Chronic hepatitis B (CHB) infection is a major public health problem as patients are at risk of development of liver cirrhosis with or without liver decompensation and/or hepatocellular carcinoma (HCC).¹ Treatment of hepatitis B virus (HBV) has faced challenges due to the formation of covalently closed circular DNA (cccDNA), hence enabling persistent viral replication. Repeated liver injury occurs when host immune response attempts to clear the virus.^2-4 Rather than unrealistically aiming for complete eradication of HBV, a pragmatic approach is to achieve therapeutic goal of clearing hepatitis B surface antigen (HBsAg) from the serum, i.e., HBsAg seroclearance.⁵ Previous studies have documented that HBsAg seroclearance before the age of 50 years and onset of cirrhosis are associated with lower risks of cirrhosis, liver decompensation, HCC and liver-related death.^5,6 Unfortunately, the annual rate of HBsAg seroclearance is extremely low at 0.22% to 2.22%,⁷ and is affected by both host-related factors such as sex, age, presence of cirrhosis, and virus-related variables including hepatitis B e antigen (HBeAg) status, HBV DNA, and quantitative HBsAg (qHBsAg) levels.^7-9

In patients with severe acute flares of CHB (AFOCHB), higher rate of hepatocyte death is anticipated, which may potentially lead to cccDNA reduction that may be further diluted by hepatocyte regeneration. Previous studies have shown some correlation between qHBsAg levels and cccDNA.^10-13 To this end, a previous study reported that alanine aminotransferase (ALT) levels of ≥200 IU/L were associated with a more rapid decline and greater annual reduction of serum HBsAg levels in treatment naïve patients.¹⁴ Kim et al.¹⁵ also reported that ALT >5× upper limit of normal (ULN) was associated with HBsAg seroclearance amongst nucleos(t)ide analogues (NUC) treated patients. We postulated that with a higher ALT level indicating an intense immune response during severe AFOCHB, patients would have greater qHBsAg decline and HBsAg seroclearance rate. The aims of the present study were to examine the qHBsAg profile and HBsAg seroclearance rate after patients experienced severe AFOCHB.

1. Patients

This was a case-control study to compare treatment naïve CHB patients with spontaneous severe AFOCHB and patients without severe AFOCHB (control group). Severe AFOCHB cases were recruited from patients who were clinically admitted with AFOCHB to Division of Gastroenterology and Hepatology, Queen Mary Hospital, Hong Kong between January 2005 and September 2018. The inclusion criteria were age ≥18 years; clinical history of CHB infection and/ or known HBsAg positivity for more than 6 months; ALT ≥10× ULN and HBV DNA ≥4 log IU/mL on presentation. Patients with other causes of acute hepatitis including drug-induced liver injury, hepatitis A, hepatitis E, co-infection with hepatitis C and other identifiable secondary causes of liver damage were excluded. Of the identified 340 patients, 194 underwent liver transplantation/succumbed. Forty-five patients were followed up at other hospitals after flares. Twenty-six patients had no stored sera available. The remaining 75 patients were enrolled into the current study. On admission, all patients were started on NUC treatment except for four patients who refused treatment. Severe AFOCHB subjects were compared with age- and sex-matched controls in a 1:2 ratio. Matched controls had ALT <10× ULN. The decision to start NUC treatment for the controls was according to international guidelines for treatment of CHB.^5,16 They were consecutively recruited from approximately 10,000 CHB patients followed up at the hepatitis clinic in our center during the same study period. After an initial phase of stabilization of mostly <4 weeks in hospital, patients were evaluated every 6 months for clinical assessment and laboratory testing for liver biochemistry, alpha-fetoprotein and HBV serology. Liver cirrhosis was diagnosed by clinical findings of splenomegaly and/or varices and/or presence of ultrasonographic features, namely small nodular liver, splenomegaly, varices, and/or ascites. HCC was diagnosed by typical cross-sectional image findings of arterial enhancement and washout on portal venous and delayed phase. The study protocol was approved by the Institutional Review Board of the University of Hong Kong and Hospital Authority Hong Kong West Cluster (IRB number: UW 21-162). Written informed consent was waived.

