Posttreatment Nonalcoholic Fatty Liver Disease Fibrosis Scores for Predicting Liver-related Complications in Patients With Chronic Hepatitis C Receiving Direct-acting Antiviral Agents

Wei-Fan Hsu; Hsueh-Chou Lai; Po-Heng Chuang; Wen-Pang Su; Sheng-Hung Chen; Hung-Yao Chen; Hung-Wei Wang; Guan-Tarn Huang; Cheng-Yuan Peng

Disclosures

J Viral Hepat. 2022;29(9):785-79. 

In This Article

Results

Baseline Characteristics

A total of 590 patients were investigated; the median age was 59 (51–66) years, 252 (42.7%) patients were male, and the median BMI was 24.02 (21.77–26.73) kg/m2. Of the investigated patients, 166 (28.1%), 123 (20.8%), 60 (10.2%) and 111 (18.8%) were diagnosed with hypertension, DM, fatty liver and LC, respectively. The baseline AST, ALT, triglyceride, total cholesterol, HDL and low-density lipoprotein levels were 49 (33–80) U/L, 59 (36–101) U/L, 84 (63–114) mg/dl, 170 (148–191) mg/dl, 48.1 (37.8–59.2) mg/dl and 95.1 (76.7–115.6) mg/dl, respectively. The median platelet count was 163 (118–207) × 109/L. The median HCV RNA level was 6.68 (6.01–7.14) log10 IU/ml, and 385 (65.3%), 163 (27.6%), 3 (0.5%), 37 (6.3%) and 2 (0.3%) patients were infected with HCV genotypes 1, 2, 3, 6 and mixed infections, respectively. The median follow-up duration was 25.27 (17.77–43.26) months. The regimens used and their treatment durations are listed in Table S1. The median FIB-4 and NFS values were 1.88 (1.26–2.97) and −0.89 (−1.84 to −0.10), respectively. Of 590 patients, 31 (5.3%) patients had LRCs (Table 1).

Of the 590 patients, 188 and 402 patients had MS and did not have MS, respectively. Patients with MS were older; had higher BMI; higher AST, ALT, triglyceride, alpha-fetoprotein (AFP) and creatinine levels; higher FIB-4 and NFS values; and lower HDL levels; and were more likely to be male and have hypertension, DM, fatty liver, LC and LRCs (Table 1). In addition, 18 of 111 (16.2%) patients with LC had LRCs, and 13 of 479 (2.7%) patients without LC had LRCs. Therefore, patients with LC had a higher risk of developing LRCs than those without LC (p < .001).

Of the 906 patients who were eligible for the investigation, 590 and 316 patients were enrolled and excluded, respectively. The excluded patients were younger and had lower AST, ALT, total bilirubin and AFP levels; lower FIB-4 values; shorter follow-up durations; higher platelet counts; and were less likely to have LC (Table S2).

Posttreatment NFS for Predicting LRCs and HCC Occurrence in all Patients and Those With MS

We investigated the predictors of LRCs in all patients with CHC and those with MS. The univariate Cox regression analysis revealed male sex, DM, albumin, total bilirubin, FIB-4, NFS and AFP at PW12 as the significantly associated factors in all patients. The multivariate Cox regression analysis identified total bilirubin (hazard ratio [HR]: 1.964, 95% confidence interval [CI]: 1.026–3.758, p = .042), NFS (HR: 2.125, 95% CI: 1.058–4.267, p = .034) and AFP at PW12 (HR: 1.071, 95% CI: 1.005–1.142, p = .034) as the independent predictors of LRCs (Table 2). The multivariate Cox regression analysis using variables collected at baseline identified male sex (HR: 3.102, 95% CI: 1.367–7.037, p = .007) and NFS at baseline (HR: 2.010, 95% CI: 1.185–3.410, p = .010) as the independent predictors of LRCs (Table S3).

