Teriparatide for acceleration of fracture repair in humans: A prospective, randomized, double-blind study of 102 postmenopausal women with distal radial fractures

Authors


  • Data included in this manuscript were presented in part at the 35th ECTS Congress, Barcelona, Spain, 24–28 May 2008. Abstract publication: Aspenberg P, Genant HK, Johansson T, Nino AJ, See K, Krohn K, Garcia P, Recknor CP, Einhorn TA, Dalsky GP, Lakshmanan M. Effects of teriparatide on distal radial fracture healing in postmenopausal women: a randomized double-blinded study [abstract no. Su-P417]. Calcif Tissue Int. 2008;82:S218.

Abstract

Animal experiments show a dramatic improvement in skeletal repair by teriparatide. We tested the hypothesis that recombinant teriparatide, at the 20 µg dose normally used for osteoporosis treatment or higher, would accelerate fracture repair in humans. Postmenopausal women (45 to 85 years of age) who had sustained a dorsally angulated distal radial fracture in need of closed reduction but no surgery were randomly assigned to 8 weeks of once-daily injections of placebo (n = 34) or teriparatide 20 µg (n = 34) or teriparatide 40 µg (n = 34) within 10 days of fracture. Hypotheses were tested sequentially, beginning with the teriparatide 40 µg versus placebo comparison, using a gatekeeping strategy. The estimated median time from fracture to first radiographic evidence of complete cortical bridging in three of four cortices was 9.1, 7.4, and 8.8 weeks for placebo and teriparatide 20 µg and 40 µg, respectively (overall p = .015). There was no significant difference between the teriparatide 40 µg versus placebo groups (p = .523). In post hoc analyses, there was no significant difference between teriparatide 40 µg versus 20 µg (p = .053); however, the time to healing was shorter in teriparatide 20 µg than placebo (p = .006). The primary hypothesis that teriparatide 40 µg would shorten the time to cortical bridging was not supported. The shortened time to healing for teriparatide 20 µg compared with placebo still may suggest that fracture repair can be accelerated by teriparatide, but this result should be interpreted with caution and warrants further study. © 2010 American Society for Bone and Mineral Research

Introduction

An estimated 9.0 million osteoporotic fractures occurred worldwide in 2000, half of them in the Americas and Europe, according to a recent report from the World Health Organization (www.who.int/entity/chp/topics/Osteoporosis.pdf). Johnell and colleagues1 estimated the lifetime risk of an osteoporotic fracture to be 40% to 50% in women and 13% to 22% in men. Annual direct-care costs attributable to osteoporotic fractures were estimated to range from $12.2 billion to $17.9 billion in the United States, measured in 2002 dollars (www.surgeongeneral.gov/library/bonehealth/docs/full.report.pdf), and these societal expenditures are expected to increase.2

Fractures in the elderly lead to a great and often irreversible loss of quality of life and are associated with an increase in mortality.3, 4 It is not only the fracture itself that has these detrimental effects but also the associated events and complications that appear during the healing time. Pain and immobility after osteoporotic fractures in the spine or hip lead to severe loss of quality of life and, partly in consequence, increasing general frailty. Many patients with osteoporotic fractures cannot tolerate load bearing even after surgery. In these cases, even a moderate reduction in the time to load bearing would be of benefit.

It is estimated that of the 7.9 million fractures sustained each year in the United States, 5% to 20% result in delayed or impaired healing.5, 6 However, there are a limited number of adjunctive therapies that can be used to accelerate the fracture healing process. Kristiansen and colleagues7 demonstrated that low-intensity ultrasound significantly accelerated fracture healing in adult men and women who had sustained a closed, dorsally angulated metaphyseal fracture of the distal radius. The local application of recombinant bone morphogenetic protein 2 (BMP-2) has been evaluated in patients with open tibia fractures.8 Currently, BMP-2 is available for treatment of acute open tibia fractures that have been stabilized with an intramedullar nail. BMPs have demonstrated enhanced healing in both preclinical and clinical studies, although the response in humans is less robust than that observed in animals.6, 9 No other successful pharmacologic treatment for fracture repair has been described in humans.

Reviews of preclinical studies concluded that once-daily administration of systemic recombinant teriparatide [hPTH(1-34)] enhanced the morphometric and mechanical properties of fracture calluses and accelerated fracture healing and suggested that this treatment may be used successfully in clinical practice.10–12 However, to date, no published clinical trials have investigated the effects of parathyroid hormone (PTH) treatment on fracture healing in humans.

Intermittent administration of systemic teriparatide has been shown to enhance callus formation and mechanical strength of fractures in young13–19 and old rats.20 In dose-response studies of teriparatide, a beneficial effect on fracture healing has been demonstrated.17, 19 Skripitz and colleagues18 reported that teriparatide had early and dramatic effects on new bone formation studied in a bone chamber, whereas sites undergoing normal remodeling showed hardly measurable effects after the relatively short treatment time of 6 weeks. Stimulatory effects also were found in rat models for implant fixation, bone implant contact, and callus distraction osteogenesis. These findings from preclinical studies suggest that fracture repair in humans may be stimulated by a short duration of teriparatide treatment.

