Long-term dalteparin in pregnancy not associated with a decrease in bone mineral density: substudy of a randomized controlled trial
Marc Rodger, The Ottawa Hospital, General Campus, 501 Smyth Road, Box 201, Ottawa, Canada ONT K1H 8 L6.
Tel.: +1 613 737 8899 (ext 74641); fax: +1 613 739 6102; e-mail: firstname.lastname@example.org
Summary. Background: The risk of decreased bone mineral density (BMD) with prophylactic dose long-term low-molecular-weight heparin (LMWH) is unknown. Objectives: We sought to determine whether long-term prophylactic dalteparin in pregnancy leads to loss of BMD. Patients/methods: Patients in a substudy of an ongoing multicenter randomized trial investigating the effect of antepartum dalteparin prophylaxis on pregnancy outcomes in thrombophilic pregnant women were randomized to either dalteparin 5000 U s.c. daily until 20 weeks and then 5,000 U s.c. q12 h until >37 weeks or to the control group. The primary outcome was absolute spine BMD at six weeks postpartum. Results: Of 77 patients eligible for the BMD substudy, 62 were analyzed. 33 patients received a mean of 212 days of dalteparin in the intervention group. 29 patients received a mean of 38 days of postpartum dalteparin in the control group. There was no difference in mean BMD between the intervention (1.11 g cm−2) and the control groups (1.14 g cm−2). Similarly, there was no difference in T-scores; the difference of −0.34 (95% confidence interval −0.93 to +0.25) in favor of the control group excludes a clinically important increase in fracture risk. Conclusions: Our results suggest that the use of long-term prophylactic dalteparin in pregnancy is not associated with a significant decrease in BMD.
Clinical trial registration: ISRCTN87441504 at http://www.controlled-trials.com.
Low-molecular-weight heparin (LMWH) is replacing unfractionated heparin (UFH) use in long-term heparin indications (including pregnancy) due to better bioavailability leading to more predictable dose–response, which allows less monitoring, less risk of heparin induced thrombocytopenia, and less risk of heparin-induced osteoporosis than UFH . Whether LMWH causes any loss of bone mineral density (BMD) is unknown and is best determined in a randomized controlled trial compared to no intervention.
Osteoporosis is defined by the World Health Organization as ‘a systemic skeletal disease characterized by low bone mass and micro-architectural alterations associated with increased bone fragility and consequent increase in fracture risk’ . Clinically, osteoporosis is usually silent and is diagnosed using bone mineral densitometry, which provides an important predictor of future fracture risk. BMD is expressed either as an absolute value (g cm−2) or as a T-score. The T-score measures the difference, by number of standard deviations, of an individual BMD from the young female reference mean. Osteoporosis is diagnosed when a T-score is −2.5 or less .
Many secondary causes of osteoporosis have been identified, including long-term UFH use, which has been termed heparin-induced osteoporosis (HIO) [4–9]. Up to 2.2% of pregnant women exposed to long-term UFH will develop painful and disabling osteoporotic compression fractures . Clinical trials and biologic experiments have clearly demonstrated less HIO with LMWH than with UFH [10,11]. In the most recent systematic review of LMWH exposure throughout pregnancy, one patient in 2777 published exposures developed an osteoporotic fracture (0.04%) . Given that pregnancy and perhaps lactation has been reported to lead to loss of BMD [12–15] the absolute risks of loss of BMD as well as of fracture in pregnancy attributable to LMWH are unknown.
The risks of long-term LMWH use are important to ascertain in order to counsel appropriately the growing number of patients using such treatment, including pregnant women with a variety of indications for its use (e.g. prevention of venous thromboembolism, of recurrent pregnancy loss, and of other placenta-related pregnancy complications). We have therefore sought to determine the risk of loss of BMD in pregnant women with long-term use of prophylactic dose dalteparin in a randomized controlled trial.
