A new glutamine synthetase index to evaluate hepatic lobular restoration in advanced fibrosis during anti‐HBV therapy

Hepatic lobular architecture distortion is a deleterious turning point and a crucial histological feature of advanced liver fibrosis in chronic liver diseases. Regression of fibrosis has been documented in chronic hepatitis B (CHB) patients. However, whether lobular architecture could be restored following fibrosis regression after antiviral therapy is still unclear. Glutamine synthetase (GS) is generally expressed by perivenular hepatocytes around hepatic veins (HV). In this study, we defined abnormal lobular architecture (GSPT) as GS expressing in the vicinity of portal tracts (PT), which denotes parenchymal extinction and lobular collapse. We defined normal lobular architecture (GSHV) as GS positivity area not approximating PTs. Therefore, we propose a new GS‐index, defined as the percentage of GSHV/(GSHV + GSPT), to evaluate the extent of architectural disruption and restoration. We evaluated 43 CHB patients with advanced fibrosis (Ishak stage ≥4). Posttreatment liver biopsy was performed after 78 weeks of anti‐HBV therapy. The median GS‐index improved from 7% (interquartile range [IQR]: 0%–23%) at baseline to 36% (IQR: 20%–57%) at Week 78 (p < 0.001). Totals of 22 patients (51%) had significant GS‐index improvement from 0% (IQR: 0%–13%) to 55% (IQR: 44%–81%), while the other half had almost no change between 17% (IQR: 0%–33%) to 20% (IQR: 12%–31%). When GS‐index78w ≥ 50% was used to define hepatic lobular restoration, 37% of patients (16/43) achieved lobular restoration, with much improvement in alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels (median value of ∆/Baseline in ALT: restored vs. nonrestored was 79.1% vs. 48.8%, p = 0.018; median value of ∆/Baseline in AST: restored vs. nonrestored was 69.1% vs. 32.5%, p = 0.005). More importantly, lobular restoration correlated with fibrosis regression (median value of ∆/Baseline in Ishak stage: restored vs. nonrestored was 25.0% vs. 0%, p = 0.008). Therefore, in the era of antiviral therapy for CHB, restoration of hepatic lobular architecture is achievable in patients with advanced fibrosis. GS‐index provides additional insight into fibrosis regression that goes beyond collagen degradation.

p = 0.008). Therefore, in the era of antiviral therapy for CHB, restoration of hepatic lobular architecture is achievable in patients with advanced fibrosis.
GS-index provides additional insight into fibrosis regression that goes beyond collagen degradation.

| INTRODUCTION
Hepatic lobular architecture distortion is one of the deleterious pathological changes following severe liver injury with advanced fibrosis. Normal hexagonal lobular architecture is lost due to parenchymal collapse, bridging portal tract to hepatic vein fibrous deposition, and vascular remodeling. 1 Portal hypertension and liver decompensation may ensue. 1,2 The reversibility of hepatitis B virus (HBV)-induced liver fibrosis has been documented in several landmark clinical studies and regressive histological features have been well described by pathologists. [3][4][5] Bedossa described the important events in fibrosis regression, including extracellular matrix degradation, hepatocellular regeneration, and hepatic lobular architecture restoration. 1 Wanless et al. proposed the hepatic repair complex after observing the serial biopsies of an HBV-induced cirrhotic patient before and after antiviral treatment. 5 In this study, Wanless et al. also noted that hepatocellular regeneration emerges as minute hepatocyte buds in regions of parenchymal extinction. However, whether with disease improvement and fibrosis regression, hepatic lobular architecture can be restored is uncertain.
Hepatic lobular architecture distortion is regarded as one of the morphological features of fibrosis progression in conventional fibrosis staging systems, such as Scheuer and Batts-Ludwig Systems. 6,7 However, the evaluation of architectural distortion is based on the subjective and experiential judgment of the pathologist. In this study, we propose a histological parameter employing glutamine synthetase (GS) immunohistochemical staining to quantify the degree of hepatic lobular architecture restoration. 8,9 The normal hexagonal liver lobule has peripherally-located portal tracts (PT) with hepatic veins (HV) in the center. 10 Hence, identifying HVs and estimating their distance from PTs are critical in establishing normal and abnormal lobular architecture. GS is expressed in perivenular hepatocytes around HVs in normal liver. Currently, the zonation profile of its gene, namely GLUL (central zone) in human hepatocytes, is demonstrated by single-cell RNA sequencing methods. 9 Therefore, GS staining can locate HVs in hepatic parenchyma. Based on this pathological characteristic, a histological parameter named GS-index is proposed to assess the extent of hepatic lobular architecture restoration.

