Bone 120 (2019) 1–8 Contents lists available at ScienceDirect Bone journal homepage: www.elsevier.com/locate/bone Full Length Article Effects of teriparatide on hip and upper limb fractures in patients with T osteoporosis: A systematic review and meta-analysis ⁎ Adolfo Díez-Péreza,b, , Fernando Marinc, Erik F. Eriksend, David L. Kendlere, John H. Kregef, Miguel Delgado-Rodríguezg,h a Hospital del Mar-IMIM-UAB, Department of Internal Medicine, Barcelona, Spain b CIBERFES, Institute Carlos III, Barcelona, Spain c Lilly Research Center, Windlesham, United Kingdom d Department of Clinical Endocrinology, Morbid Obesity and Preventive Medicine, Oslo University Hospital, Norway e University of British Columbia, Vancouver, Canada f Lilly Research Laboratories, Indianapolis, USA g Department of Preventive Medicine and Public Health, University of Jaén, Spain h CIBERESP, Institute Carlos III, Madrid, Spain A R T I C LE I N FO A B S T R A C T Keywords: In randomized clinical trials (RCTs) with teriparatide, the number of patients with incident hip fractures was Fractures small and insufficiently powered to show statistically significant differences between groups. We, therefore, Hip fracture conducted a systematic review and meta-analysis of the efficacy of teriparatide in the reduction of hip and upper Meta-analysis limb fractures in women and men with osteoporosis. A comprehensive search of databases until 22 November Systematic review 2017 was conducted for RCTs of at least 6-month duration that reported non-spine fractures (hip, humerus, Teriparatide forearm, wrist), either as an efficacy or safety endpoint. Only RCTs that included patients with the approved treatment indications and dose for use of teriparatide were included; trials with off-label use of teriparatide were excluded. Two independent reviewers performed study selection and data extraction. Statistical procedures included Peto's method and Mantel-Haenszel with empirical correction, as most of the RCTs reported zero events in at least one of the treatment arms. Study results are expressed as odds ratios (OR) with 95% confidence intervals (CI). Publication bias and heterogeneity were evaluated with standard statistical tests. Twenty-three RCTs were included, 19 with an active-controlled arm (representing 64.9% of the patients included in the control group) and 11 double-blind, representing data on 8644 subjects, 3893 of them treated with teriparatide. Mean age (SD) was 67.0 (4.5) years, median treatment duration 18 months (range: 6 to 24 months). A total of 34 incident hip, 31 humerus, 31 forearm, and 62 wrist fractures were included. Meta-analysis results showed an OR (95% CI) for hip fractures of 0.44 (0.22–0.87; p = 0.019) in patients treated with teriparatide compared with controls. The effects on the risk of humerus [1.02 (0.50–2.08)], forearm [0.53 (0.26–1.08)] and wrist fractures [1.21 (0.72–2.04)] were not statistically significant (p > 0.05). This meta-analysis provides evidence of efficacy of teriparatide in reducing hip fractures by 56% in patients with osteoporosis. 1. Introduction postmenopausal osteoporosis was to investigate vertebral anti-fracture efficacy, while phase 4 trials were primarily aimed to investigate the Typically, osteoporosis randomized control trials have primary effects on surrogate markers of bone strength and structure in a limited endpoints of vertebral fractures, since this is the most frequent fracture number of patients. Even though many of these studies included sub- experienced in postmenopausal women. Hip fractures are seldom pri- jects at high risk for having osteoporotic fractures, the number of in- mary endpoints due to their infrequent nature in elderly patients, who cident hip fractures was relatively small and insufficient to show sta- are not often recruited to registration clinical trials. The main objective tistically significant differences between groups. For example, the of the pivotal, phase 3 teriparatide studies in patients with teriparatide clinical trial with the most hip fractures published to date ⁎ Corresponding author at: Department of Internal Medicine, Hospital del