Teriparatide and Pelvic Fracture Healing: A Phase 2 Randomized Controlled Trial

Abstract

Purpose:

To determine if teriparatide (20 ug/day; TPTD) results in improved radiologic healing, reduced pain and improved functional outcome vs. placebo over 3 months in pelvic fracture patients.

Methods:

This randomized-placebo-controlled study enrolled 35 patients (women and men ≥50 years old) within 4 weeks of pelvic fracture and evaluated the effect of blinded TPTD versus placebo over 3 months on fracture healing. Fracture healing from CT images at 0 and 3 months was assessed as cortical bridging using a 5-point scale. The numeric rating scale (NRS) for pain was administered monthly. Physical performance was assessed monthly by Continuous Summary Physical Performance Score (based on 4m walk speed, timed repeated chair stands, and balance) and the Timed Up and Go (TUG) test.

Results:

The mean age was 82 and >80% were female. The intention to treat analysis showed no group difference in cortical bridging score and 50% of fractures in TPTD-treated and 53% of fractures in placebo-treated patients were healed at 3 months, unchanged after adjustment for age, sacral fracture, and fracture displacement. Median pain score dropped significantly in both groups with no group differences. Both CSPPS and TUG improved in the teriparatide group, whereas there was no improvement in the placebo group (group difference p<0.03 for CSPPS at 2 and 3 months).

Conclusion:

In this small randomized, blinded study, there was no improvement in radiographic healing (CT at 3 months) or pain with TPTD vs placebo, however, there was improved physical performance in TPTD-treated subjects that was not evident in the placebo group.

Mini Abstract

Pelvic fracture patients were randomized to blinded daily subcutaneous teriparatide (TPTD) or placebo to assess healing and functional outcomes over 3-months. With TPTD, there was no evidence of improved healing by CT or pain reduction, however physical performance improved with TPTD but not placebo (group difference p<0.03).

Introduction:

In recent years, there has been an increase in the incidence of high-functioning older patients with pelvic fracture [1, 2], and currently, pelvic fractures represent about 6% of all fractures in medicare beneficiaries [3]. Pelvic fracture is often accompanied by slow healing, severe pain, chronic immobility and loss of function and independence in the elderly [4, 5]. Approximately 60% of those with a pelvic fracture require an inpatient stay with a median stay of 9 days [6]. The 1-year mortality after pelvic fracture was 20%, greater than that seen for all other fractures except hip and femoral fractures [3] and mortality rates were still elevated at 3 years [6]. Pelvic fractures result in substantial utilization of healthcare resources and based on administrative claims data, they are one of the costliest osteoporosis-related fractures [7]. The current treatment strategy includes pain management, early mobilization, and the prevention of complications associated with comorbid conditions. However, the vast majority (92%) of pelvic fracture patients are not treated with osteoporosis medications. This is critically important because within 2 years, 41% of all patients sustain additional osteoporotic fractures [8]. It is possible that if an agent could be shown to accelerate healing of pelvic fractures, in addition to treating osteoporosis, could perhaps improve quality of life and perhaps ultimately, longevity.

There are strong preclinical data [9–12], as well as clinical evidence, that administration of parathyroid hormone (PTH) receptor agonists may improve bone union, hasten fracture healing, and improve physical function in patients with fractures [12–15]. In one nonrandomized, unblinded study, 100% of pelvic fracture patients given 1–84PTH were healed within 12 weeks compared to 68% of the controls, and both pain level and Timed Up and Go (TUG) performance were improved [16]. Despite these studies looking at teriparatide or other parathyroid compounds on fracture healing, the benefit of these anabolic treatments for fracture healing remains unclear. In fact, a recent meta-analysis concluded that PTH analogues may improve functional outcome for some fracture sites/types, but do not improve fracture healing rate or reduce pain, in agreement with other studies[17–20].

Therefore, the goal of this study was to determine if teriparatide (20 ug/day) over 3 months results in evidence of more complete cortical bridging by CT, reduces pain, and improves functional outcome as compared to placebo in patients with acute pelvic fracture. This study was planned to include 100 participants (80 completers) but due to the withdrawal of study drug supply during recruitment, the results should be considered exploratory.

