Osteoporosis International

, Volume 23, Issue 4, pp 1481–1487

Patients with cirrhosis and ascites have false values of bone density

Implications for the diagnosis of osteoporosis

Authors

    • Metabolic Bone Diseases Unit, Service of Rheumatology, Hospital Clínic, IDIBAPSUniversity of Barcelona
    • Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y DigestivasCIBERehd
    • Service of RheumatologyHospital Clínic
  • A. Monegal
    • Metabolic Bone Diseases Unit, Service of Rheumatology, Hospital Clínic, IDIBAPSUniversity of Barcelona
  • A. Muxi
    • Service of Nuclear Medicine, Hospital Clínic, IDIBAPSUniversity of Barcelona
  • A. Martinez-Ferrer
    • Metabolic Bone Diseases Unit, Service of Rheumatology, Hospital Clínic, IDIBAPSUniversity of Barcelona
  • R. Reyes
    • Metabolic Bone Diseases Unit, Service of Rheumatology, Hospital Clínic, IDIBAPSUniversity of Barcelona
  • J. Caballería
    • Liver Unit, Hospital Clínic, IDIBAPSUniversity of Barcelona
    • Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y DigestivasCIBERehd
  • L. Del Río
    • CETIR Centre Médic
  • P. Peris
    • Metabolic Bone Diseases Unit, Service of Rheumatology, Hospital Clínic, IDIBAPSUniversity of Barcelona
    • Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y DigestivasCIBERehd
  • F. Pons
    • Service of Nuclear Medicine, Hospital Clínic, IDIBAPSUniversity of Barcelona
  • A. Parés
    • Liver Unit, Hospital Clínic, IDIBAPSUniversity of Barcelona
    • Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y DigestivasCIBERehd
Original Article

DOI: 10.1007/s00198-011-1756-1

Cite this article as:
Guañabens, N., Monegal, A., Muxi, A. et al. Osteoporos Int (2012) 23: 1481. doi:10.1007/s00198-011-1756-1

Abstract

Summary

The effect of ascites on bone densitometry has been assessed in 25 patients with advanced cirrhosis, and it was concluded that ascites over 4 l causes inaccuracy of BMD measurements, particularly at the lumbar spine. This fact must be considered when assessing bone mass in patients with decompensated cirrhosis.

Introduction

Bone mineral density (BMD) measured by dual-energy x-ray absorptiometry (DXA) is the best procedure for assessment of osteoporosis and fracture risk, but BMD values at the central skeleton may be influenced by changes in soft tissues. Therefore, we have studied the effect of ascites on BMD.

Methods

BMD was measured by DXA at the lumbar spine, femoral neck and total hip, just before and shortly after therapeutic paracentesis in 25 patients with advanced liver cirrhosis. Changes in BMD, lean and fat mass, abdominal diameter and weight, as well as the amount of removed ascites were measured.

Results

The amount of drained ascites was 6.6 ± 0.5 l (range: 3.0 to 12.7 l). After paracentesis, BMD increased at the lumbar spine (from 0.944 ± 0.035 to 0.997 ± 0.038 g/cm2, p < 0.001) and at the total hip (from 0.913 ± 0.036 to 0.926 ± 0.036 g/cm2, p < 0.01). Patients with a volume of drained ascites higher than 4 l showed a significant increase in lumbar BMD (7.0%), compared with patients with a lower amount (1.5%) (p < 0.03). The decrease in total soft tissue mass correlated with the amount of removed ascites (r = 0.951, p < 0.001). Diagnosis of osteoporosis or osteopenia changed after paracentesis in 12% of patients.

Conclusion

Ascites over 4 l causes inaccuracy of BMD measurements, particularly at the lumbar spine. This fact must be considered when assessing bone mass in patients with advanced cirrhosis.

Keywords

DensitometryLiver diseaseLiver transplantationOsteopeniaParacentesis

Abbreviations

BMD

Bone mineral density

DXA

Dual-energy x-ray absorptiometry

LSC

Least significant change

Introduction

Low bone mass is a frequent complication of chronic liver diseases, mainly in patients with advanced cholestatic liver diseases and alcohol abuse, and in cirrhotic patients [1]. It has been reported that up to 26% of patients with end-stage liver disease have osteoporosis through densitometric criteria, with low bone mineral density (BMD) values at the lumbar spine and at the femur [2]. Also, medical attention has been paid to the rapid and severe bone loss following liver transplantation [3]. In fact, transplant recipients have a high prevalence of fractures with important rates of bone loss during the first months [4]. Both previous fractures and pretransplant low bone mass have been considered to be the main risk factors for developing fractures after liver transplantation [4, 5].

