General characteristics of PDAC patients
Over the two years of subject eligibility, we identified 357 patients with histologically confirmed metastatic PDAC, of whom 37 had missing values in critical variables. Thus, our analyses included 320 patients. Major characteristics of the patients are described and compared in Table 1. The mean diagnosis age of patients was 65.3 years, males and females comparable. Median survival was 177 days. Although serum GGT can mildly vary by age and sex, in clinical practice, a uniform cut-off of 48 units/liter (U/L) is the most commonly used threshold for defining GGT elevation. We chose this value a priori to dichotomize the PDAC patients based on baseline serum GGT level. We found that, except for age, sex and serum FPG, the various ascertained characteristics were all significantly different between the two groups: compared to patients with normal baseline serum GGT, patients with elevated levels had much shorter median survival (138 versus 281 days), as well as generally higher other blood markers.
Baseline serum GGT and OS of metastatic PDAC
Product-limit survival curves of elevated and normal baseline serum GGT patients are displayed in Figure 1. OS of the elevated GGT group was notably inferior to survival of the normal GGT group (log-rank statistic: 23.52, p=10-6). Univariate Cox proportional-hazards models identified 4 potential prognostic covariates: age at diagnosis, palliative chemotherapy, baseline FPG and baseline GGT. With multivariate adjustment, only age at diagnosis, FPG and GGT remained significant. Age at diagnosis was positively associated with mortality: the adjusted hazard ratio (HR) was 1.08 (95%CI 1.01-1.15) per 5 years increase; elevated baseline FPG and serum GGT were associated with 1.39- (95%CI: 1.08-1.79) and 1.53- (95%CI: 1.19-1.97) fold mortality, respectively (Table 2).
We divided PDAC patients into 4 strata by quartile of baseline serum GGT: Q1 (GGT < 30.0 U/L), Q2 (30.0 U/L ≤ GGT < 85.5 U/L), Q3 (85.5 U/L ≤ GGT < 338.0 U/L), and Q4 (GGT ≥ 338.0 U/L). By using Q1 as the reference group, controlling for age at diagnosis, palliative chemotherapy, baseline FPG and baseline ALB, we found that the adjusted HRs for Q2 through Q4 were 1.36 (95%CI: 0.96-1.93), 1.53 (95%CI: 1.07-2.19), and 1.76 (95%CI: 1.24-2.49), respectively. The multiplicative continuous dose-response association between GGT and OS was statistically significant: every 10-fold increase in GGT was associated with a HR of 1.33 (95%CI: 1.09-1.61), and the p value for this continuous trend was 0.0043 (Figure 2).
Subgroup analysis
We further performed a small series of subgroup analyses based on GGT stratification by categories of palliative chemotherapy, baseline FPG and NLR. No obvious interaction was found between palliative chemotherapy, baseline NLR and serum GGT. However, an appreciable difference in the GGT-OS association was found when metastatic PDAC patients were dichotomized by baseline FPG: in patients with elevated baseline FPG (defined as ≥ 7.0mmol/L), GGT was not associated with OS, but in patients with normal baseline FPG (defined as < 7.0mmol/L), elevated serum GGT was associated with 2.14-fold mortality (95%CI: 1.48-3.09) (Table 3). This interaction did not reach statistical significance however (p =0.07).
Among the 320 PDAC patients, 76 and 97 were additionally measured for baseline C-reactive protein (CRP) and carbohydrate antigen 19-9 (CA19-9), respectively. Correlation analysis revealed that serum GGT was positively associated with CA19-9 (r=0.43, p=10-3.9), whereas the relationship between GGT and CRP was negligible (r=-0.01, p=0.97). Considering the appreciable correlation between GGT and CA19-9, we fitted a multivariate Cox regression model that included both CA19-9 and GGT in the subset of the 97 PDAC patients with baseline CA19-9 measurements: elevated GGT was still a significant prognostic factor (HR=1.53, 95%CI: 1.10-2.13), and CA19-9 was not significantly associated with OS (HR=1.50, 95%CI: 0.78-2.85).