European Journal of Cardio-Thoracic Surgery Advance Access published January 5, 2015
ORIGINAL ARTICLE
European Journal of Cardio-Thoracic Surgery (2015) 1–7 doi:10.1093/ejcts/ezu515
Influence of visceral pleural invasion on survival in completely resected non-small-cell lung cancer† Hiroyuki Adachi*, Masahiro Tsuboi, Teppei Nishii, Taketsugu Yamamoto, Takuya Nagashima, Kohei Ando, Yoshihiro Ishikawa, Tekkan Woo, Katsuya Watanabe, Yutaka Kumakiri, Takamitsu Maehara, Takao Morohoshi, Haruhiko Nakayama and Munetaka Masuda Department of Surgery, Yokohama City University, Yokohama, Japan * Corresponding author. The Department of General Thoracic Surgery, Yokosuka Kyosai Hospital, 1-16, Yonegahama-dori, Yokosuka, Kanagawa 238-8558, Japan. Tel: +81-468-222710; fax: +81-468-252103; e-mail: [email protected] (H. Adachi). Received 10 September 2014; received in revised form 11 November 2014; accepted 18 November 2014
Abstract OBJECTIVES: Although the prognostic implications of visceral pleural invasion (VPI) are well established, it remains controversial whether the extent of VPI affects survival in patients with completely resected non-small-cell lung cancer (NSCLC). In addition, the impact of VPI according to nodal status is unclear. We evaluated the influence of the extent of pleural invasion on survival by analysing a multicentre retrospective database of patients who had undergone surgery for NSCLC. METHODS: We retrospectively reviewed the clinicopathological characteristics and outcomes of 639 patients with NSCLC who underwent anatomic complete resection from 2005 to 2007 at nine hospitals affiliated with the Yokohama Consortium of Thoracic Surgeons. RESULTS: The median follow-up was 65.0 months. The extent of pleural invasion was PL0 in 462 patients, PL1 in 135 and PL2 in 42. The 5-year overall survival rate was significantly higher in patients with PL0 tumours (75.9%) than in those with PL1 (63.6%) or PL2 tumours (54.1%). On subgroup analysis according to nodal status, PL0 was associated with a higher survival rate than that of PL1 or PL2 tumours in patients with N0 or N1 metastasis, but not in those with N2 metastasis. There was no difference between PL1 and PL2 in any subgroup. CONCLUSIONS: Our results suggest that the presence of VPI, rather than the extent, has an impact on postoperative survival in patients with NSCLC who have N0 or N1 metastasis. Because very few previous studies have addressed the effects of VPI in patients with N1 disease, further re-evaluation of the prognostic impact of VPI is necessary in this subgroup of patients. Keywords: Non-small-cell lung cancer • Postoperative • Visceral pleura • Visceral pleural invasion
INTRODUCTION Because the prognostic implications of visceral pleural invasion (VPI) are well established, tumours with VPI were up-staged to T2 even if they were 3 cm or less in diameter according to the sixth edition of the TNM staging system for lung cancer issued by the International Union Against Cancer (UICC-6) [1]. Although Hammar [2] classified VPI into three clinically significant stages (P0, P1 and P2), the UICC-6 did not precisely define VPI. It was therefore unclear whether P1 status should be included as a T2 descriptor. In the newest seventh edition of TNM staging system for lung cancer by the International Union Against Cancer (UICC-7), revised in 2010 [3], the degree of VPI was clearly † Presented at the 28th Annual Meeting of the European Association for CardioThoracic Surgery, Milan, Italy, 11–15 October 2014.
classified into four stages (PL0, PL1, PL2 and PL3) according to the extent of tumour invasion, and both PL1 and PL2 were defined as VPI. However, the UICC-7 has stated that the definition of VPI, especially the degree of VPI and its classification of PL status, remains to be validated because the International Association for the Study of Lung Cancer (IASLC) team has not yet analysed the effect of VPI on survival because no detailed pathological information on VPI were submitted to their project [4]. It thus remains controversial whether the extent of VPI affects postoperative survival in patients with completely resected non-small-cell lung cancer (NSCLC). In addition, few series have addressed the relationship between VPI and nodal metastases with respect to survival, although VPI is considered strongly related to nodal metastases [5–7]. The present study was designed to evaluate the impact of the extent of VPI on postoperative survival according to the status of lymphnode metastasis in patients with completely resected NSCLC.
© The Author 2015. Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery. All rights reserved.
