Patient Baseline Characteristics
A total of 149 eligible patients (82CD, 67UC) were enrolled, and 134 patients (73CD, 61UC) were in active, 15 patients were in remission. The baseline characteristics of all patients were shown in Table 1. The median disease duration of all patients was 3 years (interquartile range [IQR]:1.0-4.0yr). The patients with CD were younger than UC (Median age 28 .0 vs 46.0 years, P<0.001). There was no statistical difference in gender and smoking history between patients with UC and CD. Based on the Montreal classification, most of the patients with CD (55.65%) had an ileocolonic disease (L3) and 64.18% of the patients with UC had an extended colitis (E3).
Correlation Between Biomarkers and Endoscopic Score in CD Patients
Correlation coefficient r between biomarkers and endoscopic score of disease severity of CD were shown in Table 2. As shown in the table, DAO and ETX were demonstrated to have a better correlation with SES-CD (r=0.532 and r=0.468, respectively; P<0.0001 for both) compared to WBC, ESR and CRP (r=0.146, P=0.191; r=0.346, P=0.0015; r=0.250, P=0.023), while there was no correlation between D-lactate and SES-CD (r=0.167, P=0.133).
Correlation Between Biomarkers and Endoscopic Score in UC Patients
Correlation coefficient r between biomarkers and endoscopic score of disease severity of UC were shown in Table 3. Similar to CD, DAO and ETX were demonstrated to have better correlation with UCEIS (r=0.600 and r=0.505, respectively; P<0.0001 for both) compared with WBC, ESR and CRP (r=0.151, P>0.05; r=0.285, P=0.021; r=0.334, P=0.006). There was significant but weak correlation between D-lactate and UCEIS (r=0.407, P=0.0006). The relationship of DAO, D-lactate and ETX with endoscopic score of UC were shown to be better than those of CD.
Value of Intestinal Barrier Marker in Predicting Endoscopic Severity in CD
As was shown in Table 4 and Figure 1, the DAO levels of severe groups were higher than those of moderate groups (P<0.05, Figure 1A), and the DAO levels of moderate groups were also higher than those of mild groups (P<0.001, Figure 1A). The ETX levels of severe groups were higher than those of moderate groups (P<0.05, Figure 1B), and the ETX levels of moderate groups were also higher than those of mild group (P<0.01, Figure 1B). In brief, as severity of endoscopy increased, the levels of DAO and ETX increased. There was no difference in D-lactate levels among three groups of active CD patients (Figure 1C).
The DAO and ETX had diagnostic utility in differentiating CD patients with different endoscopic activities (Figure 2 and Table 4). For DAO, the optimum discriminative cutoff threshold for distinguishing endoscopic mild disease was 18.54u /L, with an AUC of 0.80, P<0.001(sensitivity 71.93%, specificity 71.93%, PPV 87.2%, NPV 54.3%and accuracy 71.95%). For ETX, the optimum discriminative cutoff threshold for distinguishing endoscopic mild disease was was 10.51u /L, AUC was 0.76(P<0.001), sensitivity was 70.18%, specificity was 80.0%, corresponding PPV, NPV and accuracy were 88.9%, 54.1% and 73.3%, respectively. Further research has found that the optimum discriminative cutoff of DAO threshold for severe disease in CD (SES-CD≥16) was 18.63 U/L (AUC=0.75, P=0.001). The test parameters of DAO were as follows: sensitivity 89.47%, specificity 53.79%, PPV 37.0%, NPV 94.4%, and accuracy 62.2%. For ETX, the optimum discriminative cutoff for severe disease was 22.38 U/L (sensitivity 57.89%, specificity 82.54%, PPV 50.0%, NPV 86.7%, and accuracy 76.83%, respectively), the AUC was 0.70, P=0.010 (SES-CD≥16). The D-lactate, as well as WBC, ESR or CRP, failed to assess endoscopic severity in CD(P>0.05 for all).
Value of intestinal barrier marker in Predicting Endoscopic Remission in CD
Although we have demonstrated that DAO and ETX were positively correlated with SES-CD, there was no significant difference in level of DAO, D-lactate and ETX between active and inactive groups in CD patients (P>0.05, respectively; Table 5). The multivariate analyses demonstrated that no optimal threshold could be identified that distinguished active CD from inactive CD, as the NPV of test was not satisfied(Figure 3 and Table 6).
Value of intestinal barrier marker in Predicting Endoscopic Severity in UC
In active UC patients, mild and moderate activity was merged into one group. There was no significant difference in DAO and D-lactate between mild-moderate and severe patients (P>0.05 for both, Figure 4), only levels of ETX for severe group were higher than mild-moderate group (P<0.05, Figure 4). The result is quite different with CD. ROC curves analysis was shown in Figure 5. No endoscopic disease category was individually distinguishable by DAO, D-lactate, ETX, WBC, ESR and CRP. Although the levels of ETX of severe group were higher than mild-moderate group, we have not found an optimum discriminative cutoff of ETX to distinguish the severity of endoscopy.
Value of intestinal barrier marker in Predicting Endoscopic Remission in UC
Significant differences were found in DAO, D-lactate and ETX levels between active and inactive UC patients (P= 0.004, P= 0.008 and P= 0.003; Table 8). On the whole, the levels of DAO, D-lactate and ETX of the active group were higher than that of the inactive group (UCEIS≤1). ROC analysis shown that DAO, D-lactate and ETX have a certain diagnostic utility in predicting endoscopic remission (Table 9 and Figure 6). The AUC of DAO for endoscopic disease thresholds of UCEIS≤1 was 0.86, P<0.001. The optimum discriminative cutoff of DAO threshold was 15.97 U/L, predicting endoscopic remission with 83.93% sensitivity and 90.91% specificity, the PPV, NPV and accuracy of this threshold were 97.9%, 52.6% and 85.07%. The AUC for D-lactate and ETX for endoscopic disease thresholds of UCEIS≤1 was 0.80, P= 0.002 and 0.80, P= 0.002, the corresponding cutoff of D-lactate and ETX are 36.53 mg/l and 17.55 U/L. The sensitivity, specificity, PPV, NPV and accuracy of D-lactate in predicting remission of this thresholds are 75%,81.82%, 95.5%, 39.1% and 76.12%. The sensitivity, specificity, PPV, NPV and accuracy of ETX in predicting remission of this thresholds are 62.5%, 92.91%, 97.2%, 32.3% and 67.16%.
We further repeated the experiment, and the results showed that if UCEIS=0 was defined as remission, there was no statistically significant difference in the level of DAO d-lactic acid and ETX of UC patients with active patients (both P>0.05). ROC curve analysis also showed that DAO, D-lactate and ETX could not predict endoscopic UC remission when UCEIS=0 was defined as remission.