A total of 939 patients with pT3 ESCC were included, but follow-up was complete in 838 patients (89.2%). Of these 838 patients, the male to female ratio was 4.1:1 (672:166) with the median age of 60 years at surgery (range 31-82 years). CRM positive (R1) was found in 59 (7%) patients according to the CAP criteria and 317 (37.8%) patients according to the RCP criteria. Lymphovascular invasion (LVI), perineural invasion (PNI), and lymph node metastasis were found in 446 (53.2%), 521 (62.2%), and 406 (48.4%) patients, respectively. The clinicopathologic parameters of the entire cohort were summarized in Table 1.
Table 1
Summary of clinical and histopathological characteristics of the 838 pT3N0-3M0 esophageal squamous cell carcinoma patients, the 5-year (5-yr), 10-year (10-yr) overall survival (OS) rate (%) and the 3-year (3-yr), 5-year (5-yr) disease-free survival (DFS) rate (%)
Characteristic | | N (%) | 5-yr OS (%) | 10-yr OS (%) | P-value | 3-yr DFS (%) | 5-yr DFS (%) | P-value |
Sex | Male | 672(80.2) | 45.7 | 39.9 | 0.881 | 46.2 | 40.9 | 0.704 |
Female | 166(19.8) | 44.5 | 33.3 | | 49.4 | 37.5 | |
Age (years) | ≤60 | 464(55.4) | 48.9 | 44.2 | 0.007 | 50.3 | 44.0 | 0.029 |
>60 | 374(44.6) | 41.2 | 31.7 | | 42.5 | 35.3 | |
Tumor location | Upper thoracic | 119(14.2) | 47.6 | 39.6 | 0.694 | 49.6 | 41.9 | 0.893 |
Middle thoracic | 486(58.0) | 45.9 | 41.7 | | 46.8 | 40.2 | |
Lower thoracic | 233(27.8) | 43.7 | 33.3 | | 45.6 | 39.2 | |
Degree of differentiation | Well | 376(44.9) | 49.1 | 41.8 | 0.001 | 53.1 | 44.9 | <0.001 |
Moderate | 392(46.8) | 44.8 | 37.6 | | 43.5 | 37.9 | |
Poor | 70(8.4) | 29.9 | 27.8 | | 32.6 | 27.9 | |
LVI | No | 392(46.8) | 55.9 | 48.5 | <0.001 | 57.7 | 50.9 | <0.001 |
Yes | 446(53.2) | 36.4 | 29.7 | | 37.5 | 30.9 | |
PNI | No | 317(37.8) | 54.3 | 46.2 | <0.001 | 55.5 | 48.0 | 0.001 |
Yes | 521(62.2) | 40.1 | 34.0 | | 41.5 | 35.3 | |
pN | N0 | 405(48.3) | 61.6 | 52.6 | <0.001 | 63.8 | 56.1 | <0.001 |
N1 | 238(28.4) | 40.3 | 34.2 | | 41.1 | 34.6 | |
N2 | 140(16.7) | 21.1 | 16.9 | | 22.6 | 16.9 | |
N3 | 55(6.6) | 13.7 | 13.7 | | 9.0 | 6.8 | |
CRM CAP criteria | R0 | 779(93.0) | 47.2 | 39.8 | <0.001 | 48.5 | 41.6 | <0.001 |
R1 | 59(7.0) | 22.9 | 22.9 | | 26.0 | 20.8 | |
CRM RCP criteria | R0 | 521(62.2) | 50.6 | 44.2 | <0.001 | 51.4 | 44.8 | 0.001 |
R1 | 317(37.8) | 37.0 | 29.3 | | 39.3 | 32.5 | |
CRM three-tier criteria | ༞1 mm | 521 (62.2) | 50.6 | 44.2 | <0.001 | 51.4 | 44.8 | <0.001 |
0-1 mm | 258(30.8) | 40.3 | 30.6 | | 42.4 | 35.2 | |
0 mm | 59 (7.0) | 22.9 | 22.9 | | 26.0 | 20.8 | |
CAP, College of American Pathologists; CRM: circumferential resection margin; LVI: lymphovascular invasion; RCP, Royal College of Pathologists; PNI: perineural invasion. |
Overall Survival And Crm Status
The median follow-up time for the entire study population was 44 months (95%CI 34.2-53.8 months), and the 5- and 10-year OS rates were 45.5% and 38.6%, respectively. The median OS of patients who were diagnosed as R0 and R1 according to CAP criteria were 49 months (95%CI 35.7-62.3 months) and 15 months (95%CI 11.7-18.3 months), respectively (P < 0.001). Median OS of patients who were diagnosed as R0 and R1 according to the RCP criteria were 66 months (95%CI 39.3-92.7 months) and 29 months (95%CI 23.0-35.0 months), respectively (P < 0.001). The Kaplan-Meier survival curve is presented in Figure 2. Patients with R1 had a significantly shorter OS than those with R0, according to either RCP or CAP criteria used (P < 0.001, both; log-rank test). When applying the 3-tier stratification system for CRM status, the median OS of patients with CRM 0mm, 0-1 mm, and >1 mm were 15 months (95%CI 11.7-18.3 months), 33 months (95%CI 24.3-41.7 months), and 66 months (95%CI 39.3-92.7 months), respectively (P < 0.001). And there was significant difference between groups with CRM 0 mm versus CRM 0-1 mm, CRM 0 mm versus CRM >1 mm, and CRM 0-1 mm versus CRM >1 mm (P = 0.002, P < 0.001, and P = 0.001, respectively, log-rank test) (Fig. 2c).
