Risk factors for fractures following liver transplantation: a population-based cohort study

Abstract

Background

Liver transplant recipients have an increased risk of osteoporosis and fractures. The aim of this study was to identify risk factors for fractures after liver transplant in a Taiwanese population.

Methods

We identified newly diagnosed liver transplant recipients from the National Health Insurance Research Database in Taiwan between 2003 and 2015. Risk factors of post-transplant fractures were analyzed using a Cox proportional hazards model.

Results

A total of 4821 patients underwent liver transplantation, of whom 419 (8.7%) had post-transplant fractures. Independent predictors of post-transplant fractures were age ≥65 years at transplantation (hazard ratio (HR): 1.566; 95% confidence interval (CI) 1.122–2.186), female sex (HR: 1.648; 95% CI 1.319–2.057), fractures within 1 year prior to transplant (HR: 3.664; 95% CI 2.503–5.364), hepatitis C carriers (HR: 1.594; 95% CI 1.289–1.970), alcoholism (HR: 1.557; 95% CI 1.087–2.230) and daily prednisolone dose >1.61–3.78 mg/day (HR: 1.354; 95% CI 1.005–1.824), >3.78–9.18 mg (HR: 4.182; 95% CI 3.155–5.544) and >9.18 mg (HR: 13.334; 95% CI 9.506–18.703). Post-transplant fractures were inversely correlated with tacrolimus (HR: 0.617; 95% CI 0.417–0.913) and sirolimus/everolimus (HR: 0.504; 95% CI 0.391–0.650) treatment.

Conclusions

The liver transplant recipients, and especially those who were aged ≥65 years, female, hepatitis C carriers, had a history of fractures within 1 year prior to transplant, alcoholism, and higher daily prednisolone dose were associated with an increased risk of post-transplant fractures. Conversely, the use of tacrolimus and sirolimus/everolimus was associated with a decreased risk of fractures.

Key Messages

1. This study identified risk factors for fractures after liver transplant in a population-based study in an area with high prevalence of hepatitis B and hepatitis C.

2. Recipients who were aged ≥65 years, female, hepatitis C carriers, had a history of fractures within 1 year prior to transplant, alcoholism, and higher daily prednisolone dose were independent risk factors for post-transplant fractures.

3. Our findings highlight the importance of identifying individuals at high risk of fractures and concomitant tacrolimus and sirolimus/everolimus treatment to avoid the use of high-dose steroids and prevent post-transplant fractures.

Keywords:

• Fracture
• glucocorticoid
• hepatitis B
• hepatitis C
• immunosuppressant
• liver transplantation

Introduction

With advances in surgical techniques and immunosuppressant agents, graft and patient survival after liver transplantation have improved dramatically in the past decades. One of the major challenges facing the field of liver transplantation today is the increasing incidence of bone diseases including osteoporosis and fractures which cause significant long-term morbidity [1,2]. Thus, understanding the general, pre- and post-transplant specific risk factors associated with post-transplant fractures may help with an early diagnosis and prompt interventions for optimal patient outcomes.

Patients with chronic liver diseases such as non-alcoholic chronic fatty liver disease and those with cholestatic disease are at an increased risk of osteoporosis [3,4]. In addition, decompensated cirrhosis with excess unconjugated bilirubin may interfere with the differentiation of primary osteoblasts [5]. Other known risk factors for osteoporosis include vitamin D and vitamin K deficiency, malnutrition, decreased body weight and hypomagnesemia, which are extremely common among patients evaluated for liver transplantation [6–8]. In general, more severe liver disease and cholestasis are associated with more severe bone loss. Therefore, many patients awaiting liver transplantation already have osteopenia or osteoporosis. Post-liver transplantation factors known to contribute to osteoporosis and fractures include high-dose glucocorticoids, other immunosuppressive agents, particularly cyclosporine and tacrolimus, secondary hyperparathyroidism and hypogonadism [9]. In addition, estrogen deficiency during menopause has also been shown to increase the risk of bone loss. Osteoporosis is common in patients with primary biliary cholangitis, partly because of the high prevalence in perimenopausal women. Post-liver transplant bone loss is most rapid in the first three to six months. The rate of bone loss then slows as the dose of glucocorticoids is reduced and pre-transplant conditions are resolved.

