1 INTRODUCTION

Hepatitis C virus (HCV) infection is one of the leading causes of chronic hepatitis, liver cirrhosis, and hepatocellular carcinoma (HCC) worldwide. In the era of interferon-based therapy, low efficacy of regimens and high intolerance of treatment-related adverse effects have led to an unsatisfactory sustained virological response (SVR) rate of around 45%–80%.1 Moreover, with the progression of liver disease, liver-related decompensations and mortality have become crucial health issues, although fortunately in recent years, direct-acting antivirals (DAAs) have emerged as the new standard of care for patients with chronic HCV infection providing excellent effectiveness with high SVR rates above 95% regardless of HCV viremia and genotypes.2 Additionally, good safety profiles have allowed extended treatment for patients in any liver stage from mild fibrosis to decompensated cirrhosis, which was historically contraindicated to interferon-based antiviral therapy.

This has contributed to an improvement in liver function reserve and portal hypertension as well as a reduction of the risk for liver-related complications (LRC) such as HCC, hepatic encephalopathy, ascites, and variceal bleeding.3-5; however, the occurrence of LRC, including HCC, is still not well-controlled for some patients with advanced fibrosis during follow-up after HCV eradication, especially for those patients contracting cirrhosis before DAA treatment,6, 7 so current HCC international guidelines still recommend lifelong surveillance for HCV cirrhotic patients, even following HCV eradication,8, 9 although long-term HCC surveillance in more and more HCV-cured patients with cirrhosis or advanced fibrosis will result in an increased burden on health-care costs. Consequently, under the consideration of limited health budget and cost-effectiveness of surveillance policies, it is crucial to identify which patients will more readily develop LRC including HCC, so still requiring lifelong surveillance. On the contrary, patients at low risk may have their numbers and the interval of their surveillance adjusted to save unnecessary costs.

Several studies have demonstrated that noninvasive markers of liver fibrosis including liver stiffness (LS) measurement and fibrosis-4 (Fib-4) index at the time of commencement of DAA treatment in patients with cirrhosis could predict the incidence of HCC development during follow-up after achieving SVR10-13; however, levels of these markers usually reduce significantly after the improvement of liver inflammation following HCV eradication; consequently, parameters before DAA treatment appear insufficient to reflect the real liver status after total clearance of HCV viremia. Currently, using the markers at the time of SVR to predict the risk of LRC in HCV patients following SVR remains undetermined, so the present study intended to evaluate the incidence of LRC among chronic HCV patients after the achievement of SVR induced by DAA and also establish a simple scoring system using noninvasive parameters at SVR to predict the occurrence of LRC.

2 PATIENTS AND METHODS 2.1 Patients

This study included consecutive chronic HCV patients receiving DAAs in our institute, Kaohsiung Chang Gung Memorial Hospital, from January 2016 until January 2019. The inclusion criteria were (1) confirmed SVR 12 weeks after completing DAA therapy, (2) Child–Pugh class A cirrhotics or non-cirrhotics; and (3) those receiving LS measurement at SVR. Patients were excluded if they had a history of prior LRC such as HCC, ascites, and variceal bleeding or developed LRC before SVR was confirmed, as were patients with either hepatitis B virus or human immunodeficiency virus coinfections. Cirrhosis was diagnosed based on abdominal ultrasound findings including blunted nodular liver parenchyma, uneven liver edge, or splenomegaly. The study was conducted in accordance with the Declaration of Helsinki of 1975 and was approved by the Research Ethics Committee of Chang Gung Memorial Hospital (IRB 201800972B0C102).

2.2 Methods

Data on demographics, clinical characteristics, HCV genotype, regimens of DAA, serum laboratory testing, abdominal ultrasound examination, and LS measurement of enrolled patients both at baseline and SVR were reviewed and recorded. SVR was defined as undetectable HCV-RNA at 12 weeks after completing DAA treatment using the Cobas AmpliPrep/Cobas TaqMan (Roche Molecular Diagnostics, Pleasanton, CA; lower limit of detection, 15 IU/ml). Fib-4 was calculated based on four factors including age, aspartate aminotransferase, alanine aminotransferase, and platelet count. After SVR, patients underwent regular surveillance using abdominal ultrasound and laboratory tests every 3 or 6 months based on the severity of liver disease. Once liver tumor was found during the follow-up, further abdominal computed tomography or magnetic resonance imaging was performed for the diagnosis of HCC.

