Mechanism of Bone Loss After Liver Transplantation: A Histomorphometric Analysis
Organ transplantation is associated with increased bone loss and high fracture risk, but the pathophysiological mechanisms responsible have not been established. We have performed a histomorphometric analysis of bone remodeling before and 3 months after liver transplantation in 21 patients (14 male, 7 female) aged 38–68 years with chronic liver disease. Eight-micrometer undecalcified sections of trans-iliac biopsies were assessed using image analysis. Preoperatively, bone turnover was low with a tendency toward reduced wall width and erosion depth. The bone formation rate increased from 0.021 ± 0.016 (mean ± SD) to 0.067 ± 0.055 μm2/μm/day after transplantation (p < 0.0002) and activation frequency from 0.24 ± 0.21/year−1 to 0.81 ± 0.67/year−1 (p < 0.0001). No significant change was observed in wall width, but there was a trend toward an increase in indices of resorption cavity size. There was a small increase in osteoid seam width postoperatively (p< 0.02) and decrease in mineralization lag time (p < 0.001). No significant changes in indices of cancellous bone structure were observed in the postoperative biopsies. These results demonstrate a highly significant and quantitatively large increase in bone turnover in the first 3 months after liver transplantation. Although no significant disruption of cancellous bone structure was demonstrated during the time course of the study, the observed changes in bone remodeling predispose to trabecular penetration and may thus result in long-term adverse effects on bone strength.
Osteoporosis is an important and common complication of organ transplantation, with fractures being reported in 25–30% of patients during the first year postoperatively.(1-5) Prospective studies have shown that bone loss is most rapid during the first 3 months after transplantation, with a tendency for bone mineral density to return toward preoperative levels after 1 year or more.(6) Although the pathogenesis of post-transplantation bone loss is poorly understood, it is believed that pre-existing bone disease and immunosuppressive agents, particularly glucocorticoids, are important contributory factors.(7)
The cellular pathophysiology of post-transplantation bone loss has not been clearly defined. In a prospective study of patients undergoing liver transplantation, McDonald et al.(8) reported histologic changes indicative of increased bone formation but were unable to demonstrate any increase in bone resorption. They concluded from these findings that the changes responsible for bone loss in their patients had occurred very early after transplantation and that by 3 months a compensatory increase in bone formation was occurring. In the present study, we have performed a comprehensive histomorphometric analysis of bone remodeling and structure in 21 patients with chronic liver disease before and 3 months after orthotopic liver transplantation.
MATERIALS AND METHODS
Of 70 consecutive adult patients admitted for assessment for liver transplantation to Addenbrooke's Hospital, Cambridge between April 1993 and December 1994, 21 patients (14 male and 7 female) aged 38–68 years (mean 54.9 years) were recruited into the study. The remaining 49 patients were excluded because of previous transplantation, acute liver disease, malignant disease, failure to give consent, or inability to obtain postoperative biopsies because of death or refusal to undergo a second biopsy. The cause of liver disease in the 21 patients recruited into the study was hepatitis C infection (n = 8), primary biliary cirrhosis (n = 4), alcoholic cirrhosis (n = 4), sclerosing cholangitis (n = 3), chronic active hepatitis (n = 1), and cryptogenic cirrhosis (n = 1). Five of the seven women were postmenopausal. At the time of the first biopsy, one woman was receiving tibolone and vitamin A and D tablets, one was receiving calcitriol and calcium therapy, one was receiving vitamin A and D tablets and two were being treated with 10 mg/day prednisolone, the durations of steroid therapy being approximately 2 and 5 years. None of the patients had received sodium fluoride, bisphosphonates, or calcitonin in the past.
