Increased risk of fragility fractures in patients with primary biliary cholangitis

Abstract Large-scale studies on the risk of fragility fractures in patients with primary biliary cholangitis (PBC) are limited due to low incidence. We aimed to investigate whether PBC is associated with fragility fractures using real-world nationwide data. The Korean National Health Insurance Service claims data from 2007 to 2020 were analyzed in this population-based cohort study. Patients with PBC (n = 4951) were matched with controls (n = 19 793) using a 1:4 ratio based on age, sex, and follow-up duration. The primary outcome was fragility fracture, which comprised fractures of the vertebra, hip, distal radius, and proximal humerus. The incidence rates (IRs) and hazard ratios (HRs) were determined to assess the impact of PBC on fragility fractures. During the median follow-up period of 5.37 years, 524 patients in the PBC group had fragility fractures (IR, 18.59/1000 person-years [PYs]). After adjusting for covariates, PBC increased the risk of fragility fractures by 1.63-fold (95% confidence interval, 1.20–2.22; P = .002). The vertebra and hip were particularly susceptible to fracture in patients with PBC, with adjusted HRs of 1.77 and 2.23, respectively. In the subgroup analysis, the risk of fragility fracture was 2.53-fold higher in men and 1.59-fold higher in women with PBC than that in the respective matched control groups. Considering the morbidity and mortality related to fragility fractures, increasing awareness of fragility fracture risk and implementing appropriate preventive measures in patients with PBC are imperative.


Introduction
Primary biliary cholangitis (PBC) is a chronic and progressive autoimmune liver disease characterized by nonsuppurative, granulomatous cholangitis that primarily affects the interlobular and septal bile ducts. 1,2If left untreated, PBC can lead to end-stage liver disease, such as liver cirrhosis and hepatic dysfunction.In addition to hepatic complications, patients with PBC often experience extrahepatic manifestations, including fatigue, pruritus, other autoimmune diseases, and osteoporosis.
Hepatic osteodystrophy is a widely recognized condition in which cirrhosis with poor liver function alters bone metabolism and leads to increase osteoporosis and osteoporotic fractures. 3,4Further, PBC may affect bone metabolism through various intricate and multifactorial mechanisms, making patients with PBC particularly vulnerable to osteoporosis. 5Cholestasis disturbs the absorption of vitamin D and calcium, which are essential elements for bone mineralization. 5Also, elevated bilirubin and consistent inflammation inhibit osteoblast activity. 5Indeed, PBC deteriorates bone microstructure, especially by reducing cortical thickening, which is responsible for bone strength. 6,7steoporosis and related fractures are well-recognized issues.Osteoporotic fracture increases the incidence of subsequent fractures, morbidity, and mortality, deteriorating quality of life. 8Attempts to characterize the relationship between PBC and osteoporotic fractures have been inconclusive and inconsistent. 9,10Because the prevalence of PBC is increasing, large-scale cohort studies are needed to assess the incidence of and risk factors for fragility fractures in patients with PBC. 11Therefore, this study aimed to determine the impact of PBC on fragility fractures using Korean nationwide cohort data.

