Utility of Mac-2 Binding Protein Glycosylation Isomer to Evaluate Graft Status After Liver Transplantation

Patients

We collected the clinicopathological data of 233 patients who underwent liver biopsy at Kyoto University Hospital after an LT between August 2015 and June 2019. In these patients, no organs from executed prisoners were used. With regard to patients who underwent more than 2 liver biopsies in this period, we included only the data of the first biopsy to avoid duplicates of the same patients. We obtained informed consent from all patients. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki, and it was approved by the ethics committee of Kyoto University Graduate School and Faculty of Medicine (approval code E2477).

Clinical Data Collection

We collected clinical data including sex, age, primary disease, the date of the transplantation from the operative note of the LT, and blood examination. Blood examinations were performed on the same day or the day before the biopsies. Blood examinations included the complete blood count, coagulation test, and biochemistry including measurements of aspartate aminotransferase (AST), alanine aminotransferase (ALT), total bilirubin (T-Bil), and albumin (ALB) levels. We also examined the M2BPGi values and other serum markers of liver fibrosis, ie, hyaluronic acid^{(10, 11)} and type 4 collagen.⁽¹²⁾ The M2BPGi value was measured using the HISCL M2BPGi Assay Kit (Sysmex, Kobe, Japan). The cutoff index (COI) of serum M2BPGi was calculated according to the equation reported previously.⁽²⁶⁾ The COI of M2BPGi was calculated as follows: COI = ([M2BPGi]_sample – [M2BPGi]_{negative control})/([M2BPGi]_{positive control} – [M2BPGi]_{negative control}), where [M2BPGi]_sample represents the M2BPGi value of the serum sample. The positive control was supplied as a calibration solution that was preliminarily standardized to yield COI = 1.0.⁽²⁷⁾ M2BPGi value ranges from 0.25 to 20.0 in previous studies on HBV, HCV, and biliary atresia patients.^{(21, 28)} Simultaneously, we calculated liver fibrosis indicators, ie, APRI^{(7, 8)} and FIB-4 index,⁽⁹⁾ according to previously reported equations. We also performed ultrasound-based elastography, ie, ARFI,^{(29, 30)} for liver stiffness measurement on the day of the biopsy, as reported previously.^{(19, 31)}

Liver Biopsy and Pathological Data Collection

Liver biopsy examination was undertaken when clinically indicated (eg, when the liver function test showed abnormal findings) or at designated intervals (a so-called nonepisode biopsy) with informed consent. Generally, a nonepisode biopsy is performed in pediatric recipients, recipients whose primary disease can recur (HCV, primary sclerosing cholangitis [PSC], PBC, or NASH), and recipients with abnormal findings in the previous liver biopsy, with an interval of 2 or 3 years depending on the discretion of attending physicians.

We used an 18-gauge needle for the liver biopsy. Liver specimens were fixed in formalin, embedded in paraffin, and stained with hematoxylin-eosin and Masson’s trichrome.

All histological assessments were performed by expert pathologists at Kyoto University Hospital, and we referred to the histopathological reports. Pathological assessment was based on METAVIR scores of the fibrosis stage (F0-F4) and necroinflammatory activity (A0-A3).⁽³²⁾ The severity of transplant rejection was classified by the rejection activity index (RAI).^{(33, 34)}

We performed immunostaining of some representative liver biopsy samples, 3 samples from HCV patients, 2 samples from PSC patients, 2 samples from acute liver failure patients, and 1 sample from a biliary atresia patient. Liver biopsy samples were stained with biotinylated Wisteria floribunda agglutinin (WFA; Vector Laboratories, Burlingame, CA), anti-Mac-2 binding protein (M2BP; R&D Systems, Minneapolis, MN), and anti-CD68 [KP1] (Abcam, Cambridge, MA) antibodies, followed by staining with Streptavidin DyLight–conjugated (Vector Laboratories) or Alexa Fluor–conjugated antibodies (Thermo Fisher Scientific, Waltham, MA). Images were obtained using a BZ-9000 microscope (KEYENCE Co., Osaka, Japan).

Statistical Analysis

Continuous variables are expressed as mean ± standard deviation (SD) or median (range). They were compared by the Mann-Whitney U test or Kruskal-Wallis test. Statistical associations of each index were evaluated by the Spearman rank correlation coefficients. We calculated the 95% confidence interval (CI) and P value for the difference between 2 correlation coefficients by the bootstrap method with 10,000 repetitions. The predictive values of liver fibrosis were assessed by receiver operating characteristic (ROC) analysis, and the area under the receiver operating characteristic curve (AUROC) was calculated. The ROC curves were compared using the DeLong test. A P value of <0.05 was considered as statistically significant.

