Osteoporosis Assessment among Adults with Liver Cirrhosis

Osteopenic bone disease occurs frequently in patients with chronic liver cirrhosis, which most frequently presents with hepatic osteodystrophy. Thus, the relationship between nutritional status and bone mineral density has been poorly measured in liver cirrhosis. This single-center study consisted of a group of 70 patients diagnosed with liver cirrhosis. The nutritional status was evaluated with the Controlling Nutritional Status index, and volumetric vertebral bone mineral density was measured with quantitative computed tomography. Among the 70 patients included, osteopenia and osteoporosis were found in 71% and 24.3%, respectively. Malnutrition assessed with the Controlling Nutritional Status index was observed in 56 (80%) patients and was more frequent in alcoholic cirrhosis patients than viral cirrhosis patients (87.24% vs. 65.22%). Significant positive correlation with Controlling Nutritional Status score was found with Model for End-Stage Liver Disease (rho = 0.576, p-value < 0.0001), Child–Pugh score (rho = 0.670, p-value < 0.0001), International Normalized Ratio (rho = 0.517, p-value = 0.001), aspartate aminotransferase (rho = 0.293, p-value = 0.045), and bilirubin (rho =0.395, p-value = 0.02). Among the liver cirrhosis patients, 15 had osteoporosis and 49 had osteopenia at the lumbar spine (L1-L4 vertebrae), as determined by bone mass density via quantitative computed tomography. A non-significant relationship between Controlling Nutritional Status index-assessed nutritional status and bone mass density was documented. Regarding osteoporosis, no differences were found between the viral and alcohol types of liver cirrhosis patients (p-value = 0.870). Age, obesity, grade of varices, Child–Pugh score, and Model for End-Stage Liver Disease score were associated with osteoporosis in patients with liver cirrhosis.


Introduction
The liver is considered an important center for a multitude of physiological processes [1]. It is seen as the main organ for metabolizing three major classes of molecules (protein, fat, also carbohydrate) [2]. Liver processes consist of macronutrient metabolism, breakdown of xenobiotic compounds, also a large number of current drugs, immune system support, blood volume adjusting, endocrine control of growth signaling pathways, and lipid and cholesterol homeostasis [1,3]. Liver cirrhosis (LC) is a normal outcome of all chronic liver diseases (CLD) and can be characterized by tissue fibrosis but also by a transformation of the normal liver framework into structurally abnormal nodules [4]. It may develop over a period of time through liver chronic inflammation and is acknowledged as the end-stage form of CLD that can be continued with a number of nutrition disorders [5].
Newly diagnosed patients with LC hospitalized in the gastroenterology department of the County Clinical Emergency Hospital of Craiova, Romania, were selected for our analysis. The inclusion period was between January 2019 and December 2020. We included patients diagnosed with viral B-induced LC, viral C-induced LC, and alcohol-consumptionrelated disease [22]. All patients had opportunistic QCT (i.e., computed tomography-CT acquired for other medical reasons). Exclusion criteria were (1) patients with hepatocellular carcinoma or other malignancies, autoimmune liver diseases, kidney disease, chronic gastritis [23]; (2) patients treated with oral supplements containing magnesium, calcium, vitamin D, phosphorus, diuretics, and tenofovir disoproxil [24]. If fractures, cementaugmented vertebral bodies, anatomical deformities, or implants were detected in the CT scan, the respective patients were not included in the measurements.
The present study protocol followed all regulations of the Declaration of Helsinki and was approved by the ethics committee of the University of Medicine and Pharmacy of Craiova, Romania (no. 173/29 October 2021).