The Thrombophilia in Pregnancy Prophylaxis Study (TIPPS) BMD substudy is an a priori established substudy of a randomized controlled open label trial of antenatal LMWH prophylaxis in high-risk pregnant thrombophilic women. TIPPS is being conducted in 15 tertiary care hospitals in Canada and Australia (11 of which recruited subjects who were included in this substudy).
Thrombophilic pregnant women with prior placenta mediated pregnancy complications or at high risk of venous thromboembolism were potentially eligible for participation in TIPPS and the BMD substudy (detailed inclusion/exclusion criteria for TIPPS are listed in Appendix S1). Further exclusions for the BMD substudy included a history of previous bone or thyroid disease, long-term use of corticosteroids, metabolic bone agents or anticonvulsants.
Subjects were randomized to either the intervention group or the control group. Randomization was carried out in blocks of eight (blocks were prepared using random number tables by the trial statistician). Randomization was stratified according to gestational age (<8, 8–12 and 12–20 weeks). Treatment allocation was concealed. Local study nurses faxed randomization requests that included study screening number, subject initials and eligibility confirmation. Central study staff then randomized subjects by opening the next opaque sealed consecutively numbered envelope in the appropriate stratum and communicated the results of the randomization via facsimile to local study nurses.
Subjects were randomized to receive either of the following:
- 1Intervention: Subcutaneous injection of 5000 IU of dalteparin sodium once daily until 20 weeks gestation, then 5000 IU twice daily until 37 weeks or the onset of labor (at the discretion of the subject or the investigator).
- 2Control: No antepartum intervention.
All subjects (intervention and control) received 5000 IU s.c. dalteparin sodium daily from 1 day postpartum until 6 weeks postpartum, inclusive.
The primary outcome was absolute spine BMD measured once at the end of the postpartum period (6 weeks postpartum). BMD was assessed by two-site (spine and total hip) dual X-ray absorptometry (DXA) using pencil beam machines in use at the 11 clinical sites that participated: Lunar DPX for 58 participants (34/58 at the Ottawa Hospital; precision error: spine 0.012 g cm−2; total hip 0.009 g cm−2) and Hologic for four participants. Scans were interpreted and reported according to standardized published criteria . Absolute total hip BMD values obtained from different machines (Lunar vs. Hologic) were standardized using published formulas . Similarly absolute spine BMD values obtained from different machines (Lunar vs. Hologic) were standardized using published formulas .
Secondary outcomes included spine T-scores, spine osteopenia, spine osteoporosis, total hip absolute BMD, total hip T-scores, hip osteopenia, hip osteoporosis, and clinically evident fractures.
All analyses were conducted as treated, the conservative approach when assessing safety variables. The primary hypothesis was that prolonged prophylactic dose antepartum dalteparin use does not reduce spine BMD at the end of the postpartum period. The primary analysis compared mean spine BMD measured postpartum in the intervention group exposed to antepartum dalteparin with that in the control group, using a two-tailed, unpaired Student’s t-test.
In the secondary analyses, mean total hip BMD and total hip/spine T-scores measured postpartum in the intervention group were compared with those in the control group, using a two-tailed unpaired Student’s t-test. A comparison of proportions of those with osteopenia and osteoporosis was made using Fisher’s exact tests. Statistical analysis was conducted using spss version 12 (SPSS Inc., Chicago, IL, USA).
Sample size was determined a priori based on an estimated mean (SD) spine BMD 1.208 (0.12) g cm−2 from previously published literature [19,20]: to detect a minimal clinically important difference (detectable difference) in absolute BMD of 10%, which approximates a T-score of −1 (the osteopenic threshold), with 90% power (β = 0.10), and a 5% level of significance (two-tailed α = 0.05) yielding a sample size of 26 per arm (n = 52). Given a higher number of crossovers than expected, 62 subjects were analyzed.
Only subjects providing informed consent were enrolled in the TIPPS BMD substudy. Ethics approval was obtained from the institutional review boards of each of the participating centers. The protocol is consistent with the principles of the Declaration of Helsinki.