Hepatitis B virus (HBV) and hepatitis C virus (HCV) are the major causes of cirrhosis and hepatocellular carcinoma leading to the need for liver transplantation. Prior to the introduction of the national HBV vaccination program in Taiwan in 1984, the prevalence of chronic HBV infection in the general population was approximately 15%–20% [10]. The prevalence of HCV infection in Taiwan is estimated to be about 3.3% (1.8–5.5%), which is much higher than the worldwide average [11]. Apart from liver diseases, both HBV and HCV infections have been shown to increase the risk of osteoporosis in non-cirrhotic patients [12,13]. However, few studies have investigated the frequency of fractures among liver transplant recipients in a population-based study in an area with high HBV and HCV carrier rates, such as Taiwan. Furthermore, these studies had relatively small sample sizes, were not population-based, and were conducted in Western countries. There may be differences in the epidemiology and risk factors for osteoporosis and fractures between different ethnicities, races and prevalence of HBV and HCV infection. Therefore, we conducted this large population-based cohort study to identify the frequency of fractures in Taiwanese liver transplant recipients. In addition, we also investigated the general and transplant recipient-specific risk factors for fractures using data from the National Health Insurance Research Database (NHIRD) in Taiwan.

Methods

Data source

Data were obtained from the NHIRD of the Ministry of Health and Welfare of Taiwan. Since November 2015, researchers are asked to perform on-site data analysis of the NHIRD at a Health and Welfare Data Centre. The National Health Insurance (NHI) program in Taiwan is a compulsory single-payer program launched in March 1995 and currently covers approximately 99% of the population in Taiwan. The NHIRD contains complete information on medical service utilization, including demographic information, ambulatory care claims, inpatient claims, procedures, prescription details and diagnoses based on International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes. High accuracy of the diagnosis codes for common conditions and severe diseases in the NHIRD has been demonstrated [14,15]. Under the NHI program, patients with liver transplantation can apply for a catastrophic illness certificate. Catastrophic illness certificates are reviewed by relevant experts after thorough verification of medical records, as well as relevant documents of the liver transplant procedure. As such, the Catastrophic Illness Patient Database of liver transplantation is highly accurate. Furthermore, high accuracy of the NHIRD in recording liver transplantation has been demonstrated [16].

Patient identification

The transplant recipients included in this study were patients aged >18 years who underwent a liver transplantation (ICD-9-CM codes V42.7, 996.82) between 1 January 2003 and 31 September 2015, identified through catastrophic illness certification. For domestic transplantation recipients, the index date was defined as the admission date with Taiwan NHI reimbursement order codes 75020B and 75020 A (liver implantation). The NHIRD did not contain official records of the transplantation procedure for overseas transplant recipients. Therefore, the index date for overseas transplant recipients was defined as 1 month before the patient applied for a catastrophic illness certificate, because the length of overseas post-transplantation stay has been estimated to be approximately 1 month. All of the enrolled index cases were followed from the index date until a diagnosis of fractures, death, the end of follow-up in the medical records, or 31 December 2015. The demographic data of the patients including age at transplantation, sex and age at diagnosis of a fracture were extracted. The patients were classified into two age subgroups: 18–64 and ≥65 years. This study was approved by the Institutional Review Board of Taipei Veterans General Hospital (2019-08-004BC). Then, we requested access to the Taiwan’s NHIRD through a formal application and approved by the Health and Welfare Data Science Centre of the Ministry of Health and Welfare, Taiwan.

Outcomes and types of fracture

Post-liver transplantation fractures were identified using ICD-9-CM codes 805–829 during the follow-up period. The fracture sites were identified from medical service claims based on the ICD-9-CM diagnosis. Fractures were grouped into seven categories: (1) vertebra (ICD-9-CM 805.x-806.x); (2) femur (ICD-9 820.x, 821.x); (3) upper extremity (humerus, radius and ulna) (ICD-9-CM 812.x, 813.x, 818.x); (4) wrist and hand (ICD-9-CM 814.x, 815.x, 816.x, 817.x); (5) rib/clavicle/scapula/sternum (ICD-9-CM 807.x, 810.x, 811.x); (6) lower extremity (patella, tibia, fibula, ankle and foot (ICD-9-CM 822.x, 823.x, 824.x, 825.x, 826.x, 827.x); (7) pelvis (ICD-9-CM 808.x) and others (ICD-9-CM, 809.x, 819.x, 828.x, 829.x). Skull and facial fractures were excluded from the analysis.

Chronic liver or biliary disease prior to transplantation

Acute and subacute necrosis of the liver (ICD-9-CM 570), alcoholic liver disease (alcoholic cirrhosis [ICD-9-CM 571.0, 571.1, 571.2, and 571.3]), primary biliary cholangitis (ICD-9-CM 571.6), hepatitis B (ICD-9-CM 070.2, 070.3, and V02.61), hepatitis C (ICD-9-CM 070.41, 070.44, 070.51, 070.54, V02.62, and 070.7), Wilson’s disease (ICD-9-CM 275.1) and biliary atresia (ICD-9-CM 751.61) were identified. The diagnosis of chronic non-alcoholic fatty liver disease was based on ICD-9-CM codes 571.8 and 571.9 and no coexisting causes of alcohol-related liver disease, hepatic steatosis and chronic liver diseases mentioned above [17]. We also recorded the claims-based diagnosis of liver malignancy (ICD-9-CM 155.0, 155.1, 155.2).