Taking into consideration all these facts, it is important to have reliable methods to assess the risk factors for fracture, the hallmark of osteoporosis. Up to now, the measurement of BMD by central dual-energy x-ray absorptiometry (DXA) provides the best assessment of fracture risk. Although if a patient suffers a fracture, depending on various factors such as advanced age, low body mass index or previous fracture, among others, it is accepted that the lower the BMD the more at risk the patient is of suffering a fracture [6]. Furthermore, there is plenty of evidence that bone densitometry assessed by DXA has several inaccuracies and errors that depend not only on the instrument, but also on the patient's characteristics, which may partially explain inconsistencies between BMD and bone fragility resulting in fractures [7].

Several problems may arise with the use of DXA. Thus, accuracy, which is how close the BMD measured by densitometry is to the actual calcium content of the bone, may be influenced by variable soft tissue densities, and inaccuracies may occur as a result of assumptions about body composition and soft tissues [6]. However, few studies with the aim to analyze the accuracy of DXA in changes of soft tissue mass have been performed. Most of these studies are focused on changes in DXA values during weight change [810] or on the influence of peritoneal fluid in patients receiving peritoneal dialysis [11]. In this respect, ascitic fluid may result in inaccuracies of BMD values obtained in patients with severe liver disease. Thus, two studies have analyzed the effect of ascites on BMD values with contradictory results. Labio et al. found that ascites falsely decreases BMD measurements, particularly at the lumbar spine [12], while Haderslev et al. reported nonsignificant differences in lumbar BMD after paracentesis [13].

We performed a prospective study on a group of end-stage cirrhotic patients, just before and shortly after therapeutic paracentesis, aimed at analyzing the effect of ascites on BMD measurements, and if ascites actually do affect BMD, what the involved factors are.

Patients and methods

Patients

The study was carried out in 25 consecutive patients, 19 men and 6 women, aged 63.8 ± 2.4 (range: 45–81 years) diagnosed with end-stage liver cirrhosis. The population consisted of 17 patients with cirrhosis of viral etiology, six with alcoholic cirrhosis and two patients with cirrhosis of other etiologies. All women were postmenopausal. Ambulatory patients were eligible for the study at the Liver Unit day hospital if they required therapeutic paracentesis. The ethical committee at the Hospital Clínic approved the study protocol and informed consent was obtained from all subjects.

Study procedures

Immediately before and shortly after therapeutic paracentesis, patients' assessment was performed by measuring body weight and body mass index, as well as abdominal perimeter and diameter (obtained from the distance between the arm of the absorptiometer and the umbilicus). Ascitic fluid volume and characteristics were also recorded. At the end of paracentesis, 8 g of albumin per liter of ascitic fluid removed was administered intravenously to prevent relative hypovolemia.

BMD was measured at the lumbar spine (L1–L4) in all patients and at the proximal femur (femoral neck and total hip) in 24 patients (one patient was excluded because of technical problems) by dual-energy x-ray absorptiometry (DXA) (Lunar Prodigy, Madison, WI, USA). The coefficients of variation (CV) in our center for lumbar spine and total hip are 0.80% and 0.61%, respectively. The same scan mode was always used for the pre- and postparacentesis measurements. Osteopenia and osteoporosis were diagnosed using the World Health Organization thresholds: T-score between −1 and −2.5 and ≤−2.5, respectively [14]. The least significant change (LSC) for BMD was calculated as follows: \( {\text{CV}} \times {1}{.96} \times \sqrt {2} \) [15]; in our center, LSCs for lumbar spine and total hip BMD are 2.2% and 1.7%, respectively. Also, total body was scanned by DXA and the manufacturer's software was used to discriminate body composition into three compartments: bone, fat and lean soft tissue.

Statistical analysis

All parameters were checked for normality using the Kolmogorov–Smirnov test. All values are expressed as mean ± SEM (standard error of the mean). Between-group comparisons were performed using the Student's t-test or the Mann–Whitney test, when appropriate. Differences between two repeated measurements were analyzed by Student's t-test for paired observations. Differences between proportions were assessed by the chi-square test and correlations were calculated by Pearson analysis. Values of p < 0.05 were considered significant. Statistical analyses were performed by using SPSS 16.0 software.

