Introduction

Osteopenia, osteoporosis, and fragility fractures are frequent complications of patients with chronic liver disease and cirrhosis. The prevalence of metabolic bone disease in patients with cirrhosis is estimated at 12–55%, higher than the general population of same age [1, 2], and up to 40% of patients with chronic liver disease may experience fragility fractures [3, 4]. This prevalence is even higher in patients with hemochromatosis or cholestatic liver diseases in whom [5] the bone histomorphometry can reveal a specific cholestatic osteopenia, characterized by elevated bone resorption and decreased bone formation [6].

Over the last years, the average age of liver transplantation (LT) recipients has progressively increased with a shift from viral to metabolic etiology of the underlying liver disease, with a correlated increase of LT carriers’ comorbidities, chronic medications, and higher waitlist post-transplant mortality [7]. Although post-transplant bone and mineral disorders are associated with increased morbidity and mortality, only few studies, with limited sample size and dating back to the late 1990s–early 2000s, have shown that the prevalence of osteoporosis could be related to the severity of liver disease in the transplant setting [8] and that cholestatic etiology, female sex and lower body weight are important predictors of osteoporosis [9].

The most common fracture site in patients with chronic liver disease has been reported to be vertebrae with fractures of the femoral neck relatively uncommon [10]. Since up to one-third of vertebral fractures can be asymptomatic, it has been suggested that spine X-rays is an essential tool in the clinical assessment of patients with secondary osteoporosis [10].

In patients with chronic liver disease, the balance in bone remodeling activities between osteoclasts and osteoblasts is profoundly altered by the liver disease [11]. Guidelines provided by most osteoporosis societies describe the causes of osteoporotic fractures in patients with chronic liver disease as a result of nutritional deficiencies due to the underlying organ disease [5, 12]. However, several studies have suggested that osteoporosis in cirrhotic patients is a multifactorial disease in which different mechanisms act to deteriorate bone mass, thus determining bone fragility [3]. Several etiologies may determine chronic liver disease, with different pathogenetic mechanisms [5]. For example, while hemochromatosis and cholestatic diseases are respectively characterized by significant increase in iron and bilirubin, which cause osteoblast inhibition [13], by contrast, viral hepatitis is associated with an activation of the immune response and cytokine release which in turn stimulate bone resorption [10, 14, 15]. Overall, two main pathophysiological mechanisms underlying osteoporosis in patients with chronic liver disease have been thus far recognized, similar to primary osteoporosis: decreased bone formation or increased bone resorption, or both.

Considering the significantly outdated and small sample-sized literature on this topic, as well as the lack of definitive clinical consensus on evaluation and management of bone fragility in patients with advanced liver disease, the main aim of our study was, therefore, to describe the prevalence, type and site of fragility fractures in a large single-referral center cohort of patients with advanced liver disease undergoing LT, with a complete characterization of etiologies, biochemistry and radiology and secondarily to provide sex-specific information on bone fragility in LT recipients in order to target resources and pharmacologic therapies to this specific setting of patients at exceedingly high risk of bone fracture.

Patients and methods

Study population

We performed a single-center retrospective analysis of consecutive patients who underwent LT from January 1st, 2010 to December 31st, 2015 at our referral Liver Transplant Center, IRCSS Azienda Ospedaliero-Universitaria di Bologna, Italy. All patients were evaluated and managed at our center and received a standardized clinical, laboratory and radiological evaluation, including thoracic and abdominal imaging before LT aligned with local and national guidelines. All patients had an abdominal computed tomography (CT) scan performed within 6 months from the LT. Biochemistries, including parathyroid hormone (PTH), minerals, and bone turnover markers (Beta-Cross Laps, CTX and bone-specific alkaline phosphatase, BSAP) were also recorded when available. All recorded biochemistries were taken in a fasting state, between 8.00 and 9.00 a.m. All samples were analyzed at the Unified Metropolitan Laboratory of Bologna [16].

The medical records reviewed were as follows: the Interregional Transplant Association chart (a local chart including clinical data of transplant candidates), as well as the chart of the in-hospital admission at the time of LT surgery, which included the previous relevant clinical history of the patient and comorbidities, including clinical fractures, etiology of liver disease and its complications, biochemistries, radiology reports including bone density tests (dual energy X-ray absorptiometry DXA), as well as concomitant pharmacologic treatments.

