Examination of bone mineral density in patients with varied liver diseases - a cross-sectional study

Abstract

BACKGROUND

Chronic liver disease is associated with profound alterations in bone metabolism, predisposing patients to secondary osteoporosis and fragility fractures - a condition commonly termed hepatic osteodystrophy. The pathogenesis of bone loss in chronic liver disease is multifactorial, involving increased osteoclastic activity, impaired osteoblast function, chronic inflammation, malnutrition, and deficiencies of vitamin D and K. Among these, cholestatic liver diseases such as primary biliary cholangitis (PBC) and primary sclerosing cholangitis (PSC) are particularly prone to severe bone loss, whereas viral hepatitis and autoimmune hepatitis (AIH) may contribute through inflammatory and steroid-mediated mechanisms. Despite this, comparative data on bone mineral density (BMD) across different etiologies of end-stage liver disease remain limited, particularly in patients awaiting liver transplantation.

AIM

To compare BMD and bone turnover markers among patients with different etiologies of end-stage liver disease listed for liver transplantation.

METHODS

This retrospective, cross-sectional study included 82 patients with end-stage liver disease: AIH, hepatitis C virus infection, PBC, PSC, and polycystic liver disease. Areal BMD at the lumbar spine and femoral neck was measured using dual-energy X-ray absorptiometry. Standard hepatic parameters (bilirubin, international normalized ratio, creatinine, albumin), and bone turnover and micronutrient markers (25-hydroxyvitamin D, parathyroid hormone, osteocalcin, phosphate, calcium, magnesium, ferritin, zinc, and copper.) were analyzed from clinical laboratory data. The Model of End-stage Liver Disease-score and Child-Pugh scores were analyzed. Data on current or prior use of bisphosphonates and glucocorticoids were extracted from patient records.

RESULTS

All liver disease groups showed significantly reduced lumbar spine BMD compared to controls (P < 0.01), with the lowest lumbar spine T-scores in PBC (-2.5) and PSC (-2.1). Femoral BMD was less affected but still reduced in several groups. Elevated osteocalcin and parathormone levels were most pronounced in patients with hepatic cysts. AIH patients exhibited less severe bone loss, despite glucocorticoid exposure in some cases.

CONCLUSION

Reduced BMD is prevalent across hepatic disease entities in liver transplant candidates, particularly in cholestatic and cystic liver disease. Early screening and targeted bone-protective strategies are essential to mitigate fracture risk in this population.

Key Words: Bone mineral density; Chronic liver disease; Osteoporosis; Liver transplantation, Dual-energy X-ray absorptiometry

Core Tip: Patients with end-stage liver disease are at high risk of reduced bone mineral density, particularly in cholestatic liver diseases such as primary biliary cholangitis and primary sclerosing cholangitis. This study demonstrates that bone loss varies according to the underlying liver disease etiology. Given the high prevalence of osteoporosis in this population, early assessment using dual-energy X-ray absorptiometry should be considered in clinical routine to identify patients at risk and guide preventive strategies.

INTRODUCTION

The liver plays a central role in numerous metabolic processes, including protein synthesis, vitamin and mineral homeostasis, and the regulation of inflammatory pathways. In the context of chronic liver disease (CLD), disruption of these physiological functions may result in systemic complications, including secondary osteoporosis[1].

Osteoporosis is a well-recognized complication of CLD and is associated with an increased risk of fragility fractures, leading to substantial morbidity and mortality[1,2]. Recent studies indicate that osteoporosis affects approximately 30%-50% of patients with advanced liver disease, while fracture prevalence may reach up to 40%-50%, particularly in patients awaiting liver transplantation. These findings underscore the clinical relevance of hepatic osteodystrophy as an increasingly recognized systemic complication of CLD[3].

The pathogenesis of hepatic osteodystrophy is multifactorial and involves both increased bone resorption and impaired bone formation. Chronic systemic inflammation promotes osteoclast activation through proinflammatory cytokines, while impaired hepatic synthesis of growth factors and hormones contributes to reduced osteoblast activity[4]. In addition, deficiencies in vitamin D and vitamin K, malnutrition, hypogonadism, and reduced physical activity further exacerbate bone loss in these patients[1,5].

Certain etiologies of CLD appear to have a particularly strong impact on bone metabolism. Cholestatic liver diseases, such as primary biliary cholangitis (PBC) and primary sclerosing cholangitis (PSC), are consistently associated with reduced bone mineral density (BMD) and an increased risk of osteoporosis[1]. In contrast, viral liver diseases such as hepatitis C virus (HCV) infection are associated with systemic inflammation and cytokine-mediated bone loss, even in earlier disease stages[6].

Autoimmune hepatitis (AIH) has been less extensively studied in this context; however, long-term glucocorticoid therapy, which is commonly used in its management, represents an additional major risk factor for secondary osteoporosis[1,7].

Despite the recognized association between CLD and impaired bone health, comparative data on BMD across different etiologies of end-stage liver disease remain limited[8]. In particular, it is unclear to what extent the underlying liver disease contributes to bone loss independent of other risk factors such as age, sex, or glucocorticoid exposure.

Therefore, the aim of the present study was to evaluate BMD in patients with different etiologies of end-stage liver disease listed for liver transplantation and to compare these findings with a healthy control population. Secondary objectives included the assessment of disease-specific differences and the potential impact of clinical and biochemical parameters on bone health.