2. Virological markers

Serum samples collected were stored at –20°C until tested. Baseline was defined as the time of flares for severe AFOCHB group and as the time of matching for control group. qHBsAg levels were measured at baseline and subsequently on a yearly basis. qHBsAg levels were measured using the Elecsys HBsAg II assay (Roche Diagnostic, GmbH, Mannheim, Germany),¹⁷ with lower limit of quantification of 0.05 IU/mL. Samples above 52,000 IU/mL were retested at a dilution of 1:100, according to manufacturer’s instructions. For patients with undetectable qHBsAg, the values taken for statistical analyses were 0.05 IU/mL. Serum HBsAg seroclearance was defined as undetectable qHBsAg for two samples taken at least 6 months apart. HBeAg seroconversion was defined by the loss of HBeAg and development of antibody to HBeAg. Qualitative HBeAg was measured by Abbott Laboratories (Chicago, IL, USA). HBV DNA levels were measured using the Cobas Taqman assay (Roche Diagnostics, Branchburg, NJ, USA), with the lower limit of detection of 10 IU/mL.

3. Statistical analysis

Data were expressed as number (%) or median (interquartile range). The Chi-square test was used for categorical variables. The Mann-Whitney U test was used for skewed continuous variables. The annual qHBsAg reduction rate was calculated by dividing the difference between the first and last qHBsAg level by the follow-up duration. The Kaplan-Meier survival analysis was used to determine the cumulative incidences of HBsAg seroclearance and HCC development. The univariate Cox proportional-hazards model was performed to identify factors associated with HBsAg seroclearance. Variables with p-values of <0.05 in the univariate model were tested in a multivariate Cox proportional-hazards model. All statistical analyses were performed using IBM SPSS, version 26.0 (IBM Corp., Armonk, NY, USA). A p-value of <0.05 was considered statistically significant.

1. Baseline demographics

Baseline demographics of the study population are shown in Table 1. There were no significant differences in the distribution of age, gender and HBeAg at baseline. The median HBV DNA levels were higher in severe AFOCHB group compared to control group (19.90×10⁶ IU/mL vs 1.48×10⁶ IU/mL, p=0.001) and so as for the median ALT levels (1,112 IU/L vs 43.5 IU/L, p<0.001). Patients in severe AFOCHB group had higher median bilirubin and lower median albumin levels. There were more cirrhotic patients in severe AFOCHB group than control group (13.3% vs 4%, p=0.01). The median follow-up for patients with severe AFOCHB and controls were 8.8 years (interquartile range, 4.8 to 10.7) and 10.5 years (interquartile range, 8.9 to 12.4), respectively.

Table 1 . Baseline Demographics of the Severe AFOCHB and Control Groups.

Demographics	Severe AFOCHB (n=75)	Control (n=150)	p-value
Age, yr	45.4 (35.6 to 54.3)	46.3 (36.0 to 55.3)	0.813
Male sex	56 (74.7)	111 (74.0)	0.914
HBeAg positive	26 (34.7)	43 (28.7)	0.358
HBV DNA, ×106 IU/mL	19.90 (1.01 to 160.00)	1.48 (0.00 to 77.70)	0.001
qHBsAg, IU/mL	3,127 (340 to 11,639)	1,178 (283 to 4,023)	0.076
qHBsAg reduction rate, IU/mL/yr	–242.4 (–1,098.5 to –3.7)	–47.3 (–189.3 to –3.4)	0.002
ALT, IU/L	1,112 (891 to 1,725)	43.5 (22.8 to 106.3)	<0.001
Bilirubin, µmol/L	29 (13 to 94)	10 (7 to 13)	<0.001
Albumin, g/L	39 (37 to 42)	44 (42 to 46)	<0.001
Cirrhosis	10 (13.3)	6 (4.0)	0.010
NUC treatment	71 (94.7)	75 (50.0)	<0.001
Duration of follow-up, yr	8.8 (4.8 to 10.7)	10.5 (8.9 to 12.4)	<0.001

Data are presented as median (interquartile range) or number (%)..