Materials and Methods

This randomized parallel-design phase 2 study evaluated the effect of TPTD versus placebo in acute pelvic fracture patients on fracture healing by CT, pain, and physical performance. This study was approved by the IRBs of Hospital for Special Surgery, Helen Hayes Hospital, and Weil Cornell Medical Center. The study was performed under IND 126129 and registered with Clinicaltrials.gov identifier: NCT02972424.

Inclusion and exclusion criteria

We included men and postmenopausal women ≥50 years of age with acute pelvic fractures (at least 1 pubic ramus fracture), occurring with minimal trauma, presenting to Hospital for Special Surgery, Helen Hayes Hospital or New York Presbyterian Hospital (Weill Cornell Medical Center) within one month of the onset of symptoms. Patients who had one or multiple pubic rami fractures (with or without sacral fractures) were included in the study.

Participants were excluded if they were unable to provide informed consent due to cognitive impairment, if they were non-ambulatory prior to the fracture (assistive devices were allowed), or if they had a contraindication or intolerance to teriparatide. Prior osteoporosis treatment was allowed. The first person was recruited into the study in May 2017 and the last person was recruited in August 2019.

At the screening visit, the consent form was reviewed and signed. A medical history was performed and cognitive function assessed to assure there were no exclusions. Blood samples were reviewed for levels of serum creatinine, alkaline phosphatase and calcium; serum calcium level had to be within normal range, serum creatinine could not be elevated more than 1.5 times above upper normal limit for age and alkaline phosphatase levels could not be greater than 1.5 times the upper normal limit (given the expected elevation with a fracture). Pelvic radiographs and a full CT scan were reviewed and fractures were categorized by the blinded radiologist (ML, RB, JK, JR) as: single pubic ramus, multiple rami fractures (ipsilateral or contralateral) or pubic rami and sacral fractures. Each fracture was also categorized as to whether or not it was displaced. A loading dose of vitamin D (50,000 IU) was given to minimize vitamin D deficiency and calcium supplements or dietary modifications were recommended as needed to bring total calcium intake to approximately 1000 mg daily.

Treatment assignment, randomization, and blinding

This study was performed under IND 126129. Randomization using randomly permuted blocks was stratified by gender and recent bisphosphonate use to assign subjects in a 1:1 blinded fashion to receive either teriparatide 20 mcg (TPTD) or identically appearing placebo by daily subcutaneous injection. Study drug and placebo were provided by Eli Lilly (Indianopolis, IN). The statistician (DM) developed the randomization list, but only the pharmacist was unblinded to treatment assignment. The statistician was only aware of a group A or B assignment for the purposes of safety reporting to the DSMB. Although we planned to recruit 100 people, a halt in supply of study drug led to an early termination of recruitment. Because of the much smaller sample enrolled, this study findings are considered exploratory.

Double Blind Treatment Period: Study visits and assessments

At the baseline visit subjects completed the Barthel Index of Activities of Daily Living (ADLs) and SF-36 physical function survey (for pre-fracture actvity), Numeric Rating Scale (NRS) for pain, and a survey to determine narcotic use. Subjects were seen as outpatients at 4, 8 and 12 weeks and the same outcomes were assessed at each visit (Barthel Index of ADLs, SF-36, NRS and narcotic use survey). In addition, at 4, 8 and 12 weeks, participants completed the lower extremity physical function tests, TUG and a short physical performance battery summarized as the Continuous Summary Physical Performance Score (CSPPS; 4 m walk speed, repeated chair stands and balance), which were not assessable at the baseline visit. All information was recorded by a study coordinator blinded to treatment status.

Open Label Extension:

At the completion of the 3-month visit, all participants were offered treatment with daily TPTD for 9 subsequent months. Functional and pain outcomes were repeated at follow-up outpatient visits at 6-, 9- and 12-months.