| Histological assessment
The liver specimens were formalin-fixed and routinely processed.
Five micrometer-thick paraffin-embedded histological sections were stained with hematoxylin and eosin, reticulin, Masson trichrome, and GS. 11 Immunohistochemistry for GS was performed by manual operation according to manufacturer's instructions (GS-6: 1:500, Millipore; antigen retrieval solution: citrate buffer for 20 min, Fuzhou Maixin Biotech. Co, Ltd; detection method: DAB kit, Fuzhou Maixin Biotech. Co, Ltd). Two senior experts independently assessed the necroinflammatory activity and fibrosis stage based on Ishak modified histology activity index grading system and Ishak fibrosis scoring system. 12 The samples with Ishak stage ≥3 were further evaluated according to the P-I-R score of the Beijing classification, which was developed during the antiviral therapy era (YS). 13 Henceforth, the glutamine synthetase index (GS-index) was defined as the percentage of normal HVs among all HVs, that is, GS HV /(GS HV + GS PT ) ( Figure 1B). According to the level of the GS-index at Week 78, we defined hepatic lobular architecture restoration as GS-index 78w ≥ 50%.

| Liver stiffness measurements
According to the manufacturer's instructions (Fibroscan; Echosens), liver stiffness measurements were obtained through transient elastography by experienced operators. 15 The LSM (in kPa) was considered reliable when 10 available measurements could be obtained with a success rate greater than 60% and a ratio of the  interquartile range (IQR) of liver stiffness to the median (IQR/med) not more than 0.30.
Serum HBV DNA levels were assessed using COBAS ® TaqMan ® HBV Test (Roche Molecular Systems) with a detection threshold of 20 IU/mL in the central laboratory. In addition, HBV serological markers (including HBsAg, anti-HBs, HBeAg, anti-HBe, and anti-HBc) were also detected in the central laboratory through Abbott Architect i2000 (Abbott) at baseline, Week 52 and 104.

| Statistical analysis
Continuous variables were expressed as mean ± standard deviation (SD) or median (IQR). Categorical variables were measured using counts and percentages. Wilcoxon signed-rank test was used to compare the distribution of the GS-index and the changes in clinical characteristics before and after treatment. Mann-Whitney test was used to compare the GS-index in restored and non-restored or reverse and non-reverse patients, and the dynamic changes of clinical and histological characteristics in restored and non-restored patients.
In addition, we used covariance analysis to adjust the baseline level of the GS-index and assess the dynamic changes of the GS-index in restored and non-restored or reverse and non-reverse patients. χ 2 test and likelihood ratios were used to compare the distribution of

| Histological assessment based on fibrosis staging (Ishak and P-I-R score)
The distribution of fibrosis stage was shown in Table 1
The median changes of GS-index from baseline to Week 78 were 60% T A B L E 1 Clinical and histological characteristics of chronic hepatitis B patients before and after antiviral treatment.
Dynamic changes in clinical data before and after treatment were compared between restored and non-restored patients (
Two representative cases of restored and non-restored patients with paired liver biopsies were shown in Figure 4. For the restored patient, GS-index increased dramatically from 0% (baseline) to 75% (Week 78) (GS-index 78w = 75%); Ishak fibrosis score decreased from 5 to 3; and the level of LSM decreased from 12.5 to 6.8 kPa. For the non-restored patient, the dynamic changes observed for GS-index were 0% (baseline) to 13% (Week 78) (GS-index 78w = 13%). Ishak fibrosis score remained at stage 5 after 78 weeks of treatment, although LSM slightly decreased from 30.4 (baseline) to 25.7 kPa (Week 78).

