ABSTRACT
-
Background/Aims
Anti-mitochondrial M2 antibody (AMA-M2) is a specific marker for primary biliary cholangitis (PBC) and it could be also present in non-PBC individuals.
-
Methods
A total of 72,173 Chinese health check-up individuals tested AMA-M2, of which non-PBC AMA-M2 positive individuals were performed follow-up. Baseline data of both clinical characteristics and laboratory examinations were collected in all AMA-M2-positive individuals. Least absolute shrinkage and selection operator (LASSO) regression was performed to investigate the potential variables for developing PBC.
-
Results
A total of 2,333 individuals were positive with AMA-M2. Eighty-two individuals had a medical history of PBC or fulfilled the diagnostic criteria of PBC at baseline, and 2,076 individuals were non-PBC. After a median follow-up of 6.6 years, 0.6% developed PBC, with an accumulative 5-year incidence rate of 0.5%. LASSO regression showed that levels of alkaline phosphatase (ALP), gamma-glutamyl transpeptidase (GGT), immunoglobulin M (IgM), eosinophilia proportion (EOS%), gamma globulin percentage, and hemoglobin (HGB) were potential variables for developing PBC. Multivariate Cox regression is used to construct a predictive model based on 7 selected variables, and time-dependent receiver operating characteristic analysis showed that the area under the curve of the prediction model at 3, 5, and 10 years were, respectively, 1.000, 0.875, and 0.917.
-
Conclusions
This study offers insights into the onset of PBC among individuals who tested positive for AMA-M2 during routine health check-ups. The prediction model based on ALP, GGT, IgM, EOS%, gamma globulin percentage, HGB, and sex has a certain predictive ability for the occurrence of PBC in this population.
-
Keywords: Primary biliary cholangitis; Anti-mitochondrial M2 antibody; Health check-up
Study Highlights
• AMA-M2 could be present in non-PBC individuals. Non-PBC AMA-M2 positive individuals could have a higher risk of developing PBC compared to the general population. ALP, GGT, IgM, EOS (%), gamma globulin (%), and HGB contribute to predicting the onset of PBC among AMA-M2 positive individuals without PBC at baseline.
Graphical Abstract
INTRODUCTION
Primary biliary cholangitis (PBC) is an autoimmune liver disease characterized by intrahepatic cholestasis, nonsuppurative cholangitis, and destruction of intrahepatic small bile ducts induced by autoreactive T cells [
1]. Anti-mitochondrial antibodies (AMA) are the immunological hallmark of PBC, which mainly targets E2 subunits of mitochondrial multi-enzyme complexes, the 2-oxo-acid dehydrogenase complexes comprising pyruvate dehydrogenase complex (PDC), branched-chain 2-oxo-acid dehydrogenase complex (BCOADC), and 2-oxo-glutarate dehydrogenase complex (OGDC) [
2]. According to the location of antigens on the inner or outer membrane of mitochondria, sensitivity to trypsin, and electrophoretic characteristics, the target antigens of AMA can be divided into 9 subtypes (M1 to M9) [
3]. Among these subtypes of AMA, AMA-M2 is one of the most specific serum biomarkers for PBC diagnosis, around 95% of patients with PBC showed AMA-M2 positivity [
4].
Several studies showed that AMA could be detected in those who developed PBC in the future, even more than 10 years before the rise of abnormal biochemical, histological changes, and clinical symptoms [
5]. In addition, AMA also could be detected in the general healthy population, although the prevalence of AMA positivity was rare [
6,
7]. Metcalf et al. [
8] first reported that in 29 patients with AMA positive but had normal liver function tests, 24 (83%) were developed to PBC after more than 10 years of follow-up. Dahlqvist et al. [
9] observed that patients with AMA positive AND normal ALP were prone to develop PBC in five years, although the 5-year incidence rate of PBC was only 16%. Zandanell et al. [
10] also reported that non-PBC subjects with AMA positive would infrequently develop PBC over six years. Therefore, individuals with AMA positive have a risk of developing PBC to some degree.
The prediction value of AMA-M2 in non-PBC individuals remains unclear. Previous studies only investigated the prevalence of AMA-M2 in the general health population [
6,
11]. However, the study further investigated the development of PBC in AMA-M2-positive individuals without PBC at baseline was lacked. In this retrospective cohort study, we aim to investigate the prevalence of AMA-M2 positive among the health check-up population, and then we performed follow-up in AMA-M2 positive individuals without PBC at baseline to investigate the incidence of PBC among them. In addition, a predictive model of PBC development in AMA-M2 positive individuals was constructed based on baseline data.
MATERIALS AND METHODS
Data source
The data included in this study was collected from the Peking Union Medical College Hospital-Health Management (PUMCH-HM) database, which has been reported in our previous study [
12]. All clinical data were collected when they performed a physical examination during a half-day clinic visit. Meanwhile, the blood, urine, and fecal samples of each health check-up population were collected to further detect laboratory data.
We retrieved the database from 2010 to 2022 in order to find out the health check-up population who tested AMA-M2 when they visited the Department of HM, PUMCH. The major inclusion criteria were (1) the participants were a health check-up population; (2) the ethnicity of participants was Chinese; (3) AMA-M2 positive with titer equal to or higher than 25 U/mL based on enzyme-linked immunosorbent assay (ELISA) or AMA-M2 positive with titer equal to or higher than 40 U/mL based on chemiluminescent immunoassay (CLIA). The exclusion criteria were (1) participants were from outpatient; (2) participants were not detected AMA-M2 when they performed a physical examination.
Follow up
This was a retrospective cohort study on health check-ups of those who tested positive for AMA-M2 in the Department of HM, PUMCH. The duration of follow-up was defined as the interval between the date of AMA-M2 first detected positive and the date of the last follow-up. All health check-up populations were invited to perform a telephone follow-up by trained health professionals in PUMCH. In our telephone follow-up, we focused on investigating the liver function test results of respondents from the latest examination. If the respondents reported that they had normal liver function tests, we regarded that they did not develop PBC. However, if the respondents reported that they did not know the results of the liver function test, or they did not perform the recent liver function test, they were invited to perform the liver function test and AMA-M2 antibody test at PUMCH. Furthermore, the restaging result would be evaluated by rheumatologists at PUMCH. In addition, the individuals who refused to ask the question or could not be contacted through telephone were regarded as lost to follow-up.
