Effect of teriparatide (rh-PTH 1–34) versus bisphosphonate on the healing of osteoporotic vertebral compression fracture: A retrospective comparative study

Pharmaceutical agents play a major role in the treatment of osteoporosis. Bisphosphonates, which inhibit osteoclast activity and bone resorption, are the current first-line medications for osteoporosis [1]. The robust suppression of osteoclast activity may hinder bone remodeling and maturation in the fracture healing process [2]. Although this adverse effect on fracture healing might raise concerns, such potential effects on healing have not been clinically evidenced in a meta-analysis of prospective randomized trials [3] and a systematic review [4].

Teriparatide (TPD, recombinant human parathyroid hormone (PTH) 1–34), an osteogenic osteoporosis agent, has been used increasingly for patients with severe osteoporosis and at high risk of fractures. Intermittent systemic administration of PTH induces bone formation through stimulation of osteoblast proliferation [5], prevention of osteoblast apoptosis [6] and increased osteoblast activity [5, 7]. This pharmaceutical agent may potentially enhance fracture healing.

Failure of fracture healing in osteoporotic vertebral compression fracture (OVCF) leads to intractable back pain associated with non-union [8, 9] and delayed vertebral collapse, resulting in neurological deficits [10]. Percutaneous vertebral augmentation is a widely accepted treatment option for OVCF. In particular, balloon kyphoplasty (BKP) may correct vertebral body height and local kyphotic deformity [11]. However, the differences in pain control between patients who received non-surgical care and kyphoplasty treatment were diminished at 12 months because the non-surgical group improved over time, probably as a result of fracture healing [12, 13]. Differences in the eventual outcome of patients with OVCF depend greatly on the union status. To date, however, whether teriparatide may enhance OVCF healing remains unclear. The current retrospective comparative study aimed to investigate the effect of teriparatide versus bisphosphonate on radiographic outcomes in the non-operative treatment of OVCF and to determine whether teriparatide enhances OVCF fracture healing.

Kaplan-Meier survival analysis showed a significant difference in the time-to-union between the two groups (p < 0.001, log-rank test) (Fig. 2). Cox regression analysis showed an adjusted relative hazard for TPD versus BP use of 1.86 (95% C.I.: 1.21–2.83). The radiographic union rate was 89% (34/38 patients) in the TPD group versus 68% (41/60 patients) in the BP group at six months after treatment, and the difference was significant (p = 0.026, Fisher’s exact test). At the final follow-up, 97% of patients (37/38 patients) in the TPD group and 90% of patients (54/60 patients) in the BP group achieved radiographic union (p = 0.243, Fisher’s exact test). Multiple logistic regression analyses were conducted using stepwise regression to adjust for heterogeneity (age, gender, fracture level, bone mineral density, pre-existing vertebral fracture, medication used, posterior wall fracture, and prior bisphosphonate use) between the two groups (Table 2). Regarding TPD versus BP use, the adjusted odds ratio for union at six months after treatment was 8.15 (95% C.I.: 2.02–43.33). Fracture site surgical interventions were not required in the TPD group, whereas two patients in the BP group eventually underwent surgical treatment. One patient was treated by balloon kyphoplasty for symptomatic non-union, and the other underwent posterior decompression and fusion for vertebral collapse with a neurological deficit.

The change of mid-vertebral body height was 4.4 mm in the TPD group and 3.4 mm in the BP group (p = 0.228, paired t-test) (Fig. 4). The change of local kyphosis was 4.6° in the TPD group and 3.8° in the BP group (p = 0.495, paired t-test) (Fig. 3). Progression of vertebral deformity (>15% loss of vertebral body height or >10° of kyphosis progression) was observed in 66% of the TPD group (25/38 patients) versus 60% of the BP group (36/60 patients), a difference that was not significant (p = 0.670, Fisher’s exact test). The union rate was 84% (21/25) in the TPD group, and 61% (22/36) in the BP group (p = 0.086, Fisher’s exact test). Multivariate logistic regression analysis using the stepwise method showed that teriparatide significantly was associated with an increased union rate; the adjusted odds ratio was 7.80 (95% C.I.: 1.41–70.35) in cases with vertebral deformity progression.

Union at six months was achieved in 61% (11/18 patients) of those with prior bisphosphonate use (past-BP) and in 71% (30/42 patients) of those without prior BP use in the BP group (p = 0.547, Fisher’s exact test). The union rate was 86% (18/21 patients) among those with prior BP use and 94% (16/17 patients) among those without prior BP use in the TPD group (p = 0.613, Fisher’s exact test). The union rate at final follow-up was 89% (16/18 patients) among those with prior BP use and 90% (38/42 patients) among those without prior BP use in the BP group (p = 1.000, Fisher’s exact test). Union occurred in 95% (20/21 patients) of those with prior BP use and in 100% (17/17 patients) of those without prior BP use in the TPD group (p = 1.000, Fisher’s exact test). Time-to-union significantly differed between the TPD group without prior BP use (3.4 ± 1.8 months) and the BP group without prior BP use (7.7 ± 1.1 months) (p = 0.042, t-test). Whereas prior bisphosphonate use tended to delay the time-to-union in the TPD group (3.4 ± 0.6 to 4.8 ± 0.5 months) (p = 0.070, t-test), it did not affect the time-to-union in the BP group (7.8 ± 1.4 to 8.3 ± 2.1 months) (p = 0.827) (Table 3). The duration of prior-BP use did not show a correlation with the OVCF time-to-union (Fig. 4).