The aim of this study was to determine the effect of systemic recombinant teriparatide treatment on time to radiographic repair of conservatively treated distal radial fractures in postmenopausal women.

Materials and Methods

The primary objective of this study was to compare the effect of 8 weeks of once-daily subcutaneous (SC) treatment with teriparatide 20 or 40 µg versus placebo on time to radiographic healing in postmenopausal women with a unilateral dorsally angulated distal radius (Colles') fracture. Radiographic healing was defined by cortical bridging in three of four cortices.21, 22 The Colles' fracture was chosen as the model to test the effects of teriparatide treatment on fracture healing because the distal radius includes both trabecular and cortical bone, is accessible for radiographs, has little soft tissue that can distort the radiograph, and is amenable to multiple functional endpoints. Secondary objectives include time to healing of four cortices evaluated from X-rays, time to early or partial endosteal healing, time to trabecular union, time to cortical bridging evaluated qualitatively by computed tomographic (CT) scans,22, 23 evaluation of anatomic deformity, and functional assessments.

Patients

Patients for this study were recruited from 14 clinical centers in 7 countries (Canada, Mexico, Poland, Romania, Spain, Sweden, and the United States); enrollment began in December 2004. One-hundred and thirteen ambulatory women 45 to 85 years of age who were at least 2 years postmenopausal and had sustained a unilateral dorsally angulated fracture of the distal radius (Colles') within the past 10 days were eligible to enter the study. Local investigators evaluated X-rays to assess patient eligibility. Patients must have received conservative treatment of the fracture, including closed reduction and immobilization, and were free of severe or chronically disabling conditions other than the distal radius fracture. Participants had to be willing and able to satisfactorily use a pen-type delivery system or be willing to receive daily subcutaneous injections from a care partner trained to use the injector. At screening, chemistry and hematology tests were analyzed by local laboratories. Serum PTH, 25-hydroxyvitamin D, and thyroid-stimulating hormone (TSH) were analyzed at a central laboratory. Patients were excluded from the study if they had sustained previous fractures or bone surgery in the currently fractured distal forearm or had joint diseases that affect the function of the wrist and/or hand of the injured arm. Additional exclusion criteria included diseases other than primary osteoporosis that affect bone metabolism or responses to therapy, including an elevated serum calcium based on the local laboratory reference range; serum PTH > 70 pg/mL; 25-hydroxyvitamin D < 12 ng/mL; active liver disease or clinical jaundice; or history of symptomatic nephro- or urolithiasis within 2 years. Patients were excluded if they had a known allergy to teriparatide or any form of PTH hormone or analogue, an increased baseline risk of osteosarcoma (Paget's disease of the bone, previous primary skeletal malignancy, or skeletal exposure to therapeutic irradiation), a history of a malignant neoplasm in the 5 years prior to the study (with the exception of superficial basal cell carcinoma or squamous cell carcinoma), and carcinoma in situ of the uterine cervix treated less than 1 year prior to the study.

Ethical review boards at each clinical center approved the study; study patients gave written informed consent before participating in any study procedures.

Study design

This is a phase 2, 53-week, prospective, randomized, parallel, double-blind, placebo-controlled, multicentered, multinational study. It consisted of four periods: a screening period of a maximum of 1 week from the day of the fracture to the day of randomization, a treatment period of 8 weeks (teriparatide or placebo), a follow-up period without treatment for 8 weeks, and a safety extension for an additional 36 weeks. During the screening period, patients began supplements of 1000 mg/day of elemental calcium and 800 IU/day of vitamin D that were continued throughout the study. The placebo and teriparatide treatments used in the study were provided by Eli Lilly and Company (Indianapolis, IN, USA).

Efficacy and safety variables

Visits were scheduled for assessment of efficacy variables and safety at 2-week intervals during the treatment and follow-up periods and at a final visit at the end of the safety extension. The primary efficacy variable, time to radiographic healing, was defined as the interval in days between the occurrence of the fracture and the time when bridging in three of four cortices was seen on X-ray images. A determination was made at each follow-up evaluation for the two cortices (radial and ulnar) visible on the anteroposterior X-ray film and the two (dorsal and volar) seen on the lateral film.22 Secondary objectives included radiographic and CT scan evidence of time to healing of four cortices bridging, early and partial endosteal healing, and trabecular union; CT scans also were used to assess time to healing of three of four cortices bridging.

Radiographs and CT scans were assessed by a central quality assurance and reading service (Synarc, Inc., San Francisco, CA, USA). CT scans were obtained for four visits during the treatment period at a subset of sites and read for qualitative features of healing. Once radiographic healing had been determined for two consecutive visits, further X-ray films were not performed except at the week 53 visit, the end of the safety extension period. The measurements for the deformity readings were taken at the end of the study at week 53 using last-observation-carried-forward analysis.

