Fractures and avascular necrosis before and after orthotopic liver transplantation: Long-term follow-up and predictive factors

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

With early posttransplant bone loss, orthotopic liver transplantation (OLT) recipients experience a high rate of fracturing and some avascular necrosis (AVN), but little is known about the incidence of and predictive factors for these skeletal complications. We studied 360 consecutive patients who underwent transplantation for primary biliary cirrhosis (PBC) and primary sclerosing cholangitis (PSC) and assessed both vertebral and nonvertebral (rib, pelvic, and femur) fractures in a protocolized fashion. Before OLT, 20% of the patients had experienced fracturing, and 1.4% of the patients had experienced AVN. Following OLT, there was a sharp increase in fracturing, with a 30% cumulative incidence of fractures at 1 year and 46% at 8 years after transplantation. In contrast to previous studies, there was a similar incidence of posttransplant vertebral and nonvertebral fractures. The greatest risk factors for posttransplant fracturing were pretransplant fracturing and the severity of osteopenia and posttransplant glucocorticoids. Nine percent of the liver recipients experienced AVN after OLT, and this correlated with pretransplant and posttransplant lipid metabolism, bone disease (bone mineral density and fracturing), and posttransplant glucocorticoids. A novel association between cholestasis and AVN was also identified, the mechanism for which is not known. Conclusion: Fortunately, recent years have seen an increase in the bone mass of liver recipients and, along with this, less fracturing and less AVN. Nonetheless, 25% of patients undergoing OLT for chronic cholestatic liver disease still develop de novo fractures after OLT; this situation demands an ongoing search for effective therapeutic agents for these patients. (HEPATOLOGY 2007.)

Following orthotopic liver transplantation (OLT), bone mass is rapidly lost during the early postoperative months. Although the period of bone loss is short, the extent of bone loss in already osteopenic patients leads to an increase in fracturing. Pain and immobility from skeletal complications cause morbidity in liver transplant recipients,1–8 especially in patients with chronic cholestatic liver disease (CCLD). This results most frequently from osteopenic fracturing but also, to a lesser extent, from avascular necrosis (AVN); the latter is regarded as a different etiologic entity than osteopenic fracturing and is perhaps related to insults to the vascular supply of bone, including trauma, hyperlipidemia, and glucocorticoid use.9–14 The incidence of and risk factors for posttransplant fractures and AVN are poorly defined.

We have recently confirmed, in a large population of liver transplant recipients undergoing OLT for CCLD, that bone mass is rapidly lost during the first 4 postoperative months but then starts to increase and may reach or surpass pretransplant levels by 2-3 years after OLT.15 This study aims to assess the incidence and predictive variables for pretransplant and posttransplant fractures (vertebral and nonvertebral) and AVN in this same population of liver transplant recipients followed from the pretransplant period to 8 years after OLT.

Abbreviations

25(OH)D, 25-hydroxy-vitamin D; AVN, avascular necrosis; BMD, bone mineral density; BMD-LS, bone mineral density at the lumbar spine; BMI, body mass index; CCLD, chronic cholestatic liver disease; CI, confidence interval; CTP, Child-Turcotte-Pugh; CyA, cyclosporine A; HR, hazard ratio; HRT, hormone replacement therapy; INR, international normalized ratio; MELD, model for end-stage liver disease; NS, not significant; OLT, orthotopic liver transplantation; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; PTH, parathyroid hormone.

Patients and Methods

Patients.

From 1985-2001, all adult patients undergoing OLT at the Mayo Clinic (Rochester, MN) with either primary biliary cirrhosis (PBC) or primary sclerosing cholangitis (PSC) were studied by a protocolized assessment before and after OLT (4 months, annually). Patients who underwent transplantation were sequentially assigned an OLT number. Diagnoses of PBC and PSC were made according to well-established criteria.16–18 The study was approved by the Mayo Institutional Review Board. All patients were followed until July 2002, death, or retransplantation.

Clinical Assessment.

Patients underwent a clinical and biochemical assessment at each time of evaluation. The liver function was assessed with Child-Turcotte-Pugh (CTP) and model for end-stage liver disease (MELD) scores, and the functional status was assessed with the Karnofsky performance scale.19 The menopausal status was determined by clinical symptoms and biochemical testing. The nutritional status, muscle wasting, complications of liver disease, and other illnesses were noted, and medications were recorded. Muscle wasting was assessed globally by the transplant hepatologist. General dietary instructions and oral calcium supplements (1.5 g/day) were given to all patients with vitamin D supplementation to normalize serum 25-hydroxy-vitamin D [25(OH)D] levels. Inflammatory bowel disease was diagnosed by colonoscopy and surveillance biopsies at the time of activation for OLT and annually thereafter in all PSC patients.

Immunosuppression.

From 1985-1990, standard immunosuppression was cyclosporine A (CyA) and prednisone with or without azathioprine. From 1990-1993, patients were treated with tacrolimus and prednisone (multicenter FK506 trial) or the standard triple therapy with prednisone, CyA, and azathioprine; the tacrolimus patients received only about half the total prednisone dose of CyA arm. In 1994, standard immunosuppression was changed to tacrolimus, prednisone, and azathioprine, whereas some patients were treated in the multicenter Neoral trial. In 1999, azathioprine was replaced by mycophenolate mofetil for the first 2-4 postoperative months. The CyA and tacrolimus doses were adjusted according to the desired serum levels. Acute cellular rejection of the allograft was treated with 1 g of intravenous Solu-Medrol on 3 alternate days.

Biochemical Indices.