Laboratory Assessments of Calcium and Biochemical Turnover Markers

At each visit, blood samples were obtained in the morning after an overnight fast, for a safety measurement of serum calcium via standard automated chemistry. Dietary calcium intake and supplement use was reviewed monthly to maintain total intake of 1000 mg daily. Abnormal serum calcium results were reviewed and a preplanned algorithm for hypercalcemia (defined as >10.1 mg/dl, the upper limit for our laboratory) required stopping calcium supplements or reducing dietary intake and repeating the measurement within 2 weeks. The algorithm specified that If hypercalcemia persisted, teriparatide/placebo would be withheld.

At 4-, 8- and 12-week visits, blood samples were also collected for levels of 25(OH)D and bone turnover markers (serum propeptide of type I procollagen [PINP] for bone formation and serum cross-linked c-telopeptide [CTX] for bone resorption). Blood samples were centrifuged, and serum was separated into 0.5 mL aliquots and stored in a −70C freezer. Samples were batched and all samples for individual patients were assayed in the same batch using the Immunodiagnostic Systems 25-Hydroxy Vitamin D RIA Assay, the Immunodiagnostic Systems Serum CrossLaps^® (CTX-I) ELISA, and the Aidian (Orion Diagnostica) UniQ PINP RIA. The ranges of intra- and interassay coefficients of variation, based on the analysis of control samples with high, medium, and low concentrations were: for PINP 6.5 to 10.2% (n=20) and 6.0 to 9.8% (n=10); for CTX 1.8 to 3.0% (n=10) and 2.5 to 10.9% (n=10); and for 25(OH)D 5.0 to 6.1% (n=10) and 7.3 to 8.2% (n=25).

CT imaging and review

All CT images at baseline and 3 months were reviewed by two radiologists with subspecialty training in musculoskeletal radiology, who were blinded to treatment assignment. At baseline the fracture was classified as (1) single pubic ramus or ipsilateral or contralteral pubic rami fractures, (2) presence or absence of a sacral fracture, and (3) amount of displacement as Minimal/No displacement (< 2 mm) or Mild/Moderate displacement (≥ 2 mm). After 3 months of treatment, all patients had a targeted low dose CT of the pubic rami to quantify the degree of cortical bridging. The cortical bridging outcome was independently evaluated by each of two radiologists and adjudicated by a third in the event of disagreement between raters, with the following qualification: if endpoint assessment was missing because of a drop-out/loss-to-follow-up, the endpoint was coded as “not healed”. For multifocal fractures, cortical bridging at each site was evaluated for review of individual fracture healing and a score representing the overall state of healing (0 to 4) was calculated by averaging scores for each fracture site and rounding up to the next highest integer. Rounding bridging scores from multiple sites favors the value of the site with the most progressed healing. We used a CT-modified version of the RUST scoring index [21] whereby <90 degrees of cortical bridging = 0 cortices, 90–179 degrees = 1 cortex, 180–269 degrees = 2 cortices, 270–359 degrees = 3 cortices, 360 degrees = 4 cortices [22]. In accordance with prior literature referencing the RUST scoring idex, we defined cortical bridging of 3 or 4 cortices as “healed”, 1 or 2 cortices as “partial healing”, and 0 cortices as “minimal/no healing”. All images were anonymized and stored in the research PACS managed by Imaging Data Evaluation and Analytics Laboratory (IDEAL) of Weill Cornell Medical Center.

Pain

The NRS has been tested for reliability and is the standardized tool used to assess pain in orthopedic and other medical applications [23, 24]. The NRS has a minimal detectable change of 30% that is considered clinically meaningful [25]. At each study visit, the NRS was administered by the study coordinator blinded to treatment group with assessment of current pain at the study visit and average pain during the prior week at each visit.

Physical Performance

The short physical performance battery we used (SPPB) has been shown to assess functional status, and has good reliability and validity. The tests include: 4 m walking speed, time for 5 repeated chair stand, and a balance assessment usually summarized into a quartile summary score [26–29]. The Continuous Summary Physical Performance Score (CSPPS) was used to summarize these tests and has been shown to identify meaningful differences in functional status in the elderly [30–32]. In addition, each subject completed the TUG test which has been shown to relate to fracture risk [33, 34].