Before initiating the evaluation, the two readers underwent standardized training that involved reviewing case material of a matching study population in consensus sessions with discussions of criteria for defining scores and the effect of artifacts and interfering conditions. Readings were performed in batches after patients completed all available visits. The readers were blinded to the patients' demographics and treatment assignments but not to the sequential order of the visits. Double readings were performed independently for the assessment of radiographic union. In case of discrepancies among the readers with regard to the time point at which union occurred, a third independent reader reviewed the case to break the tie; this result was considered to be final. Single readings were performed for the evaluation of anatomic deformity on radiographs and the assessment of fracture union using CT scans.

Functional assessments included the self-administered Patient-Rated Wrist Evaluation (PRWE) questionnaire24 and assessment of grip strength via a Jamar dynamometer.25 The self-administered PRWE is a 15-item questionnaire that rates wrist-related pain and disability in functional activities including six specific tasks and questions in which the patient rates his or her ability to perform the usual level of function in the domains of self-care, work, household work, and recreation. The scores range from 0 (no disability) to 100 (severest disability). A negative score, as a percent change from baseline, reflects an improvement in function. The mean of the three measurements was considered as the grip strength for a patient at that visit. To adjust for hand dominance in grip strength, if the nondominant hand was injured, the percentage was multiplied by 1.07; if the dominant hand was injured, the percentage was multiplied by 0.93.

Safety assessments included reports of adverse events (AEs), serious adverse events (SAEs), early discontinuations because of AEs, clinical laboratory tests, and physical examination.

Statistical methods

Comparisons for baseline continuous variables were based on an F test for treatment using type III sum of squares, and the model dependent variable equals treatment plus region; categorical variables were based on Fisher's exact test. Estimated median healing time from fracture to radiographic healing, as defined earlier, was evaluated using nonparametric survival analyses; primary analyses used the Gehan-Wilcoxon test,26 and secondary analyses used the bootstrap method for standard error of the difference to calculate the confidence interval (CI).27 Hypotheses for time to radiographic fracture healing for teriparatide 40 µg versus placebo and teriparatide 20 µg versus placebo were tested sequentially, each at a two-sided alpha equal to 0.05, using a gatekeeping strategy; if teriparatide 40 µg versus placebo was not significant, inference for teriparatide 20 µg versus placebo was not to be performed.28 In post hoc analyses for the teriparatide 20 µg versus placebo and teriparatide 40 µg versus 20 µg comparisons, the p value was based on the Gehan-Wilcoxon test to compare time to healing, and the bootstrap method for standard error of the median time difference was used to calculate the 95% CIs. The product-limit method was used to estimate the cumulative probability of healing with respect to time and to obtain the median healing time. The Cox regression model was used to test the treatment effect after adjusting for possible explanatory variables.

Secondary radiographic and CT scan parameters were analyzed in the same manner as time to fracture healing. Parameters to evaluate the degree of deformity (e.g., palmar tilt, radial angle, ulnar variance) were analyzed using analysis of variance (ANOVA) and nonparametric methods (Kruskal-Wallis and Wilcoxon rank-sum test). The treatment comparisons were performed separately for each measurement. In addition, for study of treatment group differences for combined subgroups in time to healing, Cox regression analyses were performed. The patients were categorized as having a poor, intermediate, or good anatomic positioning ranking based on the degree of deformity.

The changes from baseline in PRWE scores were analyzed using the mixed-model repeated-measures (MMRM) method with baseline score, treatment, region, treatment-by-region interaction, visit, and treatment-by-visit interaction as fixed effects and with the patient nested in treatment as the random effect in the model. Adjusted percentages in change from baseline for grip strength were analyzed using the MMRM method with baseline score, treatment, region, treatment-by-region interaction, visit, and treatment-by-visit interaction as fixed effects and with the patient nested in treatment as the random effect in the model. Comparisons of adverse events were made using a region-stratified Cochran Mantel Haenszel test. Efficacy and safety analyses were performed on a modified intent-to-treat population (patients receiving at least one dose of study drug); analyses of the primary efficacy variable also were performed on per protocol population (patients without major protocol violations). Longitudinal efficacy analyses were performed on patients with at least one postrandomization observation for the variable of interest. All analyses were done at two-sided alpha equal to 0.05.

Results

Data were collected between December 2004 and June 2007. A total of 113 women were screened, and 106 women were enrolled into the study; four women discontinued before receiving any treatment (Fig. 1). The remaining 102 women were randomly assigned to once-daily placebo (n = 34) or 20 µg (n = 34) or teriparatide 40 µg (n = 34) injections (see Fig. 1). Most patients in each group completed 8 weeks of treatment and 8 weeks of follow-up (placebo 27/34, teriparatide 20 µg 29/34, teriparatide 40 µg 26/34). Baseline characteristics were balanced across the three groups (Table 1).

