Presented in part at the 17th Annual Scientific Meeting of the American Society for Bone and Mineral Research, Baltimore, MD, U.S.A., 1995.
A significant proportion of patients will be long-term survivors of bone marrow transplantation (BMT) and little is known about their risk of late bony complications. We therefore evaluated bone mineral density (BMD) prior to BMT, post-transplantation changes in BMD, and mechanisms of bone loss in long-term survivors. We performed two analyses. The first was a cross-sectional study of 83 consecutive BMT patients (38 F, 45 M), examining the relationship between BMD and bone turnover, measured immediately prior to transplantation, and a number of disease and patient variables. The second was a prospective study of 39 patients (19F, 20 M) followed for a median of 30 months (range 5–64 months) following either allogeneic (allo, n = 29) or autologous (auto, n = 10) BMT to determine if bone loss was related to treatment of graft versus host disease (GVHD) with glucocorticoids and cyclosporine A, high bone turnover rates, or hypogonadism. Auto BMT recipients acted as a control group for effects of GVHD therapy on BMD. Prior to BMT, spinal and femoral neck (FN) BMDs were 8.6% and 14% lower in female auto BMT recipients than in female allo BMT recipients, respectively (p = 0.12 and p = 0.003). Urinary bone resorption markers were higher than in normal gender- and age-matched control subjects. Patients treated previously with glucocorticoids also had 8% lower FN BMD. Glucocorticoid-pretreated women with amenorrhoea had lower lumbar spine (LS) and FN BMDs than eumenorrheic women and women receiving HRT. Post-allo BMT, patients lost 11.7% of FN BMD compared with a nonsignificant decrease of 1.1% post-auto BMT (p < 0.001). Spinal BMD and total body bone mineral content (TBBMC) decreased by 3.9% and 3.5%, respectively, post-allo, compared with an increase (1.5%, p = 0.03) or nonsignificant decrease (−3.7%, p = NS), respectively, post-auto BMT. Post-allo BMT bone loss correlated best with the cumulative prednisolone dose at the LS and FN, and with average daily prednisolone dose for TBBMC. At the spine, the rate of bone loss was 4%/10 g of prednisolone, while the rate of bone loss at the FN was greater (9%/10 g of prednisolone). Bone loss was also negatively related to the duration of cyclosporine therapy for GVHD and baseline deoxypyridinoline concentrations. Avascular necrosis of the femoral head occurred in four, and vertebral and rib fractures occurred in one of the allo BMT patients, but in no auto BMT patients. In conclusion, BMT recipients are at risk of osteoporosis secondary to bone loss associated with their underlying illness and/or chemotherapy, particularly in female autograft recipients, and in allograft recipients secondary to GVHD and its treatment.
Bone marrow transplantation (BMT), or peripheral blood progenitor cell transplantation, is the treatment of choice for patients with certain hematological malignancies, many of whom will survive many years thereafter. Bone disease is a potential long-term complication.(1) While it is hitherto unknown if pretransplant skeletal demineralization occurs in BMT recipients, these patients are exposed to many processes predisposing to bone loss by some of the agents used to treat the underlying diseases. These include reduced physical activity levels; glucocorticoid-induced decreases in bone formation and 1,25-dihydroxyvitamin D3 concentrations; and hypogonadism secondary to the effects of chemotherapy, total body irradiation, and glucocorticoids.
Following BMT, an increase in biochemical bone resorption markers and a decrease in bone formation markers occur, creating circumstances promoting net bone loss.(2) In a small cross-sectional study, femoral neck (FN) and lumbar spinal (LS) bone mineral density (BMD) was decreased in both male and female patients following allogeneic (allo) BMT.(3)
There are a number of factors that may influence BMD in this context. Ovarian insufficiency occurs in the majority of women post-BMT. However, young, premenarchal women may recover ovarian function.(4,5) In men, normal spermatogenesis returns in up to 25% of post-BMT patients,(6) and in all patients hypothalamopituitary function is normal.(7) Cyclosporine, used to prevent or treat graft versus host disease (GVHD), leads to an acceleration of bone turnover and net bone loss.(1-2,8,9) Hypomagnesemia is potentiated by cyclosporine, causing bone pain and hypocalcemia.(8) Bone disease may also be a direct effect of GVHD itself on bone cells.(10) Abnormal cellular or cytokine-mediated bone marrow function may affect bone turnover and BMD post-BMT.(10-12) Post-transplantation, patients treated for chronic GVHD appear to be at greatest risk for bone loss.(12)
The aim of our study was 2-fold. First, we measured BMD and biochemical markers of bone turnover in a cohort of patients immediately prior to BMT and analyzed the effects of a number of disease and patient-specific variables on bone metabolism. Second, we prospectively studied a cohort of patients for over 2 years post-BMT to determine the incidence, severity, and time course of bone loss and, in the allo recipients, its relationship to gender and the doses of glucocorticoids and cyclosporine used to treat GVHD. The autologous (auto) BMT recipients acted as a control group to examine the effects of GVHD therapy on BMD because none received treatment with either prednisolone or cyclosporine post-BMT.