2.3 Liver stiffness measurement

LS measurement was performed using the transient elastography (TE; FibroScan-502, Echosens, Paris, France) with M- or XL-probe by one technician with more than 5 years of experience in performing TE. Patients were placed in a supine position with the right hand at the most abducted position for right-lobe liver scanning. When at least 10 measurements were obtained at ≥60% and interquartile range (IQR) of <30%, such measurements were considered valid assessments, and the calculated median value was used for analysis.14 LS was measured within 3 months of determination of SVR.

2.4 Liver-related complications

The occurrence of LRC during follow-up was recorded and was defined as the presentation of LRCs including ascites, variceal bleeding, hepatic encephalopathy, or HCC, with follow-up starting from the time of SVR confirmation. Death was also recorded and classified as liver- or non-liver-related. All patients were regularly followed up till the first occurrence of LRC, death, or until the end of October 2020.

2.5 Statistical analysis

To compare values between the two groups, χ2 tests were applied to analyze categorical variables, while Student's t test was used for continuous variables.

The result of LSM was expressed as median value with IQR in kilopascals (kPa), while quantitative variables were expressed with mean ± SD or median with a range. The cumulative incidences of LRC and HCC were analyzed using the Kaplan–Meier method with a log-rank test, while univariate and multivariate analyses were performed using Cox proportional hazards regression models. The area under receiver operating characteristic (AUROC) curve was used to assess the performances of LS measurement and Fib-4 in predicting occurrences of LRC and HCC, while optimal cutoffs were determined by Youden's index.15 Binary logistic regression was performed, and the risk score for each independent factor was determined by regression coefficients. All p values of <0.05 by two-tailed test were considered significant, with statistical analysis carried out using SPSS 22 software (SPSS Inc., Chicago, IL).

3 RESULTS 3.1 Patient characteristics

A total of 853 patients starting DAA therapy between January 1, 2016, and January 31, 2019, achieved SVR and received LS measurement before DAA therapy was initially evaluated. Among these, 697 patients (81.7%) were finally recruited in the study (Figure 1), with Table 1 showing the characteristics of these included patients at SVR. The mean age of these patients was 63.8 years, and 42.2% were male. Among them, 20.7% had diabetes mellitus (DM), 47.2% had cirrhosis, and 56.5% had HCV genotype 1. Certainly, patients with LRC or HCC had more cirrhosis, more DM, higher LS value, and higher Fib-4 value than those without; on the other hand, cirrhotic patients had significantly higher proportions of LRC (10.9% vs. 0.8%, p < 0.001) or HCC (7.9% vs. 0.5%, p < 0.001) than those non-cirrhotic patients during the follow-up.

Flowchart of study population. DAA, Direct-acting antiviral; EV, esophageal varices; FU, follow-up; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; LRC, liver-related complications; LS, liver stiffness; SVR, sustained virological response