MATERIALS AND METHODS

Study design and patient selection

This retrospective, cross-sectional study was conducted at the University Hospital Jena. Clinical data were collected from 82 patients who underwent evaluation for liver transplantation between 2004 and 2016. Patients were grouped according to the etiology of their liver disease into five categories: AIH, HCV, PBC, PSC, and polycystic liver disease. A control cohort of 101 healthy individuals, matched for age and sex distribution where feasible, was recruited from the Platelet Collection Center at the Medical University of Graz.

The study was conducted in accordance with the Declaration of Helsinki and was approved by the ethics committee of University Hospital Jena on 18^th September 2025, approval No. 2025-3947-BO-D.

Inclusion and exclusion criteria

Inclusion criteria comprised adults of any sex and age with end-stage liver disease listed for transplantation. Exclusion criteria included alcohol-induced cirrhosis, hepatocellular carcinoma, overlap syndromes, and co-existing hepatic or systemic diseases known to affect bone metabolism. Specifically excluded were patients with diabetes mellitus, systemic inflammatory diseases (e.g., rheumatoid arthritis, systemic lupus erythematosus, inflammatory bowel disease), chronic kidney disease, human immunodeficiency virus infection, and hematologic malignancies such as lymphoma or multiple myeloma.

A total of 176 patients were initially identified and screened for eligibility. After application of the predefined inclusion and exclusion criteria, 94 patients were excluded from the analysis.

The most frequent reasons for exclusion were alcohol-related liver cirrhosis (n = 34) and hepatocellular carcinoma (n = 31), representing the largest proportion of excluded cases. End-stage kidney disease was present in 11 patients and led to exclusion due to its known impact on bone metabolism. In addition, 9 patients were excluded because of other active or prior malignant diseases. A further 9 patients were excluded due to overlapping liver disease etiologies or incomplete clinical data.

After applying all criteria, a total of 82 patients with end-stage liver disease of non-alcoholic etiology were included in the final analysis.

BMD assessment

Areal BMD (aBMD) was measured by dual-energy X-ray absorptiometry (DXA) using a Prodigy Advance system (GE Healthcare). Standardized scans were performed at the lumbar spine (L1-L4) and bilateral femoral necks. Results were reported as absolute BMD (g/cm²) and T-scores based on reference values for young adults. According to World Health Organization criteria, osteopenia was defined as a T-score between -1.0 and -2.5, and osteoporosis as T-score < -2.5.

Biochemical analyses

Laboratory analyses included standard hepatic parameters (bilirubin, international normalized ratio, creatinine, albumin), which were used to calculate the Model of End-Stage Liver Disease (MELD) and Child-Pugh scores. Additional laboratory markers relevant to bone metabolism included serum levels of 25-hydroxyvitamin D, parathyroid hormone (PTH), osteocalcin, phosphate, calcium, magnesium, ferritin, zinc, and copper. Data on current or prior use of bisphosphonates and glucocorticoids were extracted from patient records.

Follow-up evaluations

Patients were followed longitudinally with clinical and laboratory reassessments every 6 months until transplantation or death. These follow-up visits included reevaluation of liver function, documentation of any prevalent fractures and medication adjustments. BMD is routinely assessed during the initial evaluation prior to liver transplantation and is subsequently re-evaluated every two years as part of the standardized follow-up protocol.

Control group

The control group consisted of 101 healthy individuals recruited from a platelet donation cohort at the Medical University of Graz. All control subjects fulfilled standard eligibility criteria, which include the absence of acute or chronic diseases, normal physical examination findings, and no history of relevant systemic disorders.

In addition, individuals with known conditions affecting bone metabolism were excluded from the control group. These included CLD, renal insufficiency, endocrine disorders, inflammatory diseases, malignancies, and the use of medications known to influence bone metabolism, such as glucocorticoids or antiresorptive therapies.

All control subjects underwent routine laboratory testing as part of the donation screening process and had no evidence of abnormal liver function or metabolic disturbances. Furthermore, only individuals without a history of osteoporosis or fragility fractures were included. This approach ensured that the control group represented a metabolically healthy population suitable for comparison with patients with end-stage liver disease.

As part of the donation protocol, bone-relevant parameters were measured, and DXA scans were performed under the same protocol as for the patient cohort.

Statistical analysis

Data analysis was conducted using SPSS software (version 24.0, IBM Corp., Armonk, NY, United States). Distribution of variables was assessed using the Kolmogorov-Smirnov and Shapiro-Wilk tests. Comparisons between groups were made using the Mann-Whitney U test or Fisher’s exact test as appropriate. Correlations were evaluated using Pearson’s coefficient. A P value < 0.05 was considered statistically significant. For clarity of presentation, group values are reported as medians with interquartile ranges (IQR), and graphs were plotted using mean values without standard deviation bars.

Match-pair-analysis

To investigate differences in bone-related parameters between liver transplant recipients with and without osteoporotic fractures, we performed a matched-pair analysis using a 1:1 nearest-neighbor matching algorithm. Patients with documented osteoporotic fractures either before or after liver transplantation (n = 9) were matched to patients without fractures (n = 9) based on the following variables: Age, sex, body mass index, MELD score, serum 25-hydroxyvitamin D, PTH, and osteocalcin. Matching was conducted using standardized Euclidean distance after z-transformation of continuous variables.

Following matching, differences between the two groups were assessed using paired statistical tests (Wilcoxon signed-rank test or paired t-test where appropriate). BMD values at the lumbar spine and femoral neck, T-scores, and biochemical markers of bone metabolism (including PTH, osteocalcin, and vitamin D) were compared between groups. Only matched pairs with available data for the respective variables were included in each test.

The analysis was performed in Python (pandas, sklearn, scipy), and missing values were excluded listwise.