Biochemical testing included markers of liver and kidney function and indices of calcium metabolism and was performed by Mayo medical laboratories. Serum 25(OH)D was measured by the method of Kao and Heser.20 Immunoreactive parathyroid hormone (PTH) has been measured by an immunochemiluminescent metric assay21 only since June 1989.

Measurements of the Bone Mineral Density at the Lumbar Spine (BMD-LS).

BMD-LS measurements were taken at the time of activation for OLT, 4 months after OLT, 1 year after transplantation, and yearly thereafter. Before April 1988, the bone mineral density (BMD) was measured by dual-photon absorptometry. Since April 1988, dual-energy X-ray absorptometry using Hologic machines has been used. Phantoms were used to cross-calibrate the 2 machines, and a conversion formula was established to convert to dual-energy X-ray absorptometry data. The bone mass was corrected for the bone size to calculate BMD (g/cm2). Measurements of BMD-LS had a reproducibility of 2.2%. In patients with compression fractures, measurements were determined on only the intact vertebrae. BMD readings were expressed as T-scores (standard deviations from the peak bone mass of a young, sex-matched reference population) and Z-scores (standard deviations from age-adjusted and sex-adjusted reference values). A T-score higher than −1.0 is considered normal, a T-score between −1 and −2.4 indicates osteopenia, and a T-score of −2.5 or lower indicates osteoporosis.22

Radiologic Follow-Up of Fractures.

Standard radiographs of the pelvis, chest, and thoracolumbar spine at a tube distance of 120 cm were obtained at the time of activation for OLT, 4 and 12 months after transplantation, and then yearly thereafter. Additional radiographs were taken as clinically indicated at the site of any bone pain, and if they were negative, a bone scan was performed. If AVN was suspected, additional magnetic resonance imaging scans were taken. All radiographs, bone scans, and magnetic resonance imaging scanning were performed in the Mayo Clinic and judged for fractures by trained radiologists. All radiologic reports (n = 7710) performed in this study population were reviewed to assess new fractures and AVN.

Statistical Analysis.

Parameters are reported as the means ± the standard deviation. We aimed to describe the incidence of posttransplant fracturing and AVN while taking into account the competing risk occurrences of death or retransplantation, using competing risk analysis.23, 24 Patients who had not experienced any of the endpoints (death, retransplantation, fractures, and AVN) were censored at their last radiologic assessment date. Cox proportional hazards modeling was used to determine which patients' characteristics were associated with posttransplant fracturing. Posttransplant variables with serial measurements were used in the time-dependent analysis to assess the association between these variables and posttransplant fractures and AVN. The backward elimination variable selection procedure was used to find the independent variables that predicted posttransplant fractures and AVN. Such variables with a P value less than 0.05 were included in the final multivariate regression model to predict low BMD. All analyses were performed with the SAS data analysis system25 and the S-PLUS analysis system.26

Results

Pretransplant Clinical and Biochemical Variables.

The clinical, biochemical, and BMD characteristics of the study population before and after OLT have previously been reported in detail.15 One hundred fifty-six PBC patients (135 females and 21 males) and 204 PSC patients (83 females and 121 males) underwent transplantation; 148 females were postmenopausal. The PBC patients were older than the PSC patients (53.2 ± 8.6 versus 46.8 ± 11.0 years), and the females were older than the males (50.8 ± 9.6 versus 47.6 ± 11.6 years). The patients had end-stage liver disease with a mean CTP score of 8.7 ± 1.6, a MELD score of 17.3 ± 8.8, abnormal liver function, and significant osteoporosis and osteopenia (Table 1). Eleven patients had diabetes mellitus; 2 patients were treated with insulin. Other pretransplant medications were cholestyramine (n = 64), ursodeoxycholic acid (n = 111), anticonvulsants (n = 5), thyroid replacement therapy (n = 45), hormone replacement therapy (HRT; n = 18), and bisphosphonates (n = 4).

Table 1. Bone Density and Biochemical Changes After OLT in 360 Patients Transplanted for PBC and PSC
 BaselineTime After OLT
4 Months1 Year4 Years8 Years
  • This table is a summary of a previously published table.15

  • *

    Laboratory changes between 2 consecutive times were significant with P < 0.05.

  • Laboratory changes between 2 consecutive times were significant with P < 0.01.

  • Laboratory changes between 2 consecutive times were significant with P < 0.001.

  • §

    Laboratory changes between 2 consecutive times were significant with P < 0.0001 (asterisk the second time).

  • Total alkaline phosphatase: males more than 19 years old, 98-251 U/L; females 24-45 years old, 81-231 U/L; females 46-60 years old, 84-257 U/L.

  • 11 Serum creatinine: females, 0.6-0.9 mg/dL; males, 0.8-1.2 mg/dL.

  • #

    25(OH)D: summer, 15-80 ng/mL; winter, 14-42 ng/mL.

  • **

    Desired level < 200.