Role of funding source
Pharmacia (now Pfizer, Kirkland, Canada) provided study drug and partial funding for the main trial (TIPPS). TIPPS was partially funded by the Heart and Stroke Foundation of Ontario (Grant NA 4849). These funding sources had no input/influence into design of the study, the data collection, analysis, and interpretation of the data, or the decision to approve publication of the finished manuscript.
Between July 1, 2001 and November 30, 2005, 1200 patients were screened for entry into TIPPS: of 198 who were eligible for TIPPS, 114 consented to participate and were randomized. Of these 114 TIPPS participants, 37 were not eligible for the BMD substudy (17 subjects from a site not participating in the BMD substudy, 20 subjects who were either pregnant or <6 weeks postpartum), and 77 were potentially eligible for the substudy. Of these 77 women, 62 had postpartum bone mineral densitometry performed (six refused BMD, nine discontinued the intervention following pregnancy loss; see Fig. 1). Of these 62, 34 had been randomized to the intervention group and 28 to the control group. Seven women crossed over, three from the antepartum dalteparin group (intervention) to no antepartum dalteparin group (control), primarily due to a fear of injecting medication during pregnancy, and four from the no antepartum dalteparin group (control) to the antepartum dalteparin group (intervention), primarily due to a strong desire to receive dalteparin to possibly improve pregnancy outcome, leaving 33 women exposed to antepartum dalteparin and 29 unexposed to antepartum dalteparin.
Subjects in each group were not different in respect of age, race, weight, number of prior pregnancies, number of prior live births, smoking history, and calcium/vitamin D use (Table 1). Body mass index (BMI) was higher in the antepartum dalteparin group. Mean antepartum LMWH exposure was 174 days in the intervention and 0 in the control group; total mean LMWH exposure in the intervention group was 212 days compared to 38 days in the control group.
Table 1. Baseline characteristics of study participants comparing women who received antepartum dalteparin prophylaxis (intervention) with women who did not receive antepartum dalteparin prophylaxis (control)
|Age, mean (SD)||30.6 (4.0)||31.5 (4.7)|
|Race [n, (%) Caucasian]||28 (85)||27 (93)|
|Prior pregnancies, mean (SD)||2.7 (1.7)||3.2 (2.2)|
|Prior live births, mean (SD)||0.5 (0.7)||0.6 (0.7)|
|Prepregnancy, body mass index mean (SD)||30.4 (10.4)||28.4 (8.7)|
|Total days of LMWH, mean (SD)||211 (44)||38 (8)|
|Antepartum days of LMWH, mean (SD)||174 (40)||0 (0)|
|Postpartum days of LMWH, mean (SD)||38 (7)||38 (8)|
|Calcium or vitamin D use, n (%)||4 (12)||2 (7)|
|Smoker, n (%) |
| Never||20 (61)||16 (55)|
| Former||10 (30)||11 (38)|
| Active||3 (9)||2 (7)|
Primary outcome analysis demonstrated no difference in mean spine BMD in the intervention group (1.11 g cm−2) compared with that in the control group (1.14 g cm−2); the difference is −0.03 g cm−2 [95% confidence interval (CI): −0.10 to +0.04 g cm−2]. Similarly, there was no difference in T-score in the intervention group (−0.32) compared with that in the control group (+0.02) (difference −0.34, 95% CI: −0.93 to +0.25).
In secondary outcome analyses, no differences were found between the intervention group and the control group in total hip BMD or T-scores (see Table 2). No differences were found between the two groups when examining the BMD results by categorical outcomes of osteopenia or osteoporosis (see Table 3). There were no clinical fractures in either group.
