3.1 Patient data and D-dimer levels in stage Ⅰ-Ⅱ GC patients
In total, 467 GC patients (361 men and 106 women, aged 24-81 years) were included in the study. Basic patient data (age, gender and other clinical characters) and plasma D-dimer levels (median and 25th-75th percentile) are listed in Table 1. Based on the results of the clinical evaluation, all the GC patients received R0 resection, while 111 patients with T3+ or N (+) underwent palliative chemotherapy additionally. Plasma D-dimer levels were tested and compared among different TNM stages, and the results are shown in Table 1. The results did not show significant difference of D-dimer levels among different G (p=0.429), TNM stages (p=0.866), T stages (p=0.633), or N stages (p=0.506). While the results displayed a significant increased D-dimer levels in the tumor embolus positive patients (p=0.011) and patients whose age over 60 years old (p=0.012). In addition, gender (p=0.056), tumor location (p=0.149) or if received chemotherapy (p=0.78) did not display a significant difference in D-dimer levels in this study. The results shown in Table 1 and Figure 1.
3.2 Identify the risk factors by overall survival analysis in stage Ⅰ-Ⅱ GC patients.
All the 467 patients were successfully examined during follow-up. The median follow up is 34 m (2m-95m). During the ten years follow-up, 86 of the 467 patients died; the mortality is 18.4% while the overall survival rate was 81.6%, one year survival rate is 86%, 3 years survival rate is 79%, and 5 years and 7 years survival rate is 75% and 73% by life table analysis, which were shown in Table 2
When the high and low groups were defined according to the plasma D-dimer levels at 1.5 mg/ml based on our previous study (1.5 folds than normal levels in clinical practice), 48/ 467(10.28%) GC patients were with high D-dimer levels, the survival time was shorter in high level groups( 38.21m (29.77-46.65m) ) than low groups(79.85m (76.55-83.16m)), and the overall survival is 60.4% and 84.0%, respectively (p<0.001) from the log-rank survival analysis, shown in Figure 2. Additionally, survival time correlated with tumor invasion (p=0.001), TNM stages (p=0.003), and FIB levels (p=0.007) based on the log-rank survival test. However, survival time was not influenced by patient gender (p=0.204), age (p=0.224), histological grade (p=0.231), lymph node invasion (p=0.530), tumor location (p=0.163), vascular cancer emboli (p=0.686), or additional chemotherapy treatment (p=0.685). Moreover, survival time was not influenced by CEA levels (p=0.542), CA199 levels (p=0.122), CA724 levels (p=0.964) or FDP levels (p=0.056) by log-rank test. Shown in Table 2.
Based on the Cox proportional hazards regression analysis, survival time was influenced by D-dimer levels (p<0.001, Risk ratio=3.60, 95%CI: 2.13-6.08), tumor invasion (p=0.001, Risk ratio=1.34, 95%CI: 1.10-1.63), but FIB levels (p=0.135, Risk ratio=1.49, 95%CI: 0.88-2.50) did not show significant difference. The results are shown in Table 3.
3.3 Identify the risk factors with Disease-Free Survival (DFS) in stage Ⅰ-Ⅱ GC patients.
Tumor recurrence and Disease-Free Survival (DFS) is the most crucial clinical event associated with poor prognosis of surgical GC before death and might be regarded as more objective than mortality. In this study, 98 of the 467 patients were found tumor recurrence within the follow up, the recurrence rate was 21.0 % overall, and the 1 year recurrence is 17%, 3 years recurrence is 25%, while 5 year and 7 year recurrence is 28%, 30% respectively. Shown in Table 4.
A significant difference in DFS was observed between the GC patients with high D-dimer levels and those with low levels (p<0.001),and the recurrence rate is 54.17% (26/ 48) in high groups and 17.18% (72/ 419) in low groups. The result was shown in Table 4 and Figure3. Besides that, univariate analysis by log-rank test showed that tumor invasion (p=0.002), TNM stage(p=0.002), Age(p=0.042), FIB levels(p=0.003), and FDP levels(p<0.001) were significant predictive factors for DFS, which was not associated with lymph node invasion(p=0.530), Gender(p=0.649), additional treatment(p=0.377), cancer emboli(p=0.273), histological grade (p=0.385) and tumor location (p=0.061), CEA levels(p=0.290), CA199 levels(p=0.203), and CA724 levels(p=0.805). Shown in Table 4.
However, multivariate analysis by Cox proportional hazards regression showed that only D-dimer levels (p<0.001, Risk ratio=5.57, 95%CI: 3.22-9.63) and tumor invasion (p=0.002, Risk ratio=1.34, 95%CI: 1.12-1.62) was significant independent risk factor, but FIB levels (p=0.11, Risk ratio=1.49, 95%CI: 0.92-2.42). FDP levels (p=0.45, Risk ratio=0.75, 95%CI: 0.35-1.59)and Age (p=0.35, Risk ratio=1.22, 95%CI: 0.81-1.85) did not show significant difference in Cox’s proportional hazards model for DFS as listed in Table 5.
3.4 Effectiveness of predict OS and PFS by plasma D-dimer level based on 1.0mg/l.
In our previous study, we defined the D-dimer group by 1.5 fold, while if the levels is 1.0mg/l, the result was tested again in this study. Of the 467 patients with TNM stage Ⅰ-Ⅱ, 76 patients showed higher d-dimer over the level 1.0mg/l. The survival time was short for patients with high D-dimer levels (p<0.001, Risk ratio=2.70 95%CI: 1.65-4.40) based on log-rank and Cox proportional hazards regression model (Table 2 and supplement Table 1). What’s more, DFS was short in patients with high D-dimer levels (p<0.001, Risk ratio=3.39, 95%CI: 2.07-5.54) (Table 4 and supplement Table 2) according to both log-rank and Cox regression analyzes as well. Shown in Figure 4.