Immunosuppression was achieved peri- and postoperatively using intravenous methylprednisolone, 10 mg/kg, followed by 1 mg/kg/day oral prednisolone, reducing to a maximum of 30 mg/day at 1 month and 5–10 mg/day at 3 months. Trough serum cyclosporine levels of 150–200 ng/ml were maintained immediately post-transplant, reducing to 125–175 ng/ml at 3 months. Rejection was treated with 3–8 g of intravenous methylprednisolone over 3–5 days, 9 patients experienced a single rejection, and in 2 patients rejection occurred twice during the study period. Immobilization was kept to a minimum, and the majority of patients were mobilized progressively from the second postoperative day onward.
Trans-iliac biopsies were obtained preoperatively and 3 months after transplantation from contralateral sides of the iliac crest, under local anesthetic using a modified Bordier trephine with an internal diameter of 8 mm. Informed written consent was obtained from all patients, and the study was approved by the local Ethics Committee. All patients received double tetracycline labeling prior to each biopsy(9) (300 mg of demeclocycline twice daily for 2 days, followed by a 10 day gap, 300 mg twice daily for 2 days, followed by the biopsy 3–5 days after the last dose). All biopsies were coded and the histomorphometric analysis was performed “blind” by the same observer. Paired biopsies from all 21 patients were available for standard histomorphometric analysis; for the structural analysis, for which the presence of both cortices is required, 14 pairs of biopsies were available.
Biopsies were embedded in LR white medium resin (London Resin Co., London, U.K.). Eight-micrometer undecalcified sections were stained by the von Kossa and toluidine blue techniques. Histomorphometric measurements were made using a “Digicad” digitizing tablet and cursor with an LED point light source (Kontron Ltd., Munchen, Germany) and an Olympus BHS-BH2 binocular transmitted light microscope with a BH2-DA drawing attachment (Olympus Optical Co. U.K. Ltd., London, U.K.). All histomorphometric data are described according to ASBMR nomenclature.(10)
Bone area/tissue area (B.Ar./T.Ar.), osteoid perimeter/bone perimeter (O. Pm./B.Pm.), and osteoid seam width (O.Wi) were measured on von Kossa–stained sections on a minimum of 25 fields from three to six sections. O.Wi was measured at four approximately equidistant points, or eight points on seams longer than 600 μm in length. A minimum of 20 seams per biopsy was measured on the same sections used for O.Pm. All seams with a width of 3 μm or more were measured.
The mean width of completed bone remodeling units (W.Wi) was measured on toluidine blue–stained sections viewed under polarised light at ×156 magnification. A minimum of 25 bone mineral units was measured on each biopsy from three to eight sections.
Tetracycline labeling was viewed by fluorescence microscopy on a minimum of six 15-μm unstained sections at ×156 magnification. Mineralizing perimeter (Md.Pm) was calculated as follows:
where dL.Pm is the double-labeled perimeter and sL.Pm is the single-labeled perimeter.
The mean distance between double labels was measured directly at ×312 magnification using the digitizing tablet and cursor. Measurements were made at approximately four equidistant points along the double labels. A minimum of 20 labels were measured for each biopsy on a minimum of six sections.
Mineral apposition rate was calculated as:
where L.Wi is the interlabel width and LP is the labeling period (12 days).
Adjusted appositional rate (Aj.AR), mineralization lag time (Mlt), and osteoid maturation period (Omt) were calculated as follows:
The tissue-based bone formation rate (BFR/B.Pm) was calculated as follows:
Activation frequency was calculated as:
Measurement of resorption cavity characteristics
The method described by Garrahan et al.(11) was adapted for use with the digitizing tablet and light cursor for measurement. Cavities were identified on toluidine blue–stained sections viewed under polarised light at ×156 magnification and measured at ×375 or ×750 magnification depending on the size of the cavity. Criteria for identification of resorption cavities included interruption of lamellae at an angle to the bone perimeter, absence of osteoid tissue, and depth > 3 μ m. A minimum of 20 cavities was assessed for each biopsy. The following indices were obtained: mean eroded depth (E.De, μm), maximum eroded depth (E.De Max, μm), eroded area (E.Ar, μm2), and mean cavity length (μm).