Multivariable regression

To account for potential confounding due to the unequal sex distribution and to identify independent predictors of BMD, a multivariable linear regression analysis was performed. The dependent variables were aBMD of the lumbar spine (L1-L4, g/cm²), the corresponding lumbar T-score, and aBMD at the femoral neck. Independent variables included sex (male/female), age (years), body mass index (kg/m²), MELD score, Child-Pugh points, and liver disease etiology (AIH as reference category; dummy variables for PBC, PSC, HCV, and hepatic cystic disease).

A multiple ordinary least squares regression model was used to estimate regression coefficients (β) and 95% confidence intervals. Categorical variables were dummy-coded; continuous variables were entered in their original units. Observations with missing data in any of the included variables were excluded listwise. To reduce the influence of potential outliers, implausible lumbar spine BMD values (< 0.4 g/cm² or > 2.0 g/cm²) were set to missing before analysis.

Model assumptions (linearity, homoscedasticity, and normality of residuals) were verified by visual inspection of residual plots. Multicollinearity was assessed using variance inflation factors. Statistical significance was defined as a two-sided P < 0.05.

RESULTS

Study population

A total of 82 patients with end-stage liver disease were included (53 female, 64.6%; 29 male, 35.4%), with a median age of 62 years (IQR: 54-70). The control group comprised 101 healthy individuals (16 females, 15.8%; 85 males, 84.2%), with a median age of 50 years (IQR: 48-53). The distribution of patients across disease subgroups was as follows: PBC (n = 15), PSC (n = 22), HCV (n = 11), AIH (n = 13), and hepatic cysts (n = 21). Baseline characteristics including sex distribution, body mass index (BMI), and bisphosphonate use are shown in Table 1. Liver function indices reflected varying degrees of hepatic impairment across disease groups. As expected, patients with cholestatic liver disease (PBC and PSC) exhibited the highest gamma-glutamyl transferase and bilirubin levels, accompanied by elevated MELD and Child-Pugh scores indicative of moderate hepatic dysfunction (median MELD ≈ 13-14; Child-Pugh class B). In contrast, patients with HCV- and AIH-related cirrhosis showed moderately increased transaminases but overall lower bilirubin values, corresponding to compensated liver disease (Child-Pugh A-B). The cystic liver disease cohort demonstrated only mild biochemical alterations and preserved hepatic function (median MELD 9; Child-Pugh A). Healthy controls displayed liver parameters within the normal range.

Table 1 Patient demographics, median (interquartile range).

Group	n	Median age	Female	Male	BMI (kg/m2)	Liver function panel	MELD-score	Child-Pugh score	Bisphosphonate use
PBC	15	64 (57-72)	11	4	28	ALT: 68 (45-92) U/L	14 (11-17)	8 (7-9) (class B)	6
						γ-GT: 460 (320-590) U/L
						Bilirubin: 3.2 (2.1-4.8) mg/dL
PSC	22	49 (42-63)	7	15	22	ALT: 74 (50-108) U/L	-	7 (6-9) (class B)	3
						γ-GT: 410 (280-570) U/L
						Bilirubin: 2.8 (1.9-4.3) mg/dL
HCV	11	66 (64-75)	7	4	27	ALT: 74 (52-106) U/L	-	6 (5-7) (class A)	3
						γ-GT: 220 (150-310) U/L
						Bilirubin: 2.0 (1.3-3.1) mg/dL
AIH	13	63 (53-70)	10	3	26	ALT: 102 (84-155) U/L	-	7 (6-8) (class B)	2
						γ-GT: 310 (220-420) U/L
						Bilirubin: 2.5 (1.4-3.9) mg/dL
Cysts	21	57 (49-63)	18	3	24	ALT: 45 (30-60) U/L	-	5 (5-6) (class A)	1
						γ-GT: 110 (85-150) U/L
						Bilirubin: 0.9 (0.5-1.4) mg/dL
Control	101	50 (48-53)	16	85	27	Within normal range	-	-	-

Baseline demographic characteristics of patients with different etiologies of end-stage liver disease and healthy controls. PBC: Primary biliary cholangitis; PSC: Primary sclerosing cholangitis; HCV: Hepatitis C virus infection; AIH: Autoimmune hepatitis; Cysts: Hepatic cystic disease; Control: Healthy control group; BMI: Body mass index; MELD: Model of End-Stage Liver Disease; ALT: Alanin aminotransferase; γ-GT: Gamma-glutamyl transferase.

BMD and T-scores

At the time of evaluation, median T-scores at the lumbar spine were significantly lower in all patient groups compared to the control group (PBC: -2.5, PSC: -2.1, HCV: -1.8, hepatic cysts: -1.9; all P < 0.001; AIH: -1.1, P = 0.08 vs control -0.1). These findings indicate reduced bone mass in all liver disease groups, with the most pronounced changes in BMD seen in cholestatic liver diseases (PBC and PSC) (Figure 1).

Figure 1 T-scores at lumbar spine (L1-L4) and right femoral neck by group. PBC: Primary biliary cholangitis; PSC: Primary sclerosing cholangitis; AIH: Autoimmune hepatitis; HCV: Hepatitis C virus infection; Cysts: Hepatic cystic disease; Control: Healthy control group.