Previous randomized controlled trials demonstrated that teriparatide accelerated fracture healing. Aspenberg et al. conducted a prospective, randomized, double-blind study of 102 postmenopausal women with distal radius fractures, and showed that the time to cortical bridging was significantly shorter in the once-daily 20-microgram teriparatide group than in the placebo control group [16]. Peichl et al. also demonstrated in their randomized controlled trial of 65 patients with osteoporotic pubic bone fractures that a once-daily injection of PTH 1–84 accelerated healing compared with treatment with vitamin D with calcium [17]. Additionally, several clinical cases were reported to show successful repair of non-union in sternum fractures [18] and dens fractures [19].

According to population-based epidemiologic studies, the prevalence of postmenopausal osteoporosis is approximately 30 to 45% in women aged 70 years [20, 21] and 40–45% in those aged 80 years [21, 22]. Osteoporotic fractures and related problems impact the quality of life in the older adult population. Failure of fracture healing in OVCF leads to intractable back pain associated with non-union [8, 9]. Patients with symptomatic vertebral non-union frequently require surgical interventions, including balloon kyphoplasty or spinal reconstruction. Furthermore, once non-union advances to osteonecrosis, the collapsed vertebrae cause a progressive kyphotic deformity and severe neural tissue compression with neurological deficits [8, 9, 23]. These pathologies can be managed only through major spinal reconstruction using instrumentation [24, 25]. Thus, the utmost priority in the treatment of OVCF is to prevent vertebral non-union and osteonecrosis.

In the current study, the risk factors for non-union were thoracolumbar fracture, decreased bone mineral density, posterior wall fracture, and prior bisphosphonate use. Vertebral non-union typically has been reported to occur mainly in the thoracolumbar zone [9, 13]. This phenomenon occurs due to the high stresses in this region generated by the transition from the rigid thoracic spine to the flexible lumbar spine along with the anterior element force transmission from the kyphotic thoracic spine to the relatively neutral thoracolumbar junction. Bone mineral density was negatively correlated with the intravertebral vacuum occurrence rate [26, 27]. Posterior wall fracture was reported to be a risk factor for progression of vertebral collapse [15, 28]. These factors might be considered related to the vulnerability of the small blood-supplying arteries in the vertebral bodies. Interruption of the blood supply and insufficient bone marrow neovascularization are the likely causes of avascular necrosis of the vertebral body, which may result in vertebral non-union [10, 13, 29]. Once vertebral necrosis occurs, union is difficult to achieve in the presence of a vertebral cleft. Peripheral bridging bone formation around an OVCF with vertebral clefting was often observed using teriparatide, as shown in Fig. 1. With peripheral bridging bone cross-linkage along the vertebral edges, the stability of the collapsed vertebra was recovered. Teriparatide was reported to promote ossification of the spinal ligaments [30] and to accelerate hyperplastic bone formation around vertebral fractures in a case of diffuse idiopathic skeletal hyperostosis [31]. In the current study, teriparatide significantly affected union in cases of vertebral deformity progression (adjusted odds ratio 7.80). Moreover, prior bisphosphonate use was an independent risk factor for non-union at six months. The substantial suppression of osteoclast activity with BP use might be implicated in the current results. Prior bisphosphonate use was associated with delayed union, particularly with teriparatide use. However, the duration of prior bisphosphonate use did not correlate with the time-to-union and the union rate of OVCF at the final follow-up. An explanation for this finding might be that bone metabolism after introducing the bisphosphonate may subside to a constant value by three months, with maintenance of that rate thereafter. Nenonen showed a rapid decrease and persistent change of bone metabolic markers after introducing a bisphosphonate [32]. Once the metabolism was suppressed by the bisphosphonate, bone formation would not ensue with the use of teriparatide.

Both the teriparatide and bisphosphonate groups were similar regarding the inhibition of vertebral deformity progression. Fractured vertebrae require several months to achieve adequate mechanical strength and a stable union. Angular kyphosis develops within several weeks or months after injury [33, 34]. Although teriparatide has the potential to increase bone mineral density and vertebral body strength [35, 36], vertebral deformity might progress before the medication has elicited its effects on vertebral mechanical strength.

Several limitations of the current study should be addressed. This is a retrospective comparative study that included different sample sizes and demographics between the two treatment groups. The inter-group difference in the ratio of prior bisphosphonate use was a potential source of selection bias in this study. To adjust for heterogeneity between the two treatment groups, we used multiple logistic regression analysis. The biological aspects of fracture healing should be examined prospectively using bone metabolic markers, vitamin D, and parathyroid hormone level status to determine whether the state of bone turnover was comparable between the TPD group and the BP group at baseline and to probe whether treatment affected bone turnover in these patients. Due to the nature of retrospective investigation, serum bone markers could be collected only in a very limited number of patients. In this retrospective study, the follow-up duration after three months following treatment differed for each doctor. Considering the follow-up duration differences, we showed the union rate on the Kaplan-Meier survival curve as the time course of vertebral union for each medication. An MRI study might help assess the early phase of fracture healing.