Study design and population
This population-based cohort study was conducted using nationwide health insurance claims data from the Korea National Health Insurance Service (NHIS).Korea has a universal health coverage system that provides insurance to almost the entire Korean population.The NHIS manages the national health insurance information database, which stores demographic data, medical records based on the International Classification of Diseases, 10th Revision (ICD-10) codes, and details of drug prescriptions, health check-ups, and use of healthcare facilities. 12Patients with rare and incurable diseases (RIDs) are registered as special estimate cases.The NHIS supports 90% of these individuals' medical expenses and provides them access to new medical technologies and drugs.Classification as a special estimate case requires a qualified physician's diagnosis based on serologic, imaging, and histologic results. 13For example, patients with at least two of the following conditions could be diagnosed with PBC: (1) cholestasis with alkaline phosphatase elevation without evidence of other causes; (2) presence of antimitochondrial antibody or PBC-specific antinuclear antibody immunofluorescence; and (3) characteristic pathology with chronic, nonsuppurative cholangitis. 1,2Based on other previous studies, the accuracy of RIDs in Korea is reliable with the accuracy exceeding 90%. 14,15e included patients newly diagnosed with PBC between 2007 and 2020.PBC diagnosis was identified using two criteria, namely the ICD-10 code for PBC (K74.3) and RID registration code (V174).We excluded the following individuals: (1) 18 patients with insufficient medical information; (2) 603 patients with other chronic liver diseases, such as chronic hepatitis B virus infection, alpha-1 antitrypsin defi-  2) 452 patients with a history of fragility fracture.In total, 4951 patients were selected for the study.Patients with PBC received treatment with ursodeoxycholic acid (UDCA), consistent with the established guidelines. 1,2e employed a 1:4 case-to-control ratio to enhance statistical power and ensure robust comparability between patients with PBC and matched controls (19 973 individuals).Controls were meticulously matched to cases based on age, sex, and follow-up duration with additional adjustments made for comorbidities and medication use to accurately assess and compare the risk of fragility fractures.This approach was designed to ensure a balanced comparison, accurately reflecting the health status and minimizing potential bias in fracture risk estimation (Figure 1).This study was conducted in accordance with the Declarations of Helsinki and Istanbul.The Institutional Review Board of Asan Medical Center approved this study (IRB No. 2021-1445), and the requirement for obtaining informed consent was waived as the data were deidentified.

Data collection
We gathered information on age, sex, socioeconomic status, death, comorbidities, treatment, prescriptions, and health check-ups.Socioeconomic status was based on household income, which was derived from the income-based contribution imposition system; individuals unable to pay insurance contributions are granted support by the medical aid system.Comorbidities claimed in the year prior to PBC diagnosis were analyzed, including hypertension, diabetes mellitus, dyslipidemia, stroke, chronic kidney disease, osteoporosis, anemia, and rheumatoid arthritis.Patients with decompensated cirrhosis were identified as those requiring treatment for ascites (paracentesis or spironolactone), variceal bleeding (endoscopic treatment, terlipressin, or somatostatin), or hepatic encephalopathy (Table S1).Medication use was defined as prescription for at least 30 consecutive days within the first 6 months of PBC diagnosis, while long-term medication use was classified as prescription for more than180 days within the first 2 years of PBC diagnosis.We assessed the use of fibrates (bezafibrate and fenofibrate), glucocorticoids (prednisone, prednisolone, methylprednisolone, deflazacort, betamethasone, dexamethasone, and hydrocortisone), and non-glucocorticoid immunosuppressive agents (azathioprine, mercaptopurine, and mycophenolate mofetil).
The NHIS provides general health check-ups to people aged ≥20 years once every 2 years for early diagnosis of diseases.This includes the collection of anthropometric data; blood pressure measurement; chest X-ray; blood testing including hemoglobin, fasting glucose, aminotransferases, r-glutamyl transpeptidase, and serum creatinine levels; urine testing; visual and hearing assessment; dental examination; and assessment of lifestyle factors, such as alcohol consumption, smoking, and physical activity. 16In 2020, 67.8% of all eligible Koreans underwent a health check-up. 17

Outcomes
The primary outcome of the study was any fragility fracture, which comprised fractures of the vertebra, hip, distal radius, and proximal humerus. 18,19Overall fractures were defined on an individual basis, capturing the first occurrence of a fragility fracture, regardless of the site.Site-specific fracture outcome was computed per lesion.For the fractures occurring at the same sites, only the initial episode was recorded.
The fragility fractures were validated by operational definitions using ICD-10 codes, procedure codes, radiographic study codes, conservative codes, and hospitalization. 18,20Our study employs an operational definition for fragility fractures that has been validated in numerous prior studies. 18,20,21able S2 details the specific definitions and codes utilized.In summary, a diagnosis of a fragility fracture was considered if a fracture code was accompanied by a procedural code (e.g., vertebroplasty or fracture reduction) or if a fracture code was accompanied with hospital admission and relevant radiographic codes.