TABLE 1. The demographics and clinical characteristics of all 697 patients and those with/without LRC/HCC development Variables, n (%) or mean ± SD Total N = 697 With LRC N = 39 Without LRC N = 658 p value With HCC N = 28 Without HCC N = 669 p value Age (years) 63.8 ± 10.3 65.1 ± 9.7 63.7 ± 10.3 0.422 65.6 ± 9.8 63.7 ± 10.3 0.347 Sex (%) 0.855 0.642 Male 294 (42.2) 17 (43.6) 277 (42.1) 13 (46.4) 281 (42.0) Female 403 (57.8) 22 (56.4) 381 (57.9) 15 (53.6) 388 (58.0) Genotype 0.226 0.194 1 394 (56.5) 26 (66.7) 368 (55.9) 20 (71.4) 374 (55.9) 2 271 (38.9) 13 (33.3) 258 (39.2) 8 (28.6) 263 (39.3) Other 32 (4.6) 0 (0.0) 32 (4.9) 0 (0.0) 32 (4.8) DAA therapy SOF-based regimens 322 (46.2) 22 (56.4) 300 (45.6) 0.032 16 (57.1) 306 (45.7) 0.172 PTV/r/O/D 170 (24.4) 9 (23.1) 161 (24.5) 6 (21.4) 164 (24.5) Gle/Pib 87 (12.5) 0 (0.0) 87 (13.2) 0 (0.0) 87 (13.0) Elb/Gra 82 (11.8) 4 (10.3) 78 (11.9) 4 (14.3) 78 (11.7) Others 36 (5.2) 4 (10.3) 32 (4.9) 2 (7.1) 34 (5.1) Cirrhosis 329 (47.2) 36 (92.3) 293 (44.5) <0.001 26 (92.9) 303 (45.3) <0.001 BMI (kg/m2) 25.0 ± 4.4 25.9 ± 6.5 24.9 ± 4.3 0.364 26.9 ± 7.0 24.9 ± 4.3 0.134 Obesity (%) 181 (26.0) 12 (30.8) 169 (25.7) 0.482 9 (32.1) 172 (25.7) 0.447 DM (%) 144 (20.7) 17 (43.6) 127 (19.3) <0.001 11 (39.3) 133 (19.9) 0.013 HTN 274 (39.3) 19 (48.7) 255 (38.8) 0.216 14 (50.0) 260 (38.9) 0.237 AST (U/L) 29.1 ± 15.1 37.9 ± 19.0 28.6 ± 14.7 0.004 36.7 ± 17.5 28.8 ± 14.9 0.006 ≦40 608 (87.2) 26 (66.7) 582 (88.4) <0.001 20 (71.4) 588 (87.9) 0.018 >40 89 (12.8) 13 (33.3) 76 (11.6) 8 (28.6) 81 (12.1) ALT (U/L) 24.5 ± 18.9 32.6 ± 22.8 24.0 ± 18.5 0.006 31.2 ± 20.0 24.2 ± 18.8 0.054 ≦40 620 (89.0) 32 (82.1) 588 (89.4) 0.183 22 (78.6) 598 (89.4) 0.112 >40 77 (11.0) 7 (17.9) 70 (10.6) 6 (21.4) 71 (10.6) Albumin (g/dl) 4.40 ± 0.36 4.01 ± 0.44 4.42 ± 0.34 <0.001 4.10 ± 0.47 4.41 ± 0.35 0.001 <3.5 20 (2.9) 4 (10.3) 16 (2.4) 0.021 3 (10.7) 17 (2.5) 0.042 ≧3.5 677 (97.1) 35 (89.7) 642 (97.6) 25 (89.3) 652 (97.5) Total bilirubin (mg/dl) 1.08 ± 0.63 1.41 ± 0.75 1.06 ± 0.62 <0.001 1.33 ± 0.72 1.06 ± 0.63 0.033 <1.4 564 (80.9) 25 (64.1) 539 (81.9) 0.006 19 (67.9) 545 (81.5) 0.073 ≧1.4 133 (19.1) 14 (35.9) 119 (18.1) 9 (32.1) 124 (18.5) Platelet count (×109/L) 171.3 ± 77.8 117.2 ± 59.6 174.5 ± 77.6 <0.001 121.3 ± 63.5 173.4 ± 77.7 <0.001 <150 275 (39.5) 27 (69.2) 248 (37.7) <0.001 18 (64.3) 257 (38.4) 0.006 ≧150 422 (60.5) 12 (30.8) 410 (62.3) 10 (35.7) 412 (61.6) AFP (ng/ml) 4.8 ± 6.6 10.1 ± 16.2 4.5 ± 5.4 0.036 11.9 ± 18.6 4.5 ± 5.4 0.045 <9 652 (93.5) 33 (84.6) 619 (94.1) 0.033 23 (82.1) 629 (94.0) 0.029 ≧9 45 (6.5) 6 (15.4) 39 (5.9) 5 (17.9) 40 (6.0) Fib-4 2.87 ± 1.99 4.90 ± 2.89 2.75 ± 1.86 <0.001 4.80 ± 2.89 2.79 ± 1.91 0.001 Median or mean (range or SD) <2.9 461 (66.1) 11 (28.2) 450 (68.4) <0.001 — — — ≧2.9 236 (33.9) 28 (71.8) 208 (31.6) — — <3.6 535 (76.8) — — — 11 (39.3) 524 (78.3) <0.001 ≧3.6 162 (23.2) — — 17 (60.7) 145 (21.7) LS (kPa) 12.0 ± 9.3 24.2 ± 14.7 11.3 ± 8.3 <0.001 21.0 ± 14.2 11.6 ± 8.8 0.002 Median or mean (range or SD) <14.5 522 (74.9) 11 (28.2) 511 (77.7) <0.001 — — — ≧14.5 175 (25.1) 28 (71.8) 147 (22.3) — — <11 432 (62.0) — — — 7 (25.0) 425 (63.5) <0.001 ≧11 265 (38.0) — — 21 (75.0) 244 (36.5) Abbreviations: AFP, alpha-fetoprotein; ALT, alanine transaminase; APRI, AST-Platelet ratio index; AST, aspartate aminotransferase; BMI, body mass index; DAA, direct-acting antiviral; DM, diabetes mellitus; Elb/Gra, elbasvir /grazoprevir; Fib-4, fibrosis 4; Gle/Pib, glecaprevir/pibrentasvir; HCC, hepatocellular carcinoma; HTN, hypertension; LRC, liver-related complications; LS, liver stiffness; PTV/r/O/D, paritaprevir/ritonavir/ombitasvir/dasabuvir; SOF, sofosbuvir; SVR, sustained virological response. 3.2 Incidence and features of LRC and HCC