The present study (no. JS-2156) was approved by the Ethics Committee of the PUMCH. All participants provided informed consent.
Laboratory examination
Routine blood tests, routine urinalysis, routine fecal tests, blood biochemical analyses, and immunological tests were performed at the baseline that first detected AMA-M2 positive. For follow-up participants, we performed the liver function test to detect alanine aminotransferase (ALT), total bilirubin (TBIL), direct bilirubin (DBIL), ALP, GGT, aspartate aminotransferase (AST), and total bile acids (TBA). In addition, we also detected immunoglobulin M (IgM) and AMA-M2 in follow-up participants. AMA-M2 was detected by ELISA (Shanghai Kexin Biotech Co., Ltd., Shanghai, China) or CLIA (HOB Biotech Group Corp., Ltd., Suzhou, China).
PBC diagnosis
The diagnosis of PBC was based on generally accepted criteria, according to the practice guidelines of the European Association for the Study of Liver Diseases for PBC [
13]. AMA-M2 positive health check-up individuals with slightly elevated GGT levels ([1–2]×upper limit of normal [ULN]) and normal ALP levels were considered suspected cases of PBC at baseline. Development of PBC in AMA-M2 positive individuals was determined by persistent elevation of ALP and/or GGT. Furthermore, the PBC diagnoses were made by two rheumatologists by consensus.
Statistical analyses were performed using IBM SPSS (version 23.0), R 4.2.1 (
https://www.r-project.org/), and Prism (version 9.0). All continuous variables were expressed as median and interquartile ranges. These data were analyzed using the Mann–Whitney U-test within two groups. Categorical variables were described as rates and percentages and assessed using the chi-square or Fisher’s exact tests. Kaplan–Meier curves were performed to calculate the cumulative rate of developing PBC. Least absolute shrinkage and selection operator (LASSO) regression was performed to investigate the risk factors for PBC. The cross-sectional association of risk factors and PBC presence using logistic regression, and the Cox regression model was performed to analyze the risk factors on PBC development. The Hosmer-Lemeshow test evaluates the fitting effect of the logistic regression model. The calibration curve and decision curve analysis separately evaluate the calibration accuracy and clinical benefit of the logistic regression model. In addition, the time-dependent receiver operating characteristic (ROC) curve was used to evaluate the predictive performance of prediction model based on multivariate Cox regression analysis. For statistical tests involving multiple comparisons, a Benjamini–Hochberg adjustment was used. A two-sided P-value of <0.05 was used to define statistical significance.
RESULTS
AMA-M2 positivity and PBC prevalence in the health check-up population
A total of 87194 AMA-M2 tests were performed in the health check-up population between 2010 and 2022 at the Department of HM, PUMCH (
Fig. 1). After removing 15,021 duplicates, 72,173 individuals were screened (
Fig. 1). The positive rate of AMA-M2 was 3.23% (n=2,333) among the health check-up population (
Fig. 2A), including 1,394 females (59.75%) and 939 males (40.25%) (
Fig. 2B). In addition, the quantitative distribution of positive results according to varied age groups is shown in Figure 2C. Furthermore, the medical history, including history of present illness and past medical history, was examined in AMA-M2 positive individuals. One hundred and seventy-five individuals were excluded due to missing physical examination results (
Fig. 1). In total, 82 individuals (3.8%) had a medical history of PBC or fulfilled the diagnostic criteria of PBC at baseline (
Fig. 1), and 2076 (96.2%) individuals were non-PBC individuals, including 24 (1.1%) patients with non-PBC digestive diseases, 44 (2.0%) patients with non-PBC autoimmune diseases, 17 (0.8%) patients with other diseases, 457 (21.1%) individuals with hyperlipidemia, 322 (14.9%) individuals with abnormal liver function tests, 99 (4.6%) individuals with hyperbilirubinemia, and 1,113 (51.6%) individuals were healthy populations (
Supplementary Fig. 1). In addition, 104 (5.0%) of non-PBC individuals were suspected PBC (
Fig. 1). The levels of AMA-M2 were compared in patients with PBC, non-PBC digestive diseases, non-PBC autoimmune diseases, other diseases, hyperlipidemia, hyperbilirubinemia, and healthy populations. The data showed that all non-PBC individuals had a lower level of AMA-M2 compared to patients with PBC (
Fig. 3).
Both clinical characteristics and laboratory examinations were performed in all AMA-M2 positive individuals. The differential indicators were compared between patients with PBC and non-PBC individuals (
Table 1), while other parameters that showed no significant difference were not shown. Our data showed that the baseline age was significantly older in patients with PBC compared with non-PBC individuals. Body mass index (BMI) was significantly higher in patients with PBC than those non-PBC individuals (
P=0.002,
Table 1). The markers of liver function that are associated with the diagnosis and disease activity of PBC, such as ALT (
P<0.001), DBIL (
P=0.012), ALP (
P<0.001), GGT (
P<0.001), and AST (
P<0.001) were significantly higher in patients with PBC compared with non-PBC individuals (
Table 1). Immunological measures such as IgM (
P<0.001), IgG (
P<0.001), IgA (
P=0.001), hs-CRP (
P<0.001), and RF (
P<0.001) indicated an increased level in patients with PBC compared with non-PBC individuals (
Table 1). However, the levels of ALB were significantly lower in patients with PBC compared with non-PBC individuals (
P=0.001,
Table 1).
In this study, we followed up the 2076 non-PBC individuals for a median duration of 6.6 (interquartile range 4.8–8.1) years, with 268 (12.9%) of them lost to follow-up. In total, 262 (12.6%) individuals refused to follow-up, and 6 (0.3%) individuals were dead during follow-up. The differential baseline indicators between PBC and non-PBC individuals were compared between individuals who completed follow-up and lost to follow-up (
Supplementary Table 1).