THORACIC
Cite this article as: Adachi H, Tsuboi M, Nishii T, Yamamoto T, Nagashima T, Ando K et al. Influence of visceral pleural invasion on survival in completely resected non-small-cell lung cancer. Eur J Cardiothorac Surg 2015; doi:10.1093/ejcts/ezu515.
2
H. Adachi et al. / European Journal of Cardio-Thoracic Surgery
MATERIALS AND METHODS Patients A total of 981 consecutive patients with NSCLC who underwent pulmonary resection at 9 hospitals affiliated with the Yokohama Consortium of Thoracic Surgeons (Yokohama City University Hospital and affiliated hospitals) from January 2005 to December 2007 were studied. Every 3–6 months, postoperative follow-up examinations were done by the authors and their colleagues at each hospital for at least 5 years after resection. Among these patients, 734 underwent complete anatomic resection (lobectomy or greater with systematic lymph-node dissection) without receiving induction chemotherapy or radiotherapy. After operation, 692 patients were classified as having pathological T1–3 N0–2 disease according to UICC-7. We excluded 44 patients with pathological PL3 tumours (interlobar invasion, parietal pleural invasion or both), 4 patients with tumours involving the main bronchus within 2 cm from the carina, 4 patients with carcinoid tumours and 1 patient with inadequate data on tumour recurrence. The remaining 639 patients were studied.
Pathological evaluation All specimens were stained with haematoxylin and eosin and were evaluated histopathologically by pathologists in each participating hospital. Application of elastic stains, such as Elastica van Gieson stain, was not mandatory, and each pathologist used such stains when it was difficult to evaluate the extent of VPI on haematoxylin and eosin staining. VPI status was defined according to the criteria described in UICC-7: PL0 represents a tumour within the subpleural lung parenchyma or superficially invading the pleural connective tissue beneath the elastic layer; PL1 represents a tumour invading beyond the elastic layer, but not exposed on the visceral pleural surface and PL2 represents a tumour invading and exposed on the visceral pleural surface, but not involving adjacent anatomic structures. PL1 or PL2 was defined as the presence of VPI, and PL0 was defined as the absence of VPI.
Statistical analysis After receiving approval from the institutional review boards of the participating hospitals, we retrospectively reviewed the medical records of each patient and recorded the following clinicopathological variables: age, gender, extent of pulmonary resection, histology, pathological maximum tumour diameter, pathological lymph-node metastasis and adjuvant chemotherapy. As for adjuvant chemotherapy, detailed data on the regimens used and reasons for treatment were unavailable for most patients. We therefore only surveyed whether or not patients received adjuvant chemotherapy. Overall survival was defined as the period between the date of surgical resection and the date of death from any cause. We initially analysed overall survival of the study group as a whole stratified according to the extent of VPI to evaluate differences in survival among patients with PL0, PL1 and PL2 tumours. Second, to evaluate the impact of VPI on nodal status, we performed subgroup analyses to assess overall survival according to nodal status in two ways. First, we stratified patients according to the extent of VPI (PL0, PL1 and PL2) to examine the relation between the extent
of VPI and the stage of lymph-node metastasis. Then, we stratified patients according the presence or absence of VPI to evaluate the validity of VPI as defined by UICC-7 in each stage of lymph-node metastasis. If a survival difference was found on these subgroup analyses, we performed additional analyses after classifying the patients into four groups according to tumour size and VPI status. Clinicopathological variables were contrasted between groups with the use of the Kruskal–Wallis test and χ 2 test. Survival curves were estimated using the Kaplan–Meier method and contrasted between groups with the log-rank test. To identify independent prognostic factors, we first plotted survival curves stratified according to each clinicopathological variable and compared the results between groups with the use of the log-rank test. Then, variables that showed statistical significance on univariate analysis were included in multivariate analysis, performed with using a Cox proportional-hazards model. Forward and backward stepwise procedures were used, with inclusion and exclusion criteria for a variable of P < 0.05 and P < 0.10, respectively, to determine the combination of factors that best predicted outcomes. All statistical analyses were performed with the use of IBM SPSS statistics 18 (IBM Corp., New York, NY, USA), and P-values of less than 0.05 were considered to indicate statistical significance.