Statistical analyses
We included all eligible individuals in the analyses.The matched control group was created using the greedy method, based on age (±2 years), sex, follow-up duration (0-1 year), and health check-up (yes/no).For the PBC group, the index date was defined as the date of RID registration for PBC, and the corresponding matched date was assigned for the control group.The follow-up duration was calculated from the index date to the event of fragility fracture or death, or the date of the last follow-up, December 31, 2020, the date of last follow-up, whichever came first.Conditional logistic regression analysis was used to compare the baseline characteristics of the PBC group and matched control.
We calculated the incidence rate (IR) of fragility fractures per 1000 person-years (PYs) and estimated 95% confidence intervals (CIs) using the Poisson distribution.Thereafter, we used Kaplan-Meier curves to present the cumulative incidence of fragility fractures.Multivariable Cox regression analysis with robust standard error was employed to compare the incidence of fragility fractures between the PBC and control groups.To minimize potential bias, we performed univariate and multivariable analyses, adjusting for following covariables: PBC, age, sex, socioeconomic status, comorbidities, decompensated cirrhosis, body mass index (BMI), smoking, alcohol consumption, exercise, health check-up, and medication use.Multivariable Cox regression was conducted using variables with P < .1 in the univariate analysis to identify the risk factors for fragility fractures in patients with PBC.When assessing the risk of overall fracture, we recorded the first fracture event per individual, regardless of the site.For sitespecific fractures, each fracture occurring in a different site was considered as a separate event.This methodology was applied consistently in both the Kaplan-Meier curves and Cox regression analysis.Furthermore, to evaluate the influence of PBC on fragility fractures stratified by sex, we conducted a subgroup analysis and calculated adjusted hazard ratios (AHRs) for each sex.We also categorized women into pre-(≤50 years old) and post-menopausal (>50 years old) groups to determine the fracture risk before and after menopause. 22tatistical significance was set at P < .05,and all statistical analyses were performed using SAS Enterprise Guide version 7.1 (SAS Institute, Inc., Cary, NC, USA).

Baseline characteristics
Table 1 presents the demographic and clinical features of the study population.The mean age of patients with PBC was 57.3 years, and 4059 (82.0%) were women.Compared with the matched control group, the PBC group had a higher prevalence of all comorbidities, with higher Charlson Comorbidity Index scores.Decompensated cirrhosis was present in 7.7% of patients in the PBC group but only in 0.2% of individuals in the control (P < .001).Additionally, medication use was more frequent in the PBC than in the matched control.Specifically, 93.9% of patients with PBC were treated with UDCAs.The use of glucocorticoids and non-glucocorticoid immunosuppressive agents was also significantly higher in the PBC group than matched controls (12.9% vs. 2.4% and 7.0% vs. 0.1%, respectively; both P < .001).In the entire study population, 75.6% of individuals underwent a health check-up.The proportion of patients in the PBC group who consumed alcohol was 14.4%, which was significantly different from the 22.2% in the non-PBC group.In contrast, no statistically significant differences in smoking or exercise habits were identified between the two groups.

Fragility fractures
The median follow-up duration for patients with PBC was 5.37 years (interquartile range [IQR], 2.88-8.86years).A total of 524 fragility fractures was observed in the PBC group, resulting in an IR of 18.59/1000 PYs (95% CI, 17.03-20.25),as shown in Table 2. Out of the 4951 patients with PBC, 524 experienced a fracture during the study period; 462 patients experienced a fracture at only one site, and 62 patients experienced fractures at 2 or more sites.Among the 19 793 matched controls, 1355 individuals experienced a fracture; 1246 patients had only one site fracture, while 109 patients had fractures at 2 or more sites.The PBC group had a higher risk of overall fragility fracture than the matched control, with a crude HR of 1.69 (95% CI, 1.53-1.87).The cumulative risk of fragility fracture for patients with PBC was 3.9%, 9.2%, 12.3%, and 16.9% at 2, 5, 7, and 10 years, respectively.By comparison, the matched control had fragility fracture rates of 2.1%, 5.4%, 7.6%, and 10.2% at the corresponding times, as illustrated in Figure 2. PBC was independently associated with the fragility fractures after adjusting for associated covariates, with an AHR of 1.63 (95% CI, 1.20-2.22;Table 3).In addition, older age, female sex, osteoporosis, rheumatoid arthritis, decompensated cirrhosis, and underweight (BMI < 18.5 kg/m 2 ) increased the risk of fragility fractures.