Figure 1.

Disposition of patients.

Table 1. Baseline Demographic, Injury Characteristics, and Functional Variables
 Placebo (n = 34)Teriparatide 20 µg (n = 34)Teriparatide 40 µg (n = 34)Total (n = 102)Overall p value
  • a

    Comparison for continuous variables based on F test for treatment using type III sum of squares, and the model-dependent variable = treatment + region.

  • b

    Comparison for categorical variables based on Fisher”s exact test.

Agea (years), mean ± SD61.7 ± 8.659.5 ± 9.662.8 ± 7.361.4 ± 8.6.315
Caucasian,bn (%)24 (71%)20 (59%)22 (65%)66 (65%).638
Current smoker,bn (%)5 (15%)9 (27%)4 (12%)18 (18%).348
Patient-Rated Wrist Evaluation (PRWE) score,a mean ± SD80.8 ± 12.478.4 ± 19.775.8 ± 16.678.3 ± 16.4.536
Energy of injury,bn (%)    .496
 Low33 (97%)32 (94%)30 (88%)95 (93%) 
 High1 (3%)2 (6%)4 (12%)7 (7%) 
Displacement status,bn (%)    .843
 Yes29 (85%)29 (85%)27 (79%)85 (83%) 
 No5 (15%)5 (15%)7 (21%)17 (17%) 
Fracture of ulnar-styloid process,bn (%)    1.000
 Yes11 (32%)11 (32%)11 (32%)33 (32%) 
 No23 (68%)23 (68%)23 (68%)69 (68%) 
Involvement of radio-ulnar joint,bn (%)    .825
 Yes17 (50%)16 (47%)14 (41%)47 (46%) 
 No17 (50%)18 (53%)20 (59%)55 (54%) 
Involvement of radiocarpal joint,bn (%)    .496
 Yes16 (47%)13 (38%)11 (32%)40 (39%) 
 No18 (53%)21 (62%)23 (68%)62 (61%) 
Comminution,bn (%)    .081
 Yes23 (68%)16 (47%)14 (41%)53 (52%) 
 No11 (32%)18 (53%)20 (59%)49 (48%) 
Impaction,bn (%)    .773
 Yes24 (71%)21 (62%)24 (71%)69 (68%) 
 No10 (29%)13 (38%)10 (29%)33 (32%) 
Days between fracture and start of treatment,a mean ± SD5.5 ± 2.75.6 ± 2.06.3 ± 2.15.8 ± 2.3.377
Days of duration of follow-up from date of fracture,a mean ± SD109 ± 31109 ± 32114 ± 27111 ± 30.787

Primary objective

The estimated median time from fracture to first radiographic evidence of complete cortical bridging in at least three of four cortices, the primary objective, was 9.1, 7.4, and 8.8 weeks in the placebo, teriparatide 20 µg, and teriparatide 40 µg groups, respectively (Fig. 2A; overall p = .015). There was good agreement between readers; of the 92 cases reviewed, there was disagreement regarding healing time in only eight cases. For seven of these cases, the difference was one visit (2 weeks) and in one case two visits (4 weeks). Per protocol, the gatekeeping design imposed a restriction that to control for type I error, the teriparatide 40 µg versus placebo comparison was tested first. Using a bootstrap approach, the time to healing was not different between the teriparatide 40 µg versus placebo group (95% CI for difference of medians −1.1 to 0.6 weeks, p = .523).

Figure 2.

Time to radiographic healing in three of four cortices. Vertical lines indicate time (weeks) when median healing was achieved in the placebo, teriparatide 20 µg, and teriparatide 40 µg groups. (A) Time to radiographic healing in all patients. Overall p = .015. 95% CI for reduction of median healing time teriparatide 40 µg versus placebo: −1.1 to 0.6 weeks, p = .523; teriparatide 20 µg versus placebo: −2.7 to −0.6 weeks, p = .006; teriparatide 40 µg versus teriparatide 20: −2.7 to −0.1 weeks, p = .053. (B) Time to radiographic healing without data from nine patients who, on reevaluation, did not meet all inclusion criteria (overall p < .001). The 95% CI for reduction of median healing time teriparatide 40 versus placebo: −1.4 to 0.5 weeks, p = .127; teriparatide 20 µg versus placebo: −2.8 to −1.2 weeks, p < .001; teriparatide 40 µg versus teriparatide µg 20: −2.7 to −0.4 weeks, p < .03.

In post hoc analyses, there was no significant difference between the teriparatide 40 µg and teriparatide 20 µg groups (95% CI −2.7 to −0.1 weeks, p = .053). Because different methods were used to calculate the hypothesis testing and CI, a slight inconsistency is noted between the p value and the CI in the teriparatide 40 µg versus 20 µg comparison. Median time to healing was shorter in the teriparatide 20 µg group than in the placebo group (95% CI −2.7 to −0.6 weeks, p = .006).