RESULTS The 639 subjects comprised 400 men and 239 women with a mean age of 67.2 years (range, 29–84). The extent of VPI was PL0 in 462 patients, PL1 in 135 and PL2 in 42. The patients’ characteristics are shown according to the extent of VPI given in Table 1. The pathological evaluations of maximum tumour diameter and lymphnode metastasis correlated with VPI status. Adjuvant chemotherapy was given to a higher proportion of patients with advanced disease. The median follow-up after lung resection was 65.0 months (range, 1–97). There was no significant difference in the follow-up period among PL0, PL1 and PL2 (P = 0.055). The patients’ outcomes are given in Table 2. A total of 193 patients had postoperative recurrences, and patients with PL1 or PL2 tumours tended to have higher incidences of distant metastases than patients with PL0, but the differences were not statistically significant (P = 0.32). The frequency of intrathoracic recurrence also did not differ significantly among PL0, PL1 and PL2. The 5-year overall survival rate of the 639 subjects was 72.2%. The survival curves for the study group as a whole are shown according to the PL status in Fig. 1. The 5-year overall survival rate was 75.9% in patients with PL0 tumours, 63.6% in those with PL1 tumours and 54.1% in those with PL2 tumours. The survival rate differed significantly between patients with PL0 tumours and those with PL1 tumours, but not between patients with PL1 tumours and those with PL2 tumours. Univariate analysis by the log-rank test revealed that age (P < 0.001), gender (P < 0.001), extent of pulmonary resection (P = 0.010), histology (P < 0.001), maximum tumour diameter (P < 0.001), VPI (P = 0.001) and lymph-node metastasis (P < 0.001) were significant predictors of overall survival. Adjuvant chemotherapy was not a significant predictor on univariate analysis (P = 0.509). On multivariate analysis, including the variables that showed statistical significance on univariate analysis, VPI remained one of the independent predictors of overall survival [hazard ratio (HR) = 1.397; P = 0.028] (Table 3). On subgroup analyses examining the relation of VPI status to survival according to nodal status, overall survival differed significantly between PL0 and PL1 in the subgroup of patients with N0
H. Adachi et al. / European Journal of Cardio-Thoracic Surgery
3
Table 1: Patient characteristics according to the extent of VPI
Age (mean ±SD; range) Gender Male Female Extent of resection Lobectomy or bilobectomy Pneumonectomy Histology Adenocarcinoma Squamous cell carcinoma Large cell carcinoma Others Maximum tumour diameter (cm) ≤3.0 3.1–5.0 5.1–7.0 >7.0 Lymph-node metastases N0 N1 N2 Adjuvant chemotherapy No Yes Unknown
No. of patients (%) Total number (n = 639)
PL0 (n = 462)
PL1 (n = 135)
PL2 (n = 42)
P-value
67.2 ± 9.1; 29–85
67.1 ± 9.1; 29–84
67.9 ± 8.8; 41–85
65.9 ± 10.3; 37–82
0.52
400 (62.6) 239 (37.4)
282 (61.0) 180 (39.0)
91 (67.4) 44 (32.6)
27 (64.3) 15 (35.7)
0.39
620 (97.0) 19 (3.0)
450 (97.4) 12 (2.6)
131 (97.0) 4 (3.0)
39 (92.9) 3 (7.1)
0.25
409 (64.0) 157 (24.6) 30 (4.7) 43 (6.7)
302 (65.4) 113 (24.5) 16 (3.4) 31 (6.7)
79 (58.5) 38 (28.1) 11 (8.2) 7 (5.2)
28 (66.7) 6 (14.3) 3 (7.1) 5 (11.9)
0.090
383 (60.0) 179 (28.0) 57 (8.9) 20 (3.1)
302 (65.4) 108 (23.4) 41 (8.9) 11 (2.3)
64 (47.4) 51 (37.8) 13 (9.6) 7 (5.2)
17 (40.5) 20 (47.6) 3 (7.1) 2 (4.8)
0.001
502 (78.6) 69 (11.0) 68 (10.4)
379 (82.0) 39 (8.5) 44 (9.5)
96 (71.1) 18 (13.3) 21 (15.6)
27 (64.3) 12 (28.6) 3 (7.1)
3.0 cm) VPI (PL1 or PL2) Lymph-node metastases (N1 or N2) Adjuvant chemotherapy (no) HR: hazard ratio; CI: confidence interval. *P-value in a Cox proportional-hazards model.