MATERIALS AND METHODS
The study population comprised 83 consecutive patients (38 F, 45 M) admitted for BMT selected over a period of 18 months. All patients, except two women of Chinese descent, were Caucasian. No patient selected had other diseases known to affect bone and mineral metabolism. None of the patients had lytic lesions of the LS or FN. Most patients undergoing BMT were self-caring and fully active. Prior to BMT, the majority of women were not taking either the oral contraceptive pill or hormone replacement therapy (HRT) with an estrogen and/or a progestogen. Details of the underlying disease and prior chemotherapy, including exposure to glucocorticoids or total body irradiation; smoking history; physical activity; and menopausal status were obtained by both a self-administered questionnaire, review of the medical record, and a telephone interview, where required.
Participants gave informed consent, and the research protocol was approved by the Ethics Committee of The Royal Melbourne Hospital Research Foundation.
Thirty-nine patients (19 F, 20 M) had a median of three (range one–six) repeat bone density scans over a median period of 30 months (range 5–64 months) following either allo (n = 29) or auto (n = 10) BMT. Thirty-one patients had a chemotherapy only conditioning regimen; seven of the allo and one of the auto recipients had conditioning which included total body irradiation.
For the first 2–4 months following BMT, all women received treatment with either medroxyprogesterone acetate (n = 15) or norethisterone acetate (n = 4) alone. All women subsequently received HRT with an estrogen and either a cyclical or continuous progestogen, except for one women with GVHD affecting the liver who received only 30 mg/day of medroxyprogesterone acetate. The type of HRT varied according to the prescribing physician and included 30–50 μg of ethinyl estradiol (n = 9), 0.625–1.25 mg of conjugated equine estrogens (n = 7), and 4 mg of oral or transdermal estradiol (n = 2).
Prophylaxis of GVHD in allo BMT patients was with intravenous cyclosporine A 3 mg/kg/day and short-course methotrexate. Oral cyclosporine at 6 mg/kg/day commenced when tolerated and continued for a period post-BMT which varied according to whether GVHD occurred and the risk of relapse. It was discontinued by day 60 in patients with high-risk disease.(13) Treatment of established GVHD was with the combination of intravenous methylprednisolone or oral prednisolone (2–3 mg/kg/day) and cyclosporine A (6 mg/kg/day). The prednisolone and cyclosporine doses were tapered when clinical control of GVHD was achieved. None of the auto BMT recipients were treated with either prednisolone or cyclosporine A post-BMT.
Biochemical markers of bone turnover
Blood samples were taken, and 2-h urine specimens were collected between 7 a.m. and 9 a.m. following an overnight fast, in order to measure serum bone formation markers and urine pyridinium cross-links, respectively. In the cross-sectional study, blood and urine samples were taken between days 1–8 of the menstrual cycle in premenopausal women with menses. All samples were stored at −70°C until analysis.
Markers of bone resorption
The urine total pyridinium cross-links, hydroxylysylpyridinoline or pyridinoline (Pyr) and lysylpyridinoline or deoxypyridinoline (Dpyr) were measured by monitoring fluorescence of eluates from high-performance liquid chromatography (HPLC).(14) The external Pyr standard used for this assay was prepared from human cortical bone and has been calibrated against purified Pyr. All pyridinium cross-link values were corrected for the individual measured recovery of the internal standard, isodesmosine.(15) Mean recoveries of Pyr and Dpyr, following cellulose extraction and HPLC, were 97% and 93%, respectively. Intra- and interassay coefficients of variation (CVs) were each 8% and 10% for Pyr and Dpyr, respectively.