The centralized radiological imaging data center (PACS) (the archive that holds all clinically obtained electronic radiological/nuclear medicine images in DICOM® format of each registered patient) was evaluated for each patient. For each patient, a lateral view of a conventional thoracic X-ray and a thoraco-abdominal CT scan (Scout-Scan) performed within 6 months from LT were selected and analyzed. Two trained physicians, bone metabolism specialists (G.Z and P.A), blinded to patient clinical data (except for sex and birth date), including chronic liver disease etiology or severity, re-reviewed all acquired spine images to screen for morphometric vertebral fractures. Inconsistent findings were solved by reaching consensus after several measurements of the vertebra. Semiquantitative visual assessment according to Genant’s criteria [17] was performed to ascertain and assess severity of vertebral deformities. Percentage reductions of either anterior, middle, or posterior vertebral heights were calculated and used to define mild (20–25%), moderate (26–40%) and severe (> 40%) vertebral fractures on lateral projections of spine imaging. Previous kyphoplasty or vertebroplasty were also documented and counted as one or multiple vertebral fractures according to their extension. Date and anatomic site of clinical fragility fractures (i.e. all fractures that would cause a patient to seek medical care, including clinical spine, due to absent or low trauma) were also recorded from the medical records and Interregional Transplant Association charts. DXA scans performed within 2 years of LT were also reviewed, and the lowest value of either femur neck, total hip or lumbar spine BMD T-score was used to classify patients according to Word Health Organization (WHO) BMD categories (normal BMD, osteopenia, or osteoporosis). Both T-scores and Z-scores were reported. Fractured vertebrae (L1–L4) were excluded from the analysis of lumbar BMD T-scores or Z-scores.

Ethical approval

This study was conducted in line with the Declarations of Helsinki and STROBE (STrengthening the Reporting of OBservational studies in Epidemiology) recommendations [18]. The local Ethical Committee approved this study (protocol code: 16/2020/Oss/AOUBo).

Statistical analysis

Absolute numbers and percentages were calculated for categorical data. The results for continuous variables were expressed as means and standard deviation (SD). A comparison of general characteristics of LT recipients was performed by Mann–Whitney U test, comparing fractured and non-fractured patients and general characteristics between sexes and within each sex. χ2 test was used to detect associations between fragility fractures, sex and other clinical data, such as diabetes or corticosteroids use, in both sexes. Multinomial logistic regression with stepwise backward elimination was used to identify risk factors for fragility fractures across the whole population, by adjusting for potential confounders. Covariates were chosen among expected major risk factors for fragility fractures, and significant or near-significant (P < 0.10) parameters in simple correlations. Statistical analyses were performed using SPSS (version 26.0). P values lower than 0.05 were considered statistically significant.

Results

Our study identified 429 consecutive patients receiving liver transplantation. After chart review, we excluded 63 patients with transplant surgery due to acute liver failure and no previous history of chronic liver failure, patients not undergoing their first LT (i.e. reoperation) and any combined transplantation (concomitant kidney or heart transplantations). High-trauma fractures (motor vehicle accidents, etc.) were not counted as fragility fractures. All the remaining fractures were considered to be due to low-trauma or osteoporosis, unless otherwise stated. Twenty-six patients (6.0% of the total cohort) had missing or inaccessible radiological imaging/reports and were also excluded. A total of 366 patients were included in the final analysis (Figure S1, supplementary).

Characteristics of liver transplant recipients: whole population and gender comparison

Of 366 LT recipients included in the study, the majority of them had viral cirrhosis—144 (39.3%)—and 94 (25.7%) had multifactorial disease (Table 1 and 2). Of the 94 patients with multifactorial disease, 61 (64.8%) had viral + alcoholic etiologies, 14 (14.9%) had viral + rare disease etiologies, 13 (13.8%) had viral + MASH etiologies, 1 (1.1%) had cholestatic + viral etiologies, 3 (3.2%) had viral + alcoholic + MASH etiologies, and 2 (2.1%) had alcoholic + MASH etiologies. The overall cohort was composed of 107 (29.3%) women and 259 (70.7%) men, with significant differences in sex prevalence among the etiology categories, with autoimmune and cholestatic disease being more prevalent in women (9.3% and 10.3% vs. 0.1% and 4.2%, respectively), while ALD and multifactorial disease being more common in men (13.9% and 28.6% in men vs. 5.6% and 18.7% in women, respectively). Clinical and anthropometric characteristics are shown in Table 1 and  2. The mean age was 52 years with no significant difference between women and men. A positive smoking history was more frequent in men (26.6% vs. 10.3%, P = 0.001). Diabetes and hepatocellular carcinoma were more frequent in men than in women, while females were more frequently exposed to glucocorticoids. Mean body mass index (BMI) was 25.6 kg/m2, with no differences between sexes. Hip BMD and estimated glomerular filtration rate (eGFR) were lower in women than in men, while hypertension was similar between the sexes (Table 1 and 2).