AFOCHB, acute flares of chronic hepatitis B; HBeAg, hepatitis B e antigen; HBV, hepatitis B virus; qHBsAg, quantitative hepatitis B surface antigen; ALT, alanine aminotransferase; NUC, nucleos(t)ide analogues..

2. NUC treatment and HBV DNA at last follow-up

Patients in severe AFOCHB group (71/75, 94.7%) and patients in control group (75/150, 50.0%) were started on NUC treatment at baseline (p<0.001). In severe AFOCHB group, 70.7% (53/75), 8.0% (6/75), 4.0% (3/75), 1.3% (1/75), and 10.7% (8/75) were treated with entecavir, lamivudine, tenofovir disoproxil fumarate, adefovir, and combination treatment, respectively. Four patients in the severe AFOCHB group refused treatment, one had HBV DNA of 49.3 IU/mL and the remaining three had undetectable HBV DNA at last follow-up. In the control group, 48.7% (73/150), 0.7% (1/150) and 0.7% (1/150) were treated with entecavir, tenofovir disoproxil fumarate and tenofovir alafenamide, respectively.

3. HBeAg seroconversion

Among the 69 HBeAg-positive patients (26 in severe AFOCHB group and 43 in control group), 11 and 16 patients underwent HBeAg seroconversion, respectively. The cumulative incidences of HBeAg seroconversion in severe AFOCHB and control groups were 35.7% and 16.3% at 5 years, and 46% and 32.8% at 10 years, respectively (p=0.253). After 12 years, cumulative incidences of both groups were similar (46.0% vs 46.5%, respectively) (Fig. 1).

Figure 1. Cumulative incidence of hepatitis B e antigen (HBeAg) seroconversion in the severe acute flares of chronic hepatitis B (AFOCHB) and control groups.

4. qHBsAg levels

The qHBsAg kinetics for individual patients with severe AFOCHB are depicted in Fig. 2. The median qHBsAg levels of severe AFOCHB and control groups were 3,127 IU/mL versus 1,178 IU/mL at baseline (p=0.076), 327.9 IU/mL versus 487.8 IU/mL (p=0.279) at 5 years, and 169.9 IU/mL versus 273.6 IU/mL (p=0.289) at 10 years, respectively (Fig. 3). The insignificant differences in median qHBsAg levels at year 5 and year 10 between two groups were likely due to the fact that severe AFOCHB group having a trend of higher baseline qHBsAg levels. Nonetheless, severe AFOCHB group had a greater annual qHBsAg reduction rate than control group (–242.4 IU/mL/yr vs –47.3 IU/mL/yr, p=0.002).

Figure 2. Quantitative hepatitis B surface antigen (qHBsAg) levels of individual patients with severe acute flares of chronic hepatitis B (AFOCHB) over time.

Figure 3. Median quantitative hepatitis B surface antigen (qHBsAg) levels of the severe acute flares of chronic hepatitis B (AFOCHB) and control groups over time.

5. HBsAg seroclearance

During the median 10.1-year follow-up period of these 225 CHB patients from both groups, 13 (5.8%) patients (seven in severe AFOCHB group and six in control group) had HBsAg seroclearance. The overall annual incidence rate was 0.6%. In severe AFOCHB group (7/75, 9.3%) versus in control group (6/150, 4.0%) achieved HBsAg seroclearance. The median age of HBsAg seroclearance was 49.6 years and 45.8 years, respectively (p=0.091). The cumulative HBsAg seroclearance rate was significantly higher in severe AFOCHB group (11.8%) compared with control group (5.0%, p=0.04) (Fig. 4). The cumulative HBsAg seroclearance rate in severe AFOCHB and control groups were 4.1% and 2% at 5 years, and 11.8% and 3.9% at 10 years, respectively. The median time to HBsAg seroclearance was 6.1 years (0.02 to 7.30) and 6.3 years (1.91 to 10.18), respectively. Among the seven patients in severe AFOCHB group with HBsAg seroclearance, two were patients who refused treatment. Among all 13 subjects who had HBsAg seroclearance, nine patients developed antibody against HBsAg.