Mar-IMIM, Universidad Autónoma Barcelona, Passeig Maritim 25-29, 08003 Barcelona, Spain. E-mail address: ADiez@parcdesalutmar.cat (A. Díez-Pérez). https://doi.org/10.1016/j.bone.2018.09.020 Received 24 May 2018; Received in revised form 18 September 2018; Accepted 24 September 2018 Available online 27 September 2018 8756-3282/ © 2018 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/BY/4.0/). A. Díez-Pérez et al. Bone 120 (2019) 1–8 reported only 9 such fractures [1], while the pivotal, phase 3 trial of 2.3. Data extraction and management teriparatide in postmenopausal osteoporosis reported only 5 hip fragi- lity fractures between the placebo and the teriparatide 20 μg treatment Data extracted from each study included participants, interventions, arms [2]. In the largest randomized clinical trial that included a ter- description of the comparators, and outcomes with the following spe- iparatide treatment arm, only 2 hip fractures were collected in the total cific items: first author, reference, year of publication, control group, clinical trial cohort [3]. participants' characteristics [gender, age], study design, clinical devel- However, results from non-controlled, observational studies with opment phase, incident selected non-vertebral fractures, type of frac- larger patient cohorts, suggest that teriparatide may reduce the risk of ture endpoint (efficacy or safety), and duration of treatment. In the case hip fractures [4–7]. Understanding the effects of treatment of ter- of forearm and wrist fractures, the fracture site term was kept as re- iparatide on hip fractures is important given the clinical impact and ported in the original publication; fractures reported at the “radius” health economic importance of this type of fractures, and the lack of were included in the forearm fractures analysis, while fractures re- data from individual randomized clinical trials adequately powered to ported at “carpal bones” were included in the wrist fractures analysis. evaluate the efficacy of this drug on hip fractures. The lack of detailed results of other individual non-vertebral fracture The primary aim of this systematic review and meta-analysis was to sites in most of the publications precluded a pooled analysis of all non- evaluate the efficacy of teriparatide in the reduction of hip fractures in vertebral fractures. women and men with osteoporosis. The secondary objective was the For consistency with the primary objective, we focused the analysis investigation of the effects of teriparatide on upper limb fractures (i.e. on low trauma, fragility upper limb fractures when that information humerus, forearm and wrist), in order to evaluate the fracture results in was explicitly disclosed in the publications. non-weight bearing bones. 2.4. Quality assessment 2. Material and methods The scale recommended by the Cochrane Collaboration [10] was 2.1. Literature search strategy used to assess the methodological quality of identified studies as well as the risk of bias of the individual included. The major criteria of the We designed this study according to the Preferred Reporting Items checklist were randomization, double blind (both patients and re- for Systematic Reviews and Meta-Analysis Guidelines [8]. The protocol searcher/assessor), comparability of treatment groups, available follow- was registered in the International Prospective Register of Systematic up information and equal treatment used for treatment groups. Each Reviews, PROSPERO (registration number: CRD42018085635). criterion might be answered in three ways: yes (adequate information), A comprehensive search strategy aimed to identify RCTs that re- no (inadequate information), or unclear information. ported non-spine fractures either as an efficacy or safety endpoint, published in English, French or Spanish until 22 November 2017. 