Markers of bone formation
Serum bone alkaline phosphatase (BAP) was measured by duplicate immunoradiometric assays(16) using two monoclonal antibodies directed toward the bone isoenzyme of alkaline phosphatase (Hybritech, San Diego, CA, U.S.A.). The interassay CV was 10% and cross-reaction with other alkaline phosphatases was 6%. Serum osteocalcin (OC, also known as bone Gla-protein) was measured by duplicate immunoradiometric assays using antibodies raised against human OC (Immutopics, Palo Alto, CA, U.S.A.). The intra- and interassay CVs were 7% and 9%, respectively.
BMDs of the spine (second to fourth lumbar vertebrae) and the proximal femur were measured by dual-energy X-ray absorptiometry using a Hologic QDR-1000 W densitometer (Waltham, MA, U.S.A.). The in vitro and in vivo CVs were 0.38% and 1% at LS and 0.38% and 1.7% at the FN, respectively.(17) Standardized BMD was determined by comparison of the individual BMD with North American Hologic reference data and was expressed as number of standard deviations (Z score) different from the age- and gender-specific mean BMD.
Patients were stratified for type of transplant (52 allo, 31 auto). In the cross-sectional study, analysis of covariance was used to model BMD measurements as a function of the two groups (allo and auto), age, and biochemical markers of bone turnover. In the longitudinal study, BMD and biochemical bone marker concentrations were compared using auto patients as a baseline group, by t-tests. The associations of BMD with duration of glucocorticoid and cyclosporine therapy, total and average daily doses of glucocorticoids, and baseline biochemical markers of bone turnover were analyzed by multiple linear regression. Those bone marker levels highly skewed to the right were log transformed. Analyses were performed by generalized linear modeling under an assumed normal error structure, using the statistical package GLIM.(18) Unless otherwise noted, a nominal level of significance of p < 0.05 was used and all statistical tests were two-tailed.
The underlying malignancies, type of transplant, gender and age distribution, body mass index (BMI), and details of prior treatment in the cross-sectional study for patients in the cross-sectional study are listed in Tables 1 and 2. Leukemia was the most common indication for allo BMT, while lymphoma was the most common underlying disease for auto-BMT. Gender, age, BMI, prior radiotherapy, the number of prior treatment regimens and the pre-BMT illness duration were similar in both groups. However, glucocorticoid use was higher in the auto-BMT group because a higher proportion of auto than allo recipients were treated with chemotherapy regimens containing high-dose (prednisolone >20 mg/day) glucocorticoids (69% vs. 33%). This was attributable to a larger number of patients with lymphoproliferative disease and multiple myeloma in the auto group receiving treatment with first-, second-, and third-line regimens containing glucocorticoids.
Table Table 1.. Underlying Diseases in 83 Patients Undergoing Bone Marrow Transplantation
Table Table 2.. Pretransplantation Variables (Mean ± SD) Relating to Age, and Factors Affecting Bone Mass in Female and Male Recipients of Allogeneic or Autologous BMT
The pretransplant BMDs, total body bone mineral content (TBBMC), and biochemical bone marker concentrations, according the type of BMT and gender, are listed in Table 3. In female recipients of auto-BMT, LS and FN BMDs were 8.6% (0.6 SD) and 14% (1.1 SDs) lower, respectively, than in female allo BMT recipients (p = 0.12 and p = 0.003). Male auto BMT recipients had similar spinal and FN BMDs to male allo recipients. In all BMT recipients, both urinary total Pyr and Dpyr were higher than in normal gender- and age-matched control subjects. All auto BMT recipients had significantly higher total urinary Pyr than allo BMT recipients. However, total urinary Dpyr was significantly higher only in male recipients of auto BMT. The mean (± SD) serum creatinine concentration was 0.8 ± 0.3 mmol/l. Only three patients had serum creatinine concentrations above the normal range (0.13–0.20 mmol/l).