Figure 4. Cumulative incidence of hepatitis B surface antigen (HBsAg) seroclearance in the severe acute flares of chronic hepatitis B (AFOCHB) and control groups.

Factors associated with HBsAg seroclearance were analyzed in all subjects as well as in the subgroup with severe AFOCHB. In the univariate analysis of baseline data of all 225 subjects, increasing age (p=0.01), lower baseline qHBsAg levels (p<0.001) were significantly associated with HBsAg seroclearance. The median baseline qHBsAg level was much lower in patients with HBsAg seroclearance than those without HBsAg seroclearance (39.6 IU/mL vs 1,453.5 IU/mL, p=0.015). Patients with severe AFOCHB had a trend of higher chance of HBsAg seroclearance compared to patients without severe AFOCHB (p=0.050). All these three factors (age, baseline qHBsAg levels, and having severe AFOCHB) were also independently predictive of HBsAg seroclearance in the multivariate analysis (hazard ratio [HR]=1.044, 95% confidence interval [CI]=1.000 to 1.090, p=0.049; HR=0.539, 95% CI=0.366 to 0.793, p=0.002; and HR=4.074, 95% CI=1.330 to 12.481, p=0.014, respectively). In addition, patients were stratified according to qHBsAg <200 or ≥200 IU/mL (this cutoff was chosen based on previous studies showing such level being predictive of HBsAg seroclearance),^14,18 and qHBsAg <200 IU/mL was also independently predictive of HBsAg seroclearance (HR=5.418, 95% CI=1.722 to 17.045, p=0.004). Annual qHBsAg reduction rate was not predictive of HBsAg seroclearance (Table 2). The baseline HBV DNA between patients with or without HBsAg seroclearance were similar (p=0.545) and were not predictive of HBsAg seroclearance (p=0.834). Among the 75 patients with severe AFOCHB, patients who have received NUC treatment had a trend of lower chance of HBsAg seroclearance (p=0.055). The baseline qHBsAg levels did not predict HBsAg seroclearance among severe AFOCHB group (Table 3).

Table 2 . Predictors of HBsAg Seroclearance Across All Groups.

Demographics	Univariate analysis			Multivariate analysis
	Hazard ratio (95% CI)	p-value		Hazard ratio (95% CI)	p-value
Age	1.056 (1.013–1.100)	0.010		1.044 (1.000–1.090)	0.049
Male sex	0.756 (0.233–2.454)	0.641
HBeAg positive	0.186 (0.024–1.431)	0.106
HBV DNA (per 1 IU/mL)	1.000 (1.000–1.000)	0.834
qHBsAg (per 1 IU/mL)	0.504 (0.334–0.739)	<0.001		0.539 (0.366–0.793)	0.002
qHBsAg reduction rate (per IU/mL/yr)	1.000 (1.000–1.000)	0.548
Severe AFOCHB vs control	2.979 (0.996–22.191)	0.050		4.074 (1.330–12.481)	0.014
Bilirubin (per 1 µmol/L)	1.004 (0.999–1.009)	0.092
Albumin (per 1 g/L)	0.895 (0.801–1.001)	0.053
Cirrhosis	2.926 (0.647–17.647)	0.163
NUC treatment	0.660 (0.222–1.963)	0.455

HBsAg, hepatitis B surface antigen; CI, confidence interval; HBeAg, hepatitis B e antigen; HBV, hepatitis B virus; qHBsAg, quantitative HBsAg; AFOCHB, acute flares of chronic hepatitis B; NUC, nucleos(t)ide analogues..

