Dr S. Pereira, Institute of Hepatology, RF & UCL Medical School, University College London, 69–75 Chenies Mews, London WC1E 6HX, UK. E-mail: email@example.com
Background : Retrospective studies have suggested that hormone replacement therapy may reduce the rate of bone loss in primary biliary cirrhosis, but no controlled data are available.
Methods : Forty-two post-menopausal women with primary biliary cirrhosis were treated with calcium and vitamin D, either alone (n = 21) or together with transdermal hormone replacement therapy (n = 21). Bone densitometry was performed at baseline and at 1 year, and serum and urinary markers of bone turnover were measured at three-monthly intervals.
Results : At entry, 17 patients (40%) had spinal or femoral osteopenia (T score − 1 to − 2.5) and nine (21%) had osteoporosis (T < − 2.5). In those given hormone replacement therapy, there was a significant decrease in the mean urinary deoxypyridinoline :creatinine ratios at 3 months (7.8 vs. 6.1 nm/mm creatinine for no hormone replacement therapy vs. hormone replacement therapy; P = 0.04) and a 48% reduction in urinary calcium excretion at 1 year (0.66 vs. 0.32 mm/mm creatinine; P = 0.01). Repeat bone densitometry at 1 year revealed a 2.25% increase in the hormone replacement therapy group (P = 0.02), compared with a non-significant 0.87% decrease in L2–L4 bone mineral density in those not given hormone replacement therapy. Both treatment regimens were well tolerated, with no increase in cholestasis.
Conclusions : Compared with calcium and vitamin D alone, supplemental treatment with transdermal hormone replacement therapy for 1 year improved the vertebral bone density and urinary markers of bone turnover in post-menopausal women with primary biliary cirrhosis.
Oestrogen deficiency is a well-defined risk factor for osteoporosis in post-menopausal women, and hormone replacement therapy (HRT) has been shown to inhibit bone resorption and stimulate new bone formation.1–4 If given for 5–10 years starting at the menopause, HRT halves the long-term risk of osteoporotic fractures.2,5 Although there have been no controlled trials of HRT in post-menopausal patients with primary biliary cirrhosis, bone loss occurs at twice the normal rate (2% vs. 1% per year),6 and it is likely that HRT would also reduce the fracture risk in this patient group. In a retrospective study of 16 post-menopausal women with primary biliary cirrhosis, low-dose oral or transdermal HRT was well tolerated and was associated with a significant increase in the lumbar spine bone mineral density at 1 year.7 However, there was no matched control group in this study, and the doses and duration of oestrogen therapy were not standardized. Given the high first-pass metabolism and cholestatic potential of oral oestrogens, there is also a concern about their use, even at low dose, in severe liver disease.8 In contrast with oral oestrogens, transdermal HRT maintains physiological blood oestrogen levels without the development of lithogenic bile or high concentrations of conjugated oestrogens in portal blood.9,10
The primary aim of this unblinded, prospective study was to compare the effects on the lumbar spine and femoral neck bone mineral density of a 1-year course of transdermal HRT plus standard treatment (oral calcium/vitamin D supplements and ursodeoxycholic acid) with standard treatment alone in a group of post-menopausal women with primary biliary cirrhosis. A related aim was to compare the results of liver biochemistry and serum and urinary markers of bone turnover in the two groups over the 1-year study period.
Patients and methods
Over a 15-month period, 42 patients (mean age, 59 years; range, 45–74 years) attending the Primary Biliary Cirrhosis Clinic at King's College Hospital were enrolled in the study. All but two of the patients were white; one was Asian and one was from the Middle East. All were post-menopausal, as determined by the absence of menses for at least 1 year and/or raised serum gonadotrophins. The diagnosis of primary biliary cirrhosis was based on standard biochemical and immunological criteria11 and confirmed by liver biopsy. Disease severity was graded according to liver histology12 and by the Mayo Clinic Risk Score calculated from the serum bilirubin and albumin, age, prothrombin time and oedema score.13,14 The median duration from the most recent liver biopsy to entry into the study was 4 years; 19 patients (45%) had histological stage I or II primary biliary cirrhosis, whilst the remainder had stage III or IV disease. None of the patients had a past history of fractures, adverse reactions to oestrogens or progesterone, gynaecological malignancy or thrombo-embolic disorders, and most had had a normal screening mammogram within the last 2 years.