Table 1 Clinical characteristics of the whole population, according to sex: continuous variables
Full size table
Table 2 Clinical characteristics of the whole population, according to sex: categorical variables
Full size table

Instead, when we compared patients without fractures (84/211) versus patients with fractures (65/155) no differences occurred about the presence of hepatocellular carcinoma in the two populations (P 0.683).

Regarding bone metabolism, fragility fractures prevalence was 155/366 (42.3%) in the overall population, with no significant differences between sexes (fracture prevalence in women was 37.4%, in men was 44.4%). Calcium and vitamin D3 supplements were equally distributed among both sexes, while a small proportion of patients were taking bisphosphonates (13/366, 3.5%). Laboratory parameters of mineral metabolism were tested in very few patients, and no significant differences could be observed between sexes.

Among patients with vertebral fractures (n = 145), mild vertebral fractures (Genant grade 1) were the most frequently observed (90/145, 62.1%), with similar prevalence between sexes. Moderate vertebral fractures (Genant grade 2) were more common in men compared to women (33.9% vs 16.7%, P < 0.05). Severe (Genant grade 3) vertebral fractures occurred more frequently in women compared to men (16.7% vs. 5.5%, P < 0.05). The overall number of patients with clinical fragility fractures (i.e. symptomatic) was 50/366 (13.7%). Of these, most (n = 43, 86%) were vertebral fractures. Other clinical fractures were at the humerus (n = 1), ribs (n = 7), femur (n = 1), and clavicle (n = 1). The median time between clinical fracture occurrence and LT was 2 months. All the remaining fractures were morphometric vertebral fractures. Fracture prevalence among transplant recipients was stable across each year of the study period (Figure S2, supplementary).

Women with fractures, compared to women without fractures had worse kidney function, lower urinary calcium, lower BMD and more commonly having alcoholic etiology (Supplementary Table a. and b.). Compared to men without fractures, men with fractures had lower 25-OH vitamin D levels, with no other noticeable significant differences in laboratory or clinical data (Supplementary Table c. and d.).

Most vertebral fractures were single fractures, although a significant proportion of patients (n = 60, 41.3%) had two or more vertebral fractures, up to a maximum of 10 vertebral fractures per patient (Fig. 1).

Fig. 1
figure 1

Distribution of single and multiple vertebral fractures across patients with vertebral fractures (N = 145)

Full size image

Characteristics of liver transplant recipients with bone fractures

Patients with bone fractures presented similar age and BMI compared to patients without fractures, although serum glutamic pyruvic transaminase (GPT) was lower and serum creatinine greater than patients with no fractures. Other parameters, both from clinical history and laboratory, were similar between groups (Table 3 and 4). The severity of liver disease was not different between groups. Fragility fractures showed similar rates across each liver disease etiology (Figure S3, supplementary). The most common vertebral sites were at T7, T8, T9 and T12 vertebrae (Fig. 2). Lower rates of fractures were observed in the lumbar spine. The most frequently observed vertebral fracture type was wedge fractures, with a minor but significant proportion of crush or biconcave fractures (Figure S4, supplementary).

Table 3 Clinical characteristics of the patients with and without fractures: continuous variables
Full size table
Table 4 Clinical characteristics of the patients with and without fractures: categorical variables
Full size table
Fig. 2
figure 2

Distribution of vertebral fractures (absolute frequencies)

Full size image

Characteristics of liver transplant recipients: effect of glucocorticoids on metabolic bone disease

Glucocorticoid administration differed across etiologies, although its impact on fragility fractures, vertebral fractures, or bone mineral density by DXA was not evident (Supplementary Table e.).

Characteristics of liver transplant recipients: effect of diabetes on metabolic bone disease

Diabetes prevalence differed according to etiology, although it was not associated with fragility fractures (P = 0.192), vertebral fractures prevalence or severity, or low bone density by DXA (Supplementary Table f.).

Predictors of bone fragility fractures

A logistic regression model including age, sex, BMI, alcohol use, eGFR, etiology (autoimmune or cholestatic disease vs. other), revealed that only BMI was negatively associated with prevalent fragility fractures (odds ratio, OR 1.058, 95% CI 1.001–1.118, P = 0.046), independent of other risk factors. In LT recipients, for each one-unit decrease of BMI, the risk of fragility fractures would increase by 5.8%, and vice versa (Table 5).