2.5. Data synthesis and statistical analysis Literature searches on MEDLINE (OVID interface, 1948 onwards) and EMBASE (OVID interface, 1974 onwards) were performed. Keywords As it was expected that many studies would report zero fractures in were identified with medical subject headings (MeSH) and text words at least one of the treatment arms, two procedures were applied to related to “teriparatide” and “parathyroid hormone (1-34)”. Additional estimate the risk of fracture: odds ratio (OR) by the Peto's method and sources included searches in clinical trial repositories (clinicaltrials.gov Mantel-Haenszel with empirical correction, with 95% confidence in- and EUDRAC), and contact with main authors and manufactures that tervals (CI) [11]. Differences were considered statistically significant owned study reports of clinical trials on teriparatide that missed some when p-value was < 0.05. If there were more than two comparators in fracture site details in the peer-reviewed publications. the same trial, the results of the comparators were pooled initially, and separated in sensitivity analysis (placebo and active comparators). A random effects model was not considered a priori, as heterogeneity is 2.2. Selection criteria very uncommon when the data are sparse. Statistical heterogeneity was tested using the Q test (significance level: 0.1) and I2 statistic (0% to The main outcome of interest was the incidence of fragility, low 40%: might not be important; 30% to 60%: may represent moderate trauma hip fractures following teriparatide treatment. Eligible studies heterogeneity; 50% to 90%: may represent substantial heterogeneity; met the following inclusion criteria: (1) RCTs with study participants 75% to 100%: considerable heterogeneity). Meta-regression procedures treated according to the approved indications for the use of teriparatide were applied to ascertain whether mean age and duration of trial had in the USA, Europe and Japan (i.e.; postmenopausal and male osteo- an influence on the risk of fracture. An assessment for publication bias porosis, glucocorticoid-induced osteoporosis); (2) minimum of 6-month was performed with a visual inspection of funnel plot asymmetry and treatment exposure; (3) a maximum treatment duration as approved also by Egger weighted regression tests [12,13]. (i.e.; 24 months), at the approved dose (20 μg/day subcutaneously) [9], A pre-planned sensitivity analysis was performed by removing the and (4) controlled with placebo or other active anti-osteoporosis drug, two clinical trials that did not report the hip fracture details in the peer- either commercially available, or under clinical development. Narrative reviewed publications, but in clinical study reports filed by the studies reviews, observational studies, comments, opinion pieces, methodolo- sponsor [14,15]. gical reports, editorials, and letters were excluded. Reviews were Analyses were performed using the Stata 15 statistical package screened to check for potential additional studies that were not pub- (College Station, TX, USA). lished as standalone papers. Following the screening of titles and abstracts, duplicates were re- moved. If multiple reports were available from the same study, fracture 2.6. Subgroup analysis data from the most recent report were included. Two authors (ADP and FM) independently analyzed the titles and abstracts and then performed To address heterogeneity among study populations, pre-specified the study selection and evaluation of quality and data extraction of the subgroup analyses were performed as follows: type of comparator full papers. Discrepancies were resolved by consensus between the re- (placebo or active-treatment osteoporosis drug), type of randomization viewer authors. (double-blind or open-label), patient characteristics (age and gender) and follow-up period duration. 2 A. Díez-Pérez et al. Bone 120 (2019) 1–8 Fig. 1. Study selection flow diagram (preferred reporting items for systematic reviews and meta-analysis). 