In a median follow-up of 21.4 months (range 1.1–48.5) after SVR, 39 patients developed LRC including 28 patients with HCC, nine with ascites, and two with variceal bleeding (Table 2). The 6-, 12-, 18-, 24-, and 30-month cumulative LRC incidences were 1.2%, 1.8%, 3.6%, 5.5%, and 7.7%, respectively, from the time of SVR. Median time to LRC development was 17.4 months (range: 1.7–33.7 months). At diagnosis of HCC, the mean size of the tumor(s) was 2.4 cm, and 17 (60.7%) patients had a single tumor. Based on the classification of Barcellola Clinic Liver Cancer (BCLC) stage, there were seven patients in BCLC stage 0, 18 in stage A, and three in stage C, respectively (Table 2). The 6-, 12-, 18-, 24-, and 30-month cumulative HCC incidences were 0.7%, 1.4%, 3%, 4.2%, and 5.1%, respectively. Median time to HCC development was 17.1 months (range: 1.7–33.7 months). Only two patients died during the follow-up, with one through esophageal varices (EV) bleeding and liver decompensation and the other due to uncontrolled HCC. No patient underwent liver transplantation during the follow-up period.

TABLE 2. Clinical characteristics of LRC at diagnosis N = 39 Ascites 9 EV bleeding 2 HCC 28 HCC size, mean ± SD (cm) 2.4 HCC number 1 17 2 8 ≧3 3 HCC in BCLC stage 0 7 A 18 B 3 Abbreviations: BCLC, Barcelona Clinical Liver Cancer; EV, esophageal varices; HCC, hepatocellular carcinoma; LRC, liver-related complications; SD, standard deviation. 3.3 Risk factors associated with LRC and HCC development

The performance of LS and Fib-4 in the prediction of LRC occurrence was 0.82 and 0.755 by AUROC (Figure 2A), although the comparison of LRC prediction between these two factors was insignificant. With the optimal cutoff value of 14.5 kPa, the sensitivity and specificity were 71.8% and 77.7% for LS in LRC prediction, respectively. Multivariate analysis showed that LS (hazard ratio [HR]: 3.97, 95% confidence interval (CI) [1.866, 8.446], p < 0.001), Fib-4 (HR: 2.558, 95% CI [1.212, 5.402], p = 0.014), and presence of DM (HR: 2.239, 95% CI [1.180, 4.249], p = 0.014) were independent factors associated with LRC development (Table 3). For the prediction of HCC development, the performance of LS and Fib-4 was 0.774 and 0.737 (Figure 2B), and although similar, the comparison of HCC prediction between these two factors was not statistically significant. Using the optimal cutoff value of 11 kPa, the sensitivity and specificity were 75% and 63.5% for LS in HCC prediction, respectively. Multivariate analysis showed that LS (HR: 2.828, 95% CI [1.123, 7.112], p = 0.027) and Fib-4 (HR: 3.007, 95% CI [1.326, 6.819], p = 0.008) were the only two predictors for HCC occurrence (Table 4).

(A) The performance of LS and Fib-4 in the prediction of LRC development assessed with area under receiver operating characteristic (AUROC) curve. (B). The performance of LS and Fib-4 in the prediction of HCC development assessed with AUROC curve. Fib-4, Fibrosis 4; HCC, hepatocellular carcinoma; LRC, liver-related complications; LS, liver stiffness; NPV, negative predictive values; PPV, positive predictive values

TABLE 3. Factors at SVR for the development of LRC after DAA treatment Variables Comparison Univariate analysis Multivariate analysis HR 95% CI p value HR 95% CI p value Age, years Increase per year 1.011 0.978–1.045 0.518 Sex Male vs. female 1.050 0.558–1.979 0.879 BMI, kg/m2 ≥27 vs. <27 1.130 0.572–2.232 0.725 AST, U/L >40 vs. ≦40 3.434 1.756–6.716 <0.001 ALT, U/L >40 vs. ≦40 1.701 0.750–3.858 0.204 Albumin, g/dl <3.5 vs. ≧3.5