Table 2. Absolute bone mineral density and T-scores 6 weeks postpartum in women who received antepartum dalteparin prophylaxis (intervention) vs. women who did not receive antepartum dalteparin prophylaxis (control)
|BMD spine (g cm−2; SD)||1.11 (0.15)||1.14 (0.12)||−0.03 (−0.10 to +0.04)||0.33|
|T-score spine (SD)||−0.32 (1.25)||0.02 (1.15)||−0.34 (−0.93 to +0.25)||0.25|
|BMD total hip (g cm−2; SD)||0.95 (0.12)||0.97 (0.11)||−0.02 (−0.08 to +0.03)||0.30|
|T-score total hip (SD)||0.07 (0.97)||0.35 (0.92)||−0.28 (−0.76 to +0.20)||0.25|
Table 3. Osteopenia or osteoporosis diagnosed on 6 weeks postpartum bone mineral densitometry in women who received antepartum dalteparin prophylaxis vs. women who did not receive antepartum dalteparin prophylaxis
| Osteopenia||7/33 (21, 9–39)||3/29 (10, 2–27)||0.25|
| Osteoporosis||1/33 (3, 0–16)||0/29 (0, 0–10)||0.35|
| Osteopenia||4/33 (12, 3–28)||1/29 (3, 0–18)||0.21|
| Osteoporosis||0/33 (0, 0–9)||0/29 (0, 0–10)||ND|
Because of differences in BMI between the antepartum dalteparin group and the no antepartum dalteparin group, we performed a linear regression analysis to control for BMI. Linear regression models with absolute spine BMD (dependent variable), antepartum dalteparin exposure (independent variable) with and without BMI (independent variable) were explored. While BMI was a significant predictor of BMD (P = 0.045), antepartum dalteparin exposure was not significant (P = 0.46) in the model with BMI and antepartum dalteparin exposure was not significant (P = 0.33) in the model without BMI.
This study demonstrates that long-term antepartum prophylactic dalteparin use does not cause significant loss of BMD when measured at 6 weeks postpartum. No previously published randomized clinical trial has examined whether long-term prophylactic LMWH leads to a reduction in bone density in treated subjects compared with a control group not exposed to such treatment. These results provide reassurance to patients and clinicians who contemplate using or recommending long-term dalteparin at prophylactic doses during pregnancy.
Osteoporosis is the most common serious side effect of long-term heparin use. HIO has been recognized for over 40 years. In 1965 Griffith reported on 115 patients who had received long-term UFH treatment . Ten patients received daily doses of UFH of 15 000–30 000 IU for at least 8 months, and six of these 10 patients sustained spontaneous fractures. The pathogenesis of HIO is not completely understood but several possible mechanisms have been postulated: heparin may have a direct effect on bone cells, possibly by increasing osteoclast activity and/or decreasing osteoblast activity; heparin may inhibit calcification through its high affinity for calcium ions; decreased levels of ionized calcium may stimulate parathyroid hormone, which in turn increases osteoclast activity and bone demineralization.
The pathogenesis may be uncertain, but the risk of osteoporosis and osteoporotic fracture with long-term heparin use is certain. HIO is particularly relevant in pregnancy, as a need for anticoagulation in pregnancy is one of the few indications for long-term heparin use. In a case review of 185 women receiving antenatal UFH prophylaxis, Dahlman reported that four women (2.2%) sustained osteoporotic fracture . A few controlled studies have examined bone density in women given antenatal and postpartum UFH therapy [6,19,20]. Two studies measured BMD at the initiation of UFH therapy and postpartum [19,20]; they suggest small (5%) but statistically significant decreases in bone density. A prospective matched cohort study measured vertebral BMD only during the postpartum period and found that bone density was significantly lower (7%) in the UFH group compared to controls .
Previous experimental and clinical research has demonstrated a lesser loss of BMD with LMWH than UFH. Work with rat models demonstrates that both UFH and LMWH cause a dose-dependent decrease in cancellous bone volume of the distal femur, although LMWH causes significantly less bone loss compared with UFH [11,21]. Furthermore, it was shown that both UFH and LMWH decrease the rate of bone formation, whereas only UFH increases the rate of bone resorption. In a randomized controlled trial in 44 pregnant subjects, Pettila et al.  demonstrated a significant reduction in bone density measured in serial postpartum lumbar spine BMD following long-term antepartum use of UFH (n = 23) compared with that following long-term antepartum use of LMWH (n = 21). Furthermore, there was no significant reduction in postpartum BMD in the LMWH group in comparison with a parallel cohort of untreated healthy postpartum women (n = 19), suggesting that LMWH does not cause HIO . Although this non-randomized comparison may have been underpowered to detect important differences, it was highly suggestive that LMWH does not cause HIO.