In patients with MS, the univariate Cox regression analysis revealed albumin, total bilirubin, FIB-4, NFS and AFP at PW12 as the significantly associated factors. The multivariate Cox regression analysis identified NFS (HR: 3.340, 95% CI: 1.171–9.523, p = .024) and AFP at PW12 (HR: 1.167, 95% CI: 1.057–1.289, p = .002) as the independent predictors of LRCs (Table 3). Another multivariate Cox regression analysis using variables collected at baseline identified total bilirubin (HR: 1.932, 95% CI: 1.048–3.560, p = .035) and NFS at baseline (HR: 2.116, 95% CI: 1.169–3.831, p = .013) as the independent predictors of LRCs (Table S4).

In all patients, the multivariate Cox regression analysis identified age (HR: 1.064, 95% CI: 1.003–1.128, p = .039) and AFP at PW12 (HR: 1.017, 95% CI: 1.317–1.182, p = .002) as the independent predictors of HCC (Table S5). In patients with MS, NFS (HR: 2.207, 95% CI: 1.127–4.324, p = .021) and AFP at PW12 (HR: 1.170, 95% CI: 1.046–1.309, p = .006) were identified as the independent predictors of HCC (Table S6).

NFS at PW12 had Higher Time-dependent AUROC Than NFS at Baseline in Patients With CHC Having MS

A time-dependent AUROC was used to assess the predictive performance of NFS for LRCs at baseline and PW12. NFS at PW12 AUROCs for 1-, 2- and 3-year LRCs were 0.896 (95% CI: 0.820–0.971; Figure 1A), 0.832 (0.710–0.953; Figure 1B) and 0.822 (0.707–0.936; Figure 1C), respectively, and the values were higher than NFS values at baseline in patients with MS, although no significant differences were detected in the DeLong test. The time-dependent AUROCs of NFS for LRCs at baseline and PW12 were similar in all patients (Figure S3A,B,C).

Figure 1.

Time-dependent AUROCs for the predictive performance of NFS for LRCs at baseline and PW12 in patients with CHC with MS. (A) AUROCs at year 1. (B) AUROCs at year 2. (C) AUROCs at year 3. AUROC, area under the receiver operating characteristic curve; LRC, liver-related complication; MS, metabolic syndrome; NFS, nonalcoholic fatty liver disease fibrosis score; PW12, 3 or 6 months after DAA therapy

Predictive Values of NFS and AFP at PW12 for LRCs

The multivariate Cox regression analysis revealed that total bilirubin, NFS and AFP at PW12 were independent predictors of LRCs in all patients. We used Youden's index to identify the optimal cut-off value of NFS at PW12 for the prediction of LRCs. Angulo et al. demonstrated that NFS < −1.455 and >0.675 were correlated with F0–2 and F3–4 of the modified Brunt's fibrosis stage, respectively.[13] In the present study, NFS = −1.455 and 0.675 had higher sensitivity and specificity, respectively, for predicting LRCs, and NFS = 0.158 had the highest Youden's index for predicting LRCs (Table S7). Therefore, we used 0.158 as the cut-off value of NFS at PW12. The Kaplan–Meier analysis revealed that LRC-free survival probability was significantly higher in patients with NFS ≤0.158 than in those with NFS >0.158 (Figure 2A). Using Youden's index, we also identified that AFP levels of 4.2 ng/ml at PW12 were the optimal cut-off for predicting LRCs. The Kaplan–Meier analysis also revealed that LRC-free survival probability was significantly different between patients with and without AFP >4.2 ng/ml at PW12 (Figure 2B). Patients with and without total bilirubin >1.3 mg/dl also had a significantly different LRC-free survival probability according to the Kaplan–Meier analysis (Figure 2C).

Figure 2.

Kaplan–Meier analyses of LRCs in patients with CHC. (A) Patients with NFS ≤0.158 or >0.158. (B) Patients with AFP ≤4.2 or >4.2 ng/ml. (C) Patients with T-bil ≤1.3 or >1.3 mg/dl. AFP, α-fetoprotein; LRC, liver-related complication; NFS, nonalcoholic fatty liver disease fibrosis score; T-bil, total bilirubin

The Kaplan–Meier analysis revealed that in patients with MS, LRC-free survival probability was significantly different between patients with and without an NFS >0.158 at PW12 (Figure S4A) and those with and without an AFP level >4.2 ng/ml at PW12 (Figure S4B).

processing....