Pediatric liver transplantation (LT) is associated with 10-year patient survival rates of up to 83%, depending on patient age at transplantation and graft type (1). Hence, health care providers need to be aware of potential extrahepatic sequelae afflicting long-term survivors of pediatric LT including renal impairment and posttransplant lymphoproliferative disease, growth, cognitive function, and quality of life (2,3). Bone health, particularly the prevalence and burden of reduced bone mineral density (BMD), is an appropriate focus of interest because most bone mineral accretion occurs in childhood and adolescence (4,5). Peak BMD is achieved by early adulthood. Therefore, identifying and addressing risk factors for poor bone health in childhood may be important to achieving a satisfactory peak BMD and preventing adult osteoporosis and subsequent morbidities (6). An association between low BMD and fracture risk in childhood has not been established, however (7). BMD has been reported to be low at the time of pediatric LT and improved for most patients with successful transplantation (8,9).
Identified risk factors for low BMD measured by dual-energy x-ray absorptiometry (DXA) include time post-LT, weight at DXA, and older age at transplantation (10,11). In particular, the effect of dosage and timing of corticosteroid exposure and their relation to other potential risk factors for poor skeletal health following pediatric LT have not been addressed adequately (8,10–12). The aim of the present study was to explore factors associated with BMD in children and adolescents following LT in a cross-sectional study at a single large pediatric transplant center. We hypothesized that corticosteroid exposure, age at transplantation, time since transplantation, and the nature of the primary liver disease would be associated with BMD and fractures in these survivors of pediatric LT.
PATIENTS AND METHODS
Consecutive patients seen in the ambulatory liver transplant clinic between July 2001 and July 2003 were invited to participate in the study. Inclusion criteria were all of the patients ages 4 to 18 years and >1 year post-LT. Patients who had received multiple liver grafts were eligible to participate provided the most recent transplant occurred >1 year before enrollment. Patient exclusion criteria included recipients of multivisceral grafts and those with non–English-speaking parents. The study was approved by the local institutional review board. Written informed consent was obtained from each participant and/or his or her guardian. Each participant completed a questionnaire on diet, exercise, and fracture history, provided a single blood sample, and underwent both a skeletal bone age assessment and DXA scan.
The dietary questionnaire was completed by both patients and/or parent, and required recall for a period of a typical week and asked about specified quantities of common dietary sources of calcium and vitamin D and dietary supplements. Total estimated daily intake of both calcium and vitamin D were calculated using the US Department of Agriculture National Nutrient Database for Standard Reference (13), and compared with recommended intakes for age (14). Serum intact parathyroid hormone (PTH), 25-hydroxyvitamin D (25-OHD), and 1,25-dihydroxyvitamin D (1,25-(OH)2D) were measured. Clinical data such as indications for LT, number of transplants, corticosteroid dosages, duration and weight at the time of dosage, glomerular filtration rate (GFR) measurements (using technetium-99m diethylenetriamine-pentaacetic acid excretion) performed for clinical purposes, and liver biochemistry were obtained retrospectively from the hospital charts. Corticosteroid exposure was calculated by multiplying daily dosage by number of days at the specific dose and dividing by the contemporaneous weight of the child. For patients taking corticosteroids continuously for >3 months, these periods were divided into 3-month periods and calculations were based on the mean weight during that period. All of the oral and parenteral corticosteroids were converted to equivalent dosages of prednisone (intravenous hydrocortisone doses were divided by 4; intravenous methylprednisolone doses were multiplied by 1.25). Height (using a fixed stadiometer) and weight were measured, and body mass index (BMI) was calculated at the time of DXA. All of the values were converted to z scores using the US NHANES dataset (15). Overweight and obesity were defined as BMI z score >85th and >95th percentiles, respectively, as defined by the American Academy of Pediatrics (16).
Immunosuppression after pediatric LT followed institutional protocols: patients receiving transplants before 1997 received cyclosporine, azathioprine, and corticosteroids, the latter being generally weaned by 1 year post-LT, with infants under 10 kg at the time of transplantation additionally receiving muronomab-CD3 (OKT3) for induction of immunosuppression. From 1997, patients with new transplants received tacrolimus and corticosteroids, who were intended to be weaned by 3 to 6 months. Acute cellular rejection was treated with a brief (1–3 days) course of intravenous corticosteroid followed by a tapering oral corticosteroid course.