Imaging
Measurements were performed using computed tomography (CT) and we had a General Electric Light Speed Series, with a helical 64-channel, Revolution, and a Siemens Biograph mCT 20 slices. Scanner settings of 130 kVp were used. CT images were retrospectively analyzed using a largely advanced medical imaging service for picture archiving and communication (Biotronics 3Dnet PACS). A dual-energy lumbar X-ray absorptiometry (DEXA) was performed on Hologic systems Horizon A (S/N200639). The reconstructions were obtained in the axial plane ( Figure 1A) of the lumbar vertebrae from L1 through L4. The region of interest (ROI) was chosen to be around 20 mm, leaving out the cortex ( Figure 1B). The CT attenuation was measured in Hounsfield units (HU) by sketching a click-and-drag elliptical ROI within the axial section of the lumbar trabecular bone. ROI avoided degenerative changes to the vertebrobasilar complex and cortical surfaces. The following formulas were used to approximate bone density values for the General Electric CT scan: QCT-value = 0.71 × HU + 13.82 mg/cm 3 , and, for the Siemens CT scan: QCT-value = 0.985 × HU + 15.516 mg/cm 3 [12,25]. Biograph mCT 20 slices. Scanner settings of 130 kVp were used. CT images were retrospectively analyzed using a largely advanced medical imaging service for picture archiving and communication (Biotronics 3Dnet PACS). A dual-energy lumbar X-ray absorptiometry (DEXA) was performed on Hologic systems Horizon A (S/N200639). The reconstructions were obtained in the axial plane ( Figure 1A) of the lumbar vertebrae from L1 through L4. The region of interest (ROI) was chosen to be around 20 mm, leaving out the cortex ( Figure 1B). The CT attenuation was measured in Hounsfield units (HU) by sketching a click-and-drag elliptical ROI within the axial section of the lumbar trabecular bone. ROI avoided degenerative changes to the vertebrobasilar complex and cortical surfaces.
We used Child-Pugh and MELD scores to evaluate prognosis in LC. We also determined the MELD score for all the patients, which is based on total bilirubin, creatinine, and INR. The Child-Pugh scoring system was also assessed for all patients. This index uses six clinical and laboratory criteria to categorize patients: ascites, hepatic encephalopathy, nutritional status, total bilirubin, albumin, and INR. All of the selected patients were divided by Child-Pugh classification: A: from 5 to 6 points, B: from 7 to 9 points, and C: from 10 to 15 points, and the survival of cirrhotic patients was reduced if the Child-Pugh scores/classes were increased [26][27][28].

Intra-Rater and Inter-Rater Reliability
To perform intra-observer precision, the QCT measurements were carried out twice by the same author (C.M.I.) blinded to any information at 2-week intervals on a random
We used Child-Pugh and MELD scores to evaluate prognosis in LC. We also determined the MELD score for all the patients, which is based on total bilirubin, creatinine, and INR. The Child-Pugh scoring system was also assessed for all patients. This index uses six clinical and laboratory criteria to categorize patients: ascites, hepatic encephalopathy, nutritional status, total bilirubin, albumin, and INR. All of the selected patients were divided by Child-Pugh classification: A: from 5 to 6 points, B: from 7 to 9 points, and C: from 10 to 15 points, and the survival of cirrhotic patients was reduced if the Child-Pugh scores/classes were increased [26][27][28].

Intra-Rater and Inter-Rater Reliability
To perform intra-observer precision, the QCT measurements were carried out twice by the same author (C.M.I.) blinded to any information at 2-week intervals on a random sample of 20 patients. For inter-observer reliability, the QCT measurements were performed by two readers (C.M.I. and T.N.S) for L1 to L4 on a random sample of 20 patients. The assessed intraclass correlation coefficients (ICC) were considered excellent for values greater than 0.90 and good for values between 0.75 and 0.90 [29]. We also calculated internal consistency by Cronbach's alpha (0.9: excellent; 0.8-0.9: good; 0.7-0.8: acceptable) [30].

Statistical Analysis
Besides descriptive statistics, different variables were compared using the Mann-Whitney U test or t-test (after checking the normality of continuous variables) or chisquared test (for categorical variables). Moreover, various correlations between variables were calculated by using Spearman coefficients and visually presented with a correlation heatmap (colors range from bright blue for strong positive correlations, to bright green, for strong negative correlations). Missing data (the case for calcium and T-score) were the type that were missing completely at random due to the retrospective type of study, and we used mean substitution in order to reduce the standard error [31]. We applied univariable and multivariable regression analyses with the stepwise method to investigate independent associations between BMD as determined by QCT (as dependent variable) and other clinical and laboratory parameters (as independent variables). The standardized linear coefficients β (+95% CI) showing linear correlations between two parameters were determined. Statistical analysis was performed with Python (version 3.10.7) at a 0.05 level of significance (two-sided).