Table Table 3.. Pretransplantation Variables (Mean ± SD) Relating to Bone Mass and Biochemical Bone Turnover Markers in Female and Male Recipients of Allogeneic or Autologous Bone Marrow Transplants
A large proportion (42%) of women had amenorrhoea of greater than 6 months' duration prior to BMT, predominantly due to chemotherapy; four women were postmenopausal. Of 24 female allo BMT recipients, 9 (40%) had amenorrhoea compared with 7 (50%) auto BMT recipients. Only a minority of women (5 out of 38; 4 allo vs. 1 auto BMT recipient, p = NS) had received HRT with estrogen and a progestogen and another 9 (7 allo and 2 auto BMT recipients) had received a progestogen alone (northisterone acetate) prior to BMT. Similar proportions of auto and allo recipients received prior radiotherapy (16% vs. 9.6%, respectively, p = NS) and only 1 auto patient received radiotherapy to the spine. Recipients of allo and auto BMTs had a similar number of prior treatment regimens being on average ± SD, 1.6 ± 1.3 and 2.0 ± 1.0, respectively. The pretransplanation illness duration was also similar in allo and auto groups, being on average 21 ± 29 and 16 ± 24 months, respectively.
Pretransplantation glucocorticoids had an adverse effect on FN BMD and bone resorption rates. Patients treated previously with glucocorticoids had a FN BMD 8% lower than those who had not received glucocorticoids (0.845 ± 0.13 vs. 0.918 ± 0.15 g/cm2, p = 0.02; or expressed as Z scores: −0.2 ± 1.1 vs. 0.5 ± 1.4, p = 0.01). The standardized spinal BMD and TBBMC values were similar in both groups (data not shown). Women treated with pre-BMT glucocorticoids were analyzed separately. Twelve out of 22 women had amenorrhoea: 5 were treated with HRT, 4 had normal menstrual cycles, and 1 was treated with a progestogen alone. Both standardized LS and FN BMDs were lower in amenorrheic women compared with the other women (−0.9 ± 0.9 vs. 1.2 ± 1.8 SDs, p < 0.01; and −0.9 ± 0.9 vs. 0.6 ± 0.9 SDs, p < 0.01, respectively). Bone resorption markers were significantly higher in glucocorticoid pretreated patients (184 ± 96 vs. 111 ± 60, p < 0.01 for Pyr; and 29 ± 15 vs. 19 ± 13 nmol/pmol Cr, p = 0.02 for Dpyr, respectively); however, marker levels were similar in women and men.
Descriptive statistics for variables relating to illness type, age, bone mass, biochemical bone turnover markers, illness duration, oral glucocorticoid use, prior exposure to chemotherapy or total body irradiation, and tobacco use prior to BMT in the 39 recipients of allo or auto BMT followed prospectively are shown in Tables 4 and 5. Similar changes to those seen in the cross-sectional cohort were observed. Recipients of auto BMT had LS and FN BMD and TBBMC values that were 15.5% (1.1 SD), 23.0% (1.6 SD) and 16.0%, respectively, lower than in allo recipients. Bone turnover was also higher in auto recipients with Pyr, Dpyr, OC and BAP concentrations being 83%, 32% (p = 0.13), 38% (p = 0.33), and 112% higher, respectively.
Table Table 4.. Underlying Diseases in the Cohort of Bone Marrow Transplant (BMT) Recipients Followed Prospectively for a Median of 30 Months
Table Table 5.. Baseline Variables (Mean ± SD) Relating to Age, Bone Mass, and Biochemical Bone Turnover Markers in BMT Recipients Followed Prospectively for 30 Months
During the study, 20 (72%) of the allo and none of the auto recipients developed GVHD requiring glucocorticoid therapy. Auto BMT recipients (n = 10) therefore acted as a control group for the impact of myeloablative chemoradiotherapy on BMD and TBBMC in allo BMT patients (n = 29). Spinal and FN BMD increased or did not change following auto BMT, suggesting that myeloablative therapy alone did not cause bone loss. However, 12 months following allo BMT, LS and FN BMD decreased by 2.0% and 10.0%, respectively (p < 0.01 and p < 0.001), and total body BMC decreased by 2.6% (p < 0.01).