Surveys

Study participants completed several questionnaires, including assessments of ability to perform activities of daily living [35] and the SF-36 physical sub-scales including: Physical Function (PF), Role-Physical (RP), Bodily Pain (BP), and General Health (GH) [36, 37]. Subjects were asked questions about general health and physical limitations over the 4 weeks prior to their fracture at baseline and for the 4 weeks prior to each subsequent visit.

Statistical analyses

Statistical analysis used SAS STAT 9.4 (Cary NC). Group differences at baseline were compared with t-tests or Wilcoxon rank sum tests and chi-square tests as appropriate (p-value <0.05 for significance). The primary outcome analysis was an ITT analysis of the 5 levels of cortical bridging seen on CT at 3-months. The treatment difference in cortical bridging was evaluated by a 2×5 (group × score) trend test [38] for the ordinal level of bridging. The Jonckheere-Terpstra test estimates the between-group difference in the ordinal outcome that is equivalent to the estimate of the Mann-Whitney U-test. Secondary analysis of fracture healing coded each individual as “healed” (score 3 or4) vs “not healed” (score 0–2). In an exploratory analysis, we also looked at each fracture individually for bridging score and healing. We assessed the influence of age and fracture characterizing covariates on healing success using logistic regression for the analysis of individuals with healed fractures and GENMOD for the analysis of individual fractures (since fractures are nested within a person). Group differences in pain score at the 3-month endpoint was evaluated with Wilcoxon signed rank test and longitudinal change from baseline, 4-, 8- and 12-weeks with sacral fracture and narcotic use as covariates with linear mixed models for repeated measures (LMMrm). The CSPPS was the primary outcome for physical performance. LMMrm was used with fixed effects of treatment group (TPTD vs. placebo), time (4-, 8- and 12-wk) and their interaction, with the baseline level of the SF-36 and presence of sacral or displaced fractures as covariates.

Results:

Of the 35 people randomized (18 to TPTD, 15 to Placebo), 2 patients never began study medication and on day one, were excluded from the analyses. Of the remaining 33 participants enrolled, 29 (88%) completed the 3 month study endpoint and 21 (64%) completed the 9-month open label portion of the study (Figure 1). There were more withdrawals from the placebo group (n=5) than the TPTD group (n=1).

Figure 1:

Consort diagram for the study

Most participants were female and the population was almost solely white. Baseline descriptive characteristics (Table 1) did not differ between the groups except for height. The mean age was 82 and the majority had had a fracture prior to the pelvic fracture. About half of the participants had had prior bisphosphonate exposure (balanced between groups) but only 20% had recent BP exposure within the preceding 2 years (also balanced between groups). Amost all participants had a past smoking history. The SF-36 physical function z-score indicated this population was about a standard deviation below average peers of their age, although they were able to complete most of their ADLs.

Table 1.

Demographics at Baseline

Baseline Characteristics
	TPTD (n=18)	PBO (n=15)
Female n (%)	16 (89%)	12 (80%)
Male n (%)	2 (11%)	3 (20%)
	Mean + SD (range)	Mean + SD (range)
Age (y)	81.6 ± 8.1 (66–99)	81.0 ± 11.5 (56–101)
Height (cm)	160 + 7.9 (147–180)	169 + 11.9 (147–188)*
Weight (kg)	56.7 + 10.6 (41.3–78.9)	64.5 + 18.1 (39.5–110.2)
BMI (kg/m2)	22.1 + 3.5 (15.6–29.3)	22.5 + 5.5 (15.4–39.3)
Prior smoker-quit> 1 year n (%)	16 (89%)	15 (100%)
History of prior fracture n (%)	12 (66.7%)	12 (80%)
Any prior BP use n (%)	10 (55.6%)	9 (60.0%)
BP use within 2 years n (%)	4 (22%)	3 (20%)
SF36 Physical Component Score and t-score	63.1 + 26.9 (10–95) −0.9 + 1.2 (−3.2 – 0.46)	54.6 + 34.9 (5–100) −1.3 + 1.5 (−3.4 – 0.68)
Barthel Index (ADLs) Score (0 to 20 points)	19.3 + 0.97 (16–20)	18.5 + 2.4 (12–20)
N (%) fractures with multiple rami fractures	18 (100%)	14 (93%)
N (%) displaced fractures	8 (48%)	9 (60%)
N (%) sacral fractures	7 (39%)	6 (40%)