Thirteen subjects (0.6%) were developed PBC during follow-up. Eleven (84.6%) of them were female, 7 (53.8%) of them had abnormal liver function tests (six individuals had abnormal GGT levels and one had elevated ALT level (>1×ULN), 4 (30.8%) of them was healthy individual, 1 (7.7%) of them was hyperlipidemia, and 1 (7.7%) of them was patients with breast cancer. We further compared the differential baseline parameters between follow-up non-PBC and follow-up PBC subjects (
Supplementary Table 2). The data showed that individuals who developed PBC had a higher level of ALT (
P=0.01), ALP (
P<0.001), GGT (
P<0.001), AST (
P=0.001), and IgM (
P<0.001) compared with those who did not develop PBC during follow-up (
Supplementary Table 2). The incident rate of PBC in AMA-M2 positive individuals was 98 (95% confidence interval [CI] 45–151) per 100,000 persons per year. Furthermore, 6 individuals who were initially suspected of having PBC were confirmed to have developed the disease during follow-up. Consequently, the incidence of PBC is significantly higher among those suspected of having PBC compared to AMA-M2 positive individuals without PBC at baseline (902 [95% CI 182–1,623] vs. 47 [95% CI 14–97] per 100,000 persons per year;
P<0.001).
The Kaplan–Meier curves showed that the overall cumulative incidence rates of PBC development at 3, 5, and 8 years were 0.2% (95% CI 0–0.4), 0.5% (95% CI 0.2–0.9), and 1.3% (95% CI 0.5–2.0), respectively (
Fig. 4A). Furthermore, we observed that the cumulative incidence rates of PBC development in female AMA-M2 positive individuals at 3, 5, and 8 years were 0.2% (95% CI 0–0.5), 0.6% (95% CI 0.1–1.2), and 2.0% (95% CI 0.7–3.2), respectively (
Fig. 4B). In addition, AMA-M2 positive individuals with abnormal levels of GGT showed significantly higher cumulative incidence rates of PBC development compared to those with a normal level of GGT (
P<0.0001,
Supplementary Fig. 2).
A training set, comprising 70% of the cohort, consisting of 82 confirmed PBC and 2076 AMA-M2 positive individuals without PBC at baseline, was used to construct a logistic regression model. This model investigates the cross-sectional association between various risk factors and the presence of PBC. Meanwhile, a sample size of 30% was used to establish the validation set of the model. In addition, 1808 AMA-M2 positive individuals who completed follow-up were used to build the prediction model based on Cox proportional hazards regression.
LASSO-based feature selection and the dataset including 2,158 AMA-M2 positive individuals were employed to select the significant predictors from the differential indicators shown in
Table 1 and sex based on a minimum criterion to obtain the most regularized and efficient prediction model. Among the 37 indicators, six indicators with non-zero coefficients were identified as the best potential predictors closely associated with PBC including ALP, GGT, IgM, eosinophils percentage (EOS%), gamma globulin percentage, and hemoglobin (HGB). Those significant indicators selected and sex were included in the multivariate logistic regression to establish the model for PBC presence among AMA-M2 positive individuals at baseline. Logistic regression analysis showed that five variables including ALP, GGT, IgM, EOS%, and gamma globulin percentage were associated with PBC presence among AMA-M2 positive individuals at baseline, and their coefficients were shown in
Supplementary Table 3. HGB and sex are also included in the model, contributes to improving the efficiency of the model.
ROC analysis was used to analyze the discriminative effect of the constructed logistic regression model. In the training set, ROC analysis showed that the area under curve (AUC) value of the model was 0.954 (95% CI 0.932–0.976,
Supplementary Fig. 3A). In validation set, ROC analysis showed that the AUC value of the model was 0.974 (95% CI 0.951–0.997,
Supplementary Fig. 3B).
To further evaluate the predictive performance of the logistic regression model, calibration curve analysis is conducted to assess the consistency between observed and predicted results. As a result, it was found that the bias-corrected curve was basically in line with the ideal curve (
Supplementary Fig. 4), indicating good consistency between the predicted and observed results of the model. In addition, the results of the Hosmer-Lemeshow test showed
P=0.185, indicating better goodness of fitting. Meanwhile, decision curves analysis showed that the model has a relatively high clinical net benefit and certain clinical applicability for the AMA-M2 positive health check-up population (
Supplementary Fig. 5). As shown in the supplementary Figure 5, when the threshold probability is in the range of 0.3–96%, the net benefit corresponding to the model is higher than the “all” line and “None” line, indicating that the model has practical value when the threshold probability is in that range. Furthermore, approximately 22% of AMA-M2 positive individuals without PBC received a net benefit when the model was applied.
To build the prediction model of developing PBC among the AMA-M2 positivity population, 1,808 AMA-M2 positive individuals who completed follow-up were used to perform Cox proportional hazards regression analysis based on 7 selected indicators validated by logistic regression. Multivariate Cox regression analysis showed that four variables including ALP, GGT, and EOS (%) were associated with PBC developing among AMA-M2 positive individuals without PBC at baseline (
Supplementary Table 4). IgM, gamma globulin percentage, HGB, and sex are also included in the prediction model, contributing to improving the efficiency of the model. To visualize the prediction model, a nomogram was constructed based on 7 selected indicators, and assigned scores to each of the 7 indicators included in the prediction model to predict the onset of PBC in the AMA-M2 positive health check-up population (
Fig. 5). The time-dependent ROC showed that the AUC value of the prediction model at 3, 5, and 10 years was, respectively, 1.000, 0.875, and 0.917 (
Fig. 6).
DISCUSSION
Our study has reported the development of PBC in the Chinese health check-up population who had AMA-M2 positivity but without PBC at baseline. We observed that AMA-M2 could be positive in non-PBC individuals who had digestive diseases, autoimmune diseases, hyperlipidemia, abnormal liver function tests, and hyperbilirubinemia. Furthermore, we found that non-PBC individuals with AMA-M2 positivity had a low rate of developing PBC during follow-up in this study.