Table 3 . Predictors of HBsAg Seroclearance in the Severe AFOCHB Group.

Demographics	Univariate analysis			Multivariate analysis
	Hazard ratio (95% CI)	p-value		Hazard ratio (95% CI)	p-value
Age	1.082 (1.019–1.149)	0.010		1.061 (0.983–1.146)	0.129
Male sex	0.885 (0.171–4.569)	0.884
HBeAg positive	0.297 (0.036–2.464)	0.261
HBV DNA (per 1 IU/mL)	1.000 (1.000–1.000)	0.179
qHBsAg (per 1 IU/mL)	0.594 (0.361–0.976)	0.040		0.659 (0.332–1.310)	0.235
qHBsAg reduction rate (per 1 IU/mL/yr)	1.000 (1.000–1.000)	0.768
Bilirubin (per 1 µmol/L)	1.002 (0.997–1.008)	0.427
Albumin (per 1 g/L)	0.998 (0.826–1.206)	0.987
Cirrhosis	3.206 (0619–16.599)	0.165
NUC treatment	0.128 (0.025–0.664)	0.014		0.121 (0.014–1.048)	0.055

HBsAg, hepatitis B surface antigen; AFOCHB, acute flares of chronic hepatitis B; CI, confidence interval; HBeAg, hepatitis B e antigen; HBV, hepatitis B virus; qHBsAg, quantitative HBsAg; NUC, nucleos(t)ide analogues..

6. Complications

Among the 13 patients who achieved HBsAg seroclearance (seven in severe AFOCHB group and six in control group), one non-cirrhotic patient who had severe AFOCHB at the age of 57 developed HCC 2 years after flare and only achieved HBsAg seroclearance 5 years after HCC development. Otherwise, no one developed HCC after HBsAg seroclearance. None of those non-cirrhotic patients at baseline developed cirrhosis at last follow-up, including the four patients in severe AFOCHB group who refused treatment.

Severe AFOCHB group (7/75, 9.3%) and control group (2/150, 1.3%) developed HCC (p=0.004). The cumulative incidence rate of HCC was significantly higher in the former group compared to the latter group (p<0.001) (Fig. 5). After excluding patients with baseline cirrhosis, there was still a significantly higher percentage of patients in severe AFOCHB group developing HCC compared to those in control group (9.2% [6/65] vs 0.7% [1/144], p=0.001). The fibrosis-4 (FIB-4) index for all subjects was calculated at 6 months after flares, as raised aspartate aminotransferase and ALT at time of flares are not representative of the fibrosis stage. Excluding those with FIB-4 >3.25 which is widely accepted as a predictor for advanced fibrosis,¹⁹ significantly more patients in severe AFOCHB group had a FIB-4 >3.25 at last follow-up (3.10% [2/65] vs 0% [0/144], p=0.034). There was no difference in the median change of platelets from baseline to last follow-up (41×10⁹/L [18 to 64] vs 32×10⁹/L [11 to 56], p=0.329). Even after excluding those with FIB-4 >3.25, significantly more patients in severe AFOCHB group developed HCC compared to control group (7.69% [5/65] vs 1.38% [2/144], p=0.019).

Figure 5. The cumulative incidence rate of hepatocellular carcinoma (HCC) in the severe acute flares of chronic hepatitis B (AFOCHB) and control groups.

The four untreated patients in severe AFOCHB group did not develop HCC. Apart from the patient who developed HBsAg seroclearance 5 years after HCC from severe AFOCHB group, the remaining eight patients who developed HCC remained HBsAg positive.