As part of standard therapy, patients received daily ursodeoxycholic acid (10–12 mg/kg), together with calcium (1 g) and vitamin D (400 IU) supplements (Calcichew D3; Shire Pharmaceuticals, Hampshire, UK), for at least 3 months before entry into the trial. Following oral and written information about the study, patients were then given a choice of either no extra treatment or transdermal HRT patches twice weekly (HRT group). The patches delivered 50 µg of estradiol per 24 h for 2 weeks, followed by 50 µg of estradiol and 250 µg of norethisterone acetate per 24 h for the subsequent 2 weeks in each month (Estracombi; Novartis Pharmaceuticals UK Ltd, Surrey, UK). In the eight patients who had undergone a previous hysterectomy, unopposed estradiol (Estraderm TTS 50) was prescribed. A structured questionnaire and physical examination were completed at the initial visit and at subsequent clinic appointments at 3, 6 and 12 months.
The study design was approved by the Research Ethics Committee of King's College Hospital, and all patients gave written informed consent before entering the study.
A dual-emission X-ray absorptiometer (XR-26, Lunar Corporation, Madison, WI, USA) was used to measure the bone mineral density of the lumbar spine (L2–L4) and femoral neck at baseline and at 1 year. The variability in quantification of the bone mineral density at the two sites in vivo was less than 1%. Osteopenia was defined according to the World Health Organization classification as a bone mineral density between 1 and 2.5 standard deviations lower than the peak bone mass (T score of − 1 to − 2.5).15 Osteoporosis was diagnosed if the T score was less than − 2.5.
Markers of bone turnover
Serum biochemical profiles and fasting urinary concentrations of calcium and creatinine were measured at baseline and at 3, 6 and 12 months using standard colorimetric methods (Hitachi 911 autoanalyser; Roche Diagnostics, Basel, Switzerland). Urinary concentrations of free deoxypyridinoline, a pyridinium cross-link of type-1 collagen, were measured in a competitive immunoassay (Pyrilinks-D, Metra Biosystems Inc, Mount View, CA, USA), as described previously,16 and expressed as a molar ratio to urinary creatinine. Serum osteocalcin was measured using a human monoclonal solid-phase immunoassay (NovoCalcin, Metra Biosystems Inc, Mount View, CA, USA), as described previously.17 Intra- and inter-assay coefficients of variation for both assays performed in different runs were less than 7%. Serum osteocalcin levels and urinary calcium : creatinine ratios were compared with those obtained from a community sample of 111 healthy post-menopausal women (mean age, 68 years; range, 47–88 years) without a history of osteoporosis or previous fracture, who acted as controls.
Differences between group means were assessed by one-way analysis of variance. Pearson's correlation coefficients were calculated to determine the strength of relations between demographic variables, markers of bone turnover and bone mineral density. Contributing factors to differences in bone mineral density at the measured sites were identified by linear regression. The primary end-point of the study was the change in bone mineral density at 1 year, with secondary end-points being changes in urinary and serum markers of bone turnover.
The two groups were well matched for age, duration of menopause (mean, 10.7 years; range, 1–26 years), body mass index (mean, 24.2 kg/m2; range, 17.3–31.8 kg/m2), histological stage, serum bilirubin level (mean, 16.9 µm; range, 4–65 µm) and Mayo Clinic R score (mean, 3.3; range, 1.0–4.6) (Table 1). The frequency of current or previous use of corticosteroids, other immunosuppressive agents and thyroid replacement was similar in the two groups and did not correlate with the degree of bone loss.
Table 1. Clinical characteristics and indices of disease severity in the two groups.
Controls (n = 21)
HRT (n = 21)
Data reported as means ± SEM unless otherwise indicated. There were no significant differences between the two groups with respect to any of the above variables.