The prevalence of AMA-M2 in health check-up individuals was varied in previous studies. Chen et al. [
11] reported that the prevalence of AMA-M2 in the 19,102 healthy check-up population was 0.73%. Guo et al. [
14] reported that the positive rate of AMA-M2 in the 20970 subjects who received health examination was 0.74%. Liu et al. [
6] showed that the prevalence of AMA-M2 was 0.23% in the 8,126 general adult individuals. However, a 9.67% AMA-M2 positive rate in 8,954 general individuals has been reported by Liang et al. [
15]. The potential reason for a slightly higher prevalence of AMA-M2 in our study might be associated with the study populations. AMA-M2 positive might be associated with abnormal liver function in subjects without PBC [
16]. Four hundred and twenty-one (29.5%) AMA-M2-positive individuals indicated that they had abnormal liver function tests and hyperbilirubinemia, which results in a higher prevalence of AMA-M2 in this study. In addition, AMA-M2 positive individuals who were more than 40 years old were up to more than 60% in the total of AMA-M2 positivity subjects. The age of onset in PBC is usually more than 40 years old [
7], and middle-aged females are prone to having PBC [
17]. In our AMA-M2 tested population, subjects whose age was more than 40 years old were up to 63.4%, which may contribute to a higher prevalence of AMA-M2.
Eighty-two (0.11%) health check individuals who were AMA-M2 positive reported that they were patients with PBC. Liu et al. [
7] reported that four (0.05%) health check-up individuals had been diagnosed with PBC in 8,126 health check-up subjects. Chen et al. [
11] observed that twenty-five (0.13%) health check-up individuals were confirmed with PBC in 19,012 health check-up populations. The above health check-up populations were from varied single centers. Environmental factors (e.g., region of residence) have been proven to be associated with the pathogenesis of PBC [
18]. Therefore, the varied prevalence of PBC in the health check-up population might be associated with the different habitations of subjects. The AMA-M2 levels (ELISA) of AMA-M2 positive health check-up individuals were compared in each group. Patients with PBC had the highest level of AMA-M2 compared with other AMA-M2-positive populations (
Fig. 3). Therefore, a low level of AMA-M2 might not be related to PBC in this study. Several clinical characteristics and laboratory examinations showed significant differences between individuals with PBC and non-PBC AMA-M2 positive individuals at baseline. Besides the biochemical indices and immunoglobulin showing significant differences within the two populations, it was noted that BMI was significantly higher in patients with PBC compared with non-PBC AMA-M2 positive individuals (
P=0.002,
Table 1). This result was consistent with a Mendelian randomization study that observed BMI was an independent causal factor for PBC [
19].
In this study, individuals with AMA-M2 positivity who underwent health check-ups showed a 5-year incidence rate of PBC at 0.5%. This data is relatively lower compared with those studies that performed follow-up in AMA positive population without PBC. Duan et al. [
20] reported that the 5-year incidence rate of PBC was 4.2% in those with AMA positive. Dahlqvist et al. found that 9 of 92 (9.8%) AMA-positive individuals at baseline developed PBC during 4.0±1.8 years of follow-up [
9]. Zandanell et al. [
10] observed that 6 of 59 (10.3%) AMA-positive individuals developed new onset PBC over six years of follow-up. The main reason for the varied incidence rate of PBC in the risk population (AMA-M2/AMA positive but without PBC) was the inconsistent composition in the included risk population at baseline. In this study, the healthy individuals were up to 51.6% (n=1,113) of 2,333 non-PBC AMA-M2 positive individuals, while the above studies included many AMA-M2 patients with liver disease or autoimmune disease but without PBC at baseline. Having liver-related diseases or autoimmune diseases are risk factors for developing PBC in those non-PBC patients with AMA positive [
21]. In addition, the incident rate of PBC was 98 (95% CI 45–151) per 100,000 persons per year in our AMA-M2 positive cohort. Several studies have shown that the incidence rate of PBC ranges from 0.03 to 0.58 per 100,000 persons per year in the general population across different races [
22]. Therefore, non-PBC AMA-M2 positive individuals could have a higher risk of developing PBC compared to the general population.
We further performed LASSO regression to find the potential variables predicting development of PBC, indicating that ALP, GGT, IgM, EOS%, gamma globulin percentage, and HGB could be used to construct a prediction model to evaluate the onset of PBC in non-PBC AMA-M2 positive individuals. The baseline levels of ALP, GGT, IgM, EOS%, and gamma globulin percentage showed significantly higher in individuals who have developed PBC compared with those not developed during follow-up. ALP and GGT are biomarkers of cholestasis, contributing to PBC diagnosis [
23]. However, GGT is not recommended for PBC diagnosis in some diagnostic criteria of PBC due to GGT could be elevated secondary to drug or alcohol exposure or fatty liver [
24,
25]. In recent years, many studies have focused on the role of elevated GGT in patients with PBC, especially in those with normal ALP. The level of GGT could be elevated earlier than the level of ALP in patients with PBC [
26]. In addition, PBC patients who had normal ALP and elevated GGT could have histology typical for PBC, indicating that GGT offers a more dynamic picture of PBC than ALP [
27,
28]. Furthermore, our data showed that individuals who were AMA-M2 positive and suspected of having PBC had a higher risk of developing PBC compared to those who were definitively without PBC at baseline. This was due to a significantly increased incidence rate observed among the suspected PBC cases. However, it should be noted that not all initially suspected of having PBC will ultimately progress to the condition. The level of serum IgM could be increased in patients with PBC [
29]. Serum IgM was also elevated in patients with PBC who had normal ALP and histologic PBC features [
30], indicating that serum IgM contributes to early diagnosis of PBC. EOS has been reported as the common and distinctive feature of patients with PBC, in which the levels of circulating eosinophilia were increased in patients with PBC [
31]. Furthermore, eosinophilic infiltration and granulomas could be early observed in in the biliary lesions of PBC [
32]. In addition, serum levels of potent chemokines for EOS such as CXCL5, CCL11, CCL13, CCL24, and CCL26 could be increased in patients with PBC [
33,
34], contributing to the recruitment and activation of EOS in PBC. Our studies also indicated that the percentage of circulating EOS has a predictive value of PBC development in those with AMA-M2 positivity, which might be associated with the increased levels of chemoattractant for EOS. Previous studies have found that serum gamma globulin levels in AMA or AMA-M2 positive PBC patients are significantly higher than those in AMA or AMA-M2 negative PBC patients [
35], and elevated gamma globulin levels suggest impaired liver function [
36]. This study observed that the percentage of baseline gamma globulin levels in AMA-M2 positive individuals who developed PBC was significantly higher than those who did not develop PBC, indicating that an increase in this indicator suggests the risk of PBC occurrence in AMA-M2 positive individuals. In addition, the HGB level of patients with PBC was significantly lower than those of AMA-M2 positive individuals without PBC in this study. A Mendelian randomization study found that HGB is a protective factor for the occurrence of PBC [
37], meaning that the higher the Hb level, the lower the risk of PBC occurrence. According to our results, non-PBC individuals with AMA-M2 positive should monitor the levels of ALP, GGT, IgM, and gamma globulin through liver function tests and immunological examination. In addition, the levels of EOS% and HGB also should be monitored through routine blood tests.