Subgroup analysis: 2-year landmark analysis
Of the 4951 patients with PBC, 4077 who were followed up for ≥2 years were included in the 2-year landmark analysis to minimize interference from incidental fractures that occurred independently of PBC.Table S4 presents the characteristics of these patients.The median follow-up duration was 6.32 years (IQR, 4.01-9.60years).The analysis revealed a significantly increased risk of overall fragility fracture in patients with PBC (AHR, 1.77; 95% CI, 1.29-2.42;Table S5).The risk of vertebral fracture was also higher in patients with PBC compared to that in matched controls (AHR, 1.82; 95% CI, 1.18-2.81;Table S6).Additionally, the risks of hip and proximal humerus fractures tended to be higher in patients with PBC than matched controls, although this difference did not reach statistical significance (Table S6).

Subgroup analysis: impact of PBC on fragility fractures according to sex and menopause
Of the 24 744 individuals, 4449 (18.0%) were men.The overall fragility fracture risk AHR for men with PBC was 2.53 (95% CI, 1.84-3.48).Women with PBC had a 1.59fold higher risk of overall fragility fracture (95% CI, 1.41-1.79)compared to matched female controls (Table S7).Nevertheless, the incidence rate, reflecting the disease burden from fragility fractures, was higher among women than men.Additionally, postmenopausal women showed a higher incidence of fragility fractures compared to premenopausal women (Table S8).

Discussion
This nationwide population-based cohort study revealed a 63% increased risk of fragility fractures associated with PBC.Patients with PBC had a 1.77-and 2.23-fold greater risk of vertebral and hip fractures, respectively, compared with matched controls.Furthermore, PBC had a greater influence on fragility fractures in men (AHR, 2.53) than in women (AHR, 1.59).
Previous studies have suggested an association between PBC and an increased risk of osteoporotic fracture. 9,10A study in the United Kingdom, which included 930 patients with PBC and 9202 matched controls, showed that patients with PBC had an approximate 2-fold increased risk of osteoporotic fractures compared to the controls. 9However, the study included patients with other chronic liver diseases, such as autoimmune hepatitis and primary sclerosing cholangitis; further, 48.7% of the patients with PBC did not receive a UDCA prescription, while 29.5% were prescribed corticosteroids.A subsequent Swedish study that assessed 3980 patients with PBC and 37 196 matched controls showed that PBC increased the risk of osteoporotic fractures by 1.6-fold over the 4.3-year follow-up. 10However, this finding might be inconclusive because more patients in the PBC population had other risk factors for osteoporotic fractures than those in the control group at enrollment, some of which were not adjusted for; these included malignancies, osteoporosis, and previous fractures. 19,23Considering the low incidence of PBC and fragility fractures, we included a larger dataset collected over a longer period and considered various other factors relevant to fragility fracture.Therefore, we believe our findings provide more definitive evidence on the risk of fractures in patients with PBC.Possible explanations for bone vulnerability in patients with PBC may originate from characteristic features of the disease itself.Cumulative injury to biliary epithelial cells leads to cholestasis, impairing the absorption of fatsoluble vitamins, essential for bone formation. 24Additionally, elevated bilirubin and bile acids inhibit osteoblastic activity.Increased levels of inflammatory cytokines further promote osteoclast activity, contributing to bone resorption. 5n our study, vertebral and hip fractures were the most common and vulnerable fracture sites in patients with PBC.The vertebra and hip are weight-bearing bones that cause serious mobility impairment when fractured.This, in turn, may lead to conditions, such as deep vein thrombosis, bedsores, or pneumonia. 25The 1-year mortality rate after weightbearing bone fractures is upto 24.2%. 26The mortality rate after osteoporotic fractures is higher in patients with PBC than in those without PBC. 3,10This may be explained by chronic hepatic inflammation, high disease burden, and other comorbidities observed in patients with PBC, which interfere with the wound healing process and increase the risk of complications. 27he impact of PBC on fragility fractures was 2.53 times higher in men compared to 1.59 times higher in women when compared to their matched controls, respectively.Additionally, having PBC increased the risk of fragility fractures by 1.89-fold in premenopausal women and 1.57fold in postmenopausal women.Nevertheless, the incidence rate, reflecting the disease burden from fragility fractures, was higher among women than men.Additionally, postmenopausal women showed a higher incidence of fragility fractures compared to premenopausal women (Table S8).Therefore, factors other than estrogen levels may have contributed to the increased vulnerability of men with PBC to osteoporosis and related fractures in our study.Because of its lower prevalence in men, osteoporosis is often overlooked and remains underdiagnosed and undertreated in men. 28However, approximately 30% of older men have osteoporosis, and bone loss is accelerated in chronic diseases such as PBC. 28Furthermore, men have a higher fracture-related mortality than women.The importance of the observed increased risk of fragility fractures in men with PBC should not be underestimated, and additional research is needed to elucidate its underlying mechanisms.
Efforts have been made to prevent osteoporosis in patients with PBC, including control of PBC itself.UDCA has survival benefits in patients with PBC, although its efficacy in preventing osteoporosis is yet to be established. 29UDCA increases osteoblast differentiation and activity at the cellular level. 30owever, the drug had no effect on osteoporotic fracture rates in previous population-based studies. 9,31Similarly, UDCA and other medications had no influence on fragility fracture incidence in our study.Further comprehensive investigation is required to confirm the role of UDCA in bone metabolism in patients with PBC.
We conducted a nationwide population-based cohort study spanning 14 years.However, the study had several limitations.First, due to the innate limitations of the claims data, we were unable to obtain information on liver function or osteoporosis indicators, such as liver biochemical test results or bone mineral density.Although we defined osteoporosis using ICD-10 codes, the actual prevalence of osteoporosis in patients with PBC is unknown.The lack of liver biochemical tests also made it challenging to assess the relationship between liver function and fragility fractures.However, by defining decompensated cirrhosis using procedure codes and prescriptions, we aimed to gain insights into the association between the severity of hepatic dysfunction and fragility fractures.Second, patients who experienced fragility fractures before PBC diagnosis were excluded from the study, which might have underestimated the risk of fragility fracture, as the first fragility fracture increases the risk of subsequent fractures.Furthermore, the attributable risk of fragility fractures directly associated with PBC could not be accurately determined in our study due to the limitations inherent in its design.Finally, asymptomatic patients with undiagnosed radiographic vertebral fractures, who did not seek medical attention, were not included in our analysis despite having fragility fractures.
In conclusion, this study showed that patients with PBC had a significantly higher risk of fragility fractures than matched controls.Specifically, patients with PBC were more vulnerable to vertebral and hip fractures, which may negatively impact quality of life and life expectancy.Considering the increased risk of fragility fractures in patients with PBC, awareness of fragility fracture risk and establishment of preventive measures is essential.