Results

The average volume of ascitic fluid removed was 6.6 ± 0.5 l (range 3 to 12.7 l) and the mean interval between first and second BMD measurements was 2.9 h (range 1.4 to 8.2). Physical and densitometric data, before and after therapeutic paracentesis, are expressed in Table 1.
Table 1

Physical and densitometric values before and after paracentesis

 

Before

After

p

Height (cm)

162.5 ± 2.1

  

Weight (kg)

71.8 ± 2.2

65.1 ± 2.1

<0.001

BMI (kg/m2)

25.8 ± 0.6

24.3 ± 0.6

<0.001

Abdominal diameter (cm)

26.8 ± 0.5

21.5 ± 0.5

<0.001

Abdominal perimeter (cm)

108.7 ± 2.1

98.9 ± 1.6

<0.001

Lumbar spine (L1–L4)

BMD (g/cm2)

0.944 ± 0.035

0.997 ± 0.038

<0.001

T-score

−2.25 ± 0.30

−1.84 ± 0.31

<0.001

Z-score

−1.47 ± 0.30

−0.90 ± 0.32

<0.001

Femoral neck BMD (g/cm2)

0.872 ± 0.032

0.871 ± 0.033

n.s.

T-score

−1.55 ± 0.24

−1.57 ± 0.23

n.s.

Z-score

−0.19 ± 0.21

−0.05 ± 0.21

<0.01

Total hip BMD (g/cm2)

0.913 ± 0.036

0.926 ± 0.036

<0.01

T-score

−1.2 ± 0.27

−1.19 ± 0.27

n.s.

Z-score

−0.31 ± 0.25

−0.10 ± 0.25

<0.01

Total body BMD (g/cm2)

1.112 ± 0.033

1.088 ± 0.033

<0.001

After therapeutic paracentesis, BMD significantly increased at the lumbar spine and total hip (Table 1 and Fig. 1). BMD at the lumbar spine increased from 0.944 ± 0.035 to 0.997 ± 0.038 g/cm2 (p < 0.001), with a mean percent change of 5.7%. The total hip BMD increased from 0.913 ± 0.036 to 0.926 ± 0.036 g/cm2 (p < 0.01) by an average of 2.2%. BMD at the femoral neck was not significantly changed (from 0.872 ± 0.032 to 0.871 ± 0.033 g/cm2, p = n.s.), while total body BMD significantly decreased by 2.2% (from 1.112 ± 0.033 to 1.088 ± 0.033 g/cm2, p < 0.001). After paracentesis, 80 and 40% of patients showed an increase in lumbar and total hip BMD higher than the LSC, respectively. Interestingly, only patients with a volume of drained ascitic fluid higher than 4 l showed a significant increase in lumbar BMD. Thus, lumbar BMD increased by 7.0% in patients with drained ascitic fluid greater than 4 l, as compared with 1.5% in patients with a lower amount (p < 0.03), without significant changes at the total hip (Table 2, Fig. 2). The amount of drained ascites resulted in a corresponding decrease in the abdominal diameter (r = 0.792, p < 0.001), and consequently, patients with an abdominal diameter higher than 27 cm., which were those with more ascites, showed a higher increase in lumbar BMD than patients with a lower abdominal diameter (9.0 vs. 3.8%, p < 0.03) after paracentesis. By contrast, changes in BMD were unrelated to baseline BMD values or to the presence of osteopenia or osteoporosis at the lumbar spine or total hip. In three patients (12%), the overall BMD diagnostic category (taking into account the lowest T-score value at the spine and proximal femur) changed after paracentesis from osteopenia to normal in two patients and from osteoporosis to osteopenia in one case. Changes in categories at the lumbar spine and total hip are expressed in Fig. 1.
https://static-content.springer.com/image/art%3A10.1007%2Fs00198-011-1756-1/MediaObjects/198_2011_1756_Fig1_HTML.gif
Fig. 1

Bone mineral density at the lumbar spine (a) and total hip (b) before and after paracentesis. The dashed lines indicate the cutoff for osteopenia and osteoporosis, according to the T-score. Patients with ascites removal <4 l are indicated in open circles

Table 2

Percent changes in physical parameters and densitometry according to the volume of removed ascites

 

Ascitis removed

≤4 l N = 6

>4 l N = 19

p

Weight

−5.9 ± 0.7

−10.4 ± 0.7

<0.01

Abdominal diameter

−11.7 ± 1.1

−22.3 ± 2.0

<0.01

Abdominal perimeter

−4.0 ± 1.1

−10.1 ± 1.9

n.s.

