The effectiveness and safety of parathyroid hormone in fracture healing: A meta-analysis

The very large economic and social burdens of fracture-related complications make rapid fracture healing a major public health goal. The role of parathyroid hormone (PTH) in treating osteoporosis is generally accepted, but the effect of PTH on fracture healing is controversial. This meta-analysis was designed to investigate the efficacy and safety of PTH in fracture healing. The EMBASE, PubMed, and Cochrane Library databases were systematically searched from the inception dates to April 26, 2018. The primary randomized clinical trials comparing PTH treatment for fracture healing with placebo or no treatment were identified. We did not gain additional information by contacting the authors of the primary studies. Two reviewers independently extracted the data and evaluated study quality. This meta-analysis was executed to determine the odds ratio, mean difference, standardized mean difference, and 95% confidence intervals with random-effects models. In total, 8 randomized trials including 524 patients met the inclusion criteria. There were significant differences in fracture healing time, pain relief and function improvement. There were no significant differences in the fracture healing rate or adverse events, including light-headedness, hypercalcemia, nausea, sweating and headache, except for slight bruising at the injection site. We determined that the effectiveness and safety of PTH in fracture healing is reasonably well established and credible.


' INTRODUCTION
It is estimated that there are 16 million cases of fractures in the United States every year. Despite most bone having excellent regenerative ability during fracture healing, approximately 5% to 10% of fractures have complications, such as delayed healing or nonunion (1)(2)(3). Delayed healing and nonunion are defined as fractures that have not healed for 6 and 9 months, respectively (4,5). The complications result in extended treatment time, reduced quality of living and potential additional remedial surgery, which give the patients a heavy burden and have serious societal implications (1,6). Fracture repair is an intricate process regulated by numerous genes and is affected by chemokines, cytokines, growth factors and other molecules (7)(8)(9). To promote bone regeneration and prevent complications, an in-depth understanding of the fracture healing process and the adoption of suitable interventions are essential.
Parathyroid hormone (PTH) plays a vital role in bone regeneration by stimulating the differentiation and proliferation of osteoblasts and osteoclasts (10,11). Osteoporosis has been treated with intact PTH (1-84) and teriparatide, which is an N-terminal fragment  of PTH (12). Currently, teriparatide is the only anabolic bone therapeutic medicine to treat osteoporosis approved by the Food and Drug Administration (FDA) (13)(14)(15)(16). In patients with high fracture risk, PTH can improve bone quality and prevent fractures (17).
As described previously, PTH treatment could increase bone formation biomarkers higher than it increases bone resorption biomarkers (18,19). Some animal studies have demonstrated that daily application of PTH (1-34) could increase bone mineral density to accelerate fracture healing and persistently affect the remodeling callus during fracture healing (20,21). Furthermore, studies have shown that PTH could improve fracture healing at different skeletal sites of patients (22)(23)(24). However, we found that recent studies in the literature were inconsistent regarding the effects of PTH treatment on fracture healing (25,26).
Therefore, a meta-analysis on randomized controlled trials (RCTs) was performed to demonstrate whether PTH can promote fracture healing by comparing PTH treatment to a placebo treatment or no treatment in patients with fracture.

' METHODS
This meta-analysis was conducted based on the Cochrane Handbook for Systematic Reviews of Interventions (27) and presented in compliance with Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines (PRISMA) (28).

Search strategy
The EMBASE, PubMed, and Cochrane Library databases were systematically searched from the inception dates to April 26, 2018. The keywords and Medical Subject Headings '' and ''PTH (1-84)'' in combination with ''Fracture Healing'', ''Healing, Fracture'', ''Healings, Fracture'' and ''Fracture Healings''. We did not restrict the language. The detailed search strategy is reported in the supplement. We identified original RCTs and did not obtain additional information by contacting authors of the primary studies.