To our knowledge, the present study is the first to examine the long-term virological and clinical outcomes in patients after AFOCHB. We found that patients with severe AFOCHB seemed to have more HBeAg seroconversion during the first 10 years of follow-up (Fig. 1). Despite the fact that patients with severe AFOCHB had a borderline higher median baseline qHBsAg level (p=0.076) compared to those without AFOCHB, they had significantly higher cumulative rates of HBsAg seroclearance and greater annual reduction of qHBsAg levels. These phenomena are in line with our hypothesis that patients with severe AFOCHB have more profound immune response compared to CHB patients without severe flare. In other words, they are having a stronger immune-mediated attack triggering the initial severe flares and subsequently bringing the HBV under better control with more suppression of qHBsAg levels afterwards. In addition, our study confirmed that qHBsAg measurements could be a tool to predict the probabilities of HBsAg seroclearance. Our results concurred with the findings of the meta-analysis that the pooled mean qHBsAg level was lower amongst patients with HBsAg seroclearance compared with those without.⁸ Nagaoka et al.¹⁴ also reported that years marked by at least one ALT flare (ALT ≥200 IU/L) was found to have a significant reduction in annual qHBsAg levels. The findings of our study suggested that having a lower baseline qHBsAg level was not the only predictive factor for HBsAg seroclearance, and that severe AFOCHB reflected a vigorous immune response and marked hepatic inflammation in an attempt towards viral clearance. Across all groups, our study validated that the cutoff value of qHBsAg <200 IU/mL was an independent predictor of HBsAg seroclearance, as suggested by Seto et al.¹⁸

Unlike other studies, the baseline HBV DNA of those achieving HBsAg seroclearance in our study was similar to those not achieving seroclearance and was not predictive of HBsAg seroclearance.^8,18 The discrepancy between baseline HBV DNA and qHBsAg levels as predictive factors may be due to the inclusion of both HBeAg-positive and negative patients in our study. Previous study had shown that qHBsAg levels only correlated well with HBV DNA in HBeAg-positive patients.²⁰

For severe AFOCHB patients, receiving NUC treatment had a trend of lower probability of HBsAg seroclearance (Table 3). This finding differed from that reported in the meta-analysis that treatment had no relationship with HBsAg seroclearance.⁸ However, van Bömmel et al.²¹ reported in a prospective, multicenter, randomized control trial on HBeAg-negative patients who were treated with NUC treatment and had HBV DNA <1,000 IU/mL for at least 4 years, eight out of 79 patients who stopped antiviral versus 0 out of 79 who continued treatment developed HBsAg seroclearance. Indeed in our study, the four patients who refused treatment in severe AFOCHB group had low levels of HBV DNA at last follow-up and two of them achieved HBsAg seroclearance. With our findings, it seems appropriate to indicate that the suppression of HBV replication by NUC treatment may hamper the effective immune control. Although suggesting no treatment for severe AFOCHB is probably unethical, this finding highlights the imminent need for new CHB treatments that would not only bring a sustained halt to viral replication but also restoration/enhancement of immune response. Nevertheless, this study only included patients who survived AFOCHB and did not require liver transplantation. The effect of enhancing immune response in all patients with AFOCHB would require further evaluation.