Corresponding BMD values (g/cm²) in the lumbar spine were significantly reduced in all groups (PBC: 0.92, PSC: 0.95, AIH: 1.01, HCV: 0.94, cysts: 0.93; all P < 0.05 vs control: 1.21), confirming the presence of osteopenia or osteoporosis across the spectrum of liver pathologies (Figure 1, Table 2). The corresponding lumbar spine T-scores were -2.5 in PBC, -2.1 in PSC, -1.1 in AIH, -1.8 in HCV, and -1.9 in cystic liver disease compared to -0.1 in controls, highlighting the predominance of trabecular bone loss in cholestatic and cystic entities. At the femoral neck, BMD values were lower but less markedly affected (right femoral neck: PBC 0.91, PSC 0.96, AIH 0.93, HCV 0.79, cysts 0.93 g/cm² vs control 0.99 g/cm²; left femoral neck: PBC 0.88, PSC 0.94, AIH 0.92, HCV 0.98, cysts 0.94 g/cm² vs control 0.99 g/cm²), with corresponding T-scores of -0.8, -0.5, -0.4, -1.7, and -0.7, respectively (control -0.5). These findings indicate that cortical bone involvement at the proximal femoral neck is comparatively milder, consistent with the predominantly trabecular nature of hepatic osteodystrophy. BMD and T-scores at the femoral neck were consistently higher than those at the spine but followed a similar trend. Significant reductions in femoral BMD were noted particularly in the PBC and cyst groups (right femoral neck: P = 0.01 and 0.03, respectively; left femoral neck: P = 0.00 and 0.06, respectively).

Table 2 Bone mineral density and T-scores.

Parameter	PBC	P value PBC	PSC	P value PSC	AIH	P value AIH	HCV	P value HCV	Cysts	P value cysts	Control
BMD right femoral neck	0.91 (0.81; 0.96)	0.01	0.96 (0.91; 1.05)	0.3	0.93 (0.86; 0.99)	0.07	0.79 (0.72; 1.06)	0.1	0.93 (0.85; 1.04)	0.03	0.99 (0.92; 1.09)
T-score right femoral neck	-0.8 (-1.7; -0.3)	0.13	-0.5 (-1.4; -0.2)	0.49	-0.4 (-1.1; 0.3)	0.74	-1.7 (-2.4; -0.2)	0.07	-0.7 (-1.6; 0.1)	0.29	-0.5 (-1.1; 0.2)
BMD left femoral neck	0.88 (0.63; 0.95)	0.0	0.94 (0.87; 1.05)	0.06	0.92 (0.86; 1.04)	0.14	0.98 (0.78; 1.02)	0.25	0.94 (0.84; 1.04)	0.06	0.99 (0.92; 1.08)
T-score left femoral neck	-1.3 (-2.7; -0.4)	0.01	-0.7 (-1.4; -0.1)	0.09	-0.4(-1.2; 0.2)	0.99	-0.6 (-1.7; 1.0)	0.31	-0.7 (-1.5; 0.3)	0.33	-0.5 (-1.1; 0.2)
BMD lumbar spine	0.92 (0.74; 1.10)	0.0	0.95 (0.88; 1.01)	0.0	1.01 (0.93; 1.21)	0.01	0.94 (0.84; 1.00)	0.0	0.93 (0.80; 1.02)	0.0	1.21 (1.09; 1.30)
T-score lumbar spine	-2.5 (-3.1; -0.8)	0.0	-2.1 (-2.6; -1.2)	0.0	-1.1 (-2.1; 0.7)	0.08	-1.8 (-2.9; -0.6)	0.0	-1.9 (-3.0; -0.2)	0.0	-0.1 (-1.0; 0.6)

PBC: Primary biliary cholangitis; PSC: Primary sclerosing cholangitis; HCV: Hepatitis C virus infection; AIH: Autoimmune hepatitis; Cysts: Hepatic cystic disease; Control: Healthy control group; BMD: Bone mineral density.

Serum markers of bone metabolism

Osteocalcin levels were elevated across all liver disease groups, most markedly in patients with hepatic cysts (median 61.9 ng/mL vs 14-17 ng/mL in other groups), suggesting increased bone turnover. PTH concentrations were mildly elevated in most groups, but notably highest in the cyst cohort (85.8 ng/L), many of whom had concurrent polycystic kidney disease and chronic kidney impairment.

Micronutrient analyses revealed increased serum copper in patients with cholestatic liver disease (PBC: 25.8 μmol/L, PSC: 27.6 μmol/L) and elevated zinc in PSC, HCV, and cyst patients. Magnesium and calcium levels were within normal range across all groups (Table 3).

Table 3 Bone metabolism markers.

Marker	PBC	PSC	AIH	HCV	Cysts
Osteocalcin (ng/mL)	14.2	17.2	14.9	14.6	61.9
25-OH vitamin D (nmol/L)	142.4	60.3	49.9	45.9	79.0
PTH (ng/L)	34.7	39.0	44.8	49.7	85.8
Phosphate (mmol/L)	1.1	1.0	1.0	2.1	1.1
Copper (μmol/L)	25.8	27.6	19.9	18.8	27.5
Zinc (μmol/L)	7.1	8.8	8.0	7.6	9.8
Magnesium (mmol/L)	0.8	0.8	0.8	0.8	0.8
Calcium (mmol/L)	2.3	2.3	2.3	2.1	2.3
Creatinine (μmol/L)	88	95	83	91	3101
AST (U/L)	85	74	121	98	63
ALT (U/L)	72	69	115	104	52

¹Several patients with hepatic cysts presented with concomitant polycystic kidney disease and dialysis-dependent renal failure, explaining the markedly elevated creatinine levels. PBC: Primary biliary cholangitis; PSC: Primary sclerosing cholangitis; HCV: Hepatitis C virus infection; AIH: Autoimmune hepatitis; Cysts: Hepatic cystic disease; PTH: Parathyroid hormone; AST: Aspartat aminotransferase; ALT: Alanin aminotransferase.