BMD, total patients     
 Osteopenia (n)131 (39%)102 (37%)102 (39%)72 (46%)38 (51%)
 Osteoporosis (n)128 (38%)141 (51%)113 (44%)44 (28%)22 (30%)
 T-score−2.0 ± 0.1−2.4 ± 0.1−2.2 ± 0.1−1.8 ± 0.2−1.8 ± 0.4
 Z-score−1.4 ± 0.1−1.8 ± 0.1−1.5 ± 0.1−1.0 ± 0.1−0.8 ± 0.4
Laboratory follow-up     
 Albumin (3.5-5.0 g/dL)3.0 ± 0.53.7 ± 0.6§4.0 ± 0.4§4.0 ± 0.43.9 ± 0.4§
 Total bilirubin (0.1-1.1 mg/dL)11.2 ± 11.02.1 ± 5.2§1.2 ± 3.01.3 ± 4.81.3 ± 2.8
 Direct bilirubin (0.0-0.3 mg/dL)7.0 ± 7.11.0 ± 2.7§0.5 ± 2.10.3 ± 0.60.6 ± 1.9
 Alkaline phosphatase (U/L)1107.8 ± 855.1299.9 ± 499.0§175.6 ± 234.2&150.5 ± 219.3207.1 ± 291.1
 Creatinine (mg/dL)1.2 ± 0.91.3 ± 0.6§1.3 ± 0.5§1.4 ± 0.5*1.4 ± 0.5
 Calcium (8.9-10.1 mg/dL)8.8 ± 0.69.1 ± 0.8§*9.2 ± 0.69.2 ± 0.59.3 ± 0.5
 25(OH)D (ng/mL)#20.2 ± 17.931.6 ± 13.8§36.2 ± 17.040.9 ± 14.9§38.9 ± 14.7
 PTH (1.0-5.2 pmol/L)2.3 ± 2.04.3 ± 4.5§4.8 ± 3.43.7 ± 2.03.0 ± 1.3
 Cholesterol (mg/dL)**211.5 ± 130.3201.7 ± 84.6193.8 ± 63.3197.9 ± 48.7209.7 ± 74.7
 Triglycerides (mg/dL)**119.1 ± 69.1161.1 ± 101.4142.1 ± 70.1148.5 ± 91.9150.6 ± 97.3
 Uric acid4.9 ± 3.96.1 ± 1.76.6 ± 1.86.6 ± 1.97.1 ± 2.1

To assess temporal changes, the study period (1985-2001) was divided into 3 periods by the OLT date; period 1, 1985-1989 (n = 93); period 2, 1990-1995 (n = 153); and period 3, 1996-2001 (n = 115). As reported in a previous article,15 from period 1 to period 3, patients became older (45.9 ± 9.1 versus 50.3 ± 9.9 versus 51.4 ± 11.7 years) and more postmenopausal (66% versus 67.4% versus 85%). There were temporal increases in the pretransplant 25(OH)D (17.0 ± 13.9 versus 18.0 ± 12.9 versus 29.6 ± 26.9 ng/mL), duration of disease before OLT (6.3 ± 4.6 versus 7.8 ± 5.4 versus 7.9 ± 5.9 years), body mass index (BMI; 23.6 ± 4.3 versus 24.1 ± 4.3 versus 25.6 ± 5.1 kg/m2), and T-scores (−2.5 ± 1.6 versus −2.0 ± 1.4 versus −1.7 ± 1.2), the last resulting in less osteoporosis (57% versus 34% versus 26%). Temporal decreases were seen in muscle wasting (87.8% versus 71.3%), poor nutritional status (16.1% versus 4.0%), alkaline phosphatase (1311.4 ± 871.6 versus 1181.9 ± 935.2 versus 845.0 ± 648.4 mg/dL), and direct bilirubin (8.2 ± 6.4 versus 7.1 ± 7.7 versus 5.8 ± 6.6 mg/dL). No significant changes were seen in the ratio of PBC to PSC, gender distribution, CTP or MELD scores, pretransplant albumin, creatinine, ionized calcium, or PTH.

Pretransplant Fractures and AVN.

Three hundred thirty-four (95.6%) of 360 patients had a pretransplant assessment of fractures for a mean time of 3.6 ± 4.9 years before OLT. Sixty-six (19%) patients developed pretransplant fractures (34 PBC and 32 PSC; 44 females and 22 males); 43 patients had spinal fractures (17 single fractures and 26 multiple fractures), 34 patients had rib fractures (13 single fractures and 21 multiple fractures), and 9 patients had other fractures (including femur, tibia, and calcaneus fractures). Five patients (1.4%) had AVN at the femur head (3 bilateral), only 1 of whom had steroid therapy before OLT.

There were no differences in the rates of fracturing or AVN according to assessments by gender or disease. In the univariate analysis, only BMD was associated with pretransplant fractures (P < 0.01). Pretransplant fracture rates decreased with time (23% versus 21% versus 12% from period 1 to period 3); this was significant (P < 0.05) only in patients with PSC [9 patients (20.0%) versus 19 patients (20.0%) versus 4 pts (6.4%)] and not in patients with PBC.

Posttransplant Clinical and Biochemical Variables.

The average hospitalization stay (during the first 4 months after transplantation) was 25.6 ± 21.0 days. Fifty-one (14.2%) patients were retransplanted after OLT at a mean of 1.3 ± 2.6 years, and 78 patients (21.7%) died after transplantation at a mean of 5.1 ± 4.3 years. Rejection occurred in 44.4% of patients, and nonanastomotic biliary strictures occurred in 11.9%. The liver function and indices of calcium metabolism improved following OLT (Table 1). HRT was used in 57 of 148 postmenopausal females after OLT, with only 27 females starting HRT during the first 2 years. Sixteen patients received bisphosphonates after OLT: etidronate in 1 patient (in the first post-OLT year), pamidronate in 2 patients (first and second years after OLT), and alendronate in 13 patients (all started after the first 2 posttransplant years). Sixty-three patients were enrolled in a randomized trial of calcitonin therapy or no treatment after OLT; this trial showed no effect of calcitonin therapy on BMD or fractures after OLT.27

Posttransplant loss in BMD during the first 4 months and subsequent gain have been reported in detail15 (Table 1). Posttransplant bone loss was greater in patients with PSC than in patients with PBC (−18.0 ± 20.0 versus −13.3 ± 17.8%/year, P < 0.05), and bone gain was less in CyA-treated patients than in tacrolimus-treated patients (5.0% ± 17.2% versus 10.4% ± 9.6%, P < 0.05). In addition, early bone loss did not change over time, but bone gain from 4-12 months was greater in period 3 (4.6 ± 13.8 versus 5.4 ± 10.2 versus 10.0 ± 19.3%/year). As reported previously,15 temporal decreases were seen in posttransplant hospitalization days, rejection episodes, nonanastomotic biliary strictures, treatment with cyclosporine (rather than tacrolimus), and cumulative prednisone doses. Significant temporal increases were seen in serum albumin and PTH, whereas serum creatinine and phosphorus decreased with time.