Figure 1 .
Figure 1.Study flowchart of patient and control selection.

Figure 2 .
Figure 2. Cumulative risk of fragility fractures in patients with primary biliary cholangitis and the matched control group.

Figure 3 .
Figure 3. Cumulative risks of site-specific fragility fractures in patients with primary biliary cholangitis and the matched control group.(A) Vertebral fracture.(B) Hip fracture.(C) Distal radius fracture.(D) Proximal humerus fracture.

Table 1 .
Baseline characteristics of patients with primary biliary cholangitis and matched controls.
Abbreviation: CCI, Charlson Comorbidity Index; IQR, interquartile range; PBC, primary biliary cholangitis; SD, standard deviation a Medication use was defined as prescription of the relevant drug for ≥30 consecutive days within the first 6 months of PBC diagnosis.bNon-glucocorticoidimmunosuppressive agents include azathioprine, mercaptopurine, and mycophenolate mofetil.cThedata for body mass index, smoking, alcohol consumption, and exercise at least once per week were available for participants who underwent health check-up.

Table 2 .
Observed cases and incidence rates of fragility fractures.

Table 3 .
Multivariable Cox regression analysis of overall fragility fracture.Model 2: adjusted for age, sex, socioeconomic status, comorbidities, decompensated cirrhosis, body mass index, smoking, alcohol, exercise, and health check-up.c Model 3: adjusted for age, sex, socioeconomic status, comorbidities, decompensated cirrhosis, body mass index, smoking, alcohol, exercise, health check-up, and medication use.d Medication use was defined as prescription of the relevant drug for ≥30 days within the first 6 months of PBC diagnosis.