Patient Characteristics
Baseline data in all patients are presented in Table 1. During the study period, CT was performed for 70 patients (65 males in the majority, 92.9%, and 5 females, 7.1%). The age ranged from 53 to 66.5 years, and the patients with viral cirrhosis were significantly older than patients with alcohol cirrhosis (p-value = 0.020). There were 55 cases (78.6%) of history of hepatic decompensation, with more cases for patients with alcoholic cirrhosis (p-value = 0.015). Twenty-four cases (34.3%) had jaundice, with more cases for patients with alcoholic cirrhosis (p-value = 0.008). Of the 70 patients, 33 (47.1%) had cirrhosis complicated by ascites (no differences between the two groups, p-value = 0.247), 59 (84.3%) had varices (no differences between the two groups, p-value = 0.096), and 14 (20%) had hepatic encephalopathy (no differences between the two groups, p-value = 0.087). There were 25 (35.7%) patients in Child-Pugh A, 26 (37.1%) patients in Child-Pugh B, and 19 (27.1%) patients in Child-Pugh C (with more viral cirrhosis patients in Child-Pugh A and more alcohol cirrhosis patients in Child-Pugh B or C). The CONUT score has values between 0 to 11 (with median 5), according to which a normal nutritional state was found in 14 (20%) patients, a mild nutritional state in 18 (25.7%), a moderate nutritional state in 25 (35.7%), and a severe malnutrition state in 13 (18.6%) (no notable differences were found between the two groups, p-value = 0.158). Only 30 (43%) were obese or overweight. Among the LC patients, 15 had osteoporosis and 49 had osteopenia at the lumbar spine (L1-L4 vertebrae), as determined by BMD via QCT. Regarding osteoporosis, no differences were found between viral and alcohol type of LC patients (p-value = 0.870), as shown in Table 2. No differences were obtained for calcium levels between patients with and without osteoporosis (p-value = 0.493). Significantly higher values were obtained for QCT score (100.17 ± 16.16 vs. 70.42 ± 9.39, p-value < 0.001) and T-score (−1.61 ± 0.57 vs. −1.94 ± 0.73, p-value = 0.035) for patients without osteoporosis, comparative to the patients with osteoporosis.
CONUT score did not correlate with BMD determined by CT (rho = 0.078, p-value = 0.520). The BMD determined by QCT was negatively correlated with age (rho = −0.645, p-value < 0.0001) and positively correlated with the varices grade (rho = 0.260, p-value = 0.030). Osteoporosis (small values of BMD) was found more significantly in patients with obesity (p-value = 0.048).
Seven potential risk factors were found to be associated with osteoporosis through univariate analysis: age, obesity, varices, grade of varices, Child-Pugh score, MELD score, and alanine aminotransferase (ALT). As in Table 3, multivariate analysis identified five significant factors associated with osteoporosis in patients with LC: age, obesity, grade of varices, Child-Pugh score, and MELD score.