^*p<0.05

Lab Assessment:

At 3 months, mean serum PINP levels were twofold greater in the TPTD group (170 mcg/L; n=13) as compared to the placebo group (84 mcg/L; n=9) group (p<0.001 group difference). Mean serum CTX levels at 3 months were 50% higher in the TPTD group (0.49 ng/ml) than in the placebp group (0.34 ng/ml (p=0.10). Similar differences in P1NP and CTX were seen between the TPTD and placebo groups at 4 and 8 weeks (data not shown). Serum 25(OH)D levels remained at approximately 100 nmol/L throughout the 3 months in both groups (data not shown).

Fracture characteristics and healing by CT

At baseline, over 93% of subjects had ipsilateral superior and inferior pubic rami fractures (Table 1) and 40% also had a sacral fracture. Displaced fractures (mild or moderate) were seen in 60% of the TPTD group and 44% of the placebo group (no significant differences). One person in the placebo-treated group had a single pubic ramus fracture and one person in the TPTD treated group had four pubic rami fractures (bilateral superior and inferior).

The intention to treat analysis (Table 2) of cortical bridging score as an ordinal outcome at 3 months, showed no difference between the placebo and the TPTD treated groups. There was also no difference in the percent of people with healed fractures: 50% of the individuals in TPTD treated and 53% of individuals in the placebo treated group were healed at 3 months (cortical bridging score 3 or 4; NSD). These results were unchanged after adjusting for age, presence of sacral fracture, or fracture displacement.

Table 2:

Analysis of Healing by CT at 3 months- Intention to Treat

	Teriparatide (n=18)	Placebo (n=15)	OR 95%CI for TPTD vs PBO	p-value
Bridging Score N (%) (% using mean of fractures/person)
0	1 (5.6%)	3 (20%)
1	4 (22.2%)	2 (13.3%)
2	4 (22.2%)	2 (13.3%)
3	6 (33.3%)	2 (13.3%)
4	3 (16.7%)	6 (40%)		0.72#
Union- N (%)	9 (50%)	8 (53%)	0.88 (0.22–3.5)	0.85*

^#Jonckheere-Terpstra, results were unchanged after adjusting for age, presence of a sacral fracture or displaced fracture.

^*results were unchanged after adjusting for age, presence of a sacral fracture or displaced fracture.

The data were also examined in a subgroup analysis of those completing 3 months of treatment. In this analysis, 53% of the TPTD group and 25% of the placebo group had displaced fractures, indicating that most drop-outs in the placebo group occurred in those with displaced fractures. In this analysis, healing was noted in 53% (9 of 17 people) of the TPTD group and 67% (8 of 12 people) of the placebo group (NSD).

Given that there were multiple fractures in most subjects, we also did an exploratory analysis using GENMOD, evaluating healing of individual fractures using the 0 – 4 bridging score in an analysis of group difference in trend. The two-tailed p-value for trend was equal to 0.51. There was also no difference when looking at the dichotomous analysis of healed vs. not healed. Adjustment for age, sacral fracture and displacement did not change the unadjusted estimate. In addition, results were similar in males and females and in those with or without bisphosphonate use. It appeared that individual fracture bridging score was higher in those with displaced vs non displaced fractures at 3 months (p=0.002 within TPTD group and 0.12 within placebo group).

There was no apparent difference based on time between the fracture diagnosis and the starting of treatment. Four of 7 who began TPTD within 14 days of fracture were healed at 3-months vs. 5/5 of those who began TPTD treatment after 14 days (OR = 1.333 (0.1909 – 9.3114), NS).

Pain

Figure 2 shows the least square mean pain score for the week prior to each visit in the 3 month placebo- controlled portion of the trial. Similar results were found after controlling for presence of a sacral fracture and use of narcotics. The primary outcome was the difference in the mean within-subject change from baseline in the two groups at the 3 months vs baseline. Mean change in the TPTD group was −1.94 + 3.42 and in the placebo group was −2.78 + 3.14 (NS). The pain was reduced from baseline at 8 and 12 weeks in the TPTD group and in the placebo group at 4 and 12 weeks (all p<0.05), with no group differences.