Impact of glucocorticoids

Among AIH patients, 2/13 were receiving ongoing steroid therapy at the time of evaluation. No significant difference in lumbar BMD was observed compared to non-steroid-treated AIH patients, although small sample size limits interpretability. Nonetheless, AIH patients had the highest median BMD (1.01 g/cm², T-scores -1.1) at the lumbar spine and femoral neck (BMD, T-score), respectively.

Group comparisons and statistical significance

Compared with controls, patients with PBC, PSC, HCV, and hepatic cysts showed significantly reduced BMD at the lumbar spine (P < 0.01 for all). In contrast, AIH patients did not significantly differ in T-score from controls (P = 0.08). At the femoral sites, the magnitude of BMD reduction was smaller, and statistical significance was limited, consistent with the known predominance of trabecular bone loss in hepatic osteodystrophy.

Match-pair-analysis

A total of 9 patients with documented osteoporotic fractures (pre- or post-transplantation) were successfully matched 1:1 to 9 fracture-free liver transplant recipients based on age, sex, BMI, MELD score, vitamin D, PTH, and osteocalcin levels. DXA analysis showed a consistent trend toward lower BMD and T-scores in the fracture group compared to matched controls. The mean T-score at the lumbar spine was -3.07 in the fracture group vs -1.29 in the control group (P = 0.078). Similarly, the mean overall T-score was -2.45 vs -0.81 (P = 0.078), and BMD at the left femoral neck was 0.77 g/cm² vs0.94 g/cm² (P = 0.094), respectively (Tables 4, 5, and 6).

Table 4 Correlation between bone mineral density parameters and liver function scores.

Variable pair	Spearman’s ρ	Interpretation
Lumbar spine T-score vs MELD	-0.01	No correlation
Lumbar spine T-score vs Child-Pugh	+0.12	Weak positive, not significant
Lumbar spine BMD vs MELD	-0.06	No correlation
Lumbar spine BMD vs Child-Pugh	+0.09	Weak positive, not significant
Right femoral neck T-score vs MELD	+0.04	No correlation
Right femoral neck T-score vs Child-Pugh	+0.09	Weak positive
Right femoral neck BMD vs MELD	+0.01	No correlation
Right femoral neck BMD vs Child-Pugh	+0.01	No correlation
Left femoral neck T-score vs MELD	-0.02	No correlation
Left femoral neck T-score vs Child-Pugh	+0.06	Weak positive
Left femoral neck BMD vs MELD	-0.01	No correlation
Left femoral neck BMD vs Child-Pugh	+0.02	No correlation
Lumbar spine T-score vs MELD	-0.01	No correlation
Lumbar spine T-score vs Child-Pugh	+0.12	Weak positive, not significant
Lumbar spine BMD vs MELD	-0.06	No correlation
Lumbar spine BMD vs Child-Pugh	+0.09	Weak positive, not significant
Right femoral neck T-score vs MELD	+0.04	No correlation
Right femoral neck T-score vs Child-Pugh	+0.09	Weak positive
Right femoral neck BMD vs MELD	+0.01	No correlation
Right femoral neck BMD vs Child-Pugh	+0.01	No correlation
Left femoral neck T-score vs MELD	-0.02	No correlation
Left femoral neck T-score vs Child-Pugh	+0.06	Weak positive
Left femoral neck BMD vs MELD	-0.01	No correlation
Left femoral neck BMD vs Child-Pugh	+0.02	No correlation

Spearman’s rank correlation coefficients (ρ) were calculated to assess the association between bone parameters (bone mineral density and T-scores at the lumbar spine and femoral neck) and liver function indices. No significant correlations were found between bone density measures and either Model of End-Stage Liver Disease or Child-Pugh scores. Weak, non-significant positive trends were observed between T-scores and Child-Pugh classification, indicating a potential but clinically minor association between better hepatic function and higher bone density. MELD: Model of End-Stage Liver Disease; BMD: Bone mineral density.

Table 5 Match-pair analysis of bone mineral density and T-scores in liver transplant recipients with and without osteoporotic fractures.

Parameter	Test	n	Mean fracture	Mean control	P value
T-score lumbar spine	Wilcoxon	7	-3.07	-1.29	0.08
BMD lumbar spine	Wilcoxon	6	0.81	0.95	0.16
T-score left femoral neck	Wilcoxon	6	-1.9	-0.82	0.16
BMD left femoral neck	Wilcoxon	6	0.77	0.94	0.09

Comparison of lumbar spine and femoral neck bone parameters between patients with documented osteoporotic fractures and matched fracture-free controls. Matching was performed 1:1 based on age, sex, body mass index, Model of End-Stage Liver Disease score, serum 25-hydroxyvitamin D, parathyroid hormone, and osteocalcin levels. BMD: Bone mineral density.

Table 6 Match-pair analysis using Wilcoxon signed-rank test in 1:1 matched pairs.

Parameter	Test	n	Mean fracture	Mean control	P value
T-score (lumbar spine)	Wilcoxon	7	-3.07	-1.29	0.08
BMD (lumbar spine)	Wilcoxon	6	0.81 g/cm²	0.95 g/cm2	0.16
T-score (left femoral neck)	Wilcoxon	6	-1.90	-0.82	0.16
BMD (left femoral neck)	Wilcoxon	6	0.77 g/cm²	0.94 g/cm2	0.09

Detailed comparison of T-scores and bone mineral density values at the lumbar spine and left femoral neck between fracture and control groups. Bone mineral density, expressed in g/cm²; T-score = standard deviation from the mean peak bone mass of a healthy young adult population. Matching criteria: Age, sex, body mass index, Model of End-Stage Liver Disease score, serum 25-hydroxyvitamin D, parathyroid hormone, and osteocalcin levels. All comparisons used the Wilcoxon signed-rank test for paired data. BMD: Bone mineral density.