Study variables

We assessed risk factors known to increase the risk of fractures in the general population, including female sex, osteoporosis (ICD-9-CM 733.0x) or prior history of fractures, rheumatoid arthritis (ICD-9-CM 714.0) and diabetes mellitus (ICD-9-CM 250). We also analyzed comorbidities that could be associated with fractures including hyperthyroidism (ICD-9-CM 242), hypertension (ICD-9-CM 401-405), systemic lupus erythematosus (ICD-9-CM 710.0), hyperlipidaemia (ICD-9-CM 272.0-272.4), coronary artery disease (ICD-9-CM 410-414), cerebrovascular disease (ICD-9-CM 430–438), chronic obstructive pulmonary disease (ICD-9-CM 491-493, 496), malnutrition (ICD-9-CM 262, 263), testicular hypogonadism in men (ICD-9-CM 257.2 and 257.9), menopausal and postmenopausal disorders in women (ICD-9-CM 627), and alcoholism (ICD-9-CM 303.x, V11.3, 305.0, 291.x, 357.5, 425.5, 535.3, 571.0, 571.2, 571.3). The diagnoses of comorbidities were considered valid based on two outpatient claims or one inpatient claim 1 year before or at the time of transplantation.

Glucocorticoids, immunosuppressive drugs and bone protective therapy

The dosage of systemic glucocorticoids including that prescribed orally at outpatient visits and that taken orally or intravenously during inpatient treatment were calculated until the date of fracture or the study end date. The glucocorticoids included betamethasone (Anatomical Therapeutic Chemical [ATC] code H02AB01), dexamethasone (ATC code H02AB02), methylprednisolone (ATC code H02AB04), paramethasone (ATC code H02AB05), prednisolone (ATC code H02AB06), triamcinolone (ATC code H02AB08), hydrocortisone (ATC code H02AB09), and cortisone (ATC code H02AB10). All glucocorticoid prescriptions were converted to prednisolone equivalents. The mean daily prednisolone dosage was classified into four groups by quartile: 0–1.61 mg/day, >1.61–3.78 mg/day, >3.78–9.18 mg/day, and >9.18 mg/day. Data on other immunosuppressive drugs used to prevent rejection including tacrolimus (ATC code L04AD02), cyclosporine (ATC code L04AD01), mycophenolate mofetil/mycophenolic acid (ATC code L04AA06) and sirolimus (ATC code L04AA10)/everolimus (ATC code L04AA18) that were prescribed within the follow-up period were extracted for analysis. Data on the drugs most often used in the treatment of osteoporosis including bisphosphonate (ATC codes M05BAxx and M05BBxx), calcium (ATC codes A12AAxx and V03AE07), vitamin D (ATC code A11CCxx) and calcium with vitamin D (ATC code A12AXxx) supplements were also recorded.

Statistical analysis

Categorical variables are presented as number and percentage, and continuous variables are presented as mean ± standard deviation. Cox proportional hazards regression analysis was performed to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs) of risk factors for fractures. First, potential risk factors were screened using a univariable Cox proportional hazards model. Candidate variables with p < 0.05 in the univariable Cox proportional hazards analysis were then included in multivariable Cox proportional hazards analysis to identify independent risk factors for fractures among the liver transplant recipients. All statistical analyses were performed using SPSS version 21.0 (IBM Inc., Armonk, NY). P values of < 0.05 were considered statistically significant.

Results

Demographics and clinical characteristics of the study population

A total of 4228 domestic and 593 overseas liver transplant recipients were identified in the NHIRD during the study period. The cohort included 3568 (73.8%) male and 1253 (26.0%) female patients. The mean age of the recipients was 52.7 ± 9.3 years. A total of 1194 patients died during the study period, and the mean time from transplantation to death was 1.3 ± 2.1 years. During the follow-up period, post-transplant clinical fractures were recorded in 419 (8.7%) patients during a mean follow-up period of 3.7 ± 3.1 years. The mean age at fracture was 56.4 ± 9.2 years, and the average time to fracture after transplantation was 3.0 ± 2.5 years. The highest rate of fractures was in the first 12 months after the transplant. Of the 419 patients with fractures, 101 (24.1%) had a first fracture in the first year, 80 (19.1%) at 1–2 years, 63 (15.0%) at 2–3 years, 49 (11.7%) at 3–4 years, and 45 (10.7%) at 4–5 years post transplant. In addition, 408 patients presented with a single fracture and 11 patients presented with two or three fractures. Among the total 431 fracture sites, fractures most commonly occurred in the upper extremities (humerus/radius/ulna) (n = 107, 24.8%), followed by the lower extremities (n = 91, 21.1%), vertebra (n = 73, 16.9%), clavicle/scapula/sternum/rib (n = 63, 14.6%), and femur (n = 58, 13.5%) (Table 1).