| DISCUSSION
In this study, we proposed a histological parameter (GS-index) to quantify hepatic lobular architecture restoration in CHB patients in the era of antiviral therapy. Patients in the restored group were characterized by better clinical and morphological improvements than in the non-restored group. In addition, we observed through the GS-index that reverse patients attain more significant hepatic architecture restoration than non-reverse patients. It is well documented that liver fibrosis can be reversed after successfully suppressing HBV replication by antiviral treatment. 17,18 Our study illustrated the relationship between hepatic lobular architecture restoration and fibrosis regression. There are corresponding clinical benefits in patients with degradation of collagen fibers and hepatic lobular zonation restoration. We also observed that the extent of restoration differs between reverse and non-reverse patients during longitudinal follow-up.
In this study, the definition of normal lobular architecture is that  regression cases with the morphological feature of veno-portal approximation, which may lead to selection bias. Compared with this cross-sectional study, our research is based on a national and multicenter cohort of patients in which sequential liver biopsies were performed during longitudinal follow-up, making our results more robust.
The GS-index ≥50% threshold of lobular restoration was proposed after long deliberation based on the following considerations: (i) based on this cutoff value, restored patients can attain additional improvement in both clinical and histological characteristics compared to non-restored patients; (ii) since there is no gold standard of hepatic lobular restoration assessment, in analogy with the previous semi-quantification assessment of liver fibrosis (Ishak system), which defined the "quantitative" indicators ("few," "some," and "most") based on the threshold of half of the portal tract or central area in routine practice, 20 we attempted to defined lobular restoration as the half of the HVs in the whole liver biopsy not approximating to PT at Week 78; (iii) besides, we also try to use fibrosis regression as the substitute standard of lobular restoration.
We found that the specificity of the cutoff value of 50% (GS-index 78w ) was higher than 80% (based on Ishak score vs. PIR score: 80% vs. 85%). Therefore, the cutoff value of GS-index 78w = 50% can help accurately identify patients who achieved fibrosis regression. Besides, the definition of hepatic lobular restoration needs to be confirmed in further investigation with the data of longer prospective follow-up and clinical endpoints.
The novelty of this study is the type of insight being proposed.
The GS-index has two important clinical implications. First, the GSindex adds a new dimension for the dynamic changes of fibrosis, which is beyond fibrosis severity and inflammatory activity (Supporting Information: Table 1). More importantly, hepatic lobular architecture restoration based on the posttreatment GS-index partly reveals the pathological mechanism of fibrosis regression during anti-HBV treatment. Second, the utility of the GS-index facilitates the quantification of hepatic lobular zonation restoration, which improves the situation that lobular architecture cannot be assessed previously. Hepatic lobular architecture distortion is one of the morphological features leading to portal hypertension and decompensations. 1 As a quantitative parameter of hepatic lobular architecture evaluation, GS-index may provide clues for disease progression, which reminds clinicians to monitor closely and to determine the necessity for further intervention.
Our study has certain limitations. First, loss of GS expression following the loss of endothelium-hepatocyte proximity or contact was observed in severe cirrhosis, especially Laennec 4B or 4C, leading to the underestimation of HVs. 8,[20][21][22][23][24][25][26] In our study, patterns of GS expression in cirrhotic patients (baseline Ishak stage 6) include GS-positive hepatocytes distributed around HVs and PTs. We did not observe the loss of GS expression among those patients, which may due to the extent of cirrhosis was not severe enough to lead to the loss of endothelium-hepatocyte proximity or contact. Second, the opening up of new HVs in cirrhosis regression is uncertain; therefore, we may have to approach this dilemma by counting the increase in the number of new veins rather than separating them from PTs. Last, GS overstaining might lead to confusion in identifying the HVs.
In conclusion, our study demonstrated that the restoration of hepatic lobular architecture is achievable after antiviral therapy.
GS-index gives a new evaluation to quantitatively assess hepatic lobular restoration for liver fibrosis regression in CHB patients.