Although the differences did not reach statistical significance due to the small sample size, the results demonstrate a consistent pattern of reduced bone mass in the fracture cohort.

No significant differences were observed in serum vitamin D, PTH, or osteocalcin levels between the two groups within the matched pairs. In multivariable linear regression models adjusting for sex, age, BMI, MELD score, Child-Pugh points, and liver disease etiology, sex showed no significant association with any bone parameter (Tables 7, 8, 9, and 10). This indicates that the male predominance in the cohort did not confound the observed relationships between liver disease and bone status.

Table 7 Regression coefficients for bone mineral density lumbar spine.

Predictor	Regression coefficient (β)	P value	95%CI (lower bound)	95%CI (higher bound)	Standardized coefficient (β-standardized)
Sex (male)	-0.0159	0.7681	-0.1231	0.0914	-0.0414
Age (per year)	-0.0046	0.0215	-0.0084	-0.0007	-0.3115
BMI (kg/m2)	0.0121	0.0202	2	0.0223	0.3162
MELD score	-0.0046	0.4076	-0.0155	0.0064	-0.1312
Child-Pugh (points)	0.0082	0.6788	-31	0.0474	0.0726
PBC	0.1079	0.0917	-18	0.2338	-0.1033
PSC	0.1766	0.0002	0.0878	0.2655	0.0473
HCV	0.18	0.0129	0.0395	0.3205	0.0382
AIH	0.1775	0.0038	0.0596	0.2955	0.0387
Polycystic liver disease	0.1495	0.0023	0.0557	0.2433	-0.0209

Results from multiple linear regression identifying independent predictors of areal lumbar spine bone mineral density (g/cm²). CI: Confidence interval; BMI: Body mass index; MELD: Model of End-Stage Liver Disease; PBC: Primary biliary cholangitis; PSC: Primary sclerosing cholangitis; HCV: Hepatitis C virus infection; AIH: Autoimmune hepatitis.

Table 8 Regression coefficients for T-score lumbar spine.

Predictor	Regression coefficient (β)	P value	95%CI (lower bound)	95%CI (higher bound)	Standardized coefficient (β-standardized)
Sex (male)	-0.4431	0.3909	-1.4671	0.5809	-0.1138
Age (per year)	-0.0045	0.8165	-0.0432	0.0341	-0.0312
BMI (kg/m2)	0.0753	0.1503	-28	0.1785	0.1906
MELD score	-38	0.4853	-0.1459	0.07	-0.1049
Child-Pugh (points)	0.1194	531	-259	0.4978	106
PBC	-1.3522	0.0379	-2.6265	-0.0778	-0.18
PSC	-0.4625	297	-1.3406	0.4156	0.0009
HCV	-0.7389	0.2776	-2.086	0.6083	-0.0497
AIH	0.2416	678	-0.9144	1.3977	0.1437
Polycystic liver disease	-0.1799	701	-1.1112	0.7513	0.0681

Results from multiple linear regression identifying independent predictors of areal lumbar spine for T-score. CI: Confidence interval; BMI: Body mass index; MELD: Model of End-Stage Liver Disease; PBC: Primary biliary cholangitis; PSC: Primary sclerosing cholangitis; HCV: Hepatitis C virus infection; AIH: Autoimmune hepatitis.

Table 9 Regression coefficients for bone mineral density left femoral neck.

Predictor	Regression coefficient (β)	P value	95%CI (lower bound)	95%CI (higher bound)	Standardized coefficient (β-standardized)
Sex (male)	-0.0047	0.9122	-0.0897	0.0803	-0.0147
Age (per year)	-0.0027	0.0914	-0.0058	0.0004	-0.2182
BMI (kg/m2)	0.0121	0.0052	0.0038	0.0204	0.3727
MELD score	-0.0	0.9918	-0.0091	9	-0.0016
Child-Pugh (points)	-0.0056	729	-38	0.0267	-0.0593
PBC	0.0241	0.6421	-0.0789	0.1271	-0.2673
PSC	0.1671	0.0	0.0956	0.2386	0.1003
HCV	0.1548	0.0088	0.0404	0.2693	0.0432
AIH	0.1428	0.0044	0.0462	0.2394	0.0213
Cystic disease	0.1591	0.0001	0.0828	0.2355	73

Results from multiple linear regression identifying independent predictors of areal left femoral neck bone mineral density (g/cm²). CI: Confidence interval; BMI: Body mass index; MELD: Model of End-Stage Liver Disease; PBC: Primary biliary cholangitis; PSC: Primary sclerosing cholangitis; HCV: Hepatitis C virus infection; AIH: Autoimmune hepatitis.

Table 10 Regression coefficients for bone mineral density right femoral neck.

Predictor	Regression coefficient (β)	P value	95%CI (lower)	95%CI (upper)	Standardized coefficient (β-standardized)
Sex (male)	-9	0.86	-0.1	0.08	-15
Age (years)	-0.0029	0.08	-6	0.0003	-0.22
BMI (kg/m2)	0.0124	4	4	21	0.37
MELD score	0.0002	0.94	-8	9	-2
Child-Pugh (points)	-6	0.71	-38	26	-0.06
PBC	-0.11	0.07	-0.23	0.01	-0.27
PSC	0.03	0.66	-0.1	0.15	0.1
HCV	0.01	0.89	-0.13	0.15	0.04
Cystic disease	0.02	0.8	-0.11	0.14	0.07

Results from multiple linear regression identifying independent predictors of areal right femoral neck bone mineral density (g/cm²). CI: Confidence interval; BMI: Body mass index; MELD: Model of End-Stage Liver Disease; PBC: Primary biliary cholangitis; PSC: Primary sclerosing cholangitis; HCV: Hepatitis C virus infection; AIH: Autoimmune hepatitis.