Bone Mineral Densitometry
Areal BMD of the lumbar vertebrae (anteroposterior view, L1-L4) was measured in each patient by DXA using the QDR-2000 model (Hologic Inc, Waltham, MA) with the array mode. The instrument was calibrated daily with an anthropomorphic hydroxyapatite spine phantom specific to this machine, and technical error was <1%. For each patient, a z score was calculated for the lumbar spine (L1-L4) BMD using a reference population (USA Reference Population for GE Lunar Prodigy). Bone age was determined from a radiograph of the left hand by a pediatric radiologist using the method of Greulich and Pyle (17). Where there was at least 1-year discrepancy between the patient's chronologic and bone age, the bone age was used to determine the BMD z scores. Low BMD was defined as a z score <2, in accordance with the definition used by the International Society for Clinical Densitometry (18).
Data on fractures extracted from the questionnaire included timing (pre- or post-LT), location, mechanism, and whether fractures were single or multiple.
Statistical analysis was performed using Stata version 8 (StataCorp, College Station, TX). The primary outcome variable was BMD z score for the lumbar spine, with the secondary outcome being reported fractures occurring since transplantation. Univariate analysis for associated factors was performed using the Student t test for dichotomous variables and the Pearson correlation coefficient for continuous variables. Continuous variables were converted to dichotomous variables when a nonlinear association was suspected. Multivariate linear regression analysis was used to assess the relative importance of factors significant in the univariate analysis.
A total of 60 pediatric LT recipients were approached to participate in the study. Seven declined to participate, and 1 patient who was recruited was transitioned to adult services before DXA could be performed for a final study cohort of 52 (24 boys) patients. Patient characteristics are summarized in Table 1. Eight children (15%) had undergone retransplantation, 2 of whom had received 3 grafts. One patient with congenital hepatic fibrosis who had polycystic kidney disease has subsequently undergone renal transplantation. Eight patients (15%) received tacrolimus as the primary maintenance immunosuppressive agent, with the remaining 44 receiving cyclosporine. Thirty-four (65%) patients received at least 1 treatment with high-dose intravenous corticosteroids for biopsy-proven acute cellular rejection, with 17 children having at least 3 episodes of treated acute rejection. None of the patients were Cushingoid in appearance at the time of BMD assessment. All of the patients had normal serum conjugated bilirubin levels (<0.1 mg/dL), and 2 of 52 (3.8%) patients had serum liver aminotransferase levels elevated greater than twice the upper limit of normal.
At the time of DXA, weight z scores were similar to those of the normal population, with a mean of −0.17 and an SD of 1.16. The mean height z score of the study cohort was lower than normal (mean −0.76, SD 1.32), giving slightly higher BMI z scores than normal (mean 0.30, SD 1.12). Eighteen children (34.6%) had BMI >85th percentile (defined as overweight) and 4 (7.7%) children had BMI percentiles >95th (obese). The children defined as overweight or obese had average height z scores of −0.53 and 0.02, respectively, and were together not shorter than the other subjects whose height z score was −0.95 (P = 0.16). Eighteen patients (37%) had a bone age that differed from chronological age by >1 year, including 8 patients with delayed bone age and 10 patients with advanced bone age. For these patients, BMD z scores were corrected for bone age for the analysis. After this correction, the mean BMD z score for the group was −0.66 (SD 1.02, range −4 to 1.6). Low BMD (z score <−2) was present in 3 patients (5.8%).
BMD z Scores: Univariate Analysis
Although there was a trend to increased BMD with greater time since LT and a stronger trend to decreased BMD with increasing age at LT, these linear associations were not statistically significant (Fig. 1). The only 3 patients who had BMD z scores <−2 were transplanted after the age of 8 years and all had chronic cholestatic conditions. Given a possible nonlinear association between age at LT and BMD z scores, age was converted to a dichotomous variable. Children receiving LT >10 years of age had a significantly lower BMD than those receiving LT <10 years of age (P = 0.03), although their mean follow-up time was significantly shorter (3.7 vs 8.1 years; P = 0.007).
Figure 2 demonstrates the relation between corticosteroid exposure and BMD. Cumulative doses of corticosteroid received in the first year after transplant, in the year before DXA, and total lifetime exposure did not correlate significantly with BMD. BMD was not different between those patients who received cyclosporine or tacrolimus as the primary maintenance immunosuppressive agent (P = 0.79).
Table 2 shows results of further univariate analysis. Overweight patients had significantly better BMD than those with lower BMI (P = 0.03). Height or weight z scores at time of DXA did not correlate well with BMD (r = 0.10 and 0.22, P = 0.48 and 0.11, respectively); however, height z scores were significantly associated with cumulative corticosteroid dose (Fig. 3). The 5 patients receiving transplants at the age of 10 years had lower height z scores than those receiving transplants at younger ages (means 1.01 vs −0.08; P = 0.08).