Statistical analyses were carried out using SAS software (JMP 14.0; SAS Institute, Cary, NC). Analysis by the bootstrap method was carried out using R, version 3.6.2 (R Foundation for Statistical Computing, Vienna, Austria).

Patient Characteristics

Patient characteristics and other clinicopathological data are shown in Table 1. The median age of patients in this study was 33 years (range, 7-75 years). Primary liver diseases were HBV (n = 18, 7.7%), HCV (n = 29, 12.4%), PBC (n = 30, 12.9%), PSC (n = 10, 4.3%), alcohol (n = 5, 2.1%), NASH (n = 2, 0.9%), biliary atresia (n = 103, 44.2%), acute liver failure (n = 14, 6.0%), and others including cryptogenic causes (n = 22, 9.4%). Among these 233 patients, 222 patients were recipients from living donors, and 11 patients were recipients from brain dead donors. Among the 233 patients, 44 patients were pediatric patients (under 20 years of age). A total of 17 patients had HCC before LT. Among the 233 liver biopsies, 24 patients underwent liver biopsies for a clinical indication (a concern of disease recurrence or rejection), whereas 209 patients underwent so-called nonepisode biopsies.

Diagnostic Performance of M2BPGi for Liver Fibrosis and Inflammation

The number of patients with METAVIR scores of F0, F1, F2, F3, and F4 were 49, 131, 39, 12, and 2, respectively. (We regrouped F3 and F4 together because the numbers of patients in F3 and F4 were too low for statistical analysis.) The median and quartile values of M2BPGi in patients with F0, F1, F2, and ≥F3 were 0.61 (0.425, 0.965), 0.76 (0.48, 1.34), 1.16 (0.52, 3.09), and 1.47 (0.51, 4.64), respectively. The M2BPGi value in patients with F2 was significantly higher than that in patients with F1 (P = 0.01) and F0 (P = 0.001). Moreover, the M2BPGi value in patients with ≥F3 was significantly higher than that in patients with F1 (P = 0.04) and F0 (P = 0.01). However, there was no statistically significant difference between F0 and F1, and between F2 and ≥F3 (Fig. 1A).

The numbers of patients with METAVIR scores of A0, A1, A2, and A3 were 120, 106, 6, and 1, respectively. (We regrouped A2 and A3 together because the number of patients in A2 and A3 was too low for statistical analysis.) The median and interquartile range values of M2BPGi in patients with A0, A1, and ≥A2 were 0.53 (0.42, 0.9325), 1.145 (0.6475, 1.9375), and 2.24 (1.56, 8.57), respectively. There were significant differences in the M2BPGi values between A0 and A1 (P < 0.001) and between A1 and ≥A2 (P = 0.006; Fig. 1B).

Spearman correlation analysis indicated a significant positive correlation between the fibrosis stage and M2BPGi value (r = 0.25, P < 0.001). Similarly, there was a significant positive correlation between the necroinflammatory index and M2BPGi value (r = 0.46, P < 0.001). The difference calculated by subtracting the Spearman rank correlation coefficient between the necroinflammatory index and the M2BPGi value from the value between the fibrosis stage and M2BPGi value was 0.21 (95% CI, 0.095-0.39; P = 0.002), indicating that the necroinflammatory index had a stronger correlation to the M2BPGi value than the fibrosis stage.

The AUROC to predict liver fibrosis is shown in Table 2. The AUROC of M2BPGi to predict ≥F1, ≥F2, and ≥F3 was 0.61, 0.66, and 0.66, respectively, which was not significantly different from that of other liver fibrosis markers or indexes.

The AUROC to predict the liver necroinflammatory index is shown in Table 3. The AUROC of M2BPGi to predict ≥A1 and ≥A2 was 0.75 and 0.88, respectively. The AUROC of M2BPGi to predict ≥A2 was significantly higher than that of some liver fibrosis markers (ie, platelet count and hyaluronic acid). However, it was not significantly different from other liver inflammation markers (ie, lactate dehydrogenase [LDH], AST, and ALT). It is noteworthy that the AUROC of M2BPGi to predict ≥A1 was significantly higher than that of any other liver fibrosis marker or index (ie, platelet count, hyaluronic acid, type 4 collagen, APRI, FIB-4 index, and ARFI) and serum liver inflammation marker (ie, LDH, AST, and ALT). The cutoff value, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (PLR), and negative likelihood ratio (NLR) for each fibrosis and necroinflammatory stage are shown in Table 4.