Inclusion criteria
Studies were included if they met the following criteria: (A) participants, aged 21 to 94 years old, were patients with fractures and treated by PTH; (B) RCTs comparing the PTH intervention with a placebo or no treatment; (C) teriparatide was subcutaneously injected at 20 mg per day, or PTH (1-84) was administered at 100 mg per day; and (D) trials provided the relevant data.
Studies were excluded if they met the following criteria: (A) the patients had been using PTH, unless they had experienced a wash-out period; (B) there were contraindications to related drugs at any time; (C) any other antiosteoporotic drug had been taken in either the experimental group or comparison group; (D) liver enzymes were more than twice the upper limit or serum calcium was higher than the reference level; (E) patients had rheumatoid arthritis, pathologic fractures, history of tumor or chemotherapy, metabolic bone disease, chronic renal failure, or any disease affecting bone metabolism; (F) the studies were published as abstracts, reviews or letters; or (G) the articles were not available or the data had already been published.

Risk-of-bias assessments
According to the Cochrane risk-of-bias criteria (27), 2 researchers (HH, YL) independently evaluated the methodological quality of the selected RCTs. Every quality item was classified into unclear risk, low risk, or high risk. Any disagreements about a trial were resolved through discussion or consultation with an expert. Seven items were used to assess bias, as shown in Figure 2. Other bias was defined as trials with dissimilar baseline characteristics between groups or with sponsorship coming from drug companies.

Data extraction
The following information from each study was extracted independently by two researchers (HH, TS): authors, publication year, participant characteristics, number of cases, type of fracture, duration of trials, explicit treatment, drug dose, fracture healing-related data and adverse events. Any disagreements were resolved through discussion. We only extracted the relevant information and data from the original articles when more than 2 groups were compared in the study.

Statistical analysis
Our meta-analysis was conducted in Revman version 5.3 from the Cochrane Collaboration. We calculated the odds ratio (OR) and 95% confidence intervals (CIs) using the Mantel-Haenszel statistical method for dichotomous outcomes, while we determined mean difference (MD) or standard mean difference (SMD) and 95% CIs using the inverse variance statistical method for continuous outcomes. The analyses were 2-tailed, and the data pooled with a random-effects model. po0.05 was considered statistically significant. Statistical heterogeneity in the summary data was assessed with I 2 statistics and p-values (I 2 450% or p-valueo0.10 was considered to indicate significant heterogeneity) (38,39). Sensitivity analyses were fulfilled by excluding the trials one by one. We performed subgroup analyses to appraise whether clinical characteristics could alter the results, to evaluate the statistical significance between the subgroups, and to estimate the publication bias with funnel plots.

Grading of evidence quality
Regarding risk of bias, inconsistency, indirectness, imprecision, and publication bias, the quality of evidence was independently assessed by two authors (HH, YL) based on the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) (40) methodology. The estimated results were classified as very low, low, moderate, or high. We constructed a summary table using the GRADE Profiler (version 3.6).

Literature search and characteristics
From the recently published studies, 121 potentially eligible RCTs were screened by titles and abstracts. There were 26 records read in full. Eventually, 8 records were in conformity with the inclusion criteria. A manual search of the reference lists within these studies did not reveal additional eligible studies. The process of screening the studies is illustrated in Figure 1.
Ultimately, in the meta-analysis, there were 524 participants from 8 RCTs, of which 79.6% were women and 20.4% were men. The mean age was 73.0 years old. For the fracture type, two trials had upper limb fractures (including distal radial fractures and proximal humeral fractures), five trials had lower limb fractures (including lower-extremity stress fracture, femoral neck fracture, trochanteric fractured neck of femur, osteoporotic intertrochanteric fractures, and hip fractures), and one trial had pelvic fracture. The general characteristics of the included studies are presented in Table 1.
The experimental groups received once-daily subcutaneous treatment with 20 mg of teriparatide or 100 mg of PTH . The anabolic effect of 20 mg teriparatide is equivalent to 100 mg PTH (1-84) because of the differences in pharmacokinetics (41). The control group received a placebo, no treatment, or other drugs. The duration of PTH treatment ranged from 4 weeks to 24 months. The study by Aspenberg et al. (33) included two experimental groups, taking 20 mg or 40 mg teriparatide. In the study by Huang et al. (34), the two experimental groups could take either 20 mg teriparatide or 20 mg teriparatide plus 70 mg alendronate. Additionally, the study by Kanakaris et al. (30) had two control groups that took either vitamin D and calcium or vitamin D and calcium plus 70 mg alendronate. We excluded those ineligible groups. The detailed characteristics are shown in Table 2.