Although those receiving transplants for cholestatic indications had similar BMD to the group with noncholestatic disease, the cholestatic group were significantly younger at LT (mean ages 2.7 vs 6.5 years, P < 0.001), and their mean time since transplant was slightly longer (8.3 vs 6.1 years, P = 0.06).
The mean measured GFR value within 18 months of DXA was 100 mL · min−1 · 1.73 m−2 (SD 36). Twenty-two children had a GFR below 90 mL · min−1 · 1.73 m−2 (defined as at least stage 2 chronic renal impairment (19)), although this was not associated with reduced BMD (P = 0.92, Table 2). Increased time post-LT was negatively correlated with reduced GFR (r = −0.44, P = 0.001).
Self-/parent-reported quantities for dietary intake of vitamin D and calcium-rich foods were available for 46 study patients. Supplement use included vitamin D (n = 3) and calcium (n = 6). BMD was not significantly different between those children who reported average daily intakes of vitamin D or calcium that was above the recommended levels and those who did not. When analyzed as continuous variables, no linear correlation was found between BMD and average daily vitamin D or calcium intakes. Thirty-three (63%) children reported at least weekly participation in physical activity outside of school physical education classes. Such extracurricular physical activity was not significantly associated with BMD z scores (P = 0.74).
There was no correlation among serum intact PTH, 25-OHD or 1,25-(OH)2D levels, and BMD z scores. Intact PTH levels were above the reference interval for 9 patients (17.6%), although these individuals did not have lower BMD or more fractures post-LT compared with those with normal PTH levels. PTH levels did not correlate with corticosteroid exposure in the year before DXA (data not shown).
BMD z Scores: Multivariate Analysis
Results of multivariate linear regression analysis are shown in Table 3. Greater time since LT and higher BMI z scores were significantly associated with greater BMD z score values (P = 0.04 and 0.02, respectively). Age at LT was not included in the model given significant correlation with time post-LT (r = −0.30, P = 0.03). The model including age at LT in place of time post-LT resulted in an inferior model with a lower R2 value (0.128 vs 0.143).
Nineteen (36.5%) children reported 27 episodes of bone fractures, with 9 patients (17.3%) experiencing 14 fractures before LT and 11 patients (21.1%) having 13 fractures post-LT, involving the upper or lower limbs (n = 21), ribs (n = 3), clavicles (n = 2), and small bones in the foot (n = 1). Trauma-induced fractures were documented as the mechanism in the majority, with 9 of the 13 fractures (69%) post-LT involving a fall from an object or while playing sports. Trauma was not documented for the remaining 4 fractures. BMD z scores were not associated with having fractures pre-LT, post-LT, or at any time (P > 0.05). BMD was also not associated with having a history of multiple fractures. The mean time to fracture post-LT was 3.4 years (range 0.8–8.8). Fracture-free survival post-LT is shown in Figure 4.
This cross-sectional study showed a prevalence of low BMD, as defined by z score <−2, of 5.8% in a cohort of 52 children and adolescent survivors of LT. This rate is slightly lower than the 7% to 15% reported in other cross-sectional studies (10–12). In both univariate and multivariate analyses, neither corticosteroid exposure nor the nature of the primary liver disease was associated with BMD. Older patient age (>10 years) at time of LT, shorter duration of time since transplantation, and lower BMI (specifically <85th percentile) at time of DXA were the most significant variables associated with decreased BMD after pediatric LT.
Chronic corticosteroid exposure is a well-described risk factor for reduced BMD, thought to be mediated predominantly by increased osteoblast apoptosis and reduced bone formation (20). Our finding that early (within the first year post-LT), recent (in the year before DXA), or total cumulative corticosteroid exposure was not associated with BMD or fracture occurrence is unexpected. The influence of corticosteroid dosage during the early post-LT period on the long-term BMD of children and adolescents has not been examined. D’Antiga et al (12) analyzed corticosteroid exposure as the presence of absence of corticosteroid exposure at the time of DXA in 16 patients, finding no association with BMD. Guthery et al (10), in the largest pediatric study of 109 patients, found that corticosteroid exposure in the year before DXA did correlate with BMD that was not corrected for bone age; however, early or cumulative steroid exposure was not examined. Valta et al (11) did not find a relation between cumulative corticosteroid exposure and BMD z score corrected for bone age in 40 children and adolescents, although early or recent exposure was not evaluated specifically. In our study, treated acute rejection episodes (any or multiple), a potential surrogate for corticosteroid exposure, were not associated with BMD or fractures post-LT. The 8 patients who received retransplantation and thus received multiple courses of induction corticosteroid therapy did not have significantly lower BMD than those with single grafts. It is difficult to compare the corticosteroid exposure in our group with other studies, given that different measurement approaches have been used. Despite this, we believe that our patients’ exposure was significant as evidenced by the strong association of exposure and height z scores. In light of these findings, it is interesting to note that serial BMD measurements of children as early as 3 months post-LT showed mean BMD increased between 1 and 3 months post-LT even while patients were still taking corticosteroids (8). Possible explanations include the fact that younger patients’ bones may be more resistant to the negative effects of corticosteroid, that bone recovery occurs more readily in children than in adults, or that correction of poor liver function has a greater effect on bone than does any deleterious effect of corticosteroid.