Figure 2:

Pain Scores – Average over last week using Least Square means at baseline 4, 8 and 12 weeks. p<0.05 versus baseline within each group; no group differences

Physical performance

The measurements for physical performance analysis began at 1 month (only 25% could complete at baseline), so there were fewer people in this analysis (placebo group n=11 and TPTD group n=17). There was an improvement (p<0.01) in both CSPPS and TUG assessments of physical performance within the TPTD group but no detectable differences within the placebo group (Figure 3). This result was unchanged after adjustment for age, presence of sacral fracture, pain, and SF-36 physical function score at baseline. The comparison of CSPPS between the TPTD and placebo groups was significantly higher in the TPTD group as compared to the placebo group at 2 and 3 months (p<0.03) after controlling for age, SF36 physical function at baseline and pain (Figure 3a). There was no difference in TUG between treatment groups at any time point (Figure 3b).

Figure 3:

Physical Function Scores (CSPPS and TUG) in Placebo vs. Teriparatide at 4, 8 and 12 months after correcting for age, baseline SF-36 score and pain. For the CSPPS measure (a) there was a significant improvement within the TPTD group, but not in the placebo group. * p<0.01 TPTD 4-week vs 8 or 12 weeks; # p<0.03 between groups. The time in seconds to complete the Timed Up and Go (b) was improved within the TPTD group and not the placebo group. * p<0.005 TPTD 4-week vs 8 or 12 weeks (no group differences).

Sensitivity Analysis

We performed a sensitivity analysis testing the influence on the efficacy estimate by providing the two early drop-outs with either the “best” or the “worst” values. No statistically significant effects became non-significant or vice versa.

Safety

There were no cases of hypercalcemia in the TPTD group and two cases in the placebo group during the blinded portion of the study. During the open label portion of the study, there were 4 cases of hypercalcemia and all resolved with a change in calcium or vitamin D supplementation. The incidence of adverse events by body system and treatment group (67% of TPTD treated and 59% of placebo treated) is shown in Table 3. There were no unexpected differences between TPTD and placebo groups. There were 8 serious adverse events (4 in each group) and none were believed to be related to study medication. There were 3 fractures during follow-up: one in the TPTD group (toe) and two in the placebo group (nose and shoulder).

Table 3.

Incidence of adverse events by body system and group

Body System and Preferred Term*	TPTD	Placebo	Total
	N (%)	N (%)	N (%)
Number of Patients	18	17	35
Number of Events	35	18	53
Number of Patients with Events	12 (67)	10 (59)	22 (63)
Head
Dizziness/faintness	1 (3)		1 (2)
Nose/Throat
Difficulty swallowing		1 (6)
Flu/Upper Respiratory Problems	1 (3)		1(2)
Sinus Condition	1 (3)	2 (11)	3 (6)
Chest
Shortness of breath		1 (6)	1 (2)
Gastrointestinal
Nausea	1 (3)	2 (11)	3 (6)
Vomiting	1 (3)		1 (2)
Diarrhea	1 (3)	1 (6)	2 (4)
Appetite decrease	1 (3)		1 (2)
Genitourinary
Painful urination	1 (3)		1 (2)
Difficulty urinating	1 (3)		1 (2)
Increased frequency of urination	2 (6)		2 (4)
Musculoskeletal
Muscle/bone/joint pain/condition	7 (20)	2(11)	9 (17)
Edema	2 (6)	1 (6)	3 (6)
Psychological/Behavioral
Tiredness/fatigue	2 (6)		2 (4)
Other
Accidental injury	5 (14)	3 (17)	8 (15)
Medical or surgical procedure		3 (17)	3 (6)
Not Otherwise Listed (or Unknown)	8 (23)	3 (17)	11 (21)

Open Label Extension:

Twenty-seven participants entered the extension and 21 completed the 12 months (12 from the TPTD group and 9 from the placebo group). In those that completed the study and took TPTD during the open label extension (10 TPTD and 7 placebo) there was no difference in TUG (11.8 + 1.5 vs. 13.3+ 1.8; p=0.5) or CSPPS (70.0 + 3.3 vs. 64.9 + 4.1; p=0.36) at 12 months between the TPTD and placebo groups respectively, after controlling for age, baseline SF-36 and pain (data not shown).