Risk-of-bias assessments
Methodological quality is shown in Figure 2. The incidence of the follow-up loss in patients was appraised, but only three trials had reported follow-up rates, which were all lower than 20%.

Radiographic assessment of fracture healing
Three trials (251 patients) compared PTH treatment with either a placebo or no treatment for the time until radiological fracture healing (31,33,34). The MD method was adopted. As shown in Figure 3, there was a statistically significant difference in fracture healing time (MD -3.05, 95% CI -5.96 to -0.14, p=0.04; I 2 of heterogeneity 97%, p-value of heterogeneity o0.00001). A sensitivity analysis showed that the heterogeneity was significantly lower after the trial by Peichl et al. (31) was excluded (Table 3).
Four trials (256 patients) compared PTH treatment with either a placebo or no treatment regarding the rate of radiological fracture healing (29)(30)(31)(32). The OR method was adopted. The results are shown in Figure 4. There was no statistically significant difference in the fracture healing rate (OR 7.84, 95% CI 0.47 to 130.27, p=0.15; I 2 of heterogeneity 85%, p-value of  heterogeneity =0.0002). In a sensitivity analysis, the heterogeneity was significantly lower after the trial by Peichl et al. (31) was excluded (Table 4).

Fracture pain degree
Five trials (320 patients) compared PTH treatment with either a placebo or no treatment on the degree of fracture pain by VAS or PRWE scores (31,33,34,36,37). Pain and numbness were evaluated by VAS scores in patients with trauma (31,34,36,37), while pain and function were evaluated by PRWE scores in patients with distal radius fractures (33). Due to the different scoring criteria, the SMD method was adopted. The results are shown in Figure 5. There was a statistically significant difference in fracture pain degree (SMD -1.42, 95% CI -2.55 to -0.29, p=0.01; I 2 of heterogeneity 94%, P value of heterogeneity o0.00001). Considering the significant heterogeneity, we did a subgroup analysis on the basis of the distinct scoring methods. In the VAS score subgroup, there was still significant heterogeneity (I 2 of heterogeneity 94%, p-value of heterogeneity o0.00001). We further performed a sensitivity analysis in the VAS score subgroup and found that       the heterogeneity was significantly lower after the trial by Peichl et al. (31) was excluded (Table 5).

Functional outcome
Four trials (178 patients) compared PTH treatment with either a placebo or no treatment on functional outcomes (30,31,33,37). The functional outcome was assessed with the TUG test or the self-administered PRWE questionnaire, DASH score or JHRQ (30,31,33,37). Due to the different methods, the SMD method was appropriate. The results are shown in Figure 6. The patients with PTH treatment were significantly superior to those with a placebo or no treatment in functional outcome (SMD -1.28, 95% CI -2.33 to -0.24, p=0.02; I 2 of heterogeneity 88%, p-value of heterogeneity o0.00001). In view of the result that there was significant heterogeneity (I 2 =88%), we performed a subgroup analysis based on the duration of treatment. One group had the treatment time being equal to 4 weeks (SMD -0.42, 95% CI -0.97 to 0.13, p=0.13; I 2 of heterogeneity 7%, p of heterogeneity =0.30) and the other group included the treatment time exceeding 4 weeks (SMD -2.17, 95% CI -2.89 to -1.45, po0.00001; I 2 of heterogeneity 57%, p of heterogeneity =0.13).
The sensitivity analysis for the functional outcomes showed that the heterogeneity was not significantly lower after any study was excluded (Table 6). Importantly, in this metaanalysis, the functional outcomes were better when the treatment times were longer than 4 weeks.

Adverse events
The trial by Peichl et al. declared that no deaths or adverse events were recorded (31). The trial by Huang et al. and Kanakaris et al. did not mention the relevant statistics about adverse events (30,34). Five trials reported adverse events with inconsistent methods (29,32,33,36,37). Therefore, we could not pool the relevant data, but describe them in Table 7. In comparing the PTH treatment group with a control group, there was no significant difference in lightheadedness, hypercalcemia, nausea, sweating, and headache, except for slight bruising at the injection site.