60.4 ± 1.6
56.4 ± 1.4
Duration of menopause (y)
10.8 ± 1.9
10.7 ± 1.5
Body mass index (kg/m2)
23.1 ± 0.7
25.3 ± 0.8
Histological stage III-IV
Serum bilirubin (μM)
17.2 ± 2.8
16.6 ± 3.1
Mayo Clinic R-score
3.4 ± 0.1
3.2 ± 0.2
Baseline bone mineral density
At entry into the study, six of the 42 patients (14%) fulfilled the World Health Organization criteria for osteoporosis (T < − 2.5) at the lumbar spine and five (12%) at the left femoral neck; two patients had osteoporosis at both sites (Figure 1). Seventeen others (40%) had osteopenia (T score of − 1 to − 2.5) of either the lumbar spine (n = 6) or femoral neck (n = 5), or at both sites (n = 6). As expected, there was a strong positive correlation between the bone mineral density values at the two sites (r = 0.547, P < 0.001). The bone mineral density also correlated positively with the body mass index of the patients (r = 0.32, P = 0.04 for the lumbar spine; r = 0.56, P < 0.001 for the femoral neck), but not with age, serum bilirubin levels or Mayo Clinic R score. There was a negative correlation between the femoral neck bone mineral density and histological stage (r = 0.45, P = 0.03).
Markers of bone turnover
At baseline, serum calcium levels adjusted for albumin concentrations were normal in all patients (mean, 2.38 mm; range, 2.19–2.60 mm). The mean (S.E.M.) osteocalcin concentration was 7.2 ± 0.4 µg/L (range, 2.8–14.9 µg/L), compared with a value of 8.3 ± 0.3 µg/L (range, 2.5–21.3 µg/L) in the 111 healthy controls (P = 0.07). In the no HRT group, the baseline serum osteocalcin level of 7.3 ± 0.5 µg/L was similar to that of the healthy controls and did not change significantly at subsequent time points. In contrast, serum osteocalcin in the HRT group decreased from 7.1 ± 0.8 µg/L at baseline to 5.7 ± 0.9 µg/L at 3 months (P = 0.003 vs. controls), with 6- and 12-month values of 6.2 ± 0.7 µg/L (P = 0.03) and 6.8 ± 0.9 µg/L (N.S.), respectively.
The mean (± S.E.M.) baseline urinary concentration of deoxypyridinoline of 7.33 ± 0.55 nm/mm creatinine was similar in the two groups, and did not change significantly with time in the patients given calcium and vitamin D alone (Figure 2a). In contrast, in the HRT group, there was a significant decrease in urinary deoxypyridinoline : creatinine excretion at 3 months, from 7.2 ± 0.8 nm/mm creatinine at baseline to 6.1 ± 0.6 nm/mm (P = 0.04). At 1 year, this level had decreased further to 5.7 ± 0.4 nm/mm, compared with a value of 7.9 ± 1.1 nm/mm in those not given HRT (P = 0.06). The mean baseline urinary calcium : creatinine concentration was 0.50 ± 0.05 mm/mm creatinine, compared with a value of 0.40 ± 0.0 mm/mm in the 111 healthy controls (P = 0.18). There was a significant decrease in urinary calcium : creatinine excretion with time in the HRT group, from 0.47 ± 0.06 mm/mm creatinine at baseline to 0.32 ± 0.04 mm/mm at 1 year (P = 0.03) (Figure 2b). This compared with a 1-year value in the no HRT group of 0.66 ± 0.13 mm/mm (P = 0.01 vs. HRT group), which was also significantly higher than that in the controls (P = 0.02). At baseline, urinary deoxypyridinoline and calcium correlated highly with each other (r = 0.565, P < 0.001), but not with serum calcium or osteocalcin.
Changes in bone mineral density
Thirty-nine of the 42 patients underwent repeat bone densitometry at 1 year. Of the three who did not complete the study, two withdrew for reasons unrelated to the treatment protocol and one was lost to follow-up.
At baseline, the two groups were well matched with respect to the bone mineral density of the spine and femoral neck. At 1 year, in the HRT group, there was a 2.25% increase in the bone mineral density of the lumbar spine (P = 0.02) and a non-significant 0.69% increase in femoral neck bone mineral density. In contrast, in those who received only calcium and vitamin D supplementation, there was a 0.87% decrease in L2–L4 bone mineral density (P = 0.05) and a non-significant 0.16% increase in femoral neck bone mineral density. In the HRT group, the lumbar spine change from baseline differed significantly from that in the no HRT group. At 1 year, the difference between the groups was 3.12% (95% CI, 0.98–6.02%; P < 0.01) at the lumbar spine and 0.53% (95% CI, − 3.22% to 1.73%; N.S.) at the femoral neck (Table 2).
