and more SCC in BVMI. A predominance of VMI in nonadenocarcinoma tumors has been mentioned by Naito and colleagues [20]. Others mentioned more LVMI and BVMI in adenocarcinomas and the prognosis was poorer [22]. In any event, VMI is a negative prognostic factor regardless of the histologic assessment. VMIs also vary with other pathologic features. LVMI has a constant frequency despite increasing tumor size. However, it is more frequent in the presence of another nodule in the same lobe, tumor cells in pleural lavage, stage III, and multistation N2. Consequently, there are fewer R0 resections. The prognosis is poorer, whatever the type of N involvement. This reflects major locoregional dissemination. BVMI is more frequent than LVMI, and its frequency increases with augmenting size of the tumor. Visceral pleural invasion is as frequent, but synchronous nodule and N2 involvement are less frequent than in LVMI. However, there is more frequent N1 invasion, which may be related to its greater association with SCC type histology, as we described earlier [25]. There are also more stage III patients; hence, there are more pneumonectomies and postoperative complications. Prognosis is also poor, with lower survival rates whatever the N involvement, but survival rates are globally better than those of LMVI, as also observed by others [6, 7, 12]. In addition, we observed that survival rates were the lowest when LMVI and BMVI were both present. These characteristics of the pathologic and T N M staging are not systematically analyzed in the literature. The other studies evaluate VMI in early T stages (Ia, Ib or IIa, IIb) [9 15,18, 20, 22, 23, 26], and the studies concerning all of the stages are rare [7, 8, 19]. Curiously, we found two studies reporting patients with only lymph node

6 1678 ARAME ET AL Ann Thorac Surg PROGNOSTIC VALUE OF VMI IN NSCLC 2012;94: involvement [24, 27]. Why most studies are of N0 patients is not explained but probably due to the wish to just study the prognostic value in a selected population. Thus, these studies result in failing to recognize that VMI is of prognostic value whatever the degree of N involvement would be, as we observed. Moreover, some authors say no association exists between LVMI and node involvement [24] and some question it [20]. In our series, association between LVMI and node involvement was established, with more N2 involvement in LVMI patients as well as more multistation N2 and less survival compared with BVMI patients who had more N1 lymph nodes and better survival. It seems that once the lymph or blood vessel invasion presents, the tumor further drains into the vascular system, which may explain the poorer outcome in initial stages. Some say that there is more distant metastasis with BVMI [24], which was not confirmed by our study. In general, patients who undergo induction or multimodal therapy are deliberately excluded. Most studies suggest that VMI patients might benefit from adjuvant chemotherapy (with few reporting data) and deserve inclusion in the T N M staging (without indicating which one). In our series, VMI was present, despite induction therapy, with the same frequency as in all patients who underwent surgical resection (Table 1). Macchiarini and colleagues [28] also observed residual VMI in 26% of patients who underwent induction therapy for T4 NSCLC, and VMI was the only significant factor influencing survival in univariate and multivariate analysis. There may be two hypotheses: induction therapy permits selecting tumors with particularly aggressive behavior, or induction therapy is not effective in such cases, which our results also showed. Suggesting adjuvant therapy as a consequence of this aggressive tumoral behavior is a common opinion, but few reports provide specific data. In a study from Saynak and colleagues [14], 8% of the patients underwent adjuvant therapy (N0-1 patients) but results are not available. In the Kelsey and colleagues [13] study, adjuvant chemotherapy and radiotherapy were ineffective. Tsuchiya and colleagues [16] demonstrated usefulness of oral uracil-tegafur chemotherapy after resection for stage IA NSCLC. However, in their retrospective study of 85 patients, results of 35 patients who also underwent other regimens of adjuvant therapy were not mentioned. In our study, neither induction nor adjuvant therapies were significant prognostic factors in multivariate analysis. Finally, most reports agree to consider VMI as a factor of poor prognosis