3. Results strontium ranelate [25], denosumab [29,30,32], abaloparatide [3] and romosozumab [31,33]. In the control group, 1667 (35.1%) patients 3.1. Summary of searches and study selection process received placebo and 3084 (64.9%) patients an active comparator (Table 2). Seventeen studies were of women with postmenopausal os- A total of 609 citations were identified through searches plus 4 on- teoporosis, 3 studies included exclusively men [16,20,28], and 3 studies file clinical trial reports, of which 404 records remained after removal included a mixed population of men and women [1,14,15] (Table 1). Of of duplicates. After screening via titles and abstracts, 372 articles and all subjects, there were 545 men (6.3% of the total subjects) (Table 1). study reports remained for further evaluation. Following further eva- Two studies were of patients with glucocorticoid-induced osteoporosis luation, 310 more articles were excluded for the following reasons: not [14,28] for a total of 520 patients (6.0% of the total included subjects). a RCT, off-label use of teriparatide, non-approved doses, testing of Overall, subjects' mean age (SD) was 67.0 (4.5) years, and the median combination therapies with other anti-osteoporosis drugs, or they were teriparatide treatment duration was 18 months (range: 6 to 24 months). non-human studies. In many instances, more than one exclusion criteria applied. Sixty-two records remained for full-text assessment, 39 of them 3.2. Risk of bias assessment were further excluded because the index drug was weekly teriparatide, no fracture results were included, or they were published in languages All included studies had a low risk of bias according to randomi- other than English, French or Spanish. Therefore, a total of 23 clinical zation, allocation concealment, and comparability of intervention trials published between 2001 and 2017 were included in the current groups at the beginning of the trial and equal treatment of intervention meta-analysis (Fig. 1). groups (Supplemental Table 1). No evidence of publication bias was Detailed information of the included studies is summarized in detected using funnel plot or Egger's test (intercept = 0.94; 95% CI: Table 1. Twenty-three RCTs were included, representing data on 8644 −3.00 to 1.12, p = 0.315). subjects, 3893 of them treated with teriparatide. Eight trials were double-blind during the full study follow-up [2,14,16–21], 12 open- 3.3. Pooled estimate of the effect of teriparatide on hip and upper limb bones label [3,22–33], and 3 trials had a mixed design with an initial double- fractures blind phase (6 to 12 months duration) followed by an open-label phase for up to 18 to 24 months [1,15,34] (Table 1). There were 4 placebo- A total of 34 incident hip, 31 humerus, 31 forearm, and 62 wrist controlled trials [2,15,16,18], 17 active-controlled trials fractures were reported (Table 2). In 14 studies, representing 1651 [1,14,17,19–34], one three-arm study [3], and one multiple-arm study patients, no hip fractures were reported [16–18,20,22–27,29–32,34] [31], where teriparatide was used open-label as an active comparator in (Table 1). Meta-analysis results showed an OR (95% CI) for hip frac- randomized clinical trials of abaloparatide and romosozumab respec- tures of 0.44 (0.22–0.87; p = 0.019) in patients treated with teripara- tively. The active-comparators were risedronate [1,19–21,24,28], tide compared with controls (Fig. 2). The effects on the risk of humerus alendronate [14,26,31], zoledronic acid [27,34], calcitonin [22,23], [OR (95%CI): 1.02 (0.50–2.08)], forearm [OR (95%CI): 0.53 3 A. Díez-Pérez et al. Bone 120 (2019) 1–8 Table 1 Characteristics of the included RCTs of teriparatide and hip and upper limb fractures included in the meta-analysis. Study alias Patients (n) Study Mean age Males (%) Comparator(s) Trial design Main study Type of Trial duration Trial population endpoint fracture (months) phase endpoint Neer 2001 [2] 1085 PMW 69.0 0 Placebo DB VFx Efficacy 21 3 Hadji 2012 [19] 710 PMW 71.0 0 Risedronate DB Back pain Efficacy 18 3b Malouf 2017 [1] 216 PMW and men 76.8 22.8 Risedronate DB + OL BMD Safety 18 4 Glüer 2013 [28] 92 GIO 56.3 100 Risedronate OL BMD Safety 18 3b Kendler 2018 [21] 1360 PMW 72.1 0 Risedronate DB VFx Efficacy 24 4 Miller 2016 [3] 2463 PMW 68.8 0 Placebo/Abaloparatide OL VFx Efficacy 18 3 Leder 2014 64 PMW 65.9 0 Denosumab OL BMD Safety 24 4 [29,30] Cosman 2011 [27] 275 PMW 65.0 0 Zoledronic Acid OL BMD Safety 12 4 Dempster 2016 69 PMW 64.5 0 Zoledronic Acid DB + OL Biopsy Safety 24 4 [34] Kung 2006 [22] 104 PMW 70.6 0 