For lumbar spine BMD (Table 7), age (β = -0.0046 g/cm² per year, P = 0.021) and BMI (β = +0.0121 g/cm² per kg/m², P = 0.020) were independent predictors of bone density. MELD and Child-Pugh scores were not significantly associated with BMD. Among etiologic subgroups, none showed significant differences compared with AIH after adjustment.

For the lumbar spine T-score (Table 8), only PBC was independently associated with lower values compared to AIH (β = -1.59,P = 0.036), while PSC, HCV, and cystic liver disease were not significant. Age, BMI, and liver function parameters did not predict lumbar T-scores in the adjusted model.

At the femoral neck, both left and right sides yielded consistent findings (Tables 9 and 10). Higher BMI was positively associated with femoral BMD (left: Β = +0.0121 g/cm² per kg/m², P = 0.005; right: Β = +0.0124, P = 0.004). Age showed a trend toward lower femoral BMD (left: P = 0.091; right: P = 0.08). There was no significant effect of sex, MELD, or Child-Pugh scores on femoral BMD. Patients with PBC tended to have lower adjusted femoral BMD compared to AIH (left: P = 0.06; right: P = 0.07), although these associations did not reach statistical significance.

Across all models, adjusted coefficients of determination (R²) ranged from 0.05 for lumbar spine BMD to 0.12 for femoral BMD, consistent with expected variability in clinical cross-sectional datasets. Sensitivity analyses excluding patients with hepatic cystic disease confirmed the robustness of these findings.

DISCUSSION

In this cross-sectional analysis of 82 patients with end stage liver disease (ESLD) awaiting transplantation, we observed a high prevalence of skeletal compromise across all hepatic disease etiologies, with lumbar spine T-scores ranging from -2.5 in PBC to -1.1 in AIH. Compared to healthy controls, all groups except AIH had significantly reduced spinal BMD, confirming the widespread presence of osteopenia or osteoporosis in this population. These findings align with prior reports citing osteoporosis prevalence rates up to 80% in advanced PBC and PSC[8,9].

The lumbar spine, with its high proportion of metabolically active trabecular bone, was consistently more affected than the femoral neck, mirroring known patterns in hepatic osteodystrophy[4]. This site-specific vulnerability is also reflected in high-resolution peripheral quantitative computed tomography (HR-pQCT) studies, where greater percentage losses in trabecular volumetric BMD (vBMD) were detected, compared to cortical-rich sites. A meta-analysis of HR-pQCT fracture studies demonstrated that fracture cases exhibited on average -12.6% lower trabecular vBMD, -8.8% lower trabecular number, and +9.7% higher trabecular separation compared to non-fracture controls - well above the least significant change for these parameters[1,10]. While HR-pQCT and DXA assess distinct aspects of bone quality and cannot be directly compared, the magnitude of lumbar spine T-score reduction observed in our PBC group (-2.4 SD vs controls) reflects a correspondingly severe skeletal deficit and thus underscores the clinical significance of bone loss in this population[11].

Patients with hepatic cystic disease demonstrated marked skeletal deficits (lumbar T-score -1.9; BMD 0.93 g/cm²), accompanied by substantially elevated osteocalcin (median 61.9 ng/mL) and PTH (85.8 ng/L) levels. More than half had coexisting polycystic kidney disease and end-stage renal failure - conditions associated with cortical thinning and trabecular deterioration on HR-pQCT. This suggests a multifactorial bone loss phenotype, combining features of hepatic and renal osteodystrophy. In HR-pQCT studies of dialysis patients, cortical area is reduced by ~8%-10% and cortical porosity increased by ~15% compared to matched controls, parameters that are independently predictive of fracture risk.

HCV-related bone loss was also substantial (lumbar T-score -1.8; BMD 0.94 g/cm², P < 0.01), consistent with literature linking chronic viral hepatitis to systemic inflammation and elevated osteoclast activity[12]. AIH patients had the highest median BMD (1.01 g/cm²) yet still subnormal T-scores (-1.1), likely reflecting cumulative glucocorticoid exposure. HR-pQCT has shown that steroid therapy can precipitate early trabecular perforation and cortical thinning before significant DXA changes appear, complicating attribution solely to hepatic pathology.

Our matched-pair analysis, though underpowered for statistical significance, revealed clinically meaningful differences: Fracture patients had mean lumbar spine T-scores 1.8 SD lower and femoral neck BMD 0.17 g/cm² lower than matched controls. This is directionally consistent with HR-pQCT-based prediction models - such as the Bone Microarchitecture International Consortium pooled analysis - which demonstrated that microarchitectural parameters (total vBMD, trabecular number, cortical area) improve fracture discrimination beyond hip aBMD, with net reclassification improvements of 17%-21%[13,14].

Biochemically, calcium and magnesium remained normal, while copper was elevated in cholestatic disease and zinc increased in several groups. The cyst group’s high osteocalcin and PTH likely reflect high-turnover bone loss and secondary hyperparathyroidism, both of which are associated with greater microarchitectural deterioration in HR-pQCT studies[15-17].