Other cohort studies have examined BMD in women given antenatal LMWH and have provided conflicting reports. Two published cohort studies assessed spine BMD postpartum in women receiving antenatal LMWH [23,24]. In one study of 26 women receiving antenatal LMWH, 30% developed osteopenia (BMD scans ≤1 SD below the mean BMD of an age-matched non-pregnant reference population) , while a similar study showed that BMD values of 11 treated women were not significantly different from those of a similar reference population . Two cohort studies examined the changes in BMD from the beginning of pregnancy to the postpartum period in women given antenatal LMWH [25,26]. In 123 pregnant women receiving either long-term UFH or long-term LMWH, a significant decrease in BMD of the spine (3.7%) and hip (0.9%) was observed between 12 weeks of gestation and immediately postpartum but no significant difference in BMD changes was seen between the group given UFH and the group given LMWH . In a second study, no significant losses in hip BMD were seen from the time of initiation of LMWH to 6–8 weeks postpartum . In a controlled cohort study of BMD changes with antenatal LMWH prophylaxis, Shefras  reported on 17 women given antenatal LMWH and eight pregnant controls; significant losses in BMD were seen between the initial scan (carried out between 6 months preconception and 12 weeks gestation) and the postpartum scan (carried out within 2 months postpartum) in treated women (5.4%) compared with controls (3.1%). No significant difference was observed in the BMD losses between the treated and the control groups.
LMWH is now widely used for thromboprophylaxis in pregnancy. This is primarily due to its ease of administration and its favorable safety profile . LMWH is increasingly being utilized not only for venous thromboembolism (VTE) prophylaxis in pregnancy but also to prevent placenta-associated pregnancy complications despite the poverty of evidence [27,28]. Although complications of LMWH use for the pregnant woman are uncommon, they can be serious and life-threatening. These include heparin-induced thrombocytopenia (HIT) [29,30], skin reaction and bleeding, in addition to the question of osteoporosis after prolonged LMWH therapy in pregnancy . Long-term LMWH has been thought to be weakly associated with osteoporosis and osteoporotic fractures; reported in only one patient (0.04%) in a recent systematic review by Greer . Our study reassures clinicians and patients that bone loss is unlikely to be a potential complication of prophylactic dose long-term LMWH in women of child-bearing age.
A significant strength of our study is that it is randomized. This is especially important, given the potential confounding effects of pregnancy-associated and lactation-associated osteoporosis. Several published reports have examined changes in bone density from prepregnancy to postpartum in normal pregnant women [12–15,31,32]. These studies demonstrate that pregnancy alone exerts minimal (<5% change in BMD), if any, effect on bone density but the possibility cannot be completely discounted.
A second strength of our study is that the sample size was powered to exclude clinically meaningful differences in BMD. This is best demonstrated by the 95% CI around the T-score differences at the hip or the spine, which excludes a difference as large as −0.93 that is outside the range for osteopenia range (T-score < −1). There are insufficient data to estimate the absolute fracture risk associated with a difference in the osteopenic range (< −1 T-score) in premenopausal women. If, however, we assume the relative fracture risk to be comparable with that in postmenopausal women, this study excludes an increase in fracture risk as high as 2-fold. Postmenopausal women with a T-score of −1.8 or less have an associated fracture risk in 1 year of about 2% . Given that the absolute risk of an osteoporotic fracture in women of child-bearing age is extremely low (<1%) , we can strongly reassure clinicians and patients that osteoporosis and osteoporotic fracture are not meaningful risks in women exposed to antepartum prophylactic dalteparin use. To answer definitively the question of whether dalteparin use increases the risk of osteoporotic fractures in this patient population would require an exceedingly large and unfeasible sample size.