Lumbar spine BMD

1.5 ± 1.3

7.0 ± 1.2

<0.03

Femoral neck BMD

0.4 ± 1.7

−0.2 ± 0.6

n.s.

Total hip BMD

2.0 ± 1.3

1.4 ± 0.5

n.s.

Total body BMD

−2.1 ± 0.3

−2.2 ± 0.4

n.s.

https://static-content.springer.com/image/art%3A10.1007%2Fs00198-011-1756-1/MediaObjects/198_2011_1756_Fig2_HTML.gif
Fig. 2

BMD percent changes after paracentesis at the lumbar spine and total hip, according to the volume of ascites removed. The grey shaded area indicates the least significant change for BMD at both sites (2.2% at the lumbar spine and 1.7% at the total hip)

Both total BMC and skeleton-projected area significantly increased after paracentesis, thus resulting in a decrease in total body BMD values, which changed from 1.112 ± 0.15 to 1.087 ± 0.15 g/cm2 (p < 0.001). Also, total fat tissue mass (FTM) significantly changed from 16.1 ± 6.1 to 15.4 ± 6.1 kg (p = 0.04) and total lean tissue mass (LTM) decreased from 51.66 ± 8 to 46.7 ± 8 kg (p < 0.001). Taken together, the soft tissue mass (lean plus fat) decreased from to 69.7 ± 2.2 to 63.1 ± 2.1 kg. This decrease was closely correlated with the amount of ascitic fluid removed (r = 0.951, p < 0.001) (Fig. 3). Significant correlations were also observed between the changes in the trunk lean, fat and soft tissue mass with ascites volume. By contrast, the volume of drained ascites did not correlate with changes in total (r = 0.118) and lumbar BMC (r = −0.311).
https://static-content.springer.com/image/art%3A10.1007%2Fs00198-011-1756-1/MediaObjects/198_2011_1756_Fig3_HTML.gif
Fig. 3

Correlation between the volume of ascites removed and the changes in the soft tissue mass (lean plus fat) (r = 0.951, p < 0.001)

Discussion

The results of this study show that ascitic fluid falsely reduces BMD values at the lumbar spine and total hip. The finding of this artefact, resulting in errors in BMD measured by central DXA, must be considered when assessing bone mass in cirrhotic patients with ascites in advanced or end-stage liver disease. Indeed, the BMD diagnostic category changed in 12% of the cases after removing the ascitic fluid, thus supporting the relevance of the presence of a large volume of fluid in the peritoneal cavity in BMD measurements at the lumbar spine. This represents an additional artefact for a patient-specific BMD measurement, since it is clear that DXA densitometry is associated with several errors and inaccuracies [7], although it is an established and internationally accepted procedure for the diagnosis of osteoporosis and osteopenia [14].

Because of the high prevalence of osteoporosis and osteopenia in patients awaiting liver transplantation [2, 16], BMD evaluation at the spine and hip has been recommended before transplantation [3, 17], and according to our results, a lower BMD than the actual value may be found. Furthermore, our findings also suggest that lumbar BMD changes after liver transplantation, (spontaneous or due to treatment) when compared to pretransplant BMD values in end-stage cirrhotic patients [4, 1820], may be less than reported if patients had a high volume of ascites. Unfortunately, the presence of ascites and its volume are not reported in the articles focused on this topic, although most patients were included in Child–Pugh B and C scores, and in these cases, they frequently have ascites.

The BMD change is relevant, particularly at the lumbar spine, since it is higher than the LSC in BMD (the smallest detectable difference that is related to the precision error of the machine) reported for most DXA scanners. Indeed, the LSCs in our center for lumbar spine and total hip BMD are 2.2% and 1.7%, respectively, and up to 80% of patients showed an increase greater than the LSC in BMD at the lumbar spine after paracentesis. In addition, a mean increase in BMD of 5.7% and 2.2% at the lumbar spine and total hip is similar to that observed in trials with bisphosphonates or denosumab in postmenopausal osteoporosis during 1 year of treatment [2124].