Calcitonin OL BMD Safety 6 4 Recker 2009 [25] 79 PMW 64.6 0 Strontium OL Biopsy Safety 6 4 Walker 2013 [20] 19 Men 52.8 100 Risedronate DB BMD Safety 18 4 Dempster 2016 69 PMW 63.4 0 Denosumab OL Biopsy Safety 6 4 [32] Saag 2007 [14] 428 PMW and men 56.7 19.4 Alendronate DB BMD Safety 18 3 GIO McClung 2005 203 PMW 66.0 0 Alendronate DB BMD Safety 18 3 [17] Orwoll 2003 [16] 298 Men 59.0 100 Placebo DB BMD Safety 11 3 Miyauchi 2010 203 PMW and men 69.6 7 Placebo DB + OL BMD Safety 24 3 [15] McClung 2014 206 PMW 66.7 0 Placebo/Alendronate/ OL BMD Safety 12 2 [31] Romosozumaba Panico 2011 [26] 81 PMW 62.5 0 Alendronate OL BMD Safety 18 4 Miyauchi 2008 77 PMW 70.7 0 Placebo DB BMD Safety 6 2 [18] Anastasilakis 2008 44 PMW 65.1 0 Risedronate OL BMD Safety 12 4 [24] Hwang 2006 [23] 63 PMW 67.5 0 Calcitonin OL BMD Safety 6 4 Langdahl 2017 436 PMW 71.5 0 Romosozumab OL BMD Safety 12 3 [33] Abbreviations: BMD: Bone Mineral Density (areal or volumetric); DB: Double Blind; GIO: Glucocorticoid-induced osteoporosis; OL: Open Label; PMW: Postmenopausal women; VFx: Vertebral Fractures. a Only the romosozumab 210 mg once monthly arm was included to match the phase 3 trials selected dose. (026–1.08)] and wrist fractures [OR (95%CI): 1.21 (0.72–2.04)] were reduced [OR (95%CI): 0.49 (0.24–0.98); p = 0.045; heterogeneity Chi- not statistically significant (p > 0.05) (Fig. 3a–c). The effect on the risk squared: 5.92, I2 = 0.0% (p = 0.432)]. of combined upper limb fractures was non-significant [OR (95%CI): 0.94 (0.65–1.35)] (p > 0.05) or by type of comparator (Supplemental Fig. S3). 4. Discussion In a subgroup analysis based on the type of comparator (placebo or active), a significant risk reduction for hip fractures was observed in the Our meta-analysis based on 8644 subjects enrolled in 23 rando- placebo stratum [OR (95%CI): 0.20 (0.05–0.81); p = 0.024; hetero- mized, controlled clinical trials of > 6-month duration, shows that geneity Chi-squared: 0.73, I2 = 0.0% (p = 0.694)], but not in the ac- teriparatide 20 μg per day reduces the risk of hip fractures by 56% after tive-comparator stratum [OR (95%CI): 0.54 (0.25–1.17); p = 0.118; a median treatment duration of 18 months. The analysis of the upper heterogeneity Chi-squared: 5.71, I2 = 12.5% (p = 0.335)] limb bones fractures showed neutral effects of teriparatide on the risk of (Supplemental Fig. S1). The analysis of the hip fracture results by study fractures in these non-weight bearing bones. blinding (double-blind vs open-label) showed a borderline non-sig- These results complement previous meta-analysis that has shown nificant reduction in both strata: the ORs (95% CI) were 0.50 the statistically significant efficacy of teriparatide in reducing the risk of [(0.24–1.03); p = 0.059] for the double-blind, and 0.44 [(0.22–0.87); vertebral and non-vertebral fractures in postmenopausal women with p = 0.089] for the open-label strata (Supplemental Fig. S2). The non- osteoporosis [35–39]. Our findings of the hip fracture efficacy also statistical differences for the individual upper limb fracture sites re- confirms previous results by Murad et al. [38] who reported in a net- mained similar in these subgroups analyses (data not included), with work meta-analysis, combining direct and indirect estimates, that ter- the exception of a statistical significant increased OR for wrist fractures iparatide had the highest probability (42%) of being ranked as most in the analysis of teriparatide versus active comparators [OR (95%CI): effective and had the highest reduction in the risk of hip fracture (OR, 1.93 (1.02–3.65); p = 0.045]. 