Table 5 Multinomial logistic regressiona
Full size table

Discussion

Our study investigated the prevalence of fragility fractures, either clinical or morphometric, in a large cohort of patients undergoing LT due to different etiologies, who were fully characterized in terms of radiology and medical history. The type and the most frequent location of such fractures were also assessed in order to generate a consistent fracture risk profile of a modern cohort of LT recipients. The prevalence of osteoporotic fractures within the cohort was 42.3%, most of which were thoracic vertebral fractures, with femur, humerus and ribs fractures uncommon despite the large sample size. Thanks to the blind re-evaluation of the X-rays of the spine and lateral Scout-CTs of the thoraco-lumbar vertebral column, a large proportion of patients with metabolic bone disease at the time of transplantation was noted. Most fractures (88%) were anterior wedge fractures of mild to moderate severity, with women having a more severe phenotype than men, although a slightly higher fracture prevalence was shown in men. For each one-unit decrease in BMI, fragility fracture risk increased by 5.8% independently of age, kidney function, etiology of liver disease and alcohol consumption.

To the best of our knowledge, thanks to the sample size, our is the largest study investigating LT candidates over the last 20 years, thus providing an updated clinical picture of modern cohorts of LT recipients. Studies carried out so far were mostly heterogeneous in terms of fracture prevalence in LT recipients, with a huge range of prevalence which was usually reported somewhere between 3 and 43%. Small studies often suffer from possible selection bias or variability in the criteria used in the definition of metabolic bone diseases such as osteoporosis or low BMD. In the study published in 1997 by Monegal A. et al. [8], conducted on 58 cirrhotic candidates to liver transplant, it was observed that 43% of the patients had osteoporosis that was diagnosed according to at least one vertebral fracture and/or a BMD of the lumbar vertebrae < 2 standard deviations compared to the mean values of healthy subjects of the same age [8]. Wariaghli G. et al., in a 2010 study conducted on 64 patients with chronic liver disease, showed that 45.5% of patients had osteoporosis and only 5.3% had vertebral fractures [9]. This study also came with the limitation of a minimal number of patients. Furthermore, the patients examined had only primary biliary cholangitis (PBC) or viral liver disease, limiting speculations on other etiologies.

Only few studies analyzing bone fracture risk among LT candidates have been published over the last 10 years. A study [19] on 128 patients found that the severity of liver cirrhosis was associated with hip fractures. This study, however, examined a cohort of elderly patients that were more than one decade older than those in our study [19]. Another registry-based study carried out in Sweden assessed fracture risk in MASH, showing a slightly higher rate of fractures [20]. We did not find the same results in our study regarding MASH etiology. A limitation could be represented by the low percentage of patients with MASH (0.2%) in our population, which contradicts the growing trend of this disease in the world population. However, considering the presence of MASH also in patients with multifactorial disease, we achieve almost 20% in the average study population. Given this consideration, we found that the liver disease etiology, however, does not represent a risk factor for an increase in fractures in our transplant cohort. This assumption is supported by the long-term fracture risk shown in the Sweden study population with MASH, which was very similar to the general population [20]. All these studies could therefore suggest that fragility fractures might be predominantly caused by liver dysfunction alone rather than a specific chronic liver disease etiology. As cirrhosis worsens, metabolic bone disease might also worsen, reaching its worst scenario right at the time of transplantation [20]. This hypothesis was confirmed by a recent study which assessed 102 patients before and after LT, finding that malnutrition and low BMI were the main determinants of osteopenia/osteoporosis, regardless of etiology, similarly to our study [21].

Over the years, several studies have investigated metabolic bone disease in subjects with PBC and primary sclerosing cholangitis (PSC). In a 1994 study by Camisaca M. et al., conducted on 25 women with PBC, a rapid BMD loss of 3.5% was observed in only 6 months [22], with no data on fracture prevalence. Eastell et al., in a study conducted on 210 women with PBC, described lower BMD compared to controls, although the prevalence of fractures was not assessed [23]. Angulo et al. in a study of 81 patients with PSC, demonstrated that the lumbar spine BMD of the patients was lower than age- and sex-matched healthy controls, as well as that 3% of patients had fragility fractures. Finally, they observed that patients with fractures had more advanced liver disease [24]. In this study the prevalence of fractures was presumably underestimated as the study also involved patients with PSC in the initial stages of the disease, thereby representing a limitation. Our study, instead, suggests that cholestatic disease might be equally important as a risk factor for fracture as other liver disease etiologies, because bone fracture prevalence was non-significantly different from other non-cholestatic disorders. This data can be explained by the increasing trend in the average age of LT candidates the last decade and, consequently, in their comorbidities [7].