Next, we examined the utility of M2BPGi to diagnose allograft rejection. We assessed rejection by calculating the RAI. The RAI was 0 in 139 (59.7%), 1 in 50 (21.5%), 2 in 25 (10.7%), 3 in 16 (6.9%), and 5 in 3 patients (1.3%).

First, we analyzed the correlation between the necroinflammatory index and RAI (Fig. 2A). Spearman correlation test revealed a significant positive correlation between the RAI and necroinflammatory index (r = 0.32, P < 0.001).

Second, on the basis of this result, we found a relationship between the RAI and the M2BPGi value. The median and quartile values of M2BPGi in RAI 0, 1, 2, 3, and 5 were 0.61 (0.44, 0.97), 1.12 (0.45, 1.895), 1.02 (0.705, 2.03), 1.365 (0.925, 1.7175), and 1.46 (1.11, 2.66), respectively. The M2BPGi value in RAI 1 was significantly higher than that in RAI 0 (P = 0.03). However, there was no statistically significant difference in M2BPGi between RAI 1 and 2, between RAI 2 and 3, and between RAI 3 and 5 (Fig. 2B).

The number of patients in RAI 3 and 5 was very low. Therefore, we regrouped the patients into groups based on RAI ≤2 and RAI ≥3. The median and quartile values of M2BPGi in RAI ≤2 and ≥3 groups were 0.695 (0.4675, 1.3325) and 1.39 (0.97, 1.77), respectively. The M2BPGi value in the RAI ≥3 group was significantly higher than that in the RAI ≤2 group (P = 0.001; Fig. 2C). The AUROC of M2BPGi to predict RAI ≥ 3 was 0.72. The cutoff value, sensitivity, specificity, PPV, NPV, PLR, and NLR in the RAI ≥ 3 group were 0.91, 0.842, 0.594, 0.155, 0.977, 2.071, and 0.266, respectively.

Relationship Between M2BPGi Value and HCV Infection

We analyzed the effect of HCV infection on the M2BPGi value. There were 29 patients whose primary disease was HCV, and all of them had recurrence of HCV after LT. At the time of the liver biopsy, 20 patients had viremia, whereas 9 patients had achieved a sustained virological response (SVR).

We compared the M2BPGi value between non-HCV patients, hepatitis C virus patients with viremia (HCV-viremia), and hepatitis C virus patients who had achieved sustained virological response (HCV-SVR). The median and quartile values of M2BPGi in non-HCV, HCV-viremia, and HCV-SVR groups were 0.69 (0.465, 1.17), 2.43 (2.18, 5.315), and 0.715 (0.35, 0.9475), respectively. The M2BPGi value in the HCV-viremia group was significantly higher than that in the non-HCV (P < 0.001) and HCV-SVR (P < 0.001) groups (Fig. 3A).

For further analysis, we stratified all patients according to the liver fibrosis score or liver necroinflammation score (Fig. 3B,C). The number of patients in non-HCV, HCV-viremia, and HCV-SVR groups for each fibrosis stage or necroinflammatory index are shown in Table 5. According to the liver fibrosis score, in F1 patients, the M2BPGi value in the HCV-viremia group was significantly higher than that in the non-HCV (P = 0.002) and HCV-SVR (P < 0.001) groups. However, there was no significant difference between these 3 groups of F0 patients. We could not perform statistical analysis of F2 and ≥F3 patients because of the small number of patients in each group.

Similarly, according to the liver necroinflammation score, in A1 patients, the M2BPGi value in the HCV-viremia group was significantly higher than that in the non-HCV (P = 0.04) and HCV-SVR (P < 0.001) groups. However, there were no significant differences between these 3 groups of A0 patients. We could not perform statistical analysis of ≥A2 patients because of the small number of patients in each group.

Expression of M2BPGi in Liver Biopsy Specimens

We performed immunostaining of liver specimens from LT patients. Both WFA- and M2BP-positive cells were found in the liver specimens (Fig. 4A). These cells were also positive for CD68, a specific maker of Kupffer cells in the liver, indicating that Kupffer cells were expressing M2BPGi in the liver (Fig. 4B).