To predict the onset of PBC among AMA-M2 positive individuals, we constructed a prediction model based on the six variables selected by LASSO regression. Due to PBC with a female sex predilection [
25], sex as a variable was also included in the prediction model to improve the predictive performance of the model. The time-dependent ROC analysis showed that the prediction model based on multivariate Cox regression has a good predictive value for PBC development among AMA-M2 positive health check-up individuals without PBC at baseline. PBC is not a common disease and the incidence rate is not high in the general population [
25]. Meanwhile, we also found that the 8-year cumulative incidence of PBC was relatively low among AMA-M2 positive health check-up individuals. Furthermore, it should be noted that only AMA-M2 positivity could not be applied to diagnose PBC. In clinical practice, clinicians who have not majored in rheumatology or gastroenterology may not be familiar with the diagnostic criteria of PBC, resulting in some clinicians diagnosing PBC only based on AMA-M2 or other PBC-specific autoantibodies. However, our data showed that only a minority of AMA-M2 positive individuals developed PBC during follow-up. It is necessary to assess the risk of developing PBC among individuals who test positive for AMA-M2, but a diagnosis should not be made solely on the basis of autoantibody positivity. Therefore, the model that we constructed with some potential variables for PBC development may contribute to early identification of the PBC risk population among AMA-M2 positive health checkup individuals.
There were several limitations in this study. Firstly, some AMA-M2-positive individuals did not finish the follow-up. Secondly, this study was a single-center study with a retrospective design. Therefore, the results of this study need to be validated in another cohort in the future study. Thirdly, the incident rate of PBC in non-PBC AMA-M2 positive individuals might be underestimated due to the liver biopsy not being applied in the AMA-M2 positive individuals. Therefore, the individuals with AMA-M2 positive but might have histologic features of PBC were not identified in this study. Additionally, we did not investigate the development of PBC in individuals who are AMA-M2 negative. More studies are needed to compare the risk of PBC between AMA-M2 positive individuals and AMA-M2 negative individuals. Finally, the age of individuals who were AMA-M2 positive and developed PBC was significantly older compared to those who did not develop PBC (
Supplementary Table 2,
P<0.001). This resulted in some younger AMA-M2 positive individuals not observing the onset of PBC with our limited follow-up duration. The incidence of PBC among AMA-M2 positive individuals might increase with extended follow-up time.
In conclusion, this study showed that AMA-M2 could be present in non-PBC individuals. The level of AMA-M2 was significantly lower in non-PBC individuals compared with established PBC at baseline. Furthermore, our study offers insights into the onset of PBC among individuals who tested positive for AMA-M2 during routine health check-ups. In addition, the baseline levels of ALP, GGT, IgM, EOS (%), gamma globulin percentage, and HGB were biomarkers for predicting the development of PBC in AMA-M2-positive individuals without PBC at baseline.
FOOTNOTES
-
Authors’ contribution
All authors significantly contributed to the manuscript and approved the final version for publication. YZ, TD and LW conceived and designed the study. HL, XW, SY, XX and YL contributed to performing follow-up and data acquisition. SL, HL, and XW carried out the statistical analysis. All authors contributed to the interpretation of data. HL wrote the first manuscript draft, which was critically revised by YZ, TD and LW.
-
Acknowledgements
This work was supported by the National High Level Hospital Clinical Research Funding (2022-PUMCH-B-124), and the National Key Research and Development Program of China (2018YFE0207300).
-
Conflicts of Interest
None declared. Graphical abstract created with BioRender.com, with permission (https://BioRender.com/d42i868).
SUPPLEMENTAL MATERIAL
Supplementary material is available at Clinical and Molecular Hepatology website (
http://www.e-cmh.org).
Supplementary Table 1.
Comparison of demographics, differential baseline clinical characteristics and laboratory examinations for AMA-M2 positive individuals completing follow-up and individuals not available to follow-up
Supplementary Table 2.
Comparison of demographics, differential baseline clinical characteristics and laboratory examinations for AMA-M2 positive individuals who developed PBC and did not develop PBC during follow-up
Supplementary Figure 1.
The baseline characteristics of health check-up individuals without PBC. AITD, autoimmune thyroid disease; PM/DM, dermatomyositis; RA, rheumatoid arthritis; SLE, systemic lupus erythematosus; SS, sjogren's syndrome; SSc, systemic sclerosis; UCTD, undifferentiated connective tissue disease.
Supplementary Figure 2.