The OS of older patients was significantly worse compared with that of young patients. Similarly, cases with poor differentiation, LVI, PNI, and higher lymph node metastasis numbers had worse outcome, and the relevant survival data were shown in Table 1.
Univariate Cox proportional hazards model indicated a significant relationship between OS and CRM status, according to the CAP, RCP, and three-tier criteria, the patient’s age, degree of tumor differentiation, LVI, PNI, and pN (Table 2). Multivariate Cox regression analysis was performed with risk factors that were statistically significant on univariate analysis. The results of multivariate Cox proportional hazards analysis suggested that the patient’s age, tumor differentiation, pN and CRM status, according to the CAP and RCP criteria, were independent prognostic factors for OS (Table 2). But the difference between CRM 0-1 mm and CRM >1 mm was not statistically significant (P = 0.117).
Table 2
Univariate and Multivariate Cox proportional hazards analysis of various prognostic factors and their relationship to overall survival
| Univariate | Multivariate (CAP) | Multivariate (RCP) | Multivariate (3-tier criteria) |
Parameter | HR (95% CI) | P value | HR (95% CI) | P value | HR (95% CI) | P value | HR (95% CI) | P value |
Sex | Male/Female | 0.983(0.783-1.233) | 0.882 | | | | | | |
Age (years) | ≤60/༞60 | 1.285(1.069-1.545) | 0.007 | 1.349(1.122-1.623) | 0.001 | 1.377(1.145-1.657) | 0.001 | 1.353(1.125-1.628) | 0.001 |
Tumor location | Upper thoracic | 1 | | | | | | | |
Middle thoracic | 0.995(0.756-1.312) | 0.974 | | | | | | |
Lower thoracic | 1.086(0.806-1.464) | 0.586 | | | | | | |
Degree of differentiation | Well and Moderate/Poor | 1.656(1.231-2.228) | 0.001 | 1.368(1.012-1.849) | 0.041 | 1.410(1.045-1.902) | 0.025 | 1.370(1.014-1.852) | 0.040 |
LVI | No/Yes | 1.724(1.426-2.084) | <0.001 | 1.206(0.982-1.480) | 0.074 | 1.212(0.988-1.486) | 0.066 | 1.195(0.973-1.467) | 0.089 |
PNI | No/Yes | 1.500(1.233-1.825) | <0.001 | 1.225(1.001-1.500) | 0.049 | 1.212(0.990-1.484) | 0.063 | 1.210(0.988-1.482) | 0.065 |
pN | N0 | 1 | | 1 | | 1 | | 1 | |
N1 | 1.783(1.420-2.237) | <0.001 | 1.725(1.369-2.175) | <0.001 | 1.658(1.315-2.091) | <0.001 | 1.702(1.349-2.147) | <0.001 |
N2 | 3.219(2.519-4.114) | <0.001 | 2.842(2.891-3.689) | <0.001 | 2.757(2.123-3.579) | <0.001 | 2.783(2.142-3.616) | <0.001 |
N3 | 4.536(3.248-6.334) | <0.001 | 3.917(2.770-5.537) | <0.001 | 3.969(2.808-5.611) | <0.001 | 3.889(2.749-5.501) | <0.001 |
CRM CAP criteria | R0/ R1 | 2.107(1.543-2.877) | <0.001 | 1.759(1.279-2.419) | 0.001 | | | | |
CRM RCP criteria | R0/ R1 | 1.517(1.260-1.826) | <0.001 | | | 1.279(1.059-1.545) | 0.011 | | |
CRM three-tier criteria | >1 mm | 1 | | | | | | 1 | |
0-1 mm | 1.378(1.128-1.683) | 0.002 | | | | | 1.176(0.960-1.441) | 0.117 |
0 mm | 2.350(1.706-3.238) | <0.001 | | | | | 1.869(1.346-2.595) | <0.001 |
CAP, College of American Pathologists; CRM, circumferential resection margin; LVI, lymphovascular invasion; RCP, Royal College of Pathologists; PNI, perineural invasion. |
Disease-free Survival