Discussion:

In this small randomized study, we found that 3 months of TPTD had no impact on pelvic fracture healing (measured by CT at 3 months) or pain, however, there was a small but significant improvement in physical performance within the TPTD group as compared to the placebo group. Poor physical function is one of the factors associated with high imminent risk of another fracture. Improving function as well as treating the underlying osteoporosis could help prevent additional disabling fractures in older individuals with pelvic fractures.

We did not see CT evidence of more complete healing with TPTD, although there is debate about how to define radiologic healing and no clear consensus concerning which method most precisely represents union [21, 22, 39, 40]. The Radiographic Union Scale for Tibia fractures (RUST) was developed to assign a numerical value to a healing tibial shaft fracture [41]. In the modified RUST score, which bases scoring on 4 categories like the standard RUST, the lowest score indicates no callus, and the highest score indicates remodeling callus. This score is repeated at each cortex on two orthogonal radiographs (two cortices on each projection for a possible total score of 16). However, the intermediate scores in this scale (i.e. 2–3 at each cortex) are subject to interobserver disagreement in grading incomplete callus formation, which could also be a confounder on CT. In our study design, we focused on the presence or absence of bridging callus at each of 4 quartiles of tubular bone with strong interobserver agreement (85%).

There is some clinical evidence that administration of parathyroid hormone (PTH) receptor agonists may improve non-union [12], improve spinal fusion [14], and hasten fracture healing at the distal radius [13], trochanter [42], pelvis [16], and vertebrae [43]. One other study of PTH(1–84) and pelvic fracture healing appears to have evaluated ipsilateral fractures as a single entitity (not evaluating each ramus independently). We observed different rates of healing at each ramus within the same patient in both the TPTD and placebo groups. Concurrent sacral fractures were excluded [16] and there was no mention about fracture displacement. There was also no mention of blinding for the pain and physical function measures, which are more likely to be subjective and influenced by lack of blinding. Furthermore, it is unclear whether the study was truly randomized since the discussion states that patients given PTH(1–84) were all treated at one center along with some controls and another center had only controls [16]. In a randomized unblinded Japenese trial using standard doses of weekly TPTD compared to weekly or monthly bisphosphonate in vertebral compression fracture, TPTD (n=19) led to fracture healing on x-ray and CT assessments in 2.8 months and bisphosphonate (n=24) in 3.9 months [43] with significant differences at 12, but not 24 weeks. In a small trial (n=40) with proximal humerus fractures randomized to one month of TPTD versus a control group, there were no radiographic signs of enhanced callus formation, improved pain or reduced disability of the arm, shoulder and hand in the group treated with teriparatide at 7 or 12 weeks [44]. An acceleration of fracture healing with TPTD vs placebo for 8 weeks, within 10 days of distal radius fracture, was reported in a randomized controlled trial of 102 postmenopausal women. The authors concluded that their result should be interpreted with caution as 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, TPTD 20 mcg and TPTD 40 mcg, respectively (showing no improvement with the 40 mcg dose). Results were confounded by the increased prevalence of surgical fixation to treat these fractures [13]. In a further analysis of this study, fractures were scored for radiographic appearance 5 weeks after fracture (n =27). In this analysis:9 of 10 patients with the highest dose and 8 of 9 with the lower dose showed a visible bony callus, and only 1 of the 8 placebo patients did, perhaps suggesting that TPTD has a strong effect on early callus formation [45]. We did not assess healing at time points earlier than 3 months and may have missed an early effect of TPTD to accelerate callus formation.