Posttransplant Fractures.

Following OLT, the mean radiologic follow-up was 5.3 ± 4.3 years: 7.9 ± 5.1 years in period 1, 5.8 ± 3.7 years in period 2, and 2.5 ± 1.9 years in period 3. The fracture rate increased sharply after OLT (Table 2), with 25% of the patients having fractures between OLT and 6 months after transplantation. During the following years, fractures continue to occur but at a lower rate, leading to a cumulative incidence of 45.9% of patients (n = 158) with fractures 8 years after transplantation. In a subpopulation of 63 patients, it was found that most fractures were symptomatic, resulting in bone pain. PBC patients had significantly more fractures than PSC patients (P < 0.01; Fig. 1A), and females had significantly more than males (P < 0.05; Fig. 1B). There was no difference in fracturing between postmenopausal women with HRT (n = 57) and those without HRT (n = 83; the cumulative incidence 8 years after transplantation was 60.8% versus 51%). Bisphosphonate therapy was used in too few patients (4 before transplantation and 16 after transplantation) to make any assessment of the effect.

Table 2. Cumulative Incidence of Posttransplant Fractures in 360 Patients with End-Stage PBC or PSC
 Number at OLTTime After OLT
4 Months After OLT12 Months After OLT2 Years After OLT4 Years After OLT8 Years After OLT
Cumulative Incidence of Fractures (n)Number at RiskCumulative Incidence of Fractures (n)Number at RiskCumulative Incidence of Fractures (n)Number at RiskCumulative Incidence of Fractures (n)Number at RiskCumulative Incidence of Fractures (n)Number at Risk
  • *

    The difference in the fractures after OLT with respect to the cumulative incidence of fractures between subpopulations was P < 0.05.

  • The difference in the fractures after OLT with respect to the cumulative incidence of fractures between subpopulations was P < 0.01.

  • c

    The difference in the fractures after OLT with respect to the cumulative incidence of fractures between subpopulations was P < 0.0001.

  • §

    Patients who had not died, had not undergone retransplantation, and had not sustained any fractures.

  • ∥, ¶, §

    Five patients did not have pretransplant radiologic screening and were excluded from the analysis.

  • #

    Patients who switched from tacrolimus to cyclosporine or vice versa were excluded from the analysis.

  • **

    The pre-OLT low BMD was less than 0.75 g/cm2, the pre-OLT average was 0.75-0.96 g/cm2, and the pre-OLT high was greater than 0.96 g/cm2; 342 of 360 patients had BMD measurements taken before OLT.

Total patients36068 (18.9%)262108 (30.1%)213123 (34.3%)189140 (39.2%)153158 (45.9%)66
PBC patients15638 (24.4%)10659 (38.0%)8268 (43.8%)6874 (47.9%)5781 (53.3%)22
PSC patients20430 (14.7%)15649 (24%)13155 (27.0%)12166 (32.6%)9677 (40.2%)44
Male patients*14219 (13.4%)11235 (24.6%)9436 (25.4%)8946 (32.9%)6954 (40.8%)26
Female patients*21849 (22.5%)15073 (33.6%)11987 (40.1%)10094 (43.4%)84104 (49.3%)40
Premenopausal females7014 (20.0%)4818 (25.7%)4220 (28.6%)3925 (36.1%)3126 (37.7%)19
Postmenopausal females14835 (23.7%)10255 (37.4%)7767 (45.5%)6169 (47%)5378 (54.9%)21
Pre-OLT fractures§,6630 (45.5%)3242 (63.6%)1845 (68.2%)1348 (73.1%)1050 (77.3%)5
No pre-OLT fractures§28937 (12.8%)22764 (22.2%)19376 (26.4%)17490 (31.5%)141106 (38.8%)60
Cyclosporine pts#19847 (23.7%)13875 (37.9%)10683 (41.9%)9392 (46.6%)83107 (54.8%)51
Tacrolimus pts11116 (14.4%)9224 (21.6%)8128 (25.2%)7335 (32.6%)4837 (35.3%)8
Low pre-OLT BMD§,**8422 (21.7%)7144 (43.5%)4453 (51.5%)3255 (55%)2357 (58.2%)10
Middle pre-OLT BMD§17239 (22.7%)12150 (29%)10857 (33.2%)9968 (39.7%)8281 (49.4%)32
High pre-OLT BMD§867 (8.1%)7014 (16.3%)6215 (17.4%)5617 (19.9%)4730 (24.4%)23
1985-19899329 (31.2%)5738 (40.9%)4643 (46.2%)3844 (47.3%)3652 (56%)24
1990-199515224 (15.8%)11446 (30.3%8953 (34.9%)8167 (44.3%)6676 (51.2%)42
1996-200111515 (13.1%)9124 (21.0%)7827 (23.6%)7029 (25.4%)51  
Figure 1.

Cumulative incidence of posttransplant fractures in 360 cholestatic patients undergoing liver transplantation for (A) PBC (156 patients) and PSC (204 patients) and (B) males (142 patients), premenopausal females (70 patients), and postmenopausal females (148 patients).