The median DFS time for the entire study population was 31 months (95%CI 26-36 months), and the 3- and 5-year DFS rates were 46.9% and 40.2%, respectively. When using the CAP criteria, the median OS was 34 months (95%CI 27.8-40.2 months) and 13 months (95%CI 8.9-17.1 months) for patients with R0 and R1 (P < 0.001), respectively, while using the RCP criteria, median OS was 39 months (95%CI 26.3-51.7 months) for R0 and 24 months (95%CI 19.2-28.8 months) for R1 (P = 0.001). The OS of patients diagnosed as R1 was significantly shorter compared with that of patients diagnosed as R0, according to either CAP or RCP criteria used (P < 0.001 and P = 0.001, respectively; log-rank test) (Fig. 3a and 3b). When applying the three-tier criteria for CRM status, the median OS for patients with CRM 0 mm, CRM 0-1 mm, and CRM >1 mm was 13 months (95%CI 8.9-17.1 months), 27 months (95%CI 21.4-32.6 months), and 39 months (95%CI 26.3-51.7 months), respectively (P < 0.001). And there was significant difference between groups with CRM 0 mm versus CRM 0-1mm, CRM 0 mm versus CRM >1 mm, and CRM 0-1 mm versus CRM >1 mm (P = 0.008, P < 0.001, and P = 0.018, respectively, log-rank test) (Fig. 3c).
The DFS time of older patients was significantly shorter compared with that of young patients. Similarly, cases with poor differentiation, LVI, PNI, and higher lymph node metastasis numbers had worse outcome, and the relevant survival data were shown in Table 1.
Univariate Cox proportional hazards model identified a significant relationship between DFS and CRM status, according to either the CAP or RCP, or the three-tier criteria, the patient’s age, degree of tumor differentiation, LVI, PNI, and pN (Table 3). Multivariate analyses of the above-mentioned prognostic factors confirmed R1 using the CAP criteria as an independent predictor for DFS. Patient’s age, LVI, pN also remained an independent prognostic factor (Table 3). CRM status, according to the CAP criteria, failed to be an independent prognostic factor. Although, CRM status, according to the three-tier criteria, was an independent prognostic factor, but the difference between groups CRM 0-1 mm versus CRM >1 mm was not statistically significant (P = 0.7).
Table 3
Univariate and Multivariate Cox proportional hazards analysis of various prognostic factors and their relationship to disease-free survival
| Univariate | Multivariate (CAP) | Multivariate (RCP) | Multivariate (3-tier criteria) |
Parameter | HR (95% CI) | P value | HR (95% CI) | P value | HR (95% CI) | P value | HR (95% CI) | P value |
Sex | Male/Female | 0.960(0.775-1.188) | 0.707 | | | | | | |
Age (years) | ≤60/༞60 | 1.213(1.018-1.446) | 0.031 | 1.292(1.083-1.541) | 0.002 | 1.301(1.091-1.552) | 0.003 | 1.292(1.084-1.541) | 0.004 |
Tumor location | Upper thoracic | 1 | | | | | | | |
Middle thoracic | 1.031(0.793-1.341) | 0.818 | | | | | | |
Lower thoracic | 1.068(0.801-1.423) | 0.655 | | | | | | |
Degree of differentiation | Well and Moderate/ Poor | 1.556(1.161-2.085) | 0.003 | 1.291(0.959-1.738) | 0.092 | 1.314(0.977-1.768) | 0.071 | 1.290(0.958-1.736) | 0.093 |
LVI | No/Yes | 1.733(1.447-2.077) | <0.001 | 1.244(1.022-1.515) | 0.029 | 1.256(1.032-1.528) | 0.023 | 1.241(1.019-1.512) | 0.031 |
PNI | No/Yes | 1.363(1.132-1.640) | 0.001 | 1.080(0.891-1.309) | 0.433 | 1.076(0.887-1.304) | 0.458 | 1.077(0.888-1.306) | 0.451 |
pN | N0 | 1 | | 1 | | 1 | | 1 | |
N1 | 1.813(1.462-2.247) | <0.001 | 1.752(1.406-2.182) | <0.001 | 1.703(1.367-2.123) | <0.001 | 1.745(1.399-2.175) | <0.001 |
N2 | 3.021(2.389-3.821) | <0.001 | 2.708(2.109-3.479) | <0.001 | 2.672(2.079-3.435) | <0.001 | 2.693(2.094-3.465) | <0.001 |
N3 | 4.882(3.529-6.753) | <0.001 | 4.257(3.028-5.985) | <0.001 | 4.243(3.016-5.969) | <0.001 | 4.241(3.014-5.966) | <0.001 |
CRM CAP criteria | R0/ R1 | 1.803(1.331-2.443) | <0.001 | 1.566(1.150-2.130) | 0.004 | | | | |