Table 1. Classification of all first fracture sites post-transplant.

Fracture site	n (%)
Vertebra	73 (16.9 %)
Femur	58 (13.5%)
Upper extremity (humerus, radius and ulna)	107 (24.8%)
Wrist and hand	32 (7.4%)
Clavicle/scapula/sternum/rib	63 (14.6%)
Lower extremity (tibia, fibula, patella, ankle, foot)	91 (21.1%)
Pelvis and others	7 (1.6%)

The demographic characteristics and clinical information of the liver transplant recipients with or without fractures are summarized in Table 2. The recipients who developed post‐transplant fractures tended to be female (38.4% vs. 24.8%), older than 65 years (9.8% vs. 7.2%), and have a fracture 1 year prior to transplantation (7.9% vs. 1.9%). The rates of HBV and HCV infection were 64.6% and 32.9%, respectively. Overall, 2870 (59.5%) and 1671 (34.7%) recipients were diagnosed with preoperative liver malignancy and alcohol-related cirrhosis, respectively. The absolute case numbers of chronic non-alcoholic fatty liver disease among the recipients with fractures were not available, as data on any finding with fewer than three patients cannot be extracted to protect their identification. The rates of alcoholic cirrhosis (37.2% vs. 34.4%), primary biliary cholangitis (11.5% vs. 7.2%), and chronic HCV infection (40.8% vs. 32.2%) before transplantation were higher in the recipients with fractures. However, the recipients with fractures had a lower rate of HBV infection (57.5% vs. 65.3%). There were high prevalence rates of diabetes mellitus (n = 1241, 25.7%), alcoholism (n = 1102, 22.9%) and hypertension (n = 889, 18.4%). Moreover, the recipients with fractures had higher rates of osteoporosis (2.6% vs. 0.5%), hypertension (20.8% vs. 18.2%), diabetes mellitus (28.6% vs. 25.5%), systemic lupus erythematosus (1.0% vs. 0.3%), hyperlipidaemia (3.8% vs. 3.2%), coronary artery disease (4.8% vs. 3.1%), cerebrovascular disease (4.5% vs. 2.0%), chronic obstructive pulmonary disease (5.3% vs. 2.7%), hypogonadism and postmenopausal disorders (0.7% vs. 0.4%), and alcoholism (26.7% vs. 22.5%) than the recipients without fractures.

Table 2. Comparison of demographics and comorbidities between the liver recipients with and without a post‐transplant fracture.

	Total	Fracture	Without fracture
	n = 4821	n = 419	n = 4402
Female recipient	1253 (26.0%)	161 (38.4%)	1092 (24.8%)
Age at transplant	52.7 ± 9.3	53.4 ± 9.2	52.6 ± 9.3
Age at transplant ≥ 65 years	356 (7.4%)	41 (9.8%)	315 (7.2%)
Overseas transplantation	593 (12.3%)	51 (12.2%)	542 (12.3%)
Fracture in 1 year prior to transplant	118 (2.4%)	33 (7.9%)	85 (1.9%)
Chronic liver or biliary disease prior to transplantation
Acute and subacute necrosis of liver	555 (11.5%)	38 (9.1%)	517 (11.7%)
Alcohol-related cirrhosis	1671 (34.7%)	156 (37.2%)	1515 (34.4%)
Primary biliary cholangitis	365 (7.6%)	48 (11.5%)	317 (7.2%)
Chronic non-alcoholic fatty liver disease	29 (0.6%)	NA	NA
Hepatitis B	3115 (64.6%)	241(57.5%)	2874 (65.3%)
Hepatitis C	1588 (32.9%)	171 (40.8%)	1417 (32.2%)
Wilson’s disease	37 (0.8%)	3 (0.7%)	34 (0.8%)
Biliary atresia	25 (0.5%)	0 (0.0%)	25 (0.6%)
Malignancy	2870 (59.5%)	243 (58.0%)	2627 (59.7%)
Comorbidity
Osteoporosis	34 (0.7%)	11 (2.6%)	23 (0.5%)
Hyperthyroidism	17 (0.4%)	0 (0.0%)	17 (0.4%)
Hypertension	889 (18.4%)	87 (20.8%)	802 (18.2%)
Diabetes mellitus	1241 (25.7%)	120 (28.6%)	1121 (25.5%)
Systemic lupus erythematous	19 (0.4%)	4 (1.0%)	15 (0.3%)
Hyperlipidemia	158 (3.3%)	16 (3.8%)	142 (3.2%)
Coronary artery disease	156 (3.2%)	20 (4.8%)	136 (3.1%)
Cerebrovascular disease	108 (2.2%)	19 (4.5%)	89 (2.0%)
Chronic obstructive pulmonary disease	139 (2.9%)	22 (5.3%)	117 (2.7%)
Malnutrition	8 (0.2%)	0 (0.0%)	8 (0.2%)
Rheumatoid arthritis	22 (0.5%)	NA	NA
Hypogonadism and postmenopausal disorder	20 (0.4%)	3 (0.7%)	17 (0.4%)
Alcoholism	1102 (22.9%)	112 (26.7%)	990 (22.5%)
Mean dosage of glucocorticoids
0–1.61 mg/day	1205 (25.0%)	90 (21.5%)	1115 (25.3%)
>1.61-3.78 mg/day	1205 (25.0%)	90 (21.5%)	1115 (25.3%)
>3.78-9.18 mg/day	1206 (25.0%)	144 (34.4%)	1062 (24.1%)
>9.18 mg/day	1205 (25.0%)	95 (22.7%)	1110 (25.2%)
Bone-protective medications
Bisphosphonate therapy	102 (2.1%)	13 (3.1%)	89 (2.0%)
Calcium	336 (7.0%)	35 (8.4%)	301 (6.8%)
Vitamin D	70 (1.5%)	5 (1.2%)	65 (1.5%)
Calcium and vitamin D	36 (0.7%)	5 (1.2%)	31 (0.7%)
Immunosuppressive medications
Tacrolimus	4599 (95.4%)	391 (93.3%)	4208 (95.6%)
Cyclosporine	384 (8.0%)	52 (12.4%)	332 (7.5%)
Mycophenolate mofetil/mycophenolic acid	4064 (84.3%)	372 (88.8%)	3692 (83.9%)
Sirolimus/everolimus	1389 (28.8%)	74 (17.7%)	1315 (29.9%)