Publication Bias and GRADE Profile Evidence
For this meta-analysis, there was no evidence showing obvious publication bias by examining the symmetry of the   funnel plots (Figure 7), but a funnel shape reference line could not be provided by the software due to the small number of studies. GRADE evidence profiles are shown in   (31) No adverse events or deaths were recorded --- Figure 7 -Publication bias summary.
trial by Bhandari et al.; these studies contrasted with the other three trials. Clearly, the most common reasons for the decreased level of evidence were the possible risk of bias and inconsistency.

' DISCUSSION
In this meta-analysis, we summarized that PTH treatment in patients with fracture was better than a placebo or no treatment based on the time for fracture healing, the degree of fracture pain, and the functional outcomes ( Figure 8). There is great clinical value in healing fractures over a shorter time, with reductions in pain and with functional improvements (42). Therefore, these results showed the effectiveness of PTH in fracture healing. However, previous studies were inconsistent with the effect of PTH on fracture healing (25,26). It was shown that PTH is effective in accelerating fracture healing in the study by Lou et al. (25), but not in the study by Shi et al. (26). Their studies might have some limitations, such as including the patients who received other drugs or different doses (25,26), only including the patients with osteoporosis (25), or only including fracture healing time and functional change (25). Our study included a total of 524 patients with fractures from 8 RCTs involving recently published primary studies. We included more patients and relevant outcomes, and tried to exclude inappropriate patients.
However, heterogeneity was obvious in this meta-analysis, so sensitivity analyses and subgroup analyses were performed. We found that the heterogeneity was significantly lower by exclusion of the trial by Peichl et al. (31). Fractures at different sites can be healed by different mechanisms (43). One possible explanation of the heterogeneity was that limb fracture healing differed from pelvic fracture healing in response to the anabolic effect of PTH, and we need to study these two types of fractures separately. Another possible explanation is that the equivalence of teriparatide and PTH (1-84) for fracture healing is questionable. Similarly, comparing 4-week treatment (30,37) with over 4-week treatment (31,33) produced statistically significant differences in terms of functional improvement following fracture. Although PTH can improve early callus formation (44), better effects may be observed with a longer duration of treatment. Therefore, a uniform duration of treatment is needed to support clinical decisions.
According to the results from the pooled data, PTH treatment accelerated fracture healing, which allows patients to return to normal life sooner and reduces the medical consumption and chronic morbidity associated with longterm treatment. Furthermore, PTH can be applied to any type of fracture, commenced at any time, and applied throughout the entire healing period. As a result, we suspect that PTH may be useful in the course of implant fixation and in the established nonunion fracture. Some studies have started to explore related issues (45)(46)(47)(48)(49)(50), but the number of these studies is still limited, and most of them are small sample size. The hypotheses discussed here still need to be addressed by highquality RCTs.
It was advantageous that our meta-analysis conformed to the recommendations of the Cochrane Collaboration (27) and PRISMA guidelines (28), and the quality of evidence for its outcomes was evaluated by GRADE system (40). However, there were still several limitations. First, there were only eight studies included, and the sample size was relatively Table 8 -GRADE evidence profiles for the outcomes small. Second, in our study, more than 79% of the fractures occurred in women, and the average age of participants was 73 years; therefore, we do not know whether the results are applicable to men or young adults. Third, it was difficult to guarantee consistent blindness because some RCTs lacked a placebo or were unclear about the ''random sequence generation'' and ''allocation concealment''.
In conclusion, we determined that the effectiveness and safety of PTH in fracture healing is reasonably well established and credible. At this time, there are not enough studies in this field, hence we must cautiously interpret these results, and more high-quality RCTs are needed to verify the differential effects of PTH on fracture healing in different populations.

' ACKNOWLEDGMENTS
This study was funded by research grant No. 81672230 from the National Natural Science Foundation of China. The authors thank Hui Tang, Yongsheng Wang and Hu Li for help in preparing the manuscript.

' AUTHOR CONTRIBUTIONS
Hong H and Deng Z designed the study. Deng Z conducted the study. Li J, Jiang Q and Song Q were responsible for the data collection. Hong H and Song T were responsible for data analysis. Hong H was responsible for the integration of the data analysis and manuscript drafting. Hong H and Liu Y were responsible for the data interpretation. Hong H, Liu Y, Li J, Jiang Q, Song T and Deng Z were responsible for revising the manuscript content. Hong H, Liu Y, Li J, Jiang Q, Song T and Deng Z approved the final version of manuscript.