As in previous reports, our study did not demonstrate a difference in BMD between patients undergoing transplant for cholestatic and noncholestatic indications (10,11). This observation may, again, reflect an excellent capacity for bones of even chronically unwell children to recover rapidly following successful transplantation.
Most reported fractures occurred within the first few years post-LT. The limited data available about the mechanism of fracture suggested that most of these fractures occurred in the context of falls from an object or in sports, so these are unlikely to have resulted from minimal impact. The possibility of selection bias cannot be excluded because the fracture history of those patients who declined to be involved in the study is not known. Valta et al (11) reported symptomatic fractures occurring in 12.5% of 40 children and adolescents post-LT, although asymptomatic vertebral fractures were detected in a further 17.5% of patients. In that study, as in ours, BMD was not associated with post-LT symptomatic fractures. Indeed, we could find no factors significantly related to symptomatic fractures post-LT in our patients.
Our patients who were overweight (BMI >85th percentile) at the time of DXA had significantly higher BMD. This finding should be interpreted with the understanding that none of our patients were Cushingoid in appearance at time of DXA. Additionally, the positive relation between BMI z score and BMD is unlikely to be explained simply by shorter stature, because our overweight and obese patients had comparable height z scores with the other patients. Other growth parameters such as height and weight z scores did not correlate directly with BMD. In the pediatric LT literature, weight-for-height, height, and BMI z scores were not significantly associated with BMD in 1 study, although it was unclear as to what the mean BMI of the cohort was, or what proportion were overweight or obese (11). A larger study found height and weight z scores to be significantly associated with BMD in children post-LT (10).
Time post-LT had a stronger association with BMD than did age at transplantation; however, these 2 were significantly correlated with each other and therefore not included together in the multivariate model. We are unable to be certain as to which may have the stronger influence on BMD because our study had relatively few patients receiving transplants later in childhood (5/52, 9% underwent LT >10 years of age). Guthery et al (10) found a significant association between BMD and time since LT but not age at LT. Age at LT was not related to symptomatic fracture occurrence in our study; however, Valta et al (11) found that asymptomatic vertebral fractures were more common in children receiving transplants >10 years of age.
A number of study limitations are acknowledged. First, there are well-recognized inherent problems with estimating pediatric BMD using DXA. Ideally, physiological changes in density that occur with skeletal maturation and puberty should be considered. Technical issues such as bone size changes will influence the calculated density using areal methods. BMD is underestimated by DXA in children with smaller bones because BMD calculations are based on areal (2-dimensional) measurements and hence the thickness of the bones is not appreciated (21). There are also age-, sex-, and race-related variations. Because reference datasets are not sufficiently large to account for all of these factors, and our patients’ mean height z score was within the normal range, we chose to correct for skeletal maturity based on bone age. The lumbar spine site is regarded as one of the most reproducible and accurate in measuring BMD in children (18). Newer techniques such as peripheral quantitative computed tomography and magnetic resonance imaging, which provide both densitometry and bone geometric data, may become more widely used as reference data become available (22). Second, although BMD as measured by DXA predicts risk of osteoporotic fractures in postmenopausal women (23), its role in predicting fracture risk in children is not proven. More relevant and also undetermined is whether reduced BMD in post-LT children translates to increased fracture risk in adult life. As peak bone mineral accrual rate occurs during adolescence, adult final adult bone mass may be expected to be influenced by any process that interferes with achieving this peak accrual rate. Long-term follow-up studies will provide answers to these questions. Third, our study group is subject to survivor bias and therefore will not reflect the BMD of all of the children post-LT, but only of survivors. It did include, however, 8 of 52 (15.3%) patients who received retransplants, although all of them had normal liver function at time of study. Finally, being a single-center study, our findings are not as generalizable as they would be from a multicenter study.