In a recent review of pelvic fractures, it was stated that “one of the primary goals of treatment is functional recovery” [46]. In our study, TPTD led to improvements in CSPPS in the first 3 months, although these changes were not apparent after the open label extension where all were offered TPTD. These results were further verified by reduced time (seconds) to complete the TUG. These functional measures did not change with placebo. The results were also robust and unchanged after controlling for baseline SF-36, presence of a sacral fracture, pain, or age. The CSPPS summary score of balance, walk speed and chair stands has been shown to relate to ability to perform activities of daily living and to the physical function portion of the SF36 [31, 32]. There is some clinical evidence that administration of PTH receptor agonists may improve physical function following pelvic [16], femoral neck [47] or intertrochanteric fractures [15]. Similar to our finding of improved TUG outcome at 8 and 12 weeks, TUG was significantly faster at multiple times during 6 months after treatment with TPTD compared to risedronate in individuals who had pertrochanteric hip fractures, however their reported difference disappeared by one year [15]. Improved mobility was also seen at 3 and 6 months, but not 12 months in TPTD-treated patients with unstable pertrochanteric fractures [42]. Among 159 femoral neck fracture patients, the secondary outcome of gait speed showed that 73% (47 of 64) treated with placebo and 89% (51 of 57) treated with TPTD could ambulate (gait speed[0.05 m/second) without decline at 12 months (OR, 3.07; 90% CI, 1.20–8.49; p = 0.021) and the proportions of patients treated with placebo and TPTD who regained their prefracture ambulatory status were 58% and 67%, respectively (OR, 1.47; 90% CI,0.79–2.73; p = 0.169) at 12 months [40]. The differences we found in the TUG score between TPTD and placebo treated individuals may be relevant based on a recent analysis that found that a slow TUG (greater than 12 seconds) predicts hip and major osteoporosis fracture risk, independent of clinical risk factors and BMD [34].

This trial had several limitations. The first is that we were underpowered to find an effect of TPTD on radiographic evidence of fracture healing due to a halt in supply of study medication. The sample size determination for the primary fracture healing outcome graded on a 5-point scale assumed the distribution of success reported previously [16] and a 2×5 Jonkheere-Terprstra trend test of treatment assignment by outcome grade observed at the 3-month endpoint provided 95% power with 1% alpha with an analyzable sample of 80 participants. With an analyzable sample of 35, the same effect size has only 66% power with 2% alpha. Therefore, our findings should be considered exploratory. Recruitment was difficult; we screened 363 patients to recruit 35 subjects. This experience is similar to a feasibility study of TPTD after trochanteric hip fracture, where 724 patients were screened, of whom only 143 (20%) were eligible for recruitment and only 29 (4%) elected to participate [48]. It is also difficult to retain an older cohort with multiple underlying comorbidities. The optimal radiological measures to assess healing are unclear and we only assessed healing at one time point (though at that time only approximately 50% had healed). Another limitation is that more than half the participants had displaced factures, and given that degree of displacement may have an impact on healing (suggestion of interaction), it would have been ideal to block randomize by presence of displacement. We were also not expecting that multiple pubic rami fractures would vary in healing times. Averaging the time to healing across different fractures might not be the most appropriate way to analyze healing in these patients. However, multiple fracture prevalence was balanced between the 2 groups. The timing of the initiation of TPTD is potentially important; although we attempted to administer TPTD as soon as possible, we allowed up to one month after the fracture to begin study medication. We have no explanation for the potential effect of TPTD on physical function if not associated with improved or accelerated fracture healing and since there was a very small number of patients, we need to be cautious about this conclusion. Lastly, it is difficult to assess pain and account for the many variables, such as the amount of physical activity and medication use, that alter the pain level, though we did control for narcotic use.

There were also several strengths to the study design, including the strict randomization, the use of a placebo and double-blind design, the use of CT images to assess healing, and the assessments of pain and physical performance while all investigators and paticipants remained blinded to treatment. We also utilized two musculoskeletal radiologist interpretations with a third radiologist if there was any discrepancy. We allowed people with sacral fractures to enter the study, making this a more real-world experience and evaluated each fracture independently.

In conclusion, in this small trial of patients with acute pelvic fractures, we found that there was no discernable difference in fracture healing by CT at 3 months or pain at any timepoint, but a significant improvement in physical performance was seen in the TPTD group that was not seen with placebo. Clearly a larger randomized blinded clinical trial is warranted to further assess the impact of PTH receptor agonist therapy on fracture healing.