On post hoc, blinded evaluation of baseline radiographs by two of the authors (PA and TJ), it was determined that nine patients did not meet the inclusion criteria because of a previous fracture (n = 1), volar dislocation (n = 3), or minimal dislocation (n = 5). In a post hoc analysis of the patients who did meet the criteria, excluding the nine who did not meet the criteria, time to healing was not different between the teriparatide 40 µg and placebo groups (95% CI for difference of medians −1.4 to 0.5 weeks, p = .127), with an overall p < .001 (see Fig. 2B). The time to healing was shorter in the teriparatide 20 µg group compared with the placebo group (95% CI −2.8 to −1.2 weeks, p < .001) and in the 20 µg versus teriparatide 40 µg group (95% CI −2.7 to −0.4 weeks, p < .03).

Secondary objectives

Radiologic and anatomic deformities

There were no statistically significant between-treatment differences in time to healing for the secondary objectives assessed by standard radiographic X-rays or CT scans. Palmar tilt, radial angle, and ulnar variance were assessed from radiographs for a total of 97 patients (Fig. 3). For 82 patients, the week 53 radiographs were scored; for 17 patients, the last available radiographs were scored. There were no significant between-group treatment differences for the displacement angles palmar tilt and radial angle (p > .80) (see Fig. 3). However, there was a statistically significant treatment difference for ulnar variance (p = .011), and the pairwise comparisons with placebo also were significant (p < .03). When combined subgroups with poor, intermediate, or good displacement rankings were analyzed, the subgroup analyses indicated that the time to healing was almost unrelated to the degree of dislocation (p > .26).

Figure 3.

Anatomic deformities—palmar tilt, radial angle, and ulnar variance. The boxplots (A, C, E) show the descriptive statistics of the displacement measurements: the median represented by the horizontal line in the box, the mean represented by the + symbol, the interquartile range (25th and 75th percentiles) represented by the length of the box, and outliers indicated by square symbols. The readers defined the direction of displacement or rotation and the degree of deviation from the normal position, as illustrated for each deformity (B, D, F). Palmar tilt represents the relative angle of the distal radial articular surface in relation to the long axis of the radius in the sagittal plane on lateral views and the direction of tilt (volar or dorsal). Radial angle was assessed as the relative angle of the distal radial articular surface on an anteroposterior film and the direction of rotation (lateral or medial). Ulnar variance depicts the difference in length between the ulnar aspect of the articular surface of the distal radius and the surface of the ulnar head (anteroposterior film) and the direction of displacement (distraction or shortening). (A) Palmar tilt, as relative angle (degrees). (B) Assessment of palmar tilt. (C) Radial angle, as relative angle (degrees). (D) Assessment of radial angle. (E) Ulnar variance, as millimeters. (F) Assessment of ulnar variance.

Secondary standard radiographic parameters showed no significant differences in median time to cortical bridging at four cortices (11.3, 10.9, and 11.0 weeks for placebo and 20 µg and teriparatide 40 µg groups, respectively; overall p = .558). Similarly, there were no significant differences in time to early endosteal healing (4.9, 5.0, and 4.9 weeks; overall p = .992), partial endosteal healing (7.0, 6.7, and 7.0 weeks; overall p = .177), and trabecular union (13.6, 12.9, and 13.1 weeks; overall p = .624) for the placebo and teriparatide 20 µg and teriparatide 40 µg groups, respectively.

CT scans were obtained in a subset of 72 patients. The estimated median time to first CT scan evidence of cortical bridging in at least three of four cortices was 9.1, 7.2, and 8.6 weeks in the placebo and teriparatide 20 µg and teriparatide 40 µg groups, respectively (overall p = .094). There were no significant between-group differences in CT scan median time to cortical bridging at four cortices (9.4, 9.0, and 9.3 weeks for the placebo and teriparatide 20 µg and teriparatide 40 µg groups, respectively; overall p = .054). There were no significant differences in median time to early endosteal healing (4.9, 4.8, and 5.0 weeks; overall p = .805) or partial endosteal healing (7.0, 6.7, and 6.7 weeks; overall p = .527) for the placebo and teriparatide 20 µg and teriparatide 40 µg groups, respectively. The data from CT scans did not permit a computation of median time to trabecular union for the placebo group. The times to trabecular union for the teriparatide 20 µg and teriparatide 40 µg groups, respectively, were 10.0 and 9.9 weeks (overall p = .308).

Functional results

Significant improvement in scores on pain and functional tests (Table 2) and grip strength (Table 3) compared with baseline occurred in each of the treatment groups, but there were no significant differences between the placebo and teriparatide treatment groups.