The bone section is viewed with a CCD camera mounted upon a light box (Sony, Tokyo, Japan), allowing the whole section to appear within a single field of view (magnification ×9). Images of the whole bone section are captured on an 386DX IBM-compatible AT-based computer system containing a Virtuoso frame store (Primagraphics, Royston, U.K.). All analysis software was written in the “C” language using a library of image processing subroutines (Foster Findlay Associates, Newcastle, U.K.). The stored images are converted to binary images which can be interactively edited to remove minor specimen preparation artefacts. The right and left corticomedullary junctions are defined automatically using a procedure that has been described previously.(12) Briefly, the width of each hole in the binary image is measured and given a value w. The computer generates an expanding contour which is equal to w/2 away from the edge of the hole. If > 75% of the expanded contour is covering bone, then the hole is filled in. If the contour covers < 75% of the bone, then the hole is left unfilled. The computer erodes all bone until only two objects remain (right and left cortices). The two objects are then dilated by the number of erosions plus two. The edge of the dilated objects is used as the corticomedullary delineation.
The upper and lower boundaries of the section are defined interactively by the operator. The upper and lower boundaries and right and left corticomedullary delineations define the “active” regions upon which all measurements are performed.
This method has been described in detail elsewhere.(13) Binary images are skeletonised to give a symmetrical axis of the original bone profile. The computer automatically identifies trabecular junctions, or nodes (Nd), and the ends of trabeculae, or termini (Tm). Individual trabeculae, or struts, are defined topologically; the indices thus generated include the ratio of nodes to termini (Nd/Tm), the terminus-to-terminus strut length (Tm.Tm), node-to-loop strut length (Nd.Lp), and node-to-terminus strut length (Nd.Tm), each of which are expressed per square millimeter of the total area included in the analysis.
Assessment of trabecular bone pattern factor
Trabecular bone pattern factor (Tb.Pf) was assessed automatically by measuring the trabecular area and perimeter within the active region of the binary image using the method originally described by Hahn et al.(14) The morphological operator, dilation, is applied to the binary image which adds a single pixel to the complete bone surface. B.Ar and B.Pm, including the dilated region, are then remeasured. The Tb.Pf is calculated from the following expression:
where P1 is the original perimeter, P2 is the dilated perimeter, A1 is the original area, and A2 is the dilated area.
Marrow space star volume
This was adapted from the original description of the method by Vesterby et al.(15,16) The active region is divided into approximately equal upper and lower areas, achieved by bisecting the active region with a line which joins the midpoints of the right and left corticomedullary junction. The computer generates a grid of sample points and lines, separated by 20 pixels, and superimposes them over the binary image. The points on every second grid line are displaced by a distance equal to half the separation of the lines. In order for the direction of the test lines to be uniformly distributed in three-dimensional space, possible angles of orientation are sine-weighted so that more angles will occur orthogonal to the vertical than the horizontal axis. A table containing 97 sine-weighted angles is stored in the memory of the computer, and the initial orientation of the grid lines is chosen by generating a random number between 1 and 97 which is used as an index to select an angle from the table. This ensures isotropically uniform random test directions in three-dimensional space. Any grid point falling on the marrow space is tracked along the grid line until it intercepts a trabecular edge or the edge of the active region. The distance between the grid point and the intercept (ln) is measured and raised to the power of three (l). Points in both upper and lower active regions are tracked toward the center of the section. The grid lines and points are rotated by incrementing the original index by 20 and using this value to select the next angle from the sine-weighted table, and the measurement procedure is repeated. The procedure is completed when the difference between the final and starting orientations is equal to or less than the value of the offset (20). Marrow space star volume (mm−3) is calculated from the expression
Log transformation was used to normalize non-normally distributed data. Paired data were compared using a two-tailed paired t-test. Data are expressed as the mean ± SD.