0.42). However, the review by Murad et al. [38] included several stu- There was no indication for a relationship between fracture risk dies with non-approved teriparatide doses for the treatment of osteo- reduction and patients' age or the clinical trial duration. Analysis by porosis (40 and 100 μg/day), and only 4 of the 23 RCTs that we have gender was not feasible given the low number of hip fractures in male included in the present systematic review were analyzed. subjects (n = 1). The findings of this meta-analysis are consistent with several other After removing the 2 hip fractures that were not described in the in lines of evidence suggesting that teriparatide is likely to reduce the risk the peer-reviewed papers [14,15], but which were included in the of hip fractures, including that teriparatide increases cancellous bone clinical study reports, the hip fracture risk remained statistically volume, improves bone architecture, and increases cortical thickness associated with increased cortical remodeling at the hip [4,40], which 4 A. Díez-Pérez et al. Bone 120 (2019) 1–8 Table 2 Sample size and number of hip, humerus, forearm and wrist fractures for each study included in the meta-analysis. Study alias Comp(s) TPTD (n) Hip Humerus Forearm Wrist COMP Hip Humerus Forearm Wrist TPTD (n) TPTD (n) TPTD (n) TPTD (n) (S) (n) COMP (n) COMP (n) COMP (n) COMP (n) Neer 2001 [2] Placebo 541 1 2 NR 2 544 4 2 NR 7 Hadji 2012 [19] Risedronate 360 5 4 0a 4 350 2 5 4a 2 Malouf 2017 [1] Risedronate 106 2 1 1a 0 110 7 1 0a 0 Glüer 2013 [28] Risedronate 45 0 0 0 0 47 1 0 1 0 Kendler 2018 [21] Risedronate 680 2 4 6 2 680 5 2 10 1 Miller 2016 [3] Placebo/ 818 0 2 0 17 821/824 2/0 3/1 4/1 15/7 Abaloparatide Leder 2014 Denosumab 31 0 NR NR NR 33 0 NR NR NR [29,30] Cosman 2011 [27] Zoledronic Acid 138 0 0 0 1 137 0 1 0 2 Dempster 2016 Zoledronic Acid 34 0 NR NR NR 35 0 NR NR NR [34] Kung 2006 [22] Calcitonin 47 0 0 0 0 57 0 0 0 0 Recker 2009 [25] Strontium 39 0 0 0 0 40 0 0 0 0 Walker 2013 [20] Risedronate 9 0 0 0 0 10 0 0 0 0 Dempster 2016 Denosumab 33 0 NR NR NR 36 0 NR NR NR [32] Saag 2007 [14]a Alendronate 214 0 0 1 0 214 1 0 0 0 McClung 2005 Alendronate 102 0 0 0 1 101 0 0 0 0 [17] Orwoll 2003 [16] Placebo 151 0 0 1 0 147 0 0 0 0 Miyauchi 2010 Placebo 136 0 1 0 0 67 1 1 1 0 [15]a McClung 2014 Placebo/Alendronate/ 54 0 0 0 0 50/51/ 0/0/0 0/0/0 0/0/0 0/0/1 [31] Romosozumab 51 Panico 2011 [26] Alendronate 42 0 0 0 0 39 0 0 0 0 Miyauchi 2008 Placebo 39 0 0 0 0 38 0 0 0 0 [18] Anastasilakis 2008 Risedronate 22 0 0 0 0 22 0 0 0 0 [24] Hwang 2006 [23] Calcitonin 34 0 0 0 0 29 0 0 0 0 Langdahl 2017 Romosozumab 218 0 1 1 0 218 1 0 0 0 [33] Total 3893 10 15 10 27 4751 24 16 21 35 COMP(S): Comparator(s); NR: Not Reported; TPTD: Teriparatide. a Fracture Data: Lilly Data on File. results in improved biomechanical properties [41]. Additionally, ter- (DANCE) observational study of 4085 men and women treated with iparatide 20 μg/day increased femoral neck and total hip BMD during teriparatide for 24 months, showed the rate of hip fractures decreased ongoing treatment through 24 months [42]. Also, an analysis of the as treatment duration increased [4]. Finally, the real-world experience Direct Assessment of Nonvertebral Fractures in Community Experience of teriparatide in reducing the risk of fractures showed hip fractures Fig. 2. Forest plot of hip fracture outcomes. 5 A. Díez-Pérez et al. Bone 120 (2019) 1–8 Fig. 3. Forest plots of upper limb fracture outcomes: (a) humerus, (b) forearm, (c) wrist. decreased as adherence and persistence to teriparatide treatment in- safety concerns raised by preclinical findings of increased risk of os- creased [6]. teosarcoma in rats [43], precluding the completion of the planned 3- Of note, the hip fracture efficacy with teriparatide was detected year treatment duration, and the likely accumulation of higher numbers after a relatively short median treatment duration of 18 months, with a of hip fractures. The approved maximum duration of teriparatide maximum treatment duration in the included studies of 24 months. The treatment of 24 months is shorter than for other osteoporosis treat- pivotal trial for teriparatide [2] was discontinued early because of ments, which limits the ability to show hip fracture efficacy in 6 A. Díez-Pérez et al. Bone 120 (2019) 1–8 individual clinical trials. Therefore, the demonstration of hip anti- Founding source fracture efficacy in this meta-analysis is important and clinically re- levant information since reduction in hip fracture risk is one of the This study was supported by Lilly. Lilly manufactures and com- major efficacy outcomes for osteoporosis treatments because of their