Univariate
Multivariate
HR
95% CI
P-value*
HR
95% CI
P-value
1.761 2.865 2.264 2.941 2.217 1.681 2.941 0.897
1.325–2.336 2.012–4.098 1.197–4.281 2.208–3.906 1.667–2.950 1.253–2.252 2.203–3.937 0.649–1.239
ORIGINAL ARTICLE
European Journal of Cardio-Thoracic Surgery (2015) 1–7 doi:10.1093/ejcts/ezu515
Influence of visceral pleural invasion on survival in completely resected non-small-cell lung cancer† Hiroyuki Adachi*, Masahiro Tsuboi, Teppei Nishii, Taketsugu Yamamoto, Takuya Nagashima, Kohei Ando, Yoshihiro Ishikawa, Tekkan Woo, Katsuya Watanabe, Yutaka Kumakiri, Takamitsu Maehara, Takao Morohoshi, Haruhiko Nakayama and Munetaka Masuda Department of Surgery, Yokohama City University, Yokohama, Japan * Corresponding author. The Department of General Thoracic Surgery, Yokosuka Kyosai Hospital, 1-16, Yonegahama-dori, Yokosuka, Kanagawa 238-8558, Japan. Tel: +81-468-222710; fax: +81-468-252103; e-mail: [email protected] (H. Adachi). Received 10 September 2014; received in revised form 11 November 2014; accepted 18 November 2014
Abstract OBJECTIVES: Although the prognostic implications of visceral pleural invasion (VPI) are well established, it remains controversial whether the extent of VPI affects survival in patients with completely resected non-small-cell lung cancer (NSCLC). In addition, the impact of VPI according to nodal status is unclear. We evaluated the influence of the extent of pleural invasion on survival by analysing a multicentre retrospective database of patients who had undergone surgery for NSCLC. METHODS: We retrospectively reviewed the clinicopathological characteristics and outcomes of 639 patients with NSCLC who underwent anatomic complete resection from 2005 to 2007 at nine hospitals affiliated with the Yokohama Consortium of Thoracic Surgeons. RESULTS: The median follow-up was 65.0 months. The extent of pleural invasion was PL0 in 462 patients, PL1 in 135 and PL2 in 42. The 5-year overall survival rate was significantly higher in patients with PL0 tumours (75.9%) than in those with PL1 (63.6%) or PL2 tumours (54.1%). On subgroup analysis according to nodal status, PL0 was associated with a higher survival rate than that of PL1 or PL2 tumours in patients with N0 or N1 metastasis, but not in those with N2 metastasis. There was no difference between PL1 and PL2 in any subgroup. CONCLUSIONS: Our results suggest that the presence of VPI, rather than the extent, has an impact on postoperative survival in patients with NSCLC who have N0 or N1 metastasis. Because very few previous studies have addressed the effects of VPI in patients with N1 disease, further re-evaluation of the prognostic impact of VPI is necessary in this subgroup of patients. Keywords: Non-small-cell lung cancer • Postoperative • Visceral pleura • Visceral pleural invasion
INTRODUCTION Because the prognostic implications of visceral pleural invasion (VPI) are well established, tumours with VPI were up-staged to T2 even if they were 3 cm or less in diameter according to the sixth edition of the TNM staging system for lung cancer issued by the International Union Against Cancer (UICC-6) [1]. Although Hammar [2] classified VPI into three clinically significant stages (P0, P1 and P2), the UICC-6 did not precisely define VPI. It was therefore unclear whether P1 status should be included as a T2 descriptor. In the newest seventh edition of TNM staging system for lung cancer by the International Union Against Cancer (UICC-7), revised in 2010 [3], the degree of VPI was clearly † Presented at the 28th Annual Meeting of the European Association for CardioThoracic Surgery, Milan, Italy, 11–15 October 2014.
classified into four stages (PL0, PL1, PL2 and PL3) according to the extent of tumour invasion, and both PL1 and PL2 were defined as VPI. However, the UICC-7 has stated that the definition of VPI, especially the degree of VPI and its classification of PL status, remains to be validated because the International Association for the Study of Lung Cancer (IASLC) team has not yet analysed the effect of VPI on survival because no detailed pathological information on VPI were submitted to their project [4]. It thus remains controversial whether the extent of VPI affects postoperative survival in patients with completely resected non-small-cell lung cancer (NSCLC). In addition, few series have addressed the relationship between VPI and nodal metastases with respect to survival, although VPI is considered strongly related to nodal metastases [5–7]. The present study was designed to evaluate the impact of the extent of VPI on postoperative survival according to the status of lymphnode metastasis in patients with completely resected NSCLC.
© The Author 2015. Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery. All rights reserved.