Indices of bone turnover and mineralization before and after transplantation are shown in Table 1. Prior to transplantation, static and dynamic indices of bone turnover were low when compared with age- and gender-matched controls, and there was also a trend toward lower values for wall width and erosion cavity depth when compared with healthy controls.(17-19)
Table Table 1.. Indices of Bone Turnover and Mineralization in Pre- and Post-Transplant Patients
Significant increases in O.Pm, Md.Pm, BFR at tissue level, and Acf were seen 3 months after transplantation. The BFR increased from 0.021 ± 0.016 μm2/μm/day (mean ± SD) to 0.067 ± 0.055 μm2/μm/day (p < 0.0002) and Acf from 0. 24 ± 0.21/year−1 to 0.81 ± 0.67/year−1 (p < 0.0001). There was also a small increase in O.Wi postoperatively (p < 0.02) and a decrease in Mlt (p < 0.001).
Indices related to remodeling balance are shown in Table 2. There was no significant change in wall width 3 months after transplantation. However, most indices of resorption cavity size showed a trend toward an increase in the postoperative biopsies.
Table Table 2.. Indices of Remodeling Balance Before and After Liver Transplantation
Cancellous bone structural indices are shown in Table 3. There were no significant changes observed in marrow star volume, Tb.Pf, or indices obtained by strut analysis. No correlations were observed between the pre-existing liver disease and changes in bone remodeling or structure following transplantation.
Table Table 3.. Indices of Cancellous Bone Microstructure Before and After Liver Transplantation
Our results demonstrate a highly significant increase in bone turnover 3 months after orthotopic liver transplantation in patients with chronic liver disease. These findings provide a pathophysiological basis for the early bone loss and high fracture incidence which occurs after organ transplantation. In the only other study of bone histomorphometry in patients undergoing liver transplantation,(8) an increase in osteoblast surface and O.Pm was seen 3 months postoperatively; although these changes were considered to indicate increased bone formation, they would also be consistent with increased bone turnover, although no increase in the surface extent of bone resorption was demonstrated. BFR and Acf were not assessed in that study and changes in remodeling balance could not be ascertained, since measurements of wall width and erosion depth were not performed.
No significant changes in wall width or erosion depth were demonstrated in the present study, although there was a trend toward an increase in most indices of resorption cavity size in the post-transplantation biopsies. Because of the relatively short study period, the values for wall width measured in the biopsies predominantly reflect those present before transplantation, and it was thus not possible to assess the effects of transplantation on wall width. In view of the lifespan of remodeling units, a period of at least 1–2 years is required before steady-state changes can be detected in remodeling balance.
In the absence of methods that directly assess dynamic indices of bone resorption, this information has to be obtained indirectly from measurements of BFRs.(20) Calculations based on this approach are only valid if bone is in a steady state and there is coupling of resorption and formation, neither of which assumption may safely be made for the post-transplantation biopsies in the current study. However, despite the lack of direct evidence for an increase in resorption rate, the increase in calculated Acf would be consistent with such a change, and the trend toward increased erosion cavity depth also supports an increase in osteoclast activity. Unfortunately, in this study urinary excretion of collagen-derived products was not measured but increased urinary hydroxyproline excretion has been reported in patients following cardiac transplantation.(21)
Despite the considerable increase in Acf, no significant changes in cancellous bone structure could be demonstrated in the postoperative biopsies. At a microanatomic level, bone loss may be accompanied by trabecular thinning or penetration of and erosion through trabeculae; these changes are to some extent interdependent and often coexist.(22) Increased bone turnover, particularly if accompanied by erosion of deeper cavities by osteoclasts, favors trabecular penetration and erosion, resulting in loss of connectivity of the cancellous bone and reduced bone strength. The absence of demonstrable structural alterations in cancellous bone in the present study may reflect the short study period and the absence of demonstrable increases in erosion depth. With respect to the former, highly significant changes in cancellous bone structure were demonstrated in a smaller number of women 6 months after the induction of estrogen deficiency with gonadotrophin-releasing hormone analogs(23); in that study, however, the changes in bone turnover and remodeling at 6 months were insufficient to explain the observed structural changes, possibly indicating that more severe abnormalities in bone remodeling had occurred earlier in the course of estrogen deficiency. These considerations highlight the problems associated with optimal timing of histomorphometric assessment in prospective studies in humans, in whom it is usually only practicable to obtain two biopsies. On the one hand, a short study period may be necessary to establish cellular mechanisms of bone loss, but on the other hand, longer periods of time may be required to demonstrate steady-state remodeling changes and the resulting structural alterations in cancellous bone.