In conclusion, the prevalence of low BMD in ambulatory children and adolescents at least 1 year post-LT, as measured by DXA, was low. Factors most strongly associated with decreased BMD were lower BMI z scores, older patient age at LT, and shorter time since LT. BMD and fracture occurrence were not associated with early, recent, or cumulative corticosteroid exposure, or with having a chronic cholestatic disease as a primary indication for LT. Patients receiving transplants later in childhood or adolescence and those with low BMI may represent groups that are most likely to benefit from preventive or therapeutic interventions. Prospective studies, ideally beginning before transplantation and using modern volumetric techniques of bone mineral estimation, are needed to clarify the effect of pediatric LT on long-term BMD.
Dr Riccardo Superina and Ms Enza DeLuca are acknowledged for the care provided to many of the study participants.
1. US Department of Health and Human Services, Health Resources and Services Administration, Healthcare Systems Bureau, Division of Transplantation. Annual Report of the U.S. Organ Procurement and Transplantation Network and the Scientific Registry of Transplant Recipients: Transplant Data 1998–2007
. Rockville, MD: US Department of Health and Human Services; 2008.
2. Ng VL, Fecteau A, Shepherd R, et al. Outcomes of 5-year survivors of pediatric liver transplantation: report on 461 children from a North American multicenter registry. Pediatrics
3. Bucuvalas JC, Alonso E, Magee JC, et al. Improving long-term outcomes after liver transplantation in children. Am J Transplant
4. Helenius I, Remes V, Salminen S, et al. Incidence and predictors of fractures in children after solid organ transplantation: a 5-year prospective, population-based study. J Bone Miner Res
5. Theintz G, Buchs B, Rizzoli R, et al. Longitudinal monitoring of bone mass accumulation in healthy adolescents: evidence for a marked reduction after 16 years of age at the levels of lumbar spine and femoral neck in female subjects. J Clin Endocrinol Metab
6. Mora S, Gilsanz V. Establishment of peak bone mass. Endocrinol Metab Clin North Am
7. Clark EM, Ness AR, Bishop NJ, et al. Association between bone mass and fractures in children: a prospective cohort study. J Bone Miner Res
8. Okajima H, Shigeno C, Inomata Y, et al. Long-term effects of liver transplantation on bone mineral density in children with end-stage liver disease: a 2-year prospective study. Liver Transpl
9. D’Antiga L, Moniz C, Buxton-Thomas M, et al. Bone mineral density and height gain in children with chronic cholestatic liver disease undergoing transplantation. Transplantation
10. Guthery SL, Pohl JF, Bucuvalas JC, et al. Bone mineral density in long-term survivors following pediatric liver transplantation. Liver Transpl
11. Valta H, Jalanko H, Holmberg C, et al. Impaired bone health in adolescents after liver transplantation. Am J Transplant
12. D’Antiga L, Ballan D, Luisetto G, et al. Long-term outcome of bone mineral density in children who underwent a successful liver transplantation. Transplantation
13. U.S. Department of Agriculture, Agricultural Research Service, Nutrient Data Laboratory. USDA National Nutrient Database for Standard Reference, Release 22 http://www.ars.usda.gov/ba/bhnrc/ndl
. Published 2009. Accessed June 1, 2010.
14. Greer FR, Krebs NF. Optimizing bone health and calcium intakes of infants, children, and adolescents. Pediatrics
16. Krebs NF, Himes JH, Jacobson D, et al. Assessment of child and adolescent overweight and obesity. Pediatrics
2007; 120 (suppl 4):S193–S228.
17. Greulich WW, Pyle SI. Radiographic Atlas of Skeletal Development of the Hand and Wrist. 2nd ed.Stanford, CA:Stanford University Press; 1959.
18. Bianchi ML, Baim S, Bishop NJ, et al. Official positions of the International Society for Clinical Densitometry (ISCD) on DXA evaluation in children and adolescents. Pediatr Nephrol
19. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis
20. Compston JE. Osteoporosis after liver transplantation. Liver Transpl
21. Carter DR, Bouxsein ML, Marcus R. New approaches for interpreting projected bone densitometry data. J Bone Miner Res
22. Ashby RL, Ward KA, Roberts SA, et al. A reference database for the Stratec XCT-2000 peripheral quantitative computed tomography (pQCT) scanner in healthy children and young adults aged 6-19 years. Osteoporos Int
23. Rosen CJ, Hochberg MC, Bonnick SL, et al. Treatment with once-weekly alendronate 70 mg compared with once-weekly risedronate 35 mg in women with postmenopausal osteoporosis: a randomized double-blind study. J Bone Miner Res