Our results are partly in agreement with those reported by Labio et al., who found increases in the spine BMD of 4.2%, and 1.3% in total hip BMD after therapeutic paracentesis in 20 cirrhotic patients, although they didn't find significant changes in total body BMD [12]. Both changes of lower magnitude in lumbar spine and total hip and the nonsignificant changes in total body BMD may be influenced by the longer interval between pre- and postparacentesis BMD measurements, and therefore result in a redistribution of retained sodium and water. It should be taken into account that the machine used for measuring BMD might have some influence on the results, since manufacturers of densitometry equipment have different algorithms for correcting BMD measurements for soft tissue and thickness. Therefore, we cannot validate that similar studies on densitometers from manufacturers other than the one used in the current study would lead to the same findings.

When assessing the mechanisms for the false reduction in BMD values induced by ascites, it has been difficult to ascertain the main contributory factor. However, an interesting finding in our study was that changes in lumbar BMD measured by DXA were observed only when the amount of drained ascitic fluid was higher than 4 l, indicating that when the volume of ascitis is low, the changes in spinal BMD are not relevant enough. This information is very significant when assessing BMD in patients with advanced cirrhosis and ascites, since the volume of accumulated fluid in the abdominal cavity comes out with imprecise results, but only in patients with ascites above 4 l. In the other patients with lower ascites volume, BMD measurements at the lumbar spine are more accurate. Certainly in these patients changes in BMD are within the least significant change, as seen in Fig. 2. In addition, it has been reported that BMD measurements with dual photon absorptiometry, using 153Gd as the source in patients with an abdominal diameter higher than 27 cm are unreliable [25]. Indeed, our study supports this observation, since a baseline abdominal diameter higher or lower than 27 cm was associated with different behavior in BMD results at the spine, reinforcing the fact that the volume of ascitic fluid is the main determinant of inaccuracies in BMD measurements at the central skeleton. The results concerning the increased BMD in the total hip are less important, since the percent increase in BMD is within the LSC expected at this site in our center. Moreover, it seems reasonable to observe changes, mainly at the lumbar spine and not at the hip, because of the accumulation of fluid in the abdomen.

DXA allows quantification of bone mineral, fat and lean soft tissues. However, an unresolved question is the influence of hydration on DXA soft tissue component measurements [26]. Indeed, Peng et al. found that hydration, measured by accurate direct methods, was increased in 65% of cases in a large series of cirrhotic patients. In addition, hydration indexes were positively correlated with Child–Pugh scores and were significantly higher in patients with ascites than in those without [27]. These observations support the overhydration of our cirrhotic patients, in addition to the accumulation of fluid in the peritoneal cavity, further inducing errors in DXA soft tissue composition estimates. In this sense, Pietrobelly et al. suggested that DXA is prone to fat estimation errors related to variation in soft tissue hydration [26]. Taking into account that ascites consists of a large amount of fluid, it is not surprising that its evacuation poses an error of large magnitude in the accuracy of the DXA technique, particularly at the lumbar spine. Certainly, the volume of ascites removed was closely associated with the diminution in soft tissue mass, including lean and fat mass, but not with the bone mass, which increased mainly at the lumbar spine.

Body composition changes, in both obesity and after voluntary weight loss, represent other conditions where potential sources of error on DXA accuracy may occur. However, changes in body composition after weight loss are slower than after therapeutic paracentesis, and adipose tissue instead of fluid accumulation is the main causative factor for DXA inaccuracies in measuring BMD. Studies assessing the influence of substantial weight loss on BMD have found contradictory results, a decrease in bone mass being the most frequent finding [10, 28, 29]. However, almost none of these studies have paid attention to the methodological problems with DXA accuracy. In this sense, isolated studies suggested that an alternative explanation for the discrepancies and contradictory results in BMD is that DXA is affected by changes in soft tissues, resulting in inaccurate BMD measurements in bone mass during weight loss [10, 29, 30].

In conclusion, ascitic fluid induces inaccuracies in BMD values, especially at the lumbar spine when its amount is higher than 4 l. This observation is of great value when the bone mass in patients with end-stage liver disease or its evolution in patients undergoing liver transplantation is considered. Therefore, in cirrhotic patients with ascites a reliable DXA densitometry should only be assessed after draining the liquid from the abdomen. Bone densitometry measurement in other sites, such as the forearm, could be an additional site.

Acknowledgments

This study was supported, in part, by Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBERehd), Instituto de Salud Carlos III, and FIS-08/0105, Ministerio de Ciencia e Innovación, Spain.

Conflicts of interest

None.

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2011