Cumulative incidence of PBC with 95% confidence interval boundaries in AMA-M2 positive individuals with normal GGT level and those with abnormal GGT level. AMA-M2, anti-mitochondrial M2 antibody; PBC, primary biliary cholangitis; GGT, gamma-glutamyl transferase.
Supplementary Figure 3.
Receiver operating characteristic (ROC) validation of PBC presence among AMA-M2 positive individuals at baseline in training set and testing set: (A) The area under the receiver operating characteristic curve (AUC) represents the discrimination performance of the model in the training set; (B) The AUC represents the discrimination performance of the model in the validation set.
Supplementary Figure 4.
The calibration curve of the logistic regression model for PBC presence among AMA-M2 positive individuals at baseline. AMA-M2, anti-mitochondrial M2 antibody; PBC, primary biliary cholangitis.
Supplementary Figure 5.
The decision curve of the logistic regression model for AMA-M2 positive health check-up population developing to PBC. AMA-M2, anti-mitochondrial M2 antibody; PBC, primary biliary cholangitis.
Figure 1.Flowchart of the study. AMA-M2, anti-mitochondrial M2 antibody; PBC, primary biliary cholangitis.
Figure 2.The prevalence of AMA-M2 in health check-up individuals. (A) The pie chart shows that the prevalence of AMA-M2 in total health check-up individuals. (B) The pie chart shows that the distribution of sex in AMA-M2 positive health check-up individuals. (C) The pie chart shows the distribution of age in AMA-M2 positive health check-up individuals. AMA-M2, anti-mitochondrial M2 antibody.
Figure 3.The levels of AMA-M2 (ELISA method) in different individuals. AMA-M2, anti-mitochondrial M2 antibody; PBC, primary biliary cholangitis. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.
Figure 4.Cumulative incidence of PBC with 95% confidence interval boundaries in the AMA-M2 positive individuals. (A) All AMA-M2 positive individuals without PBC at baseline. (B) Female and male AMA-M2 positive individuals without PBC at baseline. AMA-M2, antimitochondrial M2 antibody; PBC, primary biliary cholangitis.
Figure 5.The nomogram of the prediction model based on multivariate Cox regression analysis. HGB, hemoglobin; IgG, immunoglobulin G; EOS, eosinophils; GGT, gamma-glutamyl transferase; ALP, alkaline phosphatase.
Figure 6.Time-dependent ROC for the prediction model along with AUC obtained from the follow-up cohort (1,808 AMA-M2 positive individuals who completed follow-up) at 3, 5, and 10 years. ROC, receiver operating characteristic; AUC, area under curve; AMA-M2, anti-mitochondrial M2 antibody.
Table 1.Demographics, differential baseline clinical characteristics and laboratory examinations of AMA-M2 positive individuals
Table 1.
|
Characteristics |
PBC (n=82)
|
Non-PBC (n=2,076)
|
P-value |
Adjust P-value |
|
Available data (n) |
Median (IQR) or n (%) |
Available data (n) |
Median (IQR) or n (%) |
|
Age |
82 |
54 (46–60) |
2,076 |
45 (34–53) |
<0.001 |
<0.001 |
|
Female gender, n (%) |
82 |
55 (67.1%) |
2,076 |
1238 (59.6%) |
0.178 |
0.178 |
|
BMI (kg/m2) |
82 |
24.76 (21.97–27.97) |
2,064 |
23.57 (21.19–25.99) |
0.004 |
0.005 |
|
Height (cm) |
82 |
162.95 (155.45–168.30) |
2,064 |
165.00 (159.00–171.00) |
0.008 |
0.010 |
|
Waist (cm) |
77 |
84.00 (75.50–91.50) |
1,893 |
80.00 (72.00–89.00) |
0.010 |
0.013 |
|
ALT (U/L) |
82 |
29.50 (20.00–45.25) |
2,076 |
17.00 (12.00–25.00) |
<0.001 |
<0.001 |
|
TBIL (μmol/L) |
82 |
11.80 (8.45–14.70) |
2,076 |
10.30 (7.70–13.30) |
0.022 |
0.024 |
|
DBIL (μmol/L) |
82 |
4.30 (3.20–5.73) |
2,076 |
3.80 (3.00–4.80) |
0.012 |
0.013 |
|
ALP (U/L) |
82 |
111.00 (81.00–153.00) |
2,076 |
63.00 (52.00–76.00) |
<0.001 |
<0.001 |
|
GGT (U/L) |
82 |
118.50 (65.75–182.00) |
2,076 |
19.00 (13.00–33.00) |
<0.001 |
<0.001 |
|
AST (U/L) |
82 |
30.00 (21.00–38.25) |
2,076 |
19.00 (16.00–23.00) |
<0.001 |
<0.001 |
|
IgM (g/L) |
79 |
2.15 (1.53–3.51) |
2,036 |
1.06 (0.75–1.50) |
<0.001 |
<0.001 |
|
IgG (g/L) |
79 |
13.15 (11.29–15.67) |
2,037 |
11.66 (10.04–13.65) |
<0.001 |
<0.001 |
|
IgA (g/L) |
79 |
2.69 (1.89–3.23) |
2,036 |
2.17 (1.65–2.78) |
0.001 |
0.002 |
|
TP (g/L) |
82 |
76.00 (72.00–78.00) |
2,076 |
73.00 (70.00–76.00) |
<0.001 |
<0.001 |
|
ALB (g/L) |
82 |
45.00 (43.00–47.00) |
2,076 |
46.00 (44.00–48.00) |
0.001 |
<0.001 |
|
A/G |
82 |
1.50 (1.40–1.63) |
2,076 |
1.70 (1.50–1.90) |
<0.001 |