Gallego-Rojo F. J., et al., in a 1998 study conducted on 32 patients with cirrhosis of viral etiology, showed an osteoporosis prevalence by DXA to be up to 50%. The limitations of this study were the small sample of patients and the missing data about fractures [25]. In our study fracture prevalence in viral cirrhosis was noted to be over 40%, consistent with previous findings.

Therefore, by estimating the risk of fractures across various etiologies of liver disease, the present study suggests that liver etiology may not be as critical as it was initially thought, with the long-standing liver disease per se being the main risk factor for prevalent fractures. Fracture prevalence in patients with liver disease and cirrhosis at the time of LT was very high (approximately 42%) and fractures were mainly located in the thoracic vertebrae, independent of age and sex. It is yet uncertain why men had a slightly higher prevalence of fracture, although a predominantly male cohort may have affected this in the absence of other consistent explanations. Selection bias might be another reason behind this slight disproportion. BMI was the only independent predictor of fracture prevalence: for every one-unit decrease in BMI, risk of fractures increased proportionally.

In recent years, the hepatologists’ community has focused increasing interest on sarcopenia and frailty as prevalent complications able to predict morbidity, mortality, poor quality of life and worse post-LT outcomes in patients with cirrhosis. Osteoporosis should be considered an emerging issue associated with sarcopenia in LT candidates. Since these complications are potentially modifiable with early identification and therapy, clinicians should pay attention to accurately recognizing and evaluating both sarcopenia and osteoporosis in LT candidates and carriers [26].

Strengths and limitations

Strengths of this study are the large sample size, the consecutive enrolment of the patients, the centralized laboratory and full access to radiological imaging, with blinded review of radiological images, as well as the accuracy of a chart review study regarding correct diagnoses. Moreover, the study’s single-center nature attenuated variability in managing of chronic liver disease before LT.

Being a study grounded in clinical practice, limitation of this study is the absence of a control group and the cross-sectional design. Being ours a major referral center, fracture prevalence might be slightly greater than expected, because of a possible higher frequency of more severe chronic liver disease. Unfortunately, most patients had not a complete mineral metabolism evaluation through the laboratory or DXA. A limited sample size might affect the finding of comparable BMD between males and females, although numerically higher BMD was found in males, as expected. This limitation also prevents speculation on underlying bone metabolism and density in these patients, and, therefore, on the best anti-osteoporotic medications to choose in this setting. Moreover, the menopausal status was not available for all women, although a considerable proportion of women were likely premenopausal based on mean age of the population. Last, the retrospective nature of clinical data might also hide some unintentional bias.

Conclusions

Our study provides substantial evidence, confirming with its large sample size previous findings of a considerable fracture burden in patients awaiting and undergoing LT. These data support the need for a thorough bone metabolism evaluation and management for this category of patients, and implementation of this in future guidelines.

Osteoporotic vertebral fractures are frequent complications of the underlying end-stage disease. In this large study, most patients had one or two vertebral fractures, but a considerable proportion of patients experienced multiple vertebral fractures. The most affected sites are the thoracic vertebrae, in particular T7, T8, T9 and T12, and the most frequent fractures were anterior wedges. Furthermore, the majority of vertebral fractures were asymptomatic, making vertebral morphometry an essential tool to screen for bone fragility. Femoral fractures or other peripheral clinical fractures were uncommon possibly due to a relatively young population, based on mean age at transplantation.

The prevalence of fractures was similar across all etiologies, as opposed to previous limited-sample studies indicating cholestatic liver disorders as one of the etiologies carrying more significant fracture risk. The prevalence of fractures was also similar across different age groups, and comparable between males and females, although severe vertebral fractures were more common in women.

Our findings should be considered in multidisciplinary liver cirrhosis management. Osteoporotic fractures constitute, in fact, the leading risk factor for subsequent bone fractures and significantly influence the quality of life of these patients before and after LT [27]. Liver transplant screening should include laboratory tests relating to bone metabolism, a DXA to quantify bone mass and spine morphometry to exclude the presence of vertebral fractures. Future studies assessing the impact of these parameters on fracture incidence after transplantation will be warranted to estimate fracture risk in the post-transplantation period and its impact on the overall survival of patients receiving liver transplantation.