NA: not available.

Glucocorticoids, bone-protective medications and immunosuppressants

In the patients with fractures, 21.5%, 21.5%, 34.4% and 22.7% were in the 1st (0-1.61 mg/day), 2nd (>1.61-3.78 mg/day), 3rd (>3.78-9.18 mg/day) and 4th (>9.18 mg/day) quartiles of average daily glucocorticoid dose, respectively, compared to 25.3%, 25.3%, 24.1% and 25.2% in the patients without fractures. Overall, 2.1% of the patients received bisphosphonate treatment after transplantation. Calcium and vitamin D supplements were taken by 7.0% and 1.5% of the patients post-transplant, respectively. The most commonly prescribed immunosuppressive drugs were tacrolimus and mycophenolate mofetil/mycophenolic acid. A higher proportion of the recipients with fractures received cyclosporine (12.4% vs. 7.5%) and mycophenolate mofetil/mycophenolic acid (88.8% vs. 83.9%) compared to those without fractures. Conversely, the prescription rates of tacrolimus (93.3% vs. 95.6%) and sirolimus/everolimus (17.7% vs. 29.9%) were lower in the recipients with factures than in those without fractures.

Univariable and multivariable analyses of risk factors for the development of fractures

Table 3 shows the results of the univariable and multivariable analyses for associations between post-transplant fractures and demographic and clinical data. The univariable analysis revealed that transplantation at ≥65 years of age, female sex, a history of fractures within 1 year prior to transplantation, alcohol-related cirrhosis, primary biliary cholangitis, HCV infection, osteoporosis, hypertension, diabetes mellitus, systemic lupus erythematous, cerebrovascular disease, alcoholism and mean daily glucocorticoid dose >1.61–3.78 mg, >3.78–9.18 mg and >9.18 mg prednisolone equivalent were associated with a higher risk of developing a fracture. However, the risk of fractures was significantly lower in the patients with HBV infection. Among the prescribed immunosuppressives, the use of tacrolimus and sirolimus/everolimus was negatively correlated with post-transplant fractures. Overseas transplant, acute and subacute necrosis of the liver, chronic non-alcoholic fatty liver disease, Wilson’s disease, malignancy, biliary atresia, hyperthyroidism, hyperlipidaemia, coronary artery disease, chronic obstructive pulmonary disease, malnutrition, rheumatoid arthritis, hypogonadism and postmenopausal disorder and the use of bisphosphonate, calcium, vitamin D supplements, cyclosporine and mycophenolate mofetil/mycophenolic acid were not correlated with post-transplant fractures.

Table 3. Analysis of risk factors for post-transplant fractures in the liver transplant recipients.