<0.001 |
|
Glu (mmol/L) |
82 |
5.30 (4.90–5.90) |
2,076 |
5.00 (4.70–5.40) |
<0.001 |
<0.001 |
|
TG (mmol/L) |
81 |
1.42 (1.11–2.15) |
2,063 |
1.12 (0.79–1.67) |
<0.001 |
<0.001 |
|
hs-CRP (mg/L) |
79 |
1.67 (0.90–3.17) |
2,021 |
0.62 (0.33–1.32) |
<0.001 |
<0.001 |
|
RF (IU/mL) |
81 |
8.80 (6.00–17.25) |
2,070 |
5.50 (3.70–9.20) |
<0.001 |
<0.001 |
|
CysC (mg/L) |
70 |
0.85 (0.76–0.97) |
1,785 |
0.78 (0.70–0.88) |
<0.001 |
<0.001 |
|
Cr (μmol/L) |
82 |
63.00 (56.00–72.25) |
2,076 |
67.00 (59.00–79.00) |
0.009 |
0.011 |
|
TG-Ab (IU/L) |
81 |
17.18 (12.00–34.76) |
2,056 |
13.00 (10.00–19.91) |
<0.001 |
<0.001 |
|
TPO-Ab (IU/L) |
81 |
13.78 (10.00–18.45) |
2,056 |
11.90 (9.00–16.57) |
0.011 |
0.013 |
|
CA199 (U/mL) |
82 |
4.70 (2.33–9.30) |
2,076 |
6.10 (3.30–12.20) |
0.023 |
0.025 |
|
CEA (ng/mL) |
82 |
1.87 (1.24-2.61) |
2,076 |
1.60 (1.07-2.40) |
0.033 |
0.034 |
|
EOS% (%) |
82 |
4.15 (2.08–6.18) |
2,076 |
2.00 (1.30–3.30) |
<0.001 |
<0.001 |
|
EOS# (×109/L) |
82 |
0.21 (0.12–0.40) |
2,076 |
0.12 (0.07–0.20) |
<0.001 |
<0.001 |
|
HGB (g/L) |
82 |
139.00 (129.75–147.25) |
2,076 |
142.00 (133.00–154.00) |
0.041 |
0.041 |
|
ESR (mm/h) |
60 |
16.00 (9.00–24.50) |
1,893 |
7.00 (3.00–12.00) |
<0.001 |
<0.001 |
|
α1 (%) |
81 |
3.40 (3.25–3.75) |
2,073 |
3.30 (3.10–3.60) |
<0.001 |
<0.001 |
|
β1 (%) |
81 |
5.80 (5.35–6.30) |
2,073 |
5.60 (5.30–6.00) |
0.012 |
0.013 |
|
β2 (%) |
81 |
5.10 (4.60–5.75) |
2,073 |
4.70 (4.20–4.35) |
<0.001 |
<0.001 |
|
γ (%) |
81 |
19.10 (17.35–21.15) |
2,073 |
17.00 (15.10–18.80) |
<0.001 |
<0.001 |
|
Alb% |
81 |
57.50 (55.40–60.55) |
2,073 |
61.20 (59.10–63.40) |
<0.001 |
<0.001 |
|
A/G (Electrophoresis method) |
81 |
1.40 (1.20–1.50) |
2,073 |
1.60 (1.40–1.70) |
<0.001 |
<0.001 |
Abbreviations
Anti-mitochondrial M2 antibody
branched-chain 2-oxo-acid dehydrogenase complex
chemiluminescent immunoassay
enzyme-linked immunosorbent assay
erythrocyte sedimentation rate
gamma-glutamyl transferase
hypersensitive-c-reactive-protein
Least absolute shrinkage and selection operator
2-oxo-glutarate dehydrogenase complex
pyruvate dehydrogenase complex
primary biliary cholangitis
Peking Union Medical College Hospital
receiver operating characteristic
anti-thyroglobulin antibodies
anti-thyroid Peroxidase Antibody
REFERENCES
- 1. Nevens F, Trauner M, Manns MP. Primary biliary cholangitis as a roadmap for the development of novel treatments for cholestatic liver diseases†. J Hepatol 2023;78:430-441.
- 2. Moteki S, Leung PS, Coppel RL, Dickson ER, Kaplan MM, Munoz S, et al. Use of a designer triple expression hybrid clone for three different lipoyl domain for the detection of antimitochondrial autoantibodies. Hepatology 1996;24:97-103.
- 3. Berg PA, Klein R. Mitochondrial antigens and autoantibodies: from anti-M1 to anti-M9. Klin Wochenschr 1986;64:897-909.
- 4. Oertelt S, Rieger R, Selmi C, Invernizzi P, Ansari AA, Coppel RL, et al. A sensitive bead assay for antimitochondrial antibodies: Chipping away at AMA-negative primary biliary cirrhosis. Hepatology 2007;45:659-665.
- 5. Poupon R. Primary biliary cirrhosis: a 2010 update. J Hepatol 2010;52:745-758.
- 6. Liu HY, Wang LX, Liu YF. Frequencies of autoantibodies specific for primary biliary cirrhosis in a general adult population group. Zhonghua Gan Zang Bing Za Zhi 2008;16:922-925.
- 7. Liu H, Liu Y, Wang L, Xu D, Lin B, Zhong R, et al. Prevalence of primary biliary cirrhosis in adults referring hospital for annual health check-up in Southern China. BMC Gastroenterol 2010;10:100.
- 8. Metcalf JV, Mitchison HC, Palmer JM, Jones DE, Bassendine MF, James OF. Natural history of early primary biliary cirrhosis. Lancet 1996;348:1399-1402.
- 9. Dahlqvist G, Gaouar F, Carrat F, Meurisse S, Chazouillères O, Poupon R, et al. Large-scale characterization study of patients with antimitochondrial antibodies but nonestablished primary biliary cholangitis. Hepatology 2017;65:152-163.
- 10. Zandanell S, Strasser M, Feldman A, Tevini J, Strebinger G, Niederseer D, et al. Low rate of new-onset primary biliary cholangitis in a cohort of anti-mitochondrial antibody-positive subjects over six years of follow-up. J Intern Med 2020;287:395-404.
- 11. Chen BH, Wang QQ, Zhang W, Zhao LY, Wang GQ. Screening of anti-mitochondrial antibody subtype M2 in residents at least 18 years of age in an urban district of Shanghai, China. Eur Rev Med Pharmacol Sci 2016;20:2052-2060.