Table 2. Changes in bone mineral density of the spine and femoral neck after treatment with calcium and vitamin D alone (controls), or together with transdermal HRT.
Both treatment regimens were well tolerated. There were no adverse events attributable to treatment, apart from two patients who stopped HRT because of monthly bleeding and declined continuous combination therapy. Throughout the study period, liver biochemistry, including serum bilirubin levels, did not change significantly in either group. Nine of the 21 patients in the HRT group (43%) had raised serum bilirubin levels before treatment (mean, 29.6 ± 4.9 µm); the corresponding values at 3 and 12 months were 28.9 ± 6.4 µm and 24.4 ± 5.6 µm, respectively (N.S.).
Low bone mass is an important cause of morbidity in patients with primary biliary cirrhosis, resulting in an increased risk of fractures of the long bones and vertebrae, pain and deformity.18,19 Although the pathogenesis of low bone mass in primary biliary cirrhosis is poorly understood, iliac crest biopsies taken from patients with primary biliary cirrhosis at the time of liver transplantation showed evidence of increased bone resorption, with reduced or normal bone formation rates.6,20,21 Consistent with these findings, we previously reported a significant increase in urinary markers of bone resorption in 36 women with primary biliary cirrhosis, compared with 124 age- and sex-matched healthy controls, and an inverse correlation between total bone mineral content and primary biliary cirrhosis stage.22 In that study, bone densitometry revealed spinal osteopenia or osteoporosis in 58% of patients. Vitamin D deficiency may also play a role in the development of bone disease in cholestatic liver disease. However, histological osteomalacia is rare,23 and we22 and others23–26 have shown that serum vitamin D levels are generally normal in primary biliary cirrhosis.
The present study of 42 post-menopausal women with primary biliary cirrhosis confirms our previous findings, in a separate cohort of patients,22 of a high frequency of low bone mass, with spinal or femoral osteopenia or osteoporosis being present in 61% of patients. There was also a negative correlation between the femoral neck bone mineral density and disease severity, as reported by some6,26,27 but not all28,29 other investigators. Possible reasons for the lack of an association reported in some studies may be the small patient numbers, resulting in a type II error, or the effects of liver disease severity on bone mineral density being masked by other independent risk factors, such as age, hormonal status and low body mass index. In a large study of 176 patients with primary biliary cirrhosis, patients with histological stage III or IV disease had a 5.4-fold (95% CI, 2.0–14.5) increased risk of osteoporosis compared with patients with stage 1 or 2 disease.26 Similarly, in an earlier study of 210 women with primary biliary cirrhosis, the lumbar spine bone mineral density was inversely related to a risk score index of liver disease severity.6 Consistent with the results of previous studies,26,30 we also found that a low body mass index was an independent risk factor for reduced bone mineral density.
In patients with primary biliary cirrhosis, baseline serum osteocalcin levels were lower, and urinary concentrations of deoxypyridinoline and calcium were higher, than in healthy post-menopausal controls, consistent with both reduced new bone formation and increased bone resorption in these patients. We did not look for allelic polymorphisms of the vitamin D receptor or collagen genes, which may be genetic markers of peak bone mass in primary biliary cirrhosis,31,32 but it is very unlikely that any of our patients would have had osteomalacia as calcium and vitamin D supplements were given both before and during the study. The patients were also routinely given ursodeoxycholic acid for their liver disease. Ursodeoxycholic acid has been reported to enhance fractional calcium absorption in primary biliary cirrhosis,33 but, according to the results of one randomized controlled trial with a 3-year follow-up,34 does not reduce the rate of bone loss in primary biliary cirrhosis.