	Univariable		Multivariable
	HR (95% CI)	p value	HR (95% CI)	p value
Age at transplant ≥ 65 years	1.613 (1.168–2.227)	*0.004	1.566 (1.122–2.186)	*0.008
Female recipient	1.973 (1.620–2.403)	*<0.001	1.648 (1.319–2.057)	*<0.001
Overseas transplantation	0.773 (0.577–1.037)	0.086
Fracture in 1 year prior to transplant	5.284 (3.7000–7.547)	*<0.001	3.664 (2.503–5.364)	*<0.001
Chronic liver or biliary disease prior to transplantation
Acute and subacute necrosis of liver	0.750 (0.537–1.047)	0.091
Alcohol-related cirrhosis	1.259 (1.033–1.536)	*0.023	0.954 (0.691–1.316)	0.773
Primary biliary cholangitis	1.381 (1.022–1.866)	*0.035	1.200 (0.874–1.648)	0.259
Chronic non-alcoholic fatty liver disease	0.989 (0.246–3.968)	0.988
Hepatitis B	0.633 (0.521–0.769)	*<0.001	1.076 (0.867–1.335)	0.505
Hepatitis C	1.634 (1.344–1.986)	*<0.001	1.594 (1.289–1.970)	* <0.001
Wilson’s disease	0.744 (0.239–2.316)	0.609
Biliary atresia	0.049 (0.000–17.340)	0.314
Malignancy	0.969 (0.798–1.177)	0.751
Comorbidity
Osteoporosis	3.049 (1.675–5.551)	*<0.001	1.722 (0.909–3.261)	0.095
Hyperthyroidism	0.049 (0.000–62.887)	0.410
Hypertension	1.371 (1.082–1.736)	*0.009	1.280 (0.998–1.642)	0.052
Diabetes mellitus	1.349 (1.091–1.667)	*0.006	1.140 (0.911–1.426)	0.252
Systemic lupus erythematous	2.900 (1.083–7.765)	*0.034	1.569 (0.576–4.276)	0.379
Hyperlipidemia	0.905 (0.549–1.492)	0.696
Coronary artery disease	1.538 (0.982–2.411)	0.060
Cerebrovascular disease	2.227 (1.405–3.529)	*0.001	1.595 (0.989–2.571)	0.055
Chronic obstructive pulmonary disease	1.489 (0.969–2.289)	0.069
Malnutrition	0.050 (0.000–13389.353)	0.638
Rheumatoid arthritis	1.113 (0.277–4.466)	0.880
Hypogonadism and postmenopausal disorder	1.180 (0.379–3.676)	0.775
Alcoholism	1.387 (1.116–1.723)	*0.003	1.557 (1.087–2.230)	**0.016
Mean dosage of glucocorticoids
0–1.61 mg/day	1	1
>1.61–3.78 mg/day	1.476 (1.098–1.982)	*0.010	1.354 (1.005–1.824)	*0.047
>3.78–9.18 mg/day	4.506 (3.419–5.939)	*<0.001	4.182 (3.155–5.544)	*<0.001
>9.18 mg/day	13.938 (10.064–19.303)	*<0.001	13.334 (9.506–18.703)	*<0.001
Bone-protective medications
Bisphosphonate therapy	1.729 (0.995–3.005)	0.052
Calcium	0.852 (0.602–1.206)	0.366
Vitamin D	0.605 (0.250–1.461)	0.264
Calcium and vitamin D	1.094 (0.452–2.646)	0.842
Immunosuppressive medications
Tacrolimus	0.618 (0.420–0.908)	*0.014	0.617 (0.417–0.913)	*0.016
Cyclosporine	1.322 (0.988–1.770)	0.060
Mycophenolate mofetil/mycophenolic acid	0.987 (0.728–1.337)	0.931
Sirolimus/everolimus	0.592 (0.460–0.762)	*<0.001	0.504 (0.391–0.650)	*<0.001

CI: confidence interval; HR: hazard ratio. *p < 0.05.

Multivariable analysis demonstrated that the independent predictors of post-transplant fractures were an age at transplant ≥65 years (HR: 1.566; 95% CI 1.122-2.186, p = 0.008), female sex (HR: 1.648; 95% CI 1.319–2.057, p < 0.001), fractures in the 1 year prior to transplant (HR: 3.664; 95% CI 2.503–5.364, p < 0.001), HCV carriers (HR: 1.594; 95% CI 1.289–1.970, p < 0.001), alcoholism (HR: 1.557; 95% CI 1.087–2.230, p = 0.016), and daily prednisolone dose >1.61–3.78 mg/day (HR: 1.354; 95% CI 1.005–1.824, p = 0.047), >3.78-9.18 mg (HR: 4.182; 95% CI 3.155–5.544, p < 0.001) and >9.18 mg (HR: 13.334; 95% CI 9.506–18.703, p < 0.001). Post-transplant fractures remained inversely correlated with the use of tacrolimus (HR: 0.617; 95% CI 0.417–0.913, p = 0.016) and sirolimus/everolimus (HR: 0.504; 95% CI 0.391–0.650, p < 0.001).