This study has some limitations. Our study cannot determine whether higher dose (intermediate or full treatment dose) dalteparin or other LMWHs would lead to significant loss of BMD. Additional studies are required to answer these important questions. The study was inadequately powered to detect differences in categorical outcomes including osteopenia and osteoporosis. While mean population differences in bone mass may not be different with LMWH in comparison to no LMWH, it is possible that some patients inherently predisposed to HIO could exist and could develop osteoporosis. Indeed, in our study, 11 patients in the antenatal dalteparin group had osteopenia or osteoporosis compared to four patients in the control group. While these differences did not reach statistical significance, the trend suggests that further investigation will be required to eliminate the possibility of difference in categorical outcomes. Our control group was contaminated by postpartum dalteparin exposure and this may have biased toward a null effect. Given, however, that osteoporotic fracture has never been reported with short-term (<6 weeks) use of UFH, we are reassured that it is unlikely that dalteparin would have led to any significant bone loss in the control group with 6 weeks of dalteparin. It may not be possible to generalize our study to non-Caucasian populations as the large majority of our study participants were Caucasian. Finally, we did not conduct longer-term surveillance for bone loss, and hence it is possible that dalteparin exposure during pregnancy may predispose to osteoporosis in the longer term.
In conclusion, prolonged prophylactic dose dalteparin in pregnancy does not appear to cause significant loss of bone density measured at 6 weeks postpartum.
M. A. Rodger: Designed research, performed research, collected data, analyzed data, and wrote the paper. S. R. Kahn: Designed research, performed research, collected data, analyzed data, and wrote the paper. A. Cranney: Designed research, analyzed data, and wrote the paper. A. Hodsman: Designed research and wrote the paper. M. Kovacs: Performed research, collected data, and wrote the paper. A. M. Clement: Performed research, collected data, and wrote the paper. A. Lazo-Langner: Analyzed data and wrote the paper. W. M. Hague: Analyzed data and wrote the paper. A. Karovitch: Performed research, collected data, and wrote the paper. E. Keely: Designed research, performed research, collected data, and wrote the paper. M. Walker: Designed research, performed research, collected data, and wrote the paper. S. Robinson: Performed research and collected data. M. Blostein: Designed research, performed research, collected data, and wrote the paper. G. Smith: Performed research and collected data. J. Kingdom: Performed research and collected data. C. Demers: Performed research and collected data. S. Solymoss: Performed research and collected data. R. Khurana: Performed research and collected data. E. Rey: Performed research and collected data. P. S. Wells: Designed research, performed research, collected data, analyzed data, and wrote the paper.
M. A. Rodger is the recipient of the Maureen Andrew New Investigator Award from the Heart and Stroke Foundation of Ontario and holds a New Investigator Award from the Heart and Stroke Foundation of Canada. P. S. Wells is the recipient of a Canada Research Chair. S. R. Kahn is a recipient of a Senior Clinical Investigator Award from Fonds de la Recherche en Santé du Québec. A. Lazo-Langner is the recipient of a Graduate Scholarship from Consejo Nacional de Ciencia y Tecnología (CONACyT), México and is supported in part by an International Fellowship awarded by the University of Ottawa and by Program Grant PRG 5513 of the Heart and Stroke Foundation of Ontario. Thanks to M. Willson, N. Langlois, and C. Gagné-Rodger for assistance in preparing this manuscript.
Disclosure of Conflict of Interests
M. A. Rodger, S. R. Kahn, M. J. Kovacs, S. Solymoss and P. S. Wells have participated on Pfizer Advisory Boards and received compensation for this participation. M. A. Rodger, S. R. Kahn, S. Solymoss and P. S. Wells have received honoraria for lectures. M. A. Rodger, E. Rey and P. S. Wells have received grant funding from Pfizer. All other authors state that they have no conflict of interest.