- 12. Li J, Lin S, Deng C, Xu T. Predictive value of GAD antibody for diabetes in normal Chinese adults: A retrospective cohort study in China. Diabetes Metab Syndr Obes 2021;14:885-893.
- 13. European Association for the Study of the Liver. EASL Clinical Practice Guidelines: The diagnosis and management of patients with primary biliary cholangitis. J Hepatol 2017;67:145-172.
- 14. Guo Y, Wang C, Liu X, Liu J, Wang M, Gao Z, et al. Evaluation of markers associated with primary biliary cirrhosis in a population of anti-mitochondrial antibody-M2-positive individuals. Zhonghua Gan Zang Bing Za Zhi 2014;22:735-738.
- 15. Liang EY, Liu M, Ke PF, Han G, Zhang C, Deng L, et al. A population-based characterization study of anti-mitochondrial M2 antibodies and its consistency with anti-mitochondrial antibodies. Lab Med 2023;54:618-625.
- 16. Sun L, Wang Y, Liu Y, Zhao Y, Zhang X, Wang Y, et al. Differential characteristics of AMA-M2 autoantibody in primary biliary cirrhosis and non-PBC patients. Zhonghua Gan Zang Bing Za Zhi 2015;23:343-349.
- 17. Podda M, Selmi C, Lleo A, Moroni L, Invernizzi P. The limitations and hidden gems of the epidemiology of primary biliary cirrhosis. J Autoimmun 2013;46:81-87.
- 18. Matsumoto K, Ohfuji S, Abe M, Komori A, Takahashi A, Fujii H, et al. Environmental factors, medical and family history, and comorbidities associated with primary biliary cholangitis in Japan: a multicenter case-control study. J Gastroenterol 2022;57:19-29.
- 19. Xu H, Wu Z, Feng F, Li Y, Zhang S. Low vitamin D concentrations and BMI are causal factors for primary biliary cholangitis: A mendelian randomization study. Front Immunol 2022;13:1055953.
- 20. Duan W, Chen S, Li S, Lv T, Li B, Wang X, et al. The future risk of primary biliary cholangitis (PBC) is low among patients with incidental anti-mitochondrial antibodies but without baseline PBC. Hepatol Commun 2022;6:3112-3119.
- 21. Hirschfield GM, Dyson JK, Alexander GJM, Chapman MH, Collier J, Hübscher S, et al. The British Society of Gastroenterology/UK-PBC primary biliary cholangitis treatment and management guidelines. Gut 2018;67:1568-1594.
- 22. Lleo A, Colapietro F. Changes in the epidemiology of primary biliary cholangitis. Clin Liver Dis 2018;22:429-441.
- 23. Younossi ZM, Bernstein D, Shiffman ML, Kwo P, Kim WR, Kowdley KV, et al. Diagnosis and management of primary biliary cholangitis. Am J Gastroenterol 2019;114:48-63.
- 24. Lindor KD, Gershwin ME, Poupon R, Kaplan M, Bergasa NV, Heathcote EJ, et al. Primary biliary cirrhosis. Hepatology 2009;50:291-308.
- 25. European Association for the Study of the Liver. EASL Clinical Practice Guidelines: management of cholestatic liver diseases. J Hepatol 2009;51:237-267.
- 26. Gerussi A. Gamma-glutamyl transferase in patients with primary biliary cholangitis. Gastroenterol Hepatol (N Y) 2022;18:706-708.
- 27. Terziroli Beretta-Piccoli B, Stirnimann G, Mertens J, Semela D, Zen Y, Mazzucchelli L, et al. Primary biliary cholangitis with normal alkaline phosphatase: A neglected clinical entity challenging current guidelines. J Autoimmun 2021;116:102578.
- 28. Zhu YJ, Li J, Liu YG, Jiang Y, Cheng XJ, Han X, et al. Role of biochemical markers and autoantibodies in diagnosis of early-stage primary biliary cholangitis. World J Gastroenterol 2023;29:5075-5081.
- 29. Lleo A, Liao J, Invernizzi P, Zhao M, Bernuzzi F, Ma L, et al. Immunoglobulin M levels inversely correlate with CD40 ligand promoter methylation in patients with primary biliary cirrhosis. Hepatology 2012;55:153-160.
- 30. Ding D, Xuan G, Hu Y, Yu J, Liu Y, Guo G, et al. Immunoglobulin M: A neglected serum biomarker in treatment-naive primary biliary cholangitis with normal alkaline phosphatase. Hepatol Commun 2022;6:1403-1412.
- 31. Yamazaki K, Nakadate I, Suzuki K, Sato S, Masuda T. Eosinophilia in primary biliary cirrhosis. Am J Gastroenterol 1996;91:516-522.
- 32. Ichiki Y, Shimoda S, Ishibashi H, Gershwin ME. Is primary biliary cirrhosis a model autoimmune disease? Autoimmun Rev 2004;3:331-336.
- 33. Mu N, Lin F, Jiang Z, Liang Y, Yang Z. Characteristics of serum chemokine profile in primary biliary cholangitis. Cytokine 2020;136:155291.
- 34. Lin F, Shi H, Liu D, Zhang Z, Luo W, Mao P, et al. Association of CCL11, CCL24 and CCL26 with primary biliary cholangitis. Int Immunopharmacol 2019;67:372-377.
- 35. Wang QX, Shen L, Chen XY, Qiu DK, Ma X. Clinical and pathological features of primary biliary cirrhotic patients with negative anti-mitochondria antibody. Zhonghua Gan Zang Bing Za Zhi 2011;19:340-344.
- 36. Tomasi TB Jr, Tisdale WA. Serum gamma-globulins in acute and chronic liver diseases. Nature 1964;201:834-835.
- 37. Ke B, Li C, Shang H. Hematologic traits and primary biliary cholangitis: a Mendelian randomization study. J Hum Genet 2023;68:559-563.
, Tengda Xu4