Significant differences in posttransplant fractures were seen when patients were separated by the baseline BMD (Fig. 2A) or by pretransplant fracturing (Fig. 2B). Fracture rates decreased with time (P < 0.01; Fig. 3) in both PBC and PSC.

Figure 2.

Cumulative incidence of posttransplant fractures in cholestatic patients undergoing liver transplantation stratified according to (A) pretransplant BMD and (B) pretransplant fracturing. Sixty-six patients had pretransplant fracturing versus 294 patients who did not sustain a pretransplant fracture. Pre-OLT low BMD < 0.75 g/cm2 (n = 84); pre-OLT average BMD = 0.75-0.96 g/cm2 (n = 172); pre-OLT high BMD > 0.96 g/cm2(n = 86).

Figure 3.

Cumulative incidence of posttransplant fractures in cholestatic patients undergoing liver transplantation stratified according to the time of transplantation: 1985-1989 (n = 93), 1990-1995 (n = 152), and 1996-2001 (n = 115).

Patients with fractures were older than those without fractures (51.1 ± 10.3 versus 48.3 ± 10.5 years, P < 0.01), had more pretransplant muscle wasting (91.9% versus 74%, P < 0.01), lower BMI (23.5 ± 4.6 versus 25.1 ± 4.5 kg/m2, P < 0.01), lower BMD (0.80 ± 0.15 versus 0.91 ± 0.15 g/cm2, P < 0.01), higher total bilirubin (12.3 ± 11.2 versus 10.4 ± 10.8 mg/dL, P < 0.05), higher direct bilirubin (7.7 ± 7.1 versus 6.4 ± 7.0 mg/dL, P < 0.05), higher alkaline phosphatase (1231.5 ± 896.6 versus 1008.8 ± 809.2 mg/dL, P < 0.01), and lower vitamin D (12.3 ± 10.3 versus 15.3 ± 12.2 ng/mL, P < 0.05).

Patients with fractures had increased average daily glucocorticoid doses 1 month (147.0 ± 51.7 versus 133.8 ± 46.4 mg, P < 0.01) and 4 months after transplantation (53.9 ± 19.9 versus 45.7 ± 46.4 mg, P < 0.01), were treated with CyA rather than tacrolimus [100 (62.3%) versus 65 (32.5%), P < 0.01], had more hospitalization days (25.0 ± 20.2 versus 21.3 ± 16.7, P < 0.05), and 4 months after transplantation had lower BMD (0.75 ± 0.15 versus 0.87 ± 0.14 g/cm2, P < 0.01), lower albumin (3.7 ± 0.5 versus 3.8 ± 0.6 g/dL, P < 0.01), and higher alkaline phosphatase (309.3 ± 504.1 versus 291.2 ± 495.7 mg/dL, P < 0.05).

Posttransplant Fractures by Location.

Fractures divided by location are shown in Fig. 4 and Table 3. Of the 158 patients with fractures during the first 8 years after OLT, 99 patients (28.5%) sustained spinal fractures, which occurred on average 1.9 ± 2.0 years after OLT. Twenty-six patients had a single spinal fracture (7 lumbar and 19 thoracic fractures), and 63 patients had multiple spinal fractures (6 at the lumbar spine, 16 at the thoracic spine, and 41 in both the thoracic and lumbar regions).

Figure 4.

Cumulative incidence of posttransplant fractures (by fracture location) and AVN.

Table 3. Cumulative Incidence of Fractures by Location and AVN in the Total Patient Population
 Time After OLT
4 Months1 Year2 Years4 Years8 Years
Cumulative Incidence of Fractures (n)Number at RiskCumulative Incidence of Fractures (n)Number at RiskCumulative Incidence of Fractures (n)Number at RiskCumulative Incidence of Fractures (n)Number at RiskCumulative Incidence of Fractures (n)Number at Risk
  • *

    Including patients with femur fractures. Other locations were the humerus, clavicle, wrist, ankle, calcaneus, and metatarsal bones.

Spinal fractures36 (10%)29266 (18.4%)25080 (22.3%)22787 (24.4%)19699 (28.5%)102
Rib fractures33 (9.2%)29654 (15%)26360 (16.7%)24870 (19.6%)21378 (22.6%)107
Pelvic fractures5 (1.4%)32114 (3.9%)29819 (5.3%)28220 (5.6%)25226 (7.6%)139
Other fractures*12 (3.3%)31515 (4.2%)29818 (5%)28423 (6.5%)24932 (9.5%)136
AVN3 (0.8%)32313 (3.6%)29917 (4.7%)28521 (5.9%)25127 (8.9%)140

Seventy-eight patients (22.6%) had rib fractures: 23 patients had a single fracture, and 55 patients had multiple rib fractures. Rib fractures occurred on average 1.3 ± 1.7 years after transplantation. In addition, 26 (7.6%) patients had pelvic fractures, 12 patients (3.2%) had femoral neck fractures, and 25 patients (7.2%) had other fractures (including humerus, clavicle, wrist, ankle, calcaneus, and metatarsal bones). Pelvic fractures occurred on average 1.8 ± 1.9 years after OLT, and other fractures (including femur) occurred on average 2.3 ± 2.2 years after OLT.

Posttransplant AVN.