Discussion

Various studies have shown that a high percentage of patients with end-stage liver disease referred for liver transplantation have osteoporosis and osteoporotic fractures [18], and that this further increases the risk of fractures after transplantation. Consistent with previous studies, the well-known risk factors for fractures such as older age, female sex and previous fractures were also identified in the current study.

Chronic alcohol and nicotine exposure are important risk factors for osteoporosis, and both are common in liver transplant recipients. In addition, malnutrition and magnesium depletion are common in patients with alcoholism and end-stage liver disease. In the present study, compared to the recipients without fractures, those with fracture tended to have more comorbidities. Similar to previous studies, chronic alcoholism predisposed the recipients to a greater risk of fractures. In addition, we found that the patients with osteoporosis, hypertension, diabetes mellitus, systemic lupus erythematous, and cerebrovascular disease had an increased risk of fractures in univariable analysis, but not in multivariable analysis.

Hypogonadism due to central suppression of the hypothalamic-pituitary-gonadal axis is common in patients with end-stage liver disease [19]. After liver transplantation, persistent hypogonadism or partial reversibility of the hypothalamic-pituitary-gonadal system dysfunction has been demonstrated. In the present study, we found that hypogonadism and postmenopausal disorders were not associated with an increased risk of fractures. Oestrogen plays an important role in the growth and maturation of bone and also in the regulation of bone turnover via modulation of the production of various cytokines [20]. Exogenous oestrogen supplementation has been shown to prevent oestrogen deficiency-related osteoporosis of various causes [21]. In addition, oestrogen replacement therapy has also shown benefits in the prevention of osteoporosis and fractures following liver transplantation in postmenopausal women [22].

Vitamin D is important for calcium homeostasis and bone mineralization. Chronic obstructive liver disease may interfere with vitamin D metabolism, leading to decreased uptake of vitamin D. In addition, cirrhosis has been shown to impair 25-hydroxylation of vitamin D, consequently resulting in significantly depressed levels of 25-hydroxyvitamin D3 [23]. Previous studies have reported that prophylactic bisphosphonate treatment is effective in preventing bone loss and fractures in liver transplant recipients [24,25]. In the present study, we found that the use of bone protective medications including bisphosphonates, vitamin D supplements and calcium supplements did not have a beneficial effect on decreasing the risk of fractures. However, we only analyzed prescribed calcium and vitamin D supplements, and we did not consider those that could have been bought over the counter or obtained via dietary intake.

The pathogenesis of bone loss differs in different chronic liver diseases [26]. Hemochromatosis [27] Wilson’s disease [28] and cholestatic liver diseases are driven mainly by decreased bone formation, and viral hepatitis, transplantation, and the use of glucocorticoids are associated with increased bone resorption. In addition, vitamin K deficiency is frequently observed in patients with cholestasis. Vitamin K can suppress bone resorption by impairing osteoclast maturation and function. The mechanisms involved in osteoporosis in non-alcoholic fatty liver disease include cytokine production in an inflamed liver, vitamin D deficiency, metabolic syndrome and limited physical activity [29]. In patients with non-alcoholic fatty liver disease, significantly higher levels of tumour necrosis factor-alpha, lower levels of osteocalcin and higher levels of osteopontin have been reported. However, controversy exists regarding whether the type of underlying liver disease can predict bone loss and fractures. Some authors have reported higher rates of osteoporosis and fractures in patients with primary biliary cholangitis [30] and non-alcoholic fatty liver disease, especially in older Chinese men [31]. In contrast, Wester et al. reported a similar risk of fractures between patients with non-alcoholic fatty liver disease and the general population [32]. In the present analysis, neither chronic non-alcoholic fatty liver disease nor primary biliary cholangitis were independent factors for the development of fractures. Viral hepatitis causes an inflammatory response with increased production of bone resorption-activating cytokines such as tumour necrosis factor-alpha [33]. Previous studies have confirmed associations between osteopenia and fractures with HBV and HCV infection [13,34–36]. Both HBV and HCV infections were highly prevalent in our study cohort. We found that HBV infection did not increase the risk of fractures, whereas HCV infection was an independent risk factor for fractures.