Discussion
The results found in our study showed a superior prevalence of malnutrition (as evaluated with the CONUT score) for patients with a more advanced Child-Pugh level of cirrhosis. Moreover, taken separately, biochemistry markers such as high values of AST and total bilirubin corresponded to a more severe CONUT stage. The level of malnutrition was strongly correlated with liver function as determined by the MELD score. Moreover, the CONUT values were not outright related to QCT BMD data, but we found a good correlation between BMD and the presence of varices, grade of varices, age, obesity, Child-Pugh score, and MELD score, with all of them being statistically significant.
In patients with LC, it has been observed that bone loss is a frequent complication, one that can develop osteopenia levels up to 50% and osteoporosis levels ranging from 10 to 45% [32][33][34]. Among the mechanisms involved in the occurrence of bone disorders, hepatic cholestasis turned out to facilitate bone abnormalities in patients with LC, leading also to metabolic bone impairment and osteoporosis in later stages [35,36]. We also observed that high values of bilirubin and the presence of jaundice were matched up with higher stages of the CONUT index. The decrease in BMD in these patients can be caused by a set of etiopathogenic factors that mainly induce osteoblastic impairment, in spite of the fact that there also may be a certain level of osteoclastic hyperactivity [37,38]. Osteoporosis was detected in 24% to 38% of patients with end-stage disease in multiple etiologies [39][40][41]. Guichelaar et al. stated that osteoporosis was found in almost 38% of patients, osteopenia in 39%, but only 23% of patients had normal bone mass in a group of 360 patients with advanced cholestatic liver disease (primary biliary cholangitis and primary sclerosing cholangitis) [39,42]. In addition, patients with end-stage liver disease frequently suffer fractures [43]. QCT is useful in different stages of prediction, diagnosis, and prevention of osteoporosis and fractures [15,44]. Volumetric bone density measured by QCT is more sensitive to any modifications in BMD than DXA, which evaluates area bone density. Homologous findings by QCT measurement were also depicted in our study, in which low values of lumbar BMD were positively correlated with age and signs of portal hypertension such as presence of esophageal varices [17,45]. According with a previous study, our results demonstrated that the BMD measured by QCT was significantly lower in advanced stages of LC [46].
The present study attempted to identify the nutritional status among patients with LC with the use of the CONUT index in order to establish the association between malnutrition and bone loss. The results showed a correlation between a higher CONUT score and an advanced Child-Pugh classification of cirrhosis or biochemistry markers such as high values of AST and total bilirubin. A recent meta-analysis has shown that the nutritional status evaluated by the CONUT score was associated with prognosis of the liver disease [47]. Moreover, it has been found that the CONUT score was associated with the prognosis of various cancers and especially with HCC [16,[47][48][49]. The frequency of malnutrition in patients with liver disease ranges between 10% and 100%, depending largely on the characteristics of the patients and the methods of nutritional assessment performed [50]. Since the CONUT index is based on derivatives on the laboratory data by using blood samples, we easily and objectively evaluated the nutritional status of the patients [18,19]. In line with the literature, we found that malnutrition can be seen in all clinical stages but is visualized more frequently in advanced stages of liver disease [51]. Accordingly, alcoholic liver disease is more frequently related with malnutrition [52]. The prevalence found in clinical trials is between 20% for patients with compensated LC and 100% in hospitalized patients with acute alcoholic hepatitis superimposed on cirrhosis [53,54].
Our study also found a correlation between obesity and low bone mass. This involves the fact that fat and bone mass impart some environmental factors, which can relate to the risk of osteoporosis [55]. A previous analytic study indicated that higher fat mass is related to lower BMD [56].
Even though evidence that low calcium contributes to osteoporosis development is weak, there is contradictory data for calcium abnormalities in LC. We noticed no differences in calcium levels between patients with alcoholic liver disease or virally induced liver cirrhosis [57,58].
This study has also confirmed that there is a direct positive correlation between Tscore (BMD DEXA) and lumbar QCT. Furthermore, studies have confirmed that QCT is more efficacious than DEXA scan and that it also helps discriminate between groups of patients [59,60].
A higher CONUT score was related to an advanced Child-Pugh stage and MELD index with its laboratory variables being able to be considered warning signs regarding latter stages of bone disease such as osteopenia and osteoporosis [61].
Several limitations were found in the study and should be mentioned. The first reason encountered is that this is a retrospective observational study. The second one is linked to the idea that the study was based only on 70 patients diagnosed with viral B, viral C, or alcohol-induced LC cohort during pandemics, as well as additional studies on different liver diseases such as autoimmune are essentials to further confirm and extrapolate to other terms. From our thorough literature research, we encountered that there are no evident correlations between calcium levels and the severity of liver and bone disease [62]. Still, current outcomes from the research demonstrated that the CONUT score is well correlated with liver function and laboratory parameters such as Child-Pugh stage, MELD score, or AST and total bilirubin, which can be useful for predicting malnutrition as defined by the CONUT score.

Conclusions
Our findings showed a higher degree of malnutrition assessed with the CONUT index, considerably correlated with liver function as determined by the MELD score. We also found a greater prevalence of malnutrition among patients with LC. The CONUT values were not accurately related to QCT BMD data. Therefore, a strong correlation between BMD and the presence of varices, grade of varices, age, obesity, Child-Pugh score, and MELD score was found; moreover, all of them were statistically significant. Thus, bearing in mind that the CONUT has not been confirmed as a definitive marker of nutritional status in LC, assumptions about its link with liver function and bone density should be observantly drawn. Further research should contain more studies on the advantages and utility of the CONUT score for nutritional analysis in liver disease. There should also be more attention on verifying the BMD in opportunistic CT scan for osteoporosis assessment when diagnosing patients with LC.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board (or Ethics Committee) of the Emergency Hospital of Craiova (no. 173/29 October 2021).

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data used to support the findings of this study are available from the corresponding author upon reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.