After OLT, AVN occurred in 27 patients (16 PSC and 11 PBC; 13 males and 14 females) with a cumulative incidence of 8.9% (Fig. 4) at a mean of 2.4 ± 3.6 years after OLT. Patients with AVN had a lower BMD both before and after OLT than those without AVN (0.77 ± 0.14 versus 0.87 ± 0.16 g/cm2, P < 0.01) and more fracturing before transplantation (38.5% versus 17%, P < 0.05) and after transplantation (85.2% versus 41.4%, P < 0.05). Most patients with AVN (25 of 27) underwent transplantation during periods 1 and 2 (13 and 12 patients, respectively). The sites of posttransplant AVN were the proximal femur (22 patients), the distal femur (3 patients), the proximal humerus (1 patient), and the metatarsal bone (1 patient). Of the 22 patients with proximal femur AVN, 6 patients had unilateral AVN, and 16 had bilateral AVN; 3 had preceding femoral neck fractures.

Risk Factors for Posttransplant Fractures.

Univariate analysis indicated that many factors correlated with posttransplant fractures: older age, female gender and postmenopausal status, poor nutritional status, muscle wasting, underlying disease of PBC, low BMI, low OLT number, low BMD, the presence of pretransplant fractures, and elevated serum alkaline phosphatase levels (Table 4). Posttransplant risk factors for fracturing were the average daily dose of corticosteroids at 1 month and at 4 months, cyclosporine therapy, rejection episodes, and low BMD (Table 5). No other correlations were seen after OLT, including hospitalization days, changes in BMD (early bone loss or later bone gain), serum levels of calcineurin inhibitors (cyclosporine, tacrolimus), and all biochemical indices.

Table 4. Pretransplant Risk Factors for Posttransplant Fractures and AVN
VariableFracturesAVN
HR95% CIPHR95% CIP
  • INR indicates international normalized ratio, and NS, not significant.

  • *

    Multivariate analysis indicated that the independent positive predictor for posttransplant fractures is pretransplant fractures (HR = 2.77, 95% CI = 1.91-4.01, P < 0.0001); the independent negative predictors are PSC disease (HR = 0.67, 95% CI = 0.49-0.92, P < 0.05) and pretransplant BMD (HR = 0.98, 95% CI = 0.97-0.99, P < 0.0001).

  • Multivariate analysis indicated that the independent positive predictor for posttransplant AVN is pretransplant fractures (HR = 2.95, 95% CI = 1.31-6.61, P < 0.01); the independent negative predictors are the BMI (HR = 0.80, 95% CI = 0.70-0.92, P < 0.05), OLT number (HR = 0.83, 95% CI = 0.69-0.99, P < 0.05), and pretransplant triglycerides (HR = 0.39, 95% CI = 0.20-0.74, P < 0.01).

Age at OLT (years)1.031.01-1.04<0.010.970.94-1.01NS
PSC disease*0.700.51-0.94<0.051.160.52-2.59NS
Male gender0.730.53-1.000.051.460.66-3.20NS
Menopausal status1.591.03-2.45<0.050.720.23-2.19NS
Duration of disease (years)1.0050.98-1.03NS0.930.85-1.02NS
Poor nutritional status1.571.09-2.27<0.051.460.50-4.26NS
Muscle wasting2.951.43-6.08<0.012.820.37-21.53NS
BMI (kg/m2)0.930.89-0.96<0.0010.820.72-0.94<0.01
MELD score1.0010.99-1.02NS0.970.92-1.02NS
Child score1.040.95-1.13NS0.950.76-1.19NS
Karnofsky score1.020.92-1.31NS1.010.77-1.33NS
OLT number0.9990.998-1.00<0.0010.9980.997-1.00<0.05
Pre-OLT fractures*,5.564.0-7.8<0.0013.211.44-7.15<0.01
Pre-OLT BMD (g/cm2)*,0.0270.01-0.075<0.00010.020.0002-0.29<0.01
Albumin (g/dL)0.880.67-1.17NS1.310.63-2.72NS
Prothrombin time (INR)0.720.34-1.55NS1.160.01-2.05NS
Bilirubin total (mg/dL)1.010.99-1.02NS1.0040.97-1.04NS
Bilirubin direct (mg/dL)1.010.99-1.03NS1.0090.96-1.06NS
Alkaline phosphatase (U/L)1.011.00-1.00<0.051.001.00-1.00NS
Creatinine (mg/dL)0.990.85-1.16NS0.580.22-1.51NS
Calcium (mg/dL)1.110.88-1.41NS1.360.74-2.52NS
25(OH)D (ng/mL)0.990.98-1.002NS1.0050.99-1.03NS
PTH (pmol/L)0.960.87-1.07NS1.0070.78-1.30NS
Phosphorus (mg/dL)1.060.88-1.28NS1.110.09-1.77NS
Uric acid0.9730.92-1.03NS0.9880.87-1.12NS
Cholesterol (mg/dL)1.0011.000-1.002NS1.000.99-1.00NS
Triglycerides (mg/dL)1.0010.999-1.003NS0.990.98-1.00<0.05
Table 5. Posttransplant Risk Factors for Posttransplant Fractures and AVN
VariableFracturesAVN
HR95% CIPHR95% CIP
  • INR indicates international normalized ratio, and NS, not significant.

  • *

    Multivariate analysis indicated that the independent positive predictor for posttransplant fractures is 4 months of daily steroids (HR = 1.02, 95% CI = 1.01–1.02, P < 0.001); the independent negative predictor is posttransplant BMD (HR = 0.98, 95% CI = 0.97–0.99, P < 0.0001).

  • Multivariate analysis indicated that the independent positive predictors for posttransplant AVN are posttransplant nonanastomotic strictures (HR = 4.87, 95% CI = 1.8–13.5, P < 0.01) and posttransplant fractures (HR = 74.1, 95% CI = 27.2–201.7, P < 0.0001).