Despite our positive findings of severe AFOCHB patients having a greater decline in qHBsAg and higher rates of achieving functional cure, these patients with worse baseline liver parameters had worse liver-related outcomes even after started on NUC treatment. The rate of HCC development was significantly higher in severe AFOCHB patients compared with controls despite the former group had a significantly higher proportion of patients receiving NUC treatment (Table 1). Indeed, there were more cirrhotic patients in severe AFOCHB which may be related to HCC occurrence. However, even after excluding those with cirrhosis at baseline or high FIB-4 index, significantly more patients in severe AFOCHB group developed HCC compared to control group. This implied that although this group of patients with higher baseline viral activity as indicated by higher baseline HBV DNA and qHBsAg levels had better chance of HBsAg seroclearance after having severe AFOCHB, they had poorer long-term outcome despite NUC treatment. Nevertheless, for those who achieved HBsAg seroclearance subsequently, the risk of HCC remained low. Apart from the patient who achieved HBsAg seroclearance 5 years after HCC, none of the six severe AFOCHB patients with subsequent HBsAg seroclearance developed HCC. Overall, the data observed in this present study suggested that patients with severe AFOCHB should have more stringent surveillance for HCC development. Further study evaluating the cccDNA and other active viral markers, e.g., HBV core-related antigen in patients who had severe AFOCHB would show us more insight of the possible mechanisms of higher incidence of HCC even after excluding those with advanced fibrosis or cirrhosis. Interestingly, the four patients who refused treatment in severe AFOCHB group remained non-cirrhotic and did not develop HCC during the follow-up period. This suggested that in some patients who eventually survived from severe AFOCHB without treatment, the immune-mediated elimination of infected hepatocytes was strong enough to achieve viral suppression without complications. A study randomizing severe AFOCHB patients into continuation and discontinuation of NUC after stabilization of liver function to examine HBsAg seroclearance rate and development of long-term disease complications would be revealing in light of the above observations.

As HBsAg seroclearance is a rare outcome, our study may be underpowered due to small sample size, yet we have included all available patients at our center. The baseline cirrhotic status was not matched although cirrhosis was not predictive of HBsAg seroclearance in our study. It would be ideal to have matched HBsAg levels and HBeAg status between the two groups. HBV DNA was not a matching criterion (p=0.001 between two groups) because we believe that the cause of spontaneous severe AFOCHB is due to the high viral load, i.e., high HBV DNA levels, hence matching with this parameter would not be possible. In addition, HBV genotyping was not performed, although recent meta-analysis found no significant differences in HBsAg seroclearance across different genotypes among more than 10,000 patients.⁸

In conclusion, our study reported for the first time that severe AFOCHB as defined by ALT ≥10× ULN was associated with a greater decline in annual qHBsAg reduction rate and significantly higher rate of HBsAg seroclearance. However, severe AFOCHB was associated with a higher rate of development of HCC and long-term follow-up with close surveillance for liver-related complications is warranted in patients surviving AFOCHB.

This study was supported by the SK Yee Medical Foundation (ref number: 2141213).

K.S.C. received speaker’s fees from AstraZeneca and Janssen. W.K.S. received speaker’s fees from AstraZeneca and Mylan, is an advisory board member of CSL Behring, is an advisory board member and received speaker’s fees from AbbVie, and is an advisory board member, received speaker’s fees and research funding from Gilead Sciences. M.F.Y. serves as advisor/consultant for AbbVie, Assembly Biosciences, Aligos Therapeutics, Arbutus Biopharma, Bristol Myer Squibb, Clear B Therapeutics, Dicerna Pharmaceuticals, Finch Therapeutics, GlaxoSmithKline, Gilead Sciences, Immunocore, Janssen, Merck Sharp and Dohme, Hoffmann-La Roche and Springbank Pharmaceuticals, Vir Biotechnology and receives grant/research support from Assembly Biosciences, Aligos Therapeutics, Arrowhead Pharmaceuticals, Bristol Myer Squibb, Fujirebio Incorporation, Gilead Sciences, Immunocore, Merck Sharp and Dohme, Hoffmann-La Roche, Springbank Pharmaceuticals and Sysmex Corporation. Except for that, no potential conflict of interest relevant to this article was reported.

Study concept and design: M.F.Y. Data acquisition: K.Y.H., J.F. Data analysis and interpretation: K.Y.H., K.S.C, L.Y.M., W.K.S. Drafting of the manuscript: K.Y.H., J.F. Critical revision of the manuscript for important intellectual content: M.F.Y. Statistical analysis: K.Y.H., K.S.C., L.Y.M., W.K.S. Obtained funding: M.F.Y. Administrative, technical, or material support; study supervision: M.F.Y. Approval of final manuscript: all authors.