The demonstration in our study of increased Acf after transplantation is consistent with reported changes, in other studies, of biochemical markers of turnover. Osteocalcin, a marker of bone formation, has been most widely studied and has been shown to increase in the first few months after liver or cardiac transplantation, values remaining elevated for up to 1 year.(21,24,25) Elevated serum levels of procollagen-1, also a marker of bone formation, were reported in a cross-sectional study of patients who had undergone liver transplantation,(24) and in a prospective study of patients undergoing cardiac transplantation, Sambrook et al.(21) reported increases in urinary hydroxyproline and calcium excretion immediately after transplantation, although by 6 months these changes were no longer apparent.
The pathogenesis of post-transplantation bone loss is poorly understood but is likely to be multifactorial. Patients with chronic liver disease have an increased prevalence of osteoporosis, and histologic studies have predominantly shown low bone turnover and reduced bone formation at the cellular level,(26,27) changes which were also observed in our patients prior to transplantation. Pre-existing bone disease may thus worsen the consequences of bone loss following transplantation, although the mechanisms responsible before and after operation differ, at least with respect to changes in bone turnover.
The large doses of glucocorticoids that are given perioperatively and in the early postoperative stages are also likely to play a major role in bone loss, as may other immunosuppressive agents such as cyclosporine and tacrolimus. Investigation of the effects of cyclosporine on bone has produced conflicting results, inhibition of bone resorption being reported in vitro but stimulation of bone resorption being demonstrated in vivo, the latter resulting in osteopenia and a high turnover state.(28-30) High-dose glucocorticoids also have complex effects on calcium metabolism which include primary effects on the kidney and bone formation, with a decrease in renal phosphate tubular reabsorption, increased synthesis of 1,25-dihydroxyvitamin D, and reduced synthesis of osteocalcin.(31) Secondary hyperparathyroidism is also associated with glucocorticoid therapy and may result both from direct effects on parathyroid hormone secretion and indirect effects mediated by reduced intestinal calcium absorption and increased renal calcium excretion. We have recently reported an increase in plasma parathyroid hormone levels 1 month after liver transplantation, values returning to near baseline 3 months postoperatively(32); although the observed increases mainly occurred within the normal range, their timing would be consistent with a causal relationship to post-transplantation bone loss, and the increased Acf observed in the present study would also support a role for hyperparathyroidism in the pathogenesis of post-transplantation bone loss.
Despite recent claims to the contrary,(33) post-trans-plantation bone disease is an important and growing clinical problem which results in considerable morbidity.(34) Although the doses of glucocorticoids now used for immunosuppression are considerably lower than those used in the past and rates of bone loss, assessed by bone densitometry, are correspondingly lower, there is no evidence that these changes have had a significant impact on fracture incidence, which remains high in our experience and that of others.(35) The changes that we have demonstrated in bone remodeling are likely to result in disruption of cancellous bone architecture and adverse effects on bone strength which may persist despite the restoration of bone mass in later years. Prevention of post-transplantation bone disease has been little studied, although the effects of calcium, calcitriol, sodium monofluorophosphate, alphacalcidol, and bisphosphonates, singly or in combination, have been reported with conflicting results.(36-38) The results of our study support the use of antiresorptive agents in the prevention of bone loss following transplantation; the time course of bone loss indicates that perioperative intervention is likely to be most effective.
Technical assistance was provided by C.-C. Huang. J.E.C. is supported by the Wellcome Trust, S.G. by the Medical Research Council, and S.V. by the National Osteoporosis Society of Great Britain and Northwest Anglia Regional Health Authority, U.K.