1 month of glucocorticoids (mg)1.0041.001–1.007<0.011.0050.997–1.01NS
4 months of glucocorticoids (mg)*1.011.002–1.019<0.051.021.00–1.09<0.05
Cyclosporine serum (ng/mL)0.9970.99–1.00NS0.990.99–1.01NS
Tacrolimus serum (ng/mL)0.940.86–1.04NS2.641.10–6.32<0.05
Cyclosporine treatment1.851.22–2.79<0.012.950.99–8.76<0.05
Rejection episodes1.231.07–1.57<0.011.490.99–2.340.056
Hospitalization days1.0010.998–1.004NS0.9970.99–1.01NS
Nonanastomotic biliary strictures1.160.67–2.03NS5.092.24–11.55<0.0001
Post–OLT BMD (g/cm2)*0.0270.008–0.092<0.00010.0340.003–0.382<0.01
BMD loss (pre–OLT–4mo)0.900.002–3.77NS1.110.03–37.85NS
BMD gain (4–24 mo)0.1720.012–2.55NS0.150.01–2.36NS
Post–OLT fractures2.641.10–6.32<0.05
Albumin (g/dL)0.800.58–1.10NS1.200.54–2.66NS
Prothrombin time (INR)0.740.27–1.98NS0.130.004–4.38NS
Bilirubin total (mg/dL)1.0070.99–1.03NS1.0020.94–1.07NS
Bilirubin direct (mg/dL)1.020.99–1.05NS1.210.67–1.68NS
Alkaline phosphatase (U/L)1.001.00–1.00NS1.0011.00–1.001<0.01
Creatinine (mg/dL)1.090.90–1.32NS0.200.05–0.79<0.05
Calcium (mg/dL)1.120.85–1.47NS1.180.64–2.19NS
25 (OH)D (ng/mL)1.0030.99–1.01NS0.990.97–1.03NS
PTH (pmol/L)0.990.94–1.05NS0.920.73–1.16NS
Phosphorus (mg/dL)1.060.84–1.35NS0.780.42–1.46NS
Cholesterol (mg/dL)1.0010.99–1.00NS1.0031.00–1.005< 0.05
Triglycerides (mg/dL)1.000.99–1.00NS1.0031.00–1.01NS

Multivariate analysis indicated that the independent pretransplant predictors for posttransplant fractures were pretransplant fractures, low BMD, and underlying PBC disease. In addition, independent posttransplant predictors were glucocorticoid doses and low BMD.

Risk Factors for Posttransplant AVN.

Univariate analysis of pretransplant parameter data indicated that low BMI, low BMD, low serum triglyceride levels, low OLT number, and pretransplant fractures correlated with posttransplant AVN (Table 4). The changes in the triglyceride levels from the pretransplant period to the posttransplant period were greater in those with AVN [baseline: 93.3 ± 47.2 to 213.2 ± 181.0 mg/dL after OLT] than in patients without AVN [121.2 ± 70.3 to 156.2 ± 89.6 mg/dL, P < 0.0001]. This increase correlated univariately with AVN [hazard ratio (HR) = 0.99, 95% confidence interval (CI) = 0.98-1.00, P < 0.05]. Posttransplant risk factors for AVN (Table 5) were high 4-month average steroid doses, cyclosporine therapy, the presence of nonanastomotic biliary strictures, rejection episodes, posttransplant fractures, low BMD, elevated levels of alkaline phosphatase or cholesterol, and low levels of serum creatinine.

Multivariate analysis indicated that independent pretransplant predictors for posttransplant AVN were pretransplant fractures, low BMI, low OLT number, and low triglycerides. Independent posttransplant predictors for posttransplant AVN were nonanastomotic strictures and fractures.

Discussion

The majority (80%) of patients with CCLD in this study had osteoporosis or osteopenia before OLT, and 20% had radiologic evidence of fractures. A pretransplant radiologic assessment was performed in 95% of the patients, but not all patients had long-term radiologic screening before OLT. The 20% prevalence of fractures may therefore be an underestimation. Pretransplant fractures were associated with low BMD, and the majority occurred in trabecular bone (spine and ribs), whose higher rate of bone turnover makes it more susceptible to changes in bone metabolism. These findings are consistent with previous studies of chronic liver disease both before and after OLT.28, 29

After OLT, the fracture rate abruptly increased, with 25% of patients sustaining a new posttransplant fracture by 6 months, with an ongoing fracture rate thereafter affecting 46% of patients by 8 years after OLT. There are few previous studies with which to compare our data; Leidig-Bruchner et al.1 analyzed posttransplant vertebral fractures by Kaplan-Meier analysis, also indicating a high posttransplant vertebral fracture rate (33%) but a nonvertebral fracture rate of only 7%. In our study, nonvertebral fractures were as common as vertebral fractures. Although the spine was the commonest site of posttransplant fracturing, the cumulative incidence of rib fractures was almost as great. Overall, most fractures occurred at sites of trabecular bone, with the spine and ribs accounting for more than 90% of total fractures. The difference in the nonvertebral fracture rates in Leidig-Bruchner et al.'s study is probably related to methodologic differences in fracture screening.

Pretransplant BMD and pretransplant fracturing have been clearly identified as major risk factors for posttransplant fracturing. The severity of osteoporosis/osteopenia at the time of OLT is very important: posttransplant fractures in the first 12 months after OLT occurred in 50% of patients with pretransplant BMD less than 0.75 g/cm2 versus only 18% of patients with pretransplant BMD greater than 0.96 g/cm2. In addition, 80% of patients with pretransplant fractures sustained a new posttransplant fracture after OLT. Surprisingly, neither the rate of bone loss during the first 4 months after OLT nor the rate of bone gain thereafter correlated with fracturing.