There are few treatment modalities which have been shown in adequately designed trials to reduce or reverse bone loss in primary biliary cirrhosis. In healthy post-menopausal women, weight-bearing exercise and oral calcium supplements are of some benefit,35 but, in primary biliary cirrhosis, the effect of calcium supplementation on bone mineral density is modest.36,37 In the present study, treatment with calcium and vitamin D for 1 year did not prevent a significant 0.87% decrease in L2–L4 bone mineral density, and urinary markers of bone turnover remained unchanged. In contrast, transdermal HRT resulted in a significant 2.25% increase in the lumbar spine bone mineral density, and an associated decrease in the urinary excretion of deoxypyridinoline and calcium. There was also a non-significant 0.53% increase in the femoral neck bone mineral density over those not given HRT. In HRT trials in healthy post-menopausal women, alterations in cortical bone mineral density may take longer than 2–3 years before a significant change is seen, as the turnover of cortical bone is slower than that of trabecular bone. The precision of bone densitometry at the hip is also poorer than that at the spine, so that longer follow-up may be needed to see significant changes.1,3
This prospective controlled trial of transdermal HRT in primary biliary cirrhosis confirms the efficacy of HRT suggested by previous uncontrolled studies. In a retrospective study of 203 women with primary biliary cirrhosis, 16 post-menopausal patients treated with low-dose oral or transdermal HRT showed a significantly greater lumbar spine bone mineral density at 1 year than 91 post-menopausal women who received no oestrogen supplementation.7 In a more recent study of nine primary biliary cirrhosis patients with osteoporosis and one with osteopenia, who were given HRT for 2 years, the lumbar spine bone mineral density increased, whereas 10 age-matched untreated patients without osteoporosis showed a significant decrease in bone mineral density over the same time period.38 Similarly, in a recently published study of 46 unselected post-menopausal women with primary biliary cirrhosis receiving oestrogen replacement therapy, there was a significantly lower rate of bone loss than in age-matched untreated women with primary biliary cirrhosis, without worsening cholestasis.39 However, in another study of five women with primary biliary cirrhosis, two who were profoundly jaundiced before treatment experienced increases in serum bilirubin levels after 2–3 months of oral oestrogens, and treatment was stopped.8 In the present study, no such complications occurred, although, before treatment, nine of the 21 patients given transdermal HRT had raised serum bilirubin levels and one was clinically jaundiced (65 µm). This lack of cholestatic effect was not unexpected, as transdermal HRT exhibits substantially lower effects on hepatic markers of oestrogen action than does oral therapy.9,10 Nevertheless, as the reported experience suggests that women who are already jaundiced are most at risk during HRT, we have previously recommended that serum bilirubin and liver enzymes should be monitored after 1 month of treatment and regularly thereafter in this sub-group.40
One limitation of the present study is that we did not randomize patients to the two treatment arms, as many of our patients had strong opinions about HRT and declined initial attempts at randomization. Nevertheless, the two groups were well matched at baseline. However, we did not assess other potential risk factors for osteoporosis, such as smoking, diet or physical activity. Furthermore, bone densitometry was repeated after only 1 year of follow-up, although the results of large community studies have indicated that the initial beneficial effect of HRT on bone mineral density is maintained throughout treatment.41–43 Although our study was not designed to detect fractures, prospective community studies have shown that the risk of fracture increases two- to three-fold for each standard deviation decrease in bone mineral density.44 In one study of 75 post-menopausal women with one or more osteoporotic fractures, who were randomized to receive either transdermal estradiol or placebo, the median annual percentage improvement in bone mineral density in the oestrogen group was 5.3% at the lumbar spine, with a decrease in vertebral fractures of 48%.1
In a national survey undertaken in 1996 of the prescribing habits of 454 UK gastroenterologists seeing an estimated 4337 patients with primary biliary cirrhosis, 95% prescribed ursodeoxycholic acid, but only 15 doctors routinely gave calcium supplements and only one prescribed HRT.45 Consensus guidelines on the management of osteoporosis associated with chronic liver disease have been published recently, which recommend bone mineral density screening of all patients with cirrhosis or severe cholestatic liver disease, oral supplementation with calcium and vitamin D, and transdermal HRT as first-line therapy for women with established osteoporosis.46 Further controlled studies are needed to determine to what extent transdermal HRT improves bone mass and reduces fracture risk in this patient group in the long term. However, until further data are available, we recommend that transdermal HRT should be considered in all post-menopausal women with primary biliary cirrhosis at an increased risk of fracture (based on risk factor assessment and bone mineral density), as well as in those at any age who have primary or secondary amenorrhoea, or who have undergone oophorectomy. The cholestatic potential of transdermal HRT in primary biliary cirrhosis appears to be low.