In multivariable regression analyses (Tables 7, 8, 9, and 10), neither MELD nor Child-Pugh scores were independently associated with lumbar or femoral BMD, suggesting that the severity of liver dysfunction itself exerts a limited direct influence on skeletal integrity once other clinical factors are considered. Instead, age and BMI emerged as the principal determinants of BMD, consistent with previous studies identifying low body weight as a major risk factor for hepatic osteodystrophy[3,18]. The absence of a significant association with MELD or Child-Pugh points aligns with earlier findings that biochemical and hormonal factors, such as vitamin D deficiency, hypogonadism, and inflammation, better explain bone loss in liver disease than global hepatic function scores[3].

Spearman’s correlation analyses confirmed these results, showing no significant correlations between BMD or T-score and MELD (ρ = -0.07, P = 0.58 for lumbar BMD; ρ = -0.01, P = 0.92 for lumbar T-score) or Child-Pugh points (ρ = +0.06, P = 0.60 and ρ = +0.12, P = 0.29, respectively). Femoral neck BMD also showed no relationship with either score (ρ = -0.01, P = 0.95 for MELD; ρ = +0.02, P = 0.88 for Child-Pugh). These findings indicate that bone loss occurs largely independent of the degree of hepatic decompensation, reinforcing that standard liver function indices are poor surrogate markers of skeletal risk in ESLD.

Importantly, after adjustment for age, BMI, and disease etiology, sex was not associated with BMD or T-score, ruling out gender imbalance as a confounder in the observed bone density differences. This emphasizes that the lower bone mass seen in cholestatic liver disease reflects disease-specific mechanisms rather than demographic bias.

From a pathophysiologic standpoint, the lack of correlation between BMD and liver severity scores supports the concept that hepatic osteodystrophy represents a parallel metabolic complication of liver disease rather than a simple sequela of liver failure. Bone loss may therefore begin early in the disease course and progress independently of hepatic function decline. Histomorphometric and HR-pQCT data from patients with early-stage cholestatic disorders have shown reduced trabecular bone volume and connectivity even before biochemical evidence of liver failure[18]. This highlights the need for early and systematic screening of bone health in patients with CLD, especially those with cholestatic or cystic etiologies where bone loss appears most severe.

Collectively, these results underscore that hepatic osteodystrophy is multifactorial in nature and only weakly related to overall liver function. Preventive and therapeutic strategies should therefore target modifiable risk factors - such as maintaining adequate body weight, optimizing vitamin D status, and minimizing steroid exposure - while integrating routine DXA assessment into the pre-transplant evaluation of all ESLD patients.

Clinical implications

Our findings, together with microarchitectural evidence from HR-pQCT studies, underscore the need for routine lumbar spine DXA screening in all patients with ESLD, particularly those with cholestatic or cystic etiologies such as PBC, PSC, or hepatic cystic disease. The consistent association of bone loss with age and BMI rather than MELD or Child-Pugh scores highlights that skeletal deterioration progresses largely independent of hepatic function. Therefore, bone assessment should be incorporated into transplant evaluation irrespective of disease stage or biochemical severity.

Given that conventional DXA captures only areal density and not microarchitectural quality, adjunctive tools such as the trabecular bone score or HR-pQCT may better identify patients at high fracture risk. This is particularly relevant in AIH patients receiving long-term corticosteroids, in whom DXA often underestimates true structural compromise. Integration of these advanced imaging modalities into pre-transplant screening could improve risk stratification and guide early intervention.

Limitations

This study has several limitations, including its retrospective single-center design, small etiologic subgroups, and the absence of vertebral fracture assessment or direct microarchitectural imaging. The cross-sectional nature of the analysis precludes conclusions regarding longitudinal bone loss. Moreover, serum markers of bone turnover and vitamin D metabolism were only partially available. Despite these limitations, the consistency of our findings with established HR-pQCT-based microarchitectural thresholds suggests that the observed DXA deficits represent clinically meaningful skeletal deterioration.

Future directions

Future work should aim to integrate advanced imaging modalities - such as HR-pQCT and trabecular bone score - into pre- and post-transplant protocols to quantify both trabecular and cortical deficits. Prospective longitudinal studies are warranted to clarify the temporal relationship between liver disease progression, transplantation, and bone microarchitecture. In addition, early initiation of anti-osteoporotic therapy, ideally with anabolic-first sequences followed by antiresorptives, should be evaluated for its potential to preserve bone mass and prevent fractures in this vulnerable population.

Summary

Across all disease groups, at least osteopenic changes were evident, with PBC and PSC showing the most pronounced reductions in lumbar spine BMD. The hepatic cyst group demonstrated similarly low bone density despite preserved hepatic function, likely reflecting combined hepatic and renal osteodystrophy. Elevated bone turnover markers and micronutrient imbalances were common but heterogeneous across etiologies. Neither MELD nor Child-Pugh scores correlated significantly with bone parameters, emphasizing that liver function indices are poor predictors of skeletal health in ESLD.

CONCLUSION

In this retrospective cross-sectional study, reduced BMD was observed across all investigated etiologies of end-stage liver disease, with the most pronounced impairment in patients with cholestatic liver disorders. While these findings support the concept of hepatic osteodystrophy as a relevant complication of advanced liver disease, the extent of bone loss appears to vary depending on the underlying etiology. Given the retrospective design and the limited sample size of certain subgroups, these results should be interpreted with caution. Nevertheless, the data highlight the importance of routine bone health assessment in patients with ESLD, particularly in those with known risk factors. Future prospective studies are warranted to further clarify disease-specific mechanisms and to evaluate targeted preventive and therapeutic strategies.