Posttransplant fractures also correlated independently with age. Advanced age is associated with osteopenia, but other factors such as muscle mass, coordination, and activity may play an etiologic role in fracture occurrence. Female patients sustained more posttransplant fractures than male patients (34% versus 25% at 1 year), and this was due entirely to the high fracture rate in postmenopausal females; premenopausal female patients had fracture rates similar to those of male patients. Overall, the PBC patients had more fractures after OLT than the PSC patients, and this likely reflects an age and postmenopausal influence.

In parallel with improved pretransplant BMD over the 16-year study period,15 fewer fractures were seen in more recently transplanted patients both before and after OLT. The reduction in pretransplant fracturing with time was significant only in PSC patients, whereas fracture rates remained stable in PBC patients despite an aging and more menopausal patient population. It is likely that temporal improvements in the nutritional status, BMI, and vitamin D levels contributed to less fracturing in the PSC patients and to the stable fracture rates in the PBC patients, in the latter situation offsetting the negative effects of increased age and postmenopausal status.

Temporal improvements of posttransplant fracture rates were seen in both patients with PBC and patients with PSC. Bone loss during the first 4 posttransplant months did not change with time, and this implied that increased BMD before OLT in the more recently transplanted patients was important in causing the temporal reduction in posttransplant fracturing. The lower cumulative doses of glucocorticoids 1 and 4 months after OLT were likely responsible at least in part for this improvement. Bone histomorphometric studies have indicated the important effect of glucocorticoids early after OLT, with the main effect being to decrease bone formation rates.29, 30 As shown by others,31, 32 patients on cyclosporine have more fracturing than patients on tacrolimus, but this effect may well reflect the differences in the average steroid dose between the 2 groups. Both cyclosporine and tacrolimus cause increased bone turnover,33, 34 but these histomorphometric effects in liver recipients have not been demonstrated. Whether temporal improvement in fracturing will be maintained in the MELD era is unknown. No correlation was seen here between the fracturing and MELD score, the individual components of MELD, or the disease duration, and this is encouraging. Nonetheless, a longer wait for OLT, increasing age, and debility may take its toll.

AVN is regarded as a different etiologic entity than osteopenic fracturing and is thought to correlate with abnormalities in the vascular supply to bone, particularly at the femoral head.9, 10, 13 A well-known risk factor in the general population is femoral neck fracturing. Scant data are available about AVN after OLT.4, 35, 36 In this study, the cumulative incidence of AVN after OLT was 9%. The majority (80%) occurred at the femoral head, but only 3 patients had a femoral neck fracture preceding the diagnosis of AVN. Posttransplant AVN was not influenced by gender or disease but correlated with pretransplant and posttransplant BMD and fractures. Eighty-five percent of patients with posttransplant AVN also sustained fractures after OLT. This suggests that bones prone to fracturing also have abnormalities predisposing them to AVN.

In this study, posttransplant AVN also correlated with posttransplant glucocorticoids. Glucocorticoids are a well-recognized risk factor for AVN, with several potential etiologic mechanisms: increased fat emboli in the microvasculature of bone, glucocorticoid-induced apoptosis of osteocytes,12 or an increase in the size of intraosseous adipocytes, with a subsequent increase in the femoral head pressure and a decrease in the blood supply.10, 37 It is tempting to speculate that this latter mechanism may be involved in the observed association of AVN with cholesterol levels and with the pretransplant-to-posttransplant increase in triglyceride levels. The temporal improvement in the occurrence of AVN with increased use of tacrolimus may be explained by the use of lower doses of glucocorticoids.

An unexpected association was seen between AVN and both posttransplant nonanastomotic biliary strictures and alkaline phosphatase levels, suggesting a potential etiologic role for cholestasis. Cholestasis is well known for its suppressive effect on bone metabolism, with reductions in osteoblastogenesis and osteoblastic proliferation in vitro38 and reduced osteoblast number and function in vivo.29, 30 Osteocytes are derived from osteoblasts and may have a role in the mechanosensory function of bone. A decrease in osteocytes, perhaps potentiated by cholestasis, may cause mechanosensory disturbances with a subsequent collapse of bone, a disruption of its vascular supply, and consequently AVN.12 An etiologic connection between cholestasis and AVN may also contribute to the increased incidence of AVN (1.4%) in end-stage cholestatic patients before OLT versus an incidence of less than 0.01% for the US population according to the National Institute of Health registration.

In summary, at the time of OLT, 20% of patients with CCLD have already experienced osteopenic fractures, and 1.4% of patients have had AVN. The highest rate of fracturing occurs in the first 12 months after OLT (cumulative incidence of 30%); thereafter, there is a smaller but steady cumulative increase in fracturing, so that by 8 years, almost 46% of patients have sustained a fracture. Overall, the majority (>90%) of fractures occur at sites of trabecular bone (the spine, ribs, and pelvis). In contrast to previous studies, vertebral and nonvertebral fractures occur with similar incidences after OLT. The greatest risk factors for posttransplant fracturing are pretransplant fracturing and the severity of osteopenia, PBC disease, and posttransplant glucocorticoids. In addition to fractures, 9% of the liver recipients experienced AVN after OLT, and this correlated with pretransplant and posttransplant lipid metabolism, bone disease (BMD and fracturing), and posttransplant glucocorticoids. A novel association between cholestasis and AVN was also identified, the mechanism for which is not known. Fortunately, recent years have seen an increase in the bone mass of liver recipients and, along with this, less fracturing and less AVN. Nonetheless, 25% of patients undergoing OLT for CCLD still develop de novo fractures after OLT; this situation demands an ongoing search for effective therapeutic agents for these patients.

Ancillary