THORACIC
Cite this article as: Adachi H, Tsuboi M, Nishii T, Yamamoto T, Nagashima T, Ando K et al. Influence of visceral pleural invasion on survival in completely resected non-small-cell lung cancer. Eur J Cardiothorac Surg 2015; doi:10.1093/ejcts/ezu515.
2
H. Adachi et al. / European Journal of Cardio-Thoracic Surgery
MATERIALS AND METHODS Patients A total of 981 consecutive patients with NSCLC who underwent pulmonary resection at 9 hospitals affiliated with the Yokohama Consortium of Thoracic Surgeons (Yokohama City University Hospital and affiliated hospitals) from January 2005 to December 2007 were studied. Every 3–6 months, postoperative follow-up examinations were done by the authors and their colleagues at each hospital for at least 5 years after resection. Among these patients, 734 underwent complete anatomic resection (lobectomy or greater with systematic lymph-node dissection) without receiving induction chemotherapy or radiotherapy. After operation, 692 patients were classified as having pathological T1–3 N0–2 disease according to UICC-7. We excluded 44 patients with pathological PL3 tumours (interlobar invasion, parietal pleural invasion or both), 4 patients with tumours involving the main bronchus within 2 cm from the carina, 4 patients with carcinoid tumours and 1 patient with inadequate data on tumour recurrence. The remaining 639 patients were studied.
Pathological evaluation All specimens were stained with haematoxylin and eosin and were evaluated histopathologically by pathologists in each participating hospital. Application of elastic stains, such as Elastica van Gieson stain, was not mandatory, and each pathologist used such stains when it was difficult to evaluate the extent of VPI on haematoxylin and eosin staining. VPI status was defined according to the criteria described in UICC-7: PL0 represents a tumour within the subpleural lung parenchyma or superficially invading the pleural connective tissue beneath the elastic layer; PL1 represents a tumour invading beyond the elastic layer, but not exposed on the visceral pleural surface and PL2 represents a tumour invading and exposed on the visceral pleural surface, but not involving adjacent anatomic structures. PL1 or PL2 was defined as the presence of VPI, and PL0 was defined as the absence of VPI.
Statistical analysis After receiving approval from the institutional review boards of the participating hospitals, we retrospectively reviewed the medical records of each patient and recorded the following clinicopathological variables: age, gender, extent of pulmonary resection, histology, pathological maximum tumour diameter, pathological lymph-node metastasis and adjuvant chemotherapy. As for adjuvant chemotherapy, detailed data on the regimens used and reasons for treatment were unavailable for most patients. We therefore only surveyed whether or not patients received adjuvant chemotherapy. Overall survival was defined as the period between the date of surgical resection and the date of death from any cause. We initially analysed overall survival of the study group as a whole stratified according to the extent of VPI to evaluate differences in survival among patients with PL0, PL1 and PL2 tumours. Second, to evaluate the impact of VPI on nodal status, we performed subgroup analyses to assess overall survival according to nodal status in two ways. First, we stratified patients according to the extent of VPI (PL0, PL1 and PL2) to examine the relation between the extent
of VPI and the stage of lymph-node metastasis. Then, we stratified patients according the presence or absence of VPI to evaluate the validity of VPI as defined by UICC-7 in each stage of lymph-node metastasis. If a survival difference was found on these subgroup analyses, we performed additional analyses after classifying the patients into four groups according to tumour size and VPI status. Clinicopathological variables were contrasted between groups with the use of the Kruskal–Wallis test and χ 2 test. Survival curves were estimated using the Kaplan–Meier method and contrasted between groups with the log-rank test. To identify independent prognostic factors, we first plotted survival curves stratified according to each clinicopathological variable and compared the results between groups with the use of the log-rank test. Then, variables that showed statistical significance on univariate analysis were included in multivariate analysis, performed with using a Cox proportional-hazards model. Forward and backward stepwise procedures were used, with inclusion and exclusion criteria for a variable of P < 0.05 and P < 0.10, respectively, to determine the combination of factors that best predicted outcomes. All statistical analyses were performed with the use of IBM SPSS statistics 18 (IBM Corp., New York, NY, USA), and P-values of less than 0.05 were considered to indicate statistical significance.