with potential need to change the TNM staging. We observed that BVMI and LVMI were an independent prognostic factor of poor survival regardless of the T and N status. Our analyses according to the T and N status did not permit us to suggest that VMI might be included as a particular entity in T staging. To conclude, the prognostic value of VMI is underestimated. Unification of pathologic detecting methods and definitions are recommended. The benefit of adjuvant treatments in early-stage NSCLC with VMI need to be studied in prospective trials. References 1. American Joint Committee on Cancer (AJCC) cancer staging handbook. In: Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A 3rd, eds. AJCC cancer staging manual. 7th ed. Chicago: Springer; 2010: Collier FC, Blakemore WS, Kyle RH, Enterline HT, Kirby CK, Johnson J. Carcinoma of the lung: factors which influence five year survival with special reference to blood vessel invasion. Ann Surg 1957;146: Mosely JM, Dickson DR. Vascular invasion in lung cancer: clinical-pathologic significance. Am Rev Respir Dis 1960;82: Mountain CF, Dresler CM. Regional lymph node classification for lung cancer staging. Chest 1997;111: Kwiatkowski F, Girard M, Hacene K, Berlie J. SEM: un outil de gestion informatique et statistique adapté à la recherche en cancerologie. Bull Cancer 2000;87: Roberts TE, Hasleton PS, Musgrove C, Swindell R, Lawson RAM. Vascular invasion in non-small cell lung carcinoma. J Clin Pathol 1992;45: Fu XL, Zhu XZ, Shi DR, et al. Study of prognostic predictors for non-small cell lung cancer. Lung Cancer 1999;23: Rigau V, Molina T.J, Chaffaud C, et al. Blood vessel invasion in resected non small cell lung carcinomas is predictive of metastatic occurrence. Lung Cancer 2002;38: Hashizume S, Nagayasu T, Hayashi T, et al. Accuracy and prognostic impact of a vessel invasion grading system for stage IA non-small cell lung cancer. Lung Cancer 2009;65: Shoji F, Haro A, Yoshida T, et al. Prognostic significance of intratumoral blood vessel invasion in pathologic stage IA non-small cell lung cancer. Ann Thorac Surg 2010;89: Park SY, Lee HS, Jang HJ, Lee GK, Chung KY, Zo Ji. Tumor necrosis as a prognostic factor for stage IA non-small cell lung cancer. Ann Thorac Surg 2011;91: Hanagiri T, Takenaka M, Oka S, et al. Prognostic significance of lymphovascular invasion for patients with stage I nonsmall lung cancer. Eur Surg Res 2011;47: Kelsey CR, Marks LB, Hollis D, et al. Local recurrence after surgery for early stage lung cancer. Cancer 2009;115: Saynak M, Veeramachaneni NK, Hubbs JL, et al. Local failure after complete resection of N0-N1 non-small cell lung cancer. Lung Cancer 2011;71: Schuchert MJ, Schumacher L, Kilic A, et al. Impact of angiolymphatic and pleural invasion on surgical outcomes for stagei non-small cell lung cancer. Ann Thorac Surg 2011;91: Tsuchiya T, Akamine S, Muraoka M, et al. Stage IA nonsmall cell lung cancer: vessel invasion is a poor prognostic factor and a new target of adjuvant chemotherapy. Lung Cancer 2007;56: Rao J, Sayeed RA, Tomaszek S, Fischer S, Keshavjee S, Darling GE. Prognostic factors in resected satellite-nodule T4 non-small cell lung cancer. Ann Thorac Surg 2007;84: Macchiarini P, Fontanini G, Hardin MJ, et al. Blood vessel invasion by tumor cells predicts recurrence in completely resected T1N0M0 non-small cell lung cancer. J Thorac Cardiovasc Surg 1993;106: Kessler R, Gasser B, Massard G, et al. Blood vessel invasion is a major prognostic factor in resected non-small cell cancer. Ann Thorac Surg 1996;62: Naito Y, Goto K, Nagai K, et al. Vascular invasion is a strong prognostic factor after complete resection of node-negative non-small cell lung cancer. Chest 2010;138: Wang J, Chen J, Chen X, Wang B, Li K, Bi J. Blood vessel invasion as a strong independent prognostic indicator in non-small cell lung cancer: a systemic review and metaanalysis. PLoS ONE 2011;6:e Funai K, Sugimura H, Morita T, Shundo Y, Shimizu K, Shiija N. Lymphatic vessel invasion is a significant prognostic