Glucocorticoids are used to prevent or reverse severe rejection following liver transplantation. Low bone mineral density is a well-known risk factor for developing fractures in glucocorticoid-treated patients. The pathophysiology of glucocorticoid-induced osteoporosis includes directly decreasing bone formation by inducing osteoblast apoptosis and increasing resorption by promoting osteoclastogenesis [37,38]. Glucocorticoids also act indirectly by reducing intestinal calcium absorption and renal calcium wasting, thereby leading to secondary hyperparathyroidism. In addition, glucocorticoids decrease the secretion of androgens and oestrogens, primarily by inhibiting gonadotropin secretion. In the present study, we found that the liver transplant recipients with a daily prednisolone dose >1.61–3.78 mg/day, >3.78–9.18 mg and >9.18 mg had 1.354, 4.182- and 13.334-fold increased risks of fractures, respectively. In other words, the risk of fractures increased as the daily dose of steroids increased. Buckley and Humphrey recommended bisphosphonates as the first-line treatment for glucocorticoid-induced osteoporosis [38]. However, other studies have shown that discontinuing denosumab produced a rebound in bone turnover with an increased risk of new vertebral fractures [39]. Among the glucocorticoid-sparing immunosuppressants used to prevent rejection, calcineurin inhibitors including cyclosporine and tacrolimus have been reported to cause osteoporosis [40–42]. Cyclosporine stimulates both osteoclasts and osteoblasts, leading to dose-dependent high-turnover osteoporosis. Tacrolimus has been demonstrated to induce bone loss through an increase in bone resorption only. Monegal et al. reported that tacrolimus offered a significantly better long-term effect on femoral neck bone mineral density in liver transplantation recipients compared with cyclosporine [42]. The effects of the other immunosuppression regimens such as mycophenolate mofetil and sirolimus/everolimus on bone are still controversial [43]. In the present study, the use of tacrolimus and sirolimus/everolimus was beneficial with regards to fractures, which may be due to the lower dose of glucocorticoids. Cyclosporine and mycofenolate mofetil did not significantly influence the occurrence of fractures. Therefore, recipients with risk factors for fractures should pay attention to bone health status and consider the concomitant use of immunosuppressants to minimise the use of glucocorticoids and thereby decrease the risk of fractures.

The main strengths of this study include the large sample size, which may have avoided selection and participation bias, and the inclusion of multiple potential clinical covariates in the analysis. However, this study also has several limitations. First, information on the patients’ lifestyle such as smoking, the amount of alcohol consumption and caffeine use, dietary habits, body mass index, physical activity level, and family history of fractures is not available in the NHIRD. We validated the presence of alcoholism by using ICD-9-CM diagnosis codes of alcohol-related diseases and alcohol dependency. Second, bone mineral density and biochemistry data such as calcium, phosphate, 25-hydroxyvitamin D and magnesium were not evaluated. Third, data on the patients’ use of self-paid medications such as calcium and vitamin D supplements were also not recorded in the NHIRD. It is possible that the clinicians may have only offered osteoporosis prevention and/or therapy to the patients who were likely to have a higher fracture risk, which may have resulted in bias related to prescribing decisions. Furthermore, the diagnosis of fractures was based on ICD-9-CM codes rather than being validated by X-ray or other imaging modalities. However, to validity of the diagnosis of a new fracture, new fractures were defined as one inpatient or two outpatient clinical diagnoses of a fracture after the index date.

Conclusions

In conclusion, we found that female liver transplant recipients aged ≥65 years, patients with prior fractures, chronic HCV infection, alcoholism and a higher daily prednisolone dose were associated with a high risk of fractures. Conversely, tacrolimus and sirolimus/everolimus was associated with a decreased risk of fractures. Clinicians should carefully assess the risk of fractures, monitor bone mineral density and prescribe concomitant immunosuppressant therapy with tacrolimus or sirolimus/everolimus to avoid the use of high-dose glucocorticoids, as these factors remain the most effective way of preventing post-transplant fractures. Further studies are warranted to investigate the role of calcium and vitamin D supplementation and postmenopausal hormone replacement therapy in the prevention of osteoporosis and fractures following liver transplantation.

Authors’ contributions

JW Chang, HL Tsai, HH Yang, NC Lin, FC Kuo and TC Lin participated in the conception and design of the experiments; JW Chang and HL Tsai were involved in the data analysis and interpretation; JW Chang drafted the manuscript. HH Yang, NC Lin, FC Kuo, TC Lin and HL Tsai critically revised the intellectual content. All authors read and approved the final manuscript and agree to be accountable for all aspects of the work.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The Taiwan’s NHIRD is an anonymized database. Researchers can apply access to Taiwan’s NHIRD database after ethical approval from the institutional review board. The applicant must be Taiwanese or be affiliated with a Taiwanese research institute. After receiving permission of gaining access to Taiwan’s NHIRD database, the Health and Welfare Data Centre asks qualified investigators to conduct on-site data analysis by using provided computers and software including SAS, Stata, R, and SPSS at one branch of Health and Welfare Data Centre through remote connection to Health and Welfare Data Centre servers. No individual-level data may be taken out from centres. Therefore, data sharing is not applicable to this study due to privacy and ethical concerns.

Additional information

Funding

This study was supported by a grant from Taipei Veterans General Hospital (V109C-164).