Over the median follow-up period of 30 months, approximately half of the allo BMT patients lost >10%, of whom six (21%) lost >20% of FN BMD post-BMT. The mean loss in FN BMD was 11.7% compared with a nonsignificant decrease post-auto BMT. Spinal BMD and TBBMC decreased by 3.9% and 3.5%, respectively, post-allo, compared with nonsignificant increases (1.5%, p = 0.03) or decreases (−3.7%, p = NS), respectively, post-auto BMT (Fig. 1). Only one patient reattained their pretransplant BMD, and bone loss continued for up to 30 months post-allo BMT (Fig. 2). The greatest bone loss occurred in the first year following BMT.
The correlation coefficients for changes in LS and FN BMDs, and TBBMC and quantity and duration of prednisolone therapy, duration of cyclosporine therapy, and biochemical bone markers in recipients of allo BMT are shown in Table 6. The average ± SD daily and cumulative prednisolone doses (CPD) were 19.6 ± 16.8 mg and 5.4 ± 6.3 g, respectively, for the duration of GVHD therapy. On average, patients received prednisolone for 45 ± 46 months and cyclosporine for 40 ± 39 months. Following allo BMT, seven patients received >10 g prednisolone, while only seven patients did not require prednisolone. Bone loss at the LS (Table 4 and Fig. 3) correlated best with CPD and was 4%/10 g of prednisolone (r = −0.54, p < 0.01). Changes in LS BMD were also significantly negatively related to duration of cyclosporine therapy and to the daily prednisolone dose (DPD). Bone loss at the FN also correlated best with the CPD and was greater than for LS (9%/10 g of prednisolone; r = −0.47, p < 0.05). Changes in FN BMD were also significantly negatively related to duration of cyclosporine therapy and positively related to baseline serum concentrations of OC. Loss of TBBMC was significantly negatively related to the DPD and the duration of cyclosporine therapy.
Table Table 6.. Correlation Coefficients (r) for Changes in LS and FN BMD, and TB BMC Following Bone Marrow Transplantation and Cumulative and Average Doses of Prednisolone, Duration of Cyclosporin, and Baseline Biochemical Bone Markers
For changes in LS BMD, 27% of the variance was explained by a model with CPD and duration of cyclosporine A therapy (p < 0.01), of which CPD was the better independent predictor (p = 0.05). A model including baseline bone turnover markers was of marginal statistical significance, explaining 19% of the variance in changes in LS BMD (p = 0.12), with Dpyr (p < 0.05) being the best independent predictor. For changes in FN BMD, a model of CPD and duration of cyclosporine A therapy explained 19% of the variance (p < 0.05) with CPD again being the better independent predictor (p < 0.05). A model including baseline bone turnover markers explained 37% of the variance (p = 0.07) and Dpyr (p = 0.08) was a better independent predictor than CPD (p = NS). For TBBMC, a model including DPD, duration of cyclosporine A therapy, and baseline Dpyr accounted for 60% of the variance (p = 0.001) with duration of cyclosporine A therapy (p < 0.05) and DPD (p < 0.01) being the best independent predictors.
Avascular necrosis of the femoral head occurred in 4 and clinically detected vertebral and rib fractures occurred in another of 116 BMT patients transplanted since 1993 who survived longer than 6 months. All events occurred in patients who underwent allo BMT and none in the auto BMT patients. Avascular necrosis was bilateral in three patients. The skeletal events occurred a median of 19 months post-BMT and occurred only in men. FN bone loss and the CPD (15.0 vs. 5.5 g, p = 0.03) were greater in these men compared with the other allo BMT patients.
This study demonstrates clear differences in BMD of auto and allo BMT recipients both before and after their transplant. We have shown that these relatively young patients are at increased risk of osteoporosis. Bone loss was associated with their underlying illness, amenorrhoea, and/or chemotherapy, including glucocorticoids, in female auto recipients and with GVHD and its treatment in male and female recipients of allo BMT. No bone loss occurred after auto BMT. Prednisolone and cyclosporine both had an additive adverse effect on BMD as previously noted following cardiac transplantation.(19,20) Bone loss from the spine, proximal femur, and total body was strongly related to cumulative exposure to glucocorticoids, whether used as part of pre-BMT chemotherapy or for treatment of GVHD. Bone loss was also related to duration of exposure to cyclosporine. Proximal femoral bone loss was also associated with high levels of pre-BMT bone turnover.