7 Ann Thorac Surg ARAME ET AL 2012;94: PROGNOSTIC VALUE OF VMI IN NSCLC 1679 indicator in stage IA lung adenocarcinoma. Ann Surg Oncol 2011;18: Gabor S, Renner H, Popper H, et al. Invasion of blood vessels as significant prognostic factor in radically resected T1 3N0M0 nonsmall cell lung cancer. Eur J Cardiothorac Surg 2004;25: Bodendorf MO, Haas V, Laberke HG, Blummenstock G, Wex P, Graeter T. Prognostic value and therapeutic consequences of vascular invasion in non-small cell lung carcinoma. Lung Cancer 2009;64: Riquet M, Berna P, Fabre E, et al. Evolving characteristics of lung cancer: a surgical appraisal. Eur J Cardiothorac Surg 2012;41: Kawachi R, Tsukada H, Nakazato Y, et al. Early recurrence after surgical resection in patients with pathological stage I non-small cell lung cancer. Thorac Cardiovasc Surg 2009;57: Khan OA, Fitzgerald JJ, Field ML, et al. Histological determinants of survival in completely resected T1-2N1MO nonsmall cell cancer of the lung. Ann Thorac Surg 2004;77: Macchiarini P, Dulmet E, De Montpreville, et al. Prognostic significance of peritumoral blood and lymphatic vessel invasion by tumors cells in T4 non-small cell lung cancer following induction therapy. Surg Oncol 1994;4:91 9. INVITED COMMENTARY This retrospective analysis from two hospitals in France leverages a large patient cohort of nearly 3900 patients over 20 years to address the question of whether tumor vascular or lymphatic microinvasion or both are independent prognostic indicators in patients undergoing resection for non-small cell lung cancer. The authors [1] are to be commended for attempting to provide clarity on this topic, given that numerous articles have been published that have resulted in conflicting results in highly selected patient and tumor subsets. In brief, the authors have found that vascular microinvasions, lymphatic microinvasions, or both are independent predictors of overall survival invasion regardless of T or N status in patients undergoing resection for non-small cell lung cancer. The relationship between these pathologic variables and survival holds for R0 resections as well as for patients receiving induction or adjuvant therapies. In a thoracic oncology era, when appropriate emphasis is being placed on the identification of clinically meaningful and druggable molecular targets, this study returns to standard immunohistochemical analysis of tumor tissue to identify aggressive tumor biologic features. Given that the identification of vascular or lymphatic microinvasion is a pathologic observation, the potential of these findings to inform decisions regarding patient care plans will primarily be regulated to adjuvant systemic therapies. Post hoc analyses of the several phase III adjuvant chemotherapy trials for resectable lung cancer have not fully analyzed the pathologic variables of vascular and lymphatic microinvasion to ascertain whether they are predictors of response to adjuvant chemotherapy. This, unfortunately, is the real question. Does this pathologic assessment of tumor vascular or lymphatic microinvasion have any value as it relates to current recommendations for adjuvant doublet chemotherapy for resected node-positive and advanced-stage non-small cell lung cancer? Furthermore, are there biomarkers that correlate with these histopathologic findings that may better predict who may benefit from adjuvant chemotherapy? For instance, in situ protein analysis of excision repair cross completing group 1 gene product (ERCC1) and the regulatory subunit of ribonucleotide reductase (RRM1) in 784 patients enrolled in the International Adjuvant Lung Trial showed that both low ERCC1 and low RRM1 protein levels predicted a favorable response to adjuvant doublet platinum-based chemotherapy. The authors appropriately caution against using their findings to recommend adjuvant chemotherapy, but they do suggest that it be considered in future iterations of the lung cancer staging system. This recommendation is not without precedent in thoracic surgical oncology; the most recent esophageal cancer staging system incorporates pathologically assessed tumor grade into the staging criteria. In summary, this study, like many solid retrospective studies, raises additional questions. Larger numbers of patients from more institutions will need to be assessed. In addition, post hoc analyses of previous adjuvant chemotherapy trials need to be done to examine the clinical utility of vascular and lymphatic microinvasion pathologic assessment in patients with resected non-small cell lung cancer. David R. Jones, MD Department of Surgery Division of Thoracic and Cardiovascular Surgery University of Virginia PO Box Charlottesville, VA djones@virginia.edu Reference 1. Arame A, Mordant P, Cazes A, et al. Characteristics and prognostic value of lymphatic and blood vascular microinvasion in lung cancer. Ann Thorac Surg 2012;94: by The Society of Thoracic Surgeons /$36.00 Published by Elsevier Inc