mercializes teriparatide. associated morbidity, mortality and socioeconomic burden. Previous studies have shown that teriparatide induces a transient, Declaration of interest early decrease in bone density at the proximal femur in patients pre- treated with antiresorptive drugs [33,44,45]. Imaging and histological Dr. Díez-Pérez received honoraria, consultancies and/or research studies show increased cortical porosity in patients initiating teripara- funding from Amgen, UCB, Roche, Mereo, Lilly and owns shares of tide therapy, probably reflecting increased cortical bone remodeling Active Life Scientific. Dr. Eriksen received speaker and consulting fees [46,47]. Although none of these studies show increased hip fracture from Lilly, Amgen, Merck, Pfizer, EffRx and IDS. Dr. Kendler received risk, this finding has been a concern for some clinicians [48]. Our study honoraria, consultancies, and/or research funding from Lilly, Amgen, helps to clarify that teriparatide was not associated with any long-term Astra-Zeneca and Pfizer. Drs. F. Marin and J.H. Krege are employees safety issues regarding cortical-rich fracture sites, rather the opposite. and shareholders of Lilly. Dr. Delgado-Rodríguez has nothing to dis- However, the meta-analytical approach did not permit an evaluation of close. the time course of this reduction. Similarly, cortical bone mineral density loss at the distal radius [2,49] has also been a concern. How- Author roles ever, our meta-analysis showed a neutral effect on forearm and wrist fractures overall. Dr. Díez-Pérez, Dr. Marin and Prof. Delgado-Rodríguez designed the This meta-analysis has some limitations. The total number of hip study. Dr. Díez-Pérez and Dr. Marin searched literature databases and fractures (n = 34) is small, however, the number of fractures we col- performed the selection of studies and data collection. Prof. Delgado- lected in the teriparatide intervention arm (n = 10) vs. the comparator Rodríguez made the statistical analyses of the data. Dr. Díez-Pérez and arms (n = 24) was similar to that in the alendronate pivotal FIT-I trial Dr. Marin wrote the first manuscript draft. All authors interpreted the (n = 11 in alendronate and n = 22 hip fractures in placebo) [50], but results, reviewed and approved the final manuscript. Prof. Delgado- lower than in other pivotal studies including the HORIZON trial for Rodríguez takes responsibility for the integrity of the data analysis. zoledronate (n = 140 hip fractures) [51], the FREEDOM trial for de- nosumab (n = 70 hip fractures) [52], and the HIP trial for risedronate Acknowledgments (n = 101 hip fractures) [53]. These trials were, however, of longer treatment duration (2.9 years in FIT-I, and 3 years in the other studies), The authors would like to thank Yolanda Moreno, Department of and were compared with placebo. In the present meta-analysis, how- Medical Information, Lilly Europe, for her support in the bibliographic ever, 19 out of the 23 analyzed clinical trials had one or two active search. The authors thank, Andrea Rossi of International Scientific comparator arms with several drugs that have already shown anti- Communications, Lilly, for providing medical writing and editorial fracture efficacy at the hip, representing 64.9% of the patients included support, in accordance with Good Publication Practice (GPP3) guide- in the control group, that might have decreased the effect size of ter- lines. iparatide in the results. Another limitation is that we pool clinical trials of postmenopausal women, men, and patients on glucocorticoid References therapy where the risk for hip fractures may be different. Finally, we were unable to do a systematic review of other individual non-vertebral [1] J. Malouf-Sierra, U. Tarantino, P.A. García-Hernández, et al., Effect of teriparatide fracture sites given the incomplete reporting in most of the studies. In or risedronate in elderly patients with a recent pertrochanteric hip fracture: final results of a 78-week randomized clinical trial, J. Bone Miner. Res. 32 (5) (2017) fact, in only 4 of the 23 included clinical trials, were non-vertebral 1040–1051. fractures