BMT is now more common than renal transplantation in Australia.(21) Male allo recipients are particularly prone to serious skeletal complications, including avascular necrosis of the femoral head. The skeletal complication of avascular necrosis of the femoral head deserves special consideration because of its considerable morbidity and difficulties in its diagnosis and treatment. Our findings are consistent with those of Enright et al. who found that 10.4% of allo BMT, but no auto BMT recipients developed avascular bone necrosis a median of 12 months post-BMT.(22) Almost all cases had received glucocorticoid therapy for GVHD, and the extent of the disease was related to both cumulative gluccocorticoid dose and increasing age, but not gender.
Although our number of cases was small, the possibility that increased rates of bone loss at the FN may be associated with a higher incidence of avascular necrosis of the femoral head should be examined in larger prospective studies. Neither did we address the specific roles of hepatic GVHD and/or vitamin D deficiency, or hypogonadism in the etiology of this complication, but these factors also merit further investigation.
Increased bone turnover was present prior to BMT and may have resulted in subsequent bone loss, particularly at the proximal femur. Another smaller prospective study has shown that serum bone formation markers decrease while bone resorption markers increase post-BMT, favoring net bone loss, although BMD was not measured.(2)
It is not known whether treatment with testosterone in males, or with calcitriol, bisphosphonates, or calcitonin prevents bone loss following BMT. Both bisphosphonates and calcitonin are effective in ameliorating bone loss after cardiac transplantation.(23,24) Recently, the bisphosphonates, etidronate and alendronate, have been shown to prevent bone loss at the spine and proximal femur in patients commencing glucocorticoids.(25,26) Both may also reduce vertebral fracture rates. However, the rates of bone loss at the FN in the placebo groups of the glucocorticoid-induced osteoporosis studies were less than in our current study, although the rates of spinal bone loss were similar. Nevertheless, inhibition of bone resorption with antiresorptive drugs could therefore potentially reduce post-BMT bone loss.
Serum 1,25-dihydroxyvitamin D3 concentrations decrease acutely following BMT(8) and calcitriol alone has recently been shown to be effective in preventing proximal femoral bone loss following cardiac or single lung transplantation.(27) Calcitriol also prevents spinal bone loss following initiation of glucocorticoid therapy for rheumatic diseases.(28) There is also evidence in cardiac transplant recipients that the combination of calcitriol with bisphosphonates may prevent post-transplantation bone loss.(24)
In our study HRT with a progestogen initially, then with a combination of an estrogen and a progestogen in the conventional doses used for treatment of postmenopausal osteoporosis, did not prevent bone loss in women following BMT. This may be related to initial lone progestogen treatment not preventing bone loss, as seen in our recent study of postovariectomy bone loss,(29) or because estrogen was also unable to prevent rapid post-BMT proximal femoral bone loss. The latter is suggested by continuing FN bone loss in women after the initial 3-month phase of engraftment when progestogens alone were used. Therapy with estrogen and cyclical progestogens for 12 months commencing, on average, 13 months, following allo or auto BMT increased LS BMD in a small cohort of women.(30) However, these women may have passed through the most rapid phase of post-BMT bone loss. In addition, the effects of HRT on preventing proximal femoral bone loss were not studied.
In conclusion, auto-BMT recipients, particularly female, are at increased risk of osteoporosis secondary to pretransplant bone loss associated with their underlying illness and/or its treatment with cytotoxic drugs and glucocorticoids. Amenorrhoea is a common sequela and, whenever possible, it should be treated with HRT to prevent bone loss and to ameliorate the adverse skeletal effects of glucocorticoids. By contrast, prolonged post-BMT use of glucocorticoids and cyclosporine A for the treatment of GVHD in allo BMT recipients is associated with substantial bone loss and a significant, albeit small, risk of avascular necrosis of bone. Post-allo BMT bone loss was also related to high pre-BMT bone turnover rates and it was not prevented by HRT in women. It is in this group of BMT patients, in particular, that prospective studies of prevention of post-BMT bone loss with antiresorptive drugs should now be considered.
We thank Associate Professor John Wark for supervising the bone densitometry. We gratefully acknowledge the expert technical assistance of Mr. Bahtiyar Kaymakci, Ms. Stella Yeung, and Ms. Cathy Poon. We are indebted to the nursing staff of the former Ward 5 West Day Care Center. This work was funded by the D.W. Keir Fellowship, The Royal Melbourne Hospital, and The Australian National Health and Medical Research Council.