RESULTS The 639 subjects comprised 400 men and 239 women with a mean age of 67.2 years (range, 29–84). The extent of VPI was PL0 in 462 patients, PL1 in 135 and PL2 in 42. The patients’ characteristics are shown according to the extent of VPI given in Table 1. The pathological evaluations of maximum tumour diameter and lymphnode metastasis correlated with VPI status. Adjuvant chemotherapy was given to a higher proportion of patients with advanced disease. The median follow-up after lung resection was 65.0 months (range, 1–97). There was no significant difference in the follow-up period among PL0, PL1 and PL2 (P = 0.055). The patients’ outcomes are given in Table 2. A total of 193 patients had postoperative recurrences, and patients with PL1 or PL2 tumours tended to have higher incidences of distant metastases than patients with PL0, but the differences were not statistically significant (P = 0.32). The frequency of intrathoracic recurrence also did not differ significantly among PL0, PL1 and PL2. The 5-year overall survival rate of the 639 subjects was 72.2%. The survival curves for the study group as a whole are shown according to the PL status in Fig. 1. The 5-year overall survival rate was 75.9% in patients with PL0 tumours, 63.6% in those with PL1 tumours and 54.1% in those with PL2 tumours. The survival rate differed significantly between patients with PL0 tumours and those with PL1 tumours, but not between patients with PL1 tumours and those with PL2 tumours. Univariate analysis by the log-rank test revealed that age (P < 0.001), gender (P < 0.001), extent of pulmonary resection (P = 0.010), histology (P < 0.001), maximum tumour diameter (P < 0.001), VPI (P = 0.001) and lymph-node metastasis (P < 0.001) were significant predictors of overall survival. Adjuvant chemotherapy was not a significant predictor on univariate analysis (P = 0.509). On multivariate analysis, including the variables that showed statistical significance on univariate analysis, VPI remained one of the independent predictors of overall survival [hazard ratio (HR) = 1.397; P = 0.028] (Table 3). On subgroup analyses examining the relation of VPI status to survival according to nodal status, overall survival differed significantly between PL0 and PL1 in the subgroup of patients with N0
H. Adachi et al. / European Journal of Cardio-Thoracic Surgery
3
Table 1: Patient characteristics according to the extent of VPI
Age (mean ±SD; range) Gender Male Female Extent of resection Lobectomy or bilobectomy Pneumonectomy Histology Adenocarcinoma Squamous cell carcinoma Large cell carcinoma Others Maximum tumour diameter (cm) ≤3.0 3.1–5.0 5.1–7.0 >7.0 Lymph-node metastases N0 N1 N2 Adjuvant chemotherapy No Yes Unknown
No. of patients (%) Total number (n = 639)
PL0 (n = 462)
PL1 (n = 135)
PL2 (n = 42)
P-value
67.2 ± 9.1; 29–85
67.1 ± 9.1; 29–84
67.9 ± 8.8; 41–85
65.9 ± 10.3; 37–82
0.52
400 (62.6) 239 (37.4)
282 (61.0) 180 (39.0)
91 (67.4) 44 (32.6)
27 (64.3) 15 (35.7)
0.39
620 (97.0) 19 (3.0)
450 (97.4) 12 (2.6)
131 (97.0) 4 (3.0)
39 (92.9) 3 (7.1)
0.25
409 (64.0) 157 (24.6) 30 (4.7) 43 (6.7)
302 (65.4) 113 (24.5) 16 (3.4) 31 (6.7)
79 (58.5) 38 (28.1) 11 (8.2) 7 (5.2)
28 (66.7) 6 (14.3) 3 (7.1) 5 (11.9)
0.090
383 (60.0) 179 (28.0) 57 (8.9) 20 (3.1)
302 (65.4) 108 (23.4) 41 (8.9) 11 (2.3)
64 (47.4) 51 (37.8) 13 (9.6) 7 (5.2)
17 (40.5) 20 (47.6) 3 (7.1) 2 (4.8)
0.001
502 (78.6) 69 (11.0) 68 (10.4)
379 (82.0) 39 (8.5) 44 (9.5)
96 (71.1) 18 (13.3) 21 (15.6)
27 (64.3) 12 (28.6) 3 (7.1)
3.0 cm) VPI (PL1 or PL2) Lymph-node metastases (N1 or N2) Adjuvant chemotherapy (no) HR: hazard ratio; CI: confidence interval. *P-value in a Cox proportional-hazards model.
Univariate
Multivariate
HR
95% CI
P-value*
HR
95% CI
P-value
1.761 2.865 2.264 2.941 2.217 1.681 2.941 0.897
1.325–2.336 2.012–4.098 1.197–4.281 2.208–3.906 1.667–2.950 1.253–2.252 2.203–3.937 0.649–1.239