analyzed as an efficacy endpoint, while in most of them these [2] R.M. Neer, C.D. Arnaud, J.R. Zanchetta, et al., Effect of parathyroid hormone (1-34) fractures were reported as adverse events in the safety section. on fractures and bone mineral density in postmenopausal women with osteoporosis, N. Engl. J. Med. 344 (19) (2001) 1434–1441. The strength of this review is its comprehensive nature, including [3] P.D. Miller, G. Hattersley, B.J. Riis, et al., Effect of Abaloparatide vs placebo on new randomized, controlled trials with active comparators or placebo at the vertebral fractures in postmenopausal women with osteoporosis: a randomized approved teriparatide dose in the approved indications for its use. This clinical trial, JAMA 316 (7) (2016) 722–733. [4] E.F. Eriksen, T.M. Keaveny, E.R. Gallagher, J.H. Krege, The effects of teriparatide approach reduced the potential risk of recall and selection bias. We therapy at the hip in patients with osteoporosis, Bone 67 (2014) 246–256. assessed four fracture outcomes: hip, humerus, forearm and wrist. This [5] D.P. Beall, R.G. Feldman, M.L. Gordon, et al., Patients with prior vertebral or hip is the first systematic analysis of the efficacy of teriparatide at in- fractures treated with teriparatide in the direct assessment of nonvertebral fractures in community experience (DANCE) observational study, Osteoporos. Int. 27 (2016) dividual fracture sites of the upper limb. 1191–1198. [6] R.T. Burge, D.P. Disch, S. Gelwicks, X. Zhang, J.H. Krege, Hip and other fragility fracture incidence in real-world teriparatide-treated patients in the United States, 5. Conclusions Osteoporos. Int. 28 (2017) 799–809. [7] S.L. Silverman, K.G. Saag, B. Langdahl, et al., Hip and other fracture risk in patients receiving teriparatide in real-world clinical practice: pooled data from four pro- Our systematic review and individual-level meta-analyses of 23 spective observational studies, Osteoporos. Int. (Suppl. 1) (2017) (OC36). randomized clinical trials comparing teriparatide to placebo or other [8] L. Shamseer, D. Moher, M. Clarke, et al., Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation, anti-osteoporosis drugs in a total of 8644 subjects with osteoporosis in Br. Med. J. g7647 (2015) 349. the approved indications for its clinical use, indicate that there is a [9] US Food and Drugs Administration, FORTEO. Prescribing Information, Available at: significant reduction of hip fracture and a neutral effect on the pooled https://www.accessdata.fda.gov/drugsatfda_docs/label/2013/021318s036lbl.pdf, (2018) , Accessed date: 6 March 2018. upper limb fractures reported. Collectively, these results support con- [10] J.P.T. Higgins, D.G. Altman, P.C. Gøtzsche, et al., The Cochrane Collaboration's tool sideration of teriparatide for the treatment of osteoporosis as a first-line for assessing risk of bias in randomised trials, Br. Med. J. 343 (2011) 889–893. therapy in patients at high risk for osteoporotic fractures, although they [11] M.J. Sweeting, A.J. Sutton, P.C. Lambert, What to add to nothing? Use and avoidance of continuity corrections in meta-analysis of sparse data, Stat. Med. 23 should be interpreted in the context of the relatively small number of (9) (2004) 1351–1375. reported hip fractures. [12] M. Egger, G.D. Smith, M. Schneider, et al., Bias in meta-analysis detected by a Supplementary data to this article can be found online at https:// simple, graphical test, Br. Med. J. 315 (7109) (1997) 629–634. [13] J.L. Peters, A.J. Sutton, D.R. Jones, K.R. Abrams, L. Rushton, Comparison of two doi.org/10.1016/j.bone.2018.09.020. methods to detect publication bias in meta-analysis, JAMA 295 (6) (2006) 676–680. 7 A. Díez-Pérez et al. Bone 120 (2019) 1–8 [14] K.G. Saag, E. Shane, S. Boonen, et al., Teriparatide or alendronate in glucocorticoid- monoclonal antibody) versus teriparatide in postmenopausal women with osteo- induced osteoporosis, N. Engl. J. Med. 357 (20) (2007) 2028–2039. porosis transitioning from oral bisphosphonate therapy: a randomised, open-label, [15] A. Miyauchi, T. Matsumoto, T. Sugimoto, M. Tsujimoto, M.R. Warner, T. Nakamura, phase 3 trial