CRM RCP criteria | R0/ R1 | 1.362(1.140-1.627) | 0.001 | | | 1.124(0.937-1.348) | 0.209 | | |
CRM three-tier criteria | ༞1 mm | 1 | | | | | | 1 | |
0-1 mm | 1.255(1.036-1.521) | 0.020 | | | | | 1.039(0.855-1.264) | 0.700 |
0 mm | 1.945(1.424-2.655) | <0.001 | | | | | 1.588(1.157-2.181) | 0.004 |
CAP, College of American Pathologists; CRM: circumferential resection margin; LVI: lymphovascular invasion; RCP, Royal College of Pathologists; PNI: perineural invasion. |
Crm Status, Lymph Node Status, Os, And Dfs
As the previous results showed, lymph node metastasis was also an independent unfavorable prognostic factor, and whether it would affect the predictive value of CRM status is uncertain. Then overall survival and disease-free survival curves were analysed further for pN0, pN1-2, and pN3 subgroups using the CAP and RCP criteria, as well as the three-tier criteria. This analysis showed good separation of the OS and DFS curves within the pN0 and pN1-2 groups applying either the CAP or RCP, or the 3-tier criteria, but not the pN3 group (Fig. 4 and 5). Within the pN0 group, patients with CRM >1 mm had better survival than patients with CRM 0 mm and CRM 0-1 mm (OS, P = 0.005 and P < 0.001; DFS, P = 0.017 and P = 0.001; respectively, log-rank test) (Fig. 4 and 5). However, the difference in OS and DFS between CRM 0 mm and CRM 0-1 mm was not statistically significant within the pN0 group (P = 0.476 and P = 0.692, respectively, log-rank test). But in the pN1-2 group, patients with CRM 0 mm had worse survival than patients with CRM 0-1 mm and CRM >1 mm (OS, P < 0.001 and P < 0.001; DFS, P = 0.001 and P = 0.005; respectively, log-rank test) (Fig. 4 and 5). And there was no significant difference between CRM 0-1 mm and CRM >1 mm in OS and DFS within the pN1-2 group (P = 0.813 and P = 0.194, respectively, log-rank test). And the detailed univariate cox regression analyses data related to OS and DFS in pN0, pN1-2, and pN3 subgroups were shown in Table 4.
Table 4
CRM status and survival adjusted for lymph node status
| | N0 | N1-2 | N3 |
| | HR (95%CI) | P value | HR (95%CI) | P value | HR (95%CI) | P value |
OS | | | | | | | |
CAP criteria | R1 vs. R0 | 1.846(1.067-3.196) | 0.029 | 2.511(1.645-3.835) | <0.001 | 1.367(0.576-3.246) | 0.478 |
RCP criteria | R1 vs. R0 | 1.874(1.366-2.570) | <0.001 | 1.122(0.874-1.441) | 0.366 | 1.499(0.830-2.705) | 0.179 |
CRM 3-tier criteria | 0 mm vs. 0-1 mm | 1.236(0.686-2.228) | 0.481 | 2.521(1.590-3.996) | <0.001 | 1.112(0.437-2.833) | 0.824 |
0 mm vs. >1 mm | 2.208(1.258-3.876) | 0.006 | 2.483(1.608-3.835) | <0.001 | 1.596(0.642-3.967) | 0.314 |
0-1 mm vs. >1 mm | 1.801(1.283-2.529) | 0.001 | 0.970(0.741-1.270) | 0.825 | 1.468(0.776-2.777) | 0.238 |
DFS | | | | | | | |
CAP criteria | R1 vs. R0 | 1.615(0.952-2.740) | 0.076 | 1.936(1.271-2.949) | 0.002 | 1.369(0.611-3.071) | 0.446 |
RCP criteria | R1 vs. R0 | 1.735(1.287-2.339) | <0.001 | 0.956(0.751-1.218) | 0.716 | 1.196(0.676-2.115) | 0.538 |
CRM three-tier criteria | 0 mm vs. 0-1 mm | 1.120(0.634-1.979) | 0.695 | 2.126(1.345-3.359) | 0.001 | 1.266(0.523-3.067) | 0.601 |
0 mm vs. >1 mm | 1.884(1.097-3.232) | 0.022 | 1.820(1.184-2.800) | 0.006 | 1.440(0.613-3.380) | 0.403 |
0-1 mm vs. >1 mm | 1.700(1.233-2.345) | 0.001 | 0.845 (0.651-1.097) | 0.205 | 1.123(0.604-2.088) | 0.714 |
CAP, College of American Pathologists; CRM: circumferential resection margin; RCP, Royal College of Pathologists |