Table 2. Patient-Rated Wrist Evaluation (PRWE) Score
 PlaceboTeriparatide 20 µgTeriparatide 40 µg
nAdjusted mean ± SEnAdjusted mean ± SEnAdjusted mean ± SE
  • Note: Change from baseline in PRWE scores was analyzed using the mixed-model repeated-measures method. Adjusted mean = least-squares mean from the treatment-by-visit interaction.

  • a

    p < 0.001 change from baseline, all scores.

  • b

    p < 0.05 teriparatide 20 versus 40.

  • c

    p < 0.05 teriparatide 20 versus placebo.

PRWE total score (percent change from baselinea)
 Baseline3079.8 ± 2.72977.7 ± 2.73178.7 ± 2.6
 Week 528−20.3 ± 3.628−22.1 ± 3.728−18.5 ± 3.6
 Week 930−43.2 ± 4.227−45.8 ± 4.431−43.2 ± 4.1
 Week 1327−51.0 ± 3.927−61.2 ± 4.030−54.0 ± 3.8
 Week 1728−63.1 ± 2.828−69.5 ± 2.9b31−59.8 ± 2.7
PRWE pain score (percent change from baselinea)
 Baseline3136.5 ± 1.73033.1 ± 1.73234.7 ± 1.6
 Week 529−9.2 ± 1.829−11.8 ± 1.830−7.1 ± 1.7
 Week 931−17.8 ± 2.129-17.8 ± 2.232−17.0 ± 2.1
 Week 1328−20.9 ± 2.129−23.4 ± 2.131−20.5 ± 2.1
 Week 1729−26.1 ± 1.629−28.4 ± 1.632−24.6 ± 1.5
PRWE function score (percent change from baselinea)
 Baseline3086.4 ± 2.73087.7 ± 2.73187.4 ± 2.7
 Week 528−21.9 ± 4.429−19.9 ± 4.328−20.3 ± 4.3
 Week 930−50.6 ± 4.528−56.3 ± 4.631−51.7 ± 4.4
 Week 1327−60.3 ± 3.928−73.3 ± 3.9c31−66.4 ± 3.7
 Week 1728−73.7 ± 2.929−80.7 ± 2.931−69.9 ± 2.8
Table 3. Grip Strength as Percent of Uninjured Hand
 PlaceboTeriparatide 20 µgTeriparatide 40 µg
nAdjusted mean ± SEnAdjusted mean ± SEnAdjusted mean ± SE
  • Note: There were no significant between-group differences. Change from baseline in grip strength was analyzed using the mixed-model repeated measures method. Adjusted mean = least-squares mean from the treatment-by-visit interaction. To adjust for hand dominance, if the nondominant hand was injured, the percentage was multiplied by 1.07; if the dominant hand was injured, the percentage was multiplied by 0.93. Adjusted percentages were analyzed using the mixed-model repeated measures method.

  • a

    p < .05 change from baseline.

Grip strength as percent of uninjured handa
 Week 92444.8% ± 6.12539.3% ± 6.12342.1% ± 6.4
 Week 132554.0% ± 5.82755.4% ± 5.82757.9% ± 6.1
 Week 172963.8% ± 5.93062.7% ± 5.92965.8% ± 6.2
Grip strength in injured hand actual value analysisa
 Week 93113.0 ± 1.83012.5 ± 1.83213.1 ± 1.9
 Week 132915.6 ± 1.93115.5 ± 1.93115.4 ± 2.0
 Week 173017.8 ± 1.93017.6 ± 1.93117.6 ± 2.0

Safety

There were no significant differences across groups in the incidence of treatment-emergent adverse events (overall p = .397). Three patients in the placebo group and no teriparatide-treated patients experienced serious adverse events (overall p = .046). One patient in the placebo group developed sustained hypercalcemia. Although not statistically significant (p = .279), more patients (n = 7) in the teriparatide 40 µg group reported nausea compared with three patients each in the placebo and teriparatide 20 µg groups. Two patients discontinued treatment because of an adverse event (p = .603); one patient in the placebo group incurred a new distal radius fracture, and one patient in the teriparatide 40 µg group discontinued because of mild nausea. No abnormal fracture healing was detected through the last visit at 53 weeks. There was one death in the study: A placebo-treated patient died after 102 days of treatment.

Discussion

Although the primary hypothesis that teriparatide 40 µg would shorten the time to cortical bridging was not supported, post hoc analyses suggest that the clinically approved teriparatide 20 µg dose significantly shortened median time to healing in three of four cortices compared with placebo treatment.

The lack of effect of teriparatide 40 µg to accelerate healing compared with placebo was an unexpected finding. Preclinical studies have demonstrated a clear dose-response relationship from 10 to 800 µg/kg in the intermittent administration of recombinant teriparatide in experimental animals, with higher doses being more potent for enhancement and acceleration of fracture repair by increasing callus formation and mechanical strength.13, 14, 16–20

In a clinical trial of postmenopausal women, both teriparatide 20 µg and 40 µg induced periosteal mineral apposition and increased endocortical resorption at the distal radius that were associated with an improvement in bone geometry of the distal radius, reflected by higher axial and polar moments of inertia.29 Thus both doses should be associated with an increase in bone strength and improved resistance to fracture. However, Neer and colleagues30 reported that teriparatide 40 µg but not 20 µg resulted in a decreased mineral density in the cortex of the radial shaft, likely related to increased remodeling. Markers of bone turnover were numerically higher with teriparatide 40 µg compared to 20 µg.31 When callus appears as bridging on radiographs, it is rather compact, like cortical bone. This might explain why we found a positive effect of teriparatide 20 µg but not 40 µg and even a significant difference between the doses in a post hoc analysis. Increased remodeling by the higher dose may have made a larger and mechanically functional callus less visible on X-rays by increasing its porosity. It must be noted that the early and mechanically most important phase of repair comprises mostly bone modeling with little remodeling. During this phase, both doses may have stimulated bone formation. Later effects of the high dose are less important because fractures normally are healed mechanically before cortical continuity is seen.

Skripitz and Aspenberg12 suggested that in animal models of fracture healing and implant fixation, the effect of teriparatide appeared to be stronger on newly forming bone than on preexisting bone. A positive effect of teriparatide can be expected only after an osseous callus has been developed.32 The distal radius is considered a non-weight-bearing bone, and the stimulatory effect of teriparatide is, to some extent, exerted via an increased bone formative response to loading.33 Bone is formed rapidly during fracture repair. Conservatively treated, distal radial fractures heal by direct apposition of woven bone onto necrotic trabeculae, which is ongoing already the second week after injury. Therefore, the dose of teriparatide needed to enhance fracture healing may differ from that required for the treatment of osteoporosis.16 However, in this study, the clinically available dose performed better than a higher dose.

It appears that teriparatide 20 µg and 40 µg increased the number of early healers (see Fig. 2A, B). At 7 weeks, approximately 10% of the placebo group had healed compared with approximately 20% and 40%, respectively, for the teriparatide 40 µg and teriparatide 20 µg groups. However, at later time points these differences were no longer evident because patients who take a long time to bridge three cortices of four are likely to have an unfavorable fracture configuration for cortical healing and to be healed by internal callus before cortical continuity can be seen. Thus the longer the time to cortical bridging, the less relevant this variable might be as an index of completed repair from a mechanical or clinical point of view.

Although teriparatide might reduce the risk of nonunion, it appears from animal data as well as the present study that the main clinical advantage of using teriparatide would be an acceleration of time to fracture healing and enhanced bone formation.10–14, 16, 19–21, 32 The reduction in healing time may be important for elderly patients, who tend to deteriorate in general health during a long healing time. It also could have economic importance for the young, active patient. The 8-week treatment duration resulted in the discontinuation of teriparatide before radiographic healing occurred in all patients. However, a fracture callus is considered to grow and reach sufficient biomechanical properties well before mineralization is dense enough to show up as complete healing on radiographs. It is therefore difficult to say whether a longer treatment period would have any benefit.

Conservatively treated distal radial fractures were chosen for this study because we wanted proof of concept, and previous studies have shown that accelerated fracture healing is possible in these fractures.7, 34 We consider it a relevant and practical model for studying fracture healing per se in humans. For clinical relevance, however, other fractures may be more appropriate. The choice of outcome variable for studies on accelerated fracture repair is controversial and received much attention after this study was planned.21, 35 Radiographic healing represents a rough estimate for clinical healing, which also should include return to normal function of the limb.21 Biomechanical measurements, e.g., on externally fixated fractures, perhaps would be more to the point but cannot replace the patient's subjective function. Fractures appear healed on radiographs when the callus has reached a considerable mineralization and osseous maturity, at which time the mechanical properties are often reasonably restored. At this stage, a mechanical stimulus such as ultrasound might be more effective for further increasing the density than teriparatide, which might increase turnover, with fewer effects on radiodensity. A subset of 72 patients in the current study also underwent evaluation by CT, and these findings tend to support the results of plain radiography.

Acceleration of fracture repair by a systemic drug appears to be a generalizable phenomenon, likely to be similar in other fractures than the distal radius, although this remains to be demonstrated. Many fractures are treated by internal fixation, permitting return to near-normal function even before the fracture is healed. However, in many cases, and especially in patients with osteoporosis, internal fixation cannot provide enough stability for function. For example, intertrochanteric fractures of the femur are usually sufficiently fixed for load bearing, but each step is painful over many weeks until the fracture is healed. This is one of many possible indications for teriparatide in fracture treatment needing to be investigated. Such a study also could concentrate on variables more relevant to the patient, such as time to pain-free walking or quality of life.

This study has several limitations. First, it was planned to evaluate acceleration of fracture repair in humans as a biologic (or radiographic) phenomenon, with clinical consequences as secondary endpoints. It is likely that the clinical outcome of distal radial fractures is largely dependent on the injury to the ligamentous apparatus around the wrist, which takes a longer time to recover than the fracture healing process. To some extent, this recovery might be accelerated if teriparatide treatment allows earlier plaster removal and rehabilitation; however, in this study, patients had similar immobilization times, and a positive effect on functional parameters was not observed. A second limitation is that statistical power may have been reduced by the large variation in fracture severity and because some patients were later determined to have not met the inclusion criteria of a Colles' fracture.

Although the primary hypothesis, that the dose of teriparatide 40 µg would shorten the time to cortical bridging, was not supported in the study, post hoc analyses suggest that the clinically approved dose of teriparatide 20 µg had a highly significant effect on reducing the median time to healing in three of four cortices compared with placebo. The shortened time to healing for teriparatide 20 µg compared with placebo may suggest that fracture repair can be accelerated by teriparatide, but this result should be interpreted with caution and warrants further study.

Disclosures

Per Aspenberg serves as a consultant for Eli Lilly and Company. Torsten Johansson, Pedro A García-Hernández, and Christopher P Recknor report no conflicts of interest. Harry K Genant serves as a consultant for Eli Lilly and Company, Merck, Wyeth, GSX, Genentech, Amgen, Servier, Roche, BMS, and OsteoLign; he is a stockholder in Synarc, Inc., and Osteologix and serves as a member of Synarc board of directors and as a member of Osteologix and Osteolign S advisory boards. Thomas A Einhorn serves as a consultant for Eli Lilly and Company and Zelos Therapeutics; he has had previous grant support from Eli Lilly and Company and currently has grant support from Zelos Therapeutics. Antonio J Nino is a full-time employee of GSK. He was a full-time employee of Eli Lilly and Company at the time of the preparation of this manuscript and is a stockholder of Eli Lilly and Company. Anke Fierlinger is a full-time employee of Synarc, Inc. Kyoungah See, Kelly Krohn, Gail P Dalsky, Bruce H Mitlak, and Mark C Lakshmanan are full-time employees and stockholders of Eli Lilly and Company.

All authors had full and unrestricted access to the study data. Drs. Aspenberg, Johansson, and García-Hernández were principal investigators in the study. They, along with Drs. Genant, Recknor, Einhorn, Nino, See, Krohn, Dalsky, Mitlak, Fierlinger, and Lakshmanan, were involved in the analysis and interpretation of data, development/critical, or writing and revising the manuscript; all authors approved the final submitted version.

Acknowledgements

The authors thank the study investigators (see Appendix). This study (Registry No.: ClinicalTrials.gov, Identifier: NCT00190944; trial registration date September 12, 2005) was supported by Eli Lilly and Company, Indianapolis, IN, USA.

Appendix: List of Investigators

  • Canada: Kenneth Faber, MD, St. Joseph's Health Center, London Hand and Upper Limb Center, London, Ontario

  • Mexico: Pedro Garcia, MD, Hospital Universitario U.A.N.L., Unidad de Endocrinologia, Avenida Madero y Ronzalitos, Monterrey, Nuevo Leon

  • Poland: Jan Blacha, MD, Klinika Ortoped. Traumatologi, Akademi, Medycznej W Lublini, UI. Jaczewskiego 8 Lublin

  • Professor Andrzej Gorecki, Szpital Kliniczny Dzieciaka, Jezus, Katedra Anestezjkologii, I Klinika Ortopedii I Traumatologii, U1. Lindleya 4 Warszawa

  • Spain: Emilio Calvo-Crespo, MD, Fundacion Jimenez Diaz, Cirugia Ortopedica y Traumatologia, Adva/ Reyes Catolicos, 2 Madrid

  • Sweden: Per Aspenberg, MD, Universitetssjukhuset, SE 58185 Linköping

  • Torsten Johansson, MD, PhD, Universitetssjukhuset, SE 58185 Linköping

  • Romania: Liviu Badica, MD, Spitalul Clinic Urgenta Floreasca, Calea Floreasca, Nr. 8, Bucharest

  • Liviu Diaconescu, MD, Spitalul Universitar De Urgenta, Bucuresti, Clinica De Ortope, Str. Splaiui Independeteri Nr. 169, Sector 5, Bucuresti

  • Professor Dan Poenaru, Spitalul Judetean Nr. 1, Clinica De Ortopedie Nr. Ii Timisoa, Str. Prof. Iosif Bulbuca nr. 156, 300736, Timisoara

  • United States: Carolyn Becker, MD, Columbia Presbyterian Medical Center, New York, NY

  • Marcus Fulcher, MD, Medical College of Georgia, Augusta, GA

  • Michael Gottsman, MD, United Osteoporosis Center, Gainesville, GA

  • Louis Murdock, MD, Intermountain Orthopedics, P.A., Boise, ID

  • Douglas Regan, MD, Mercy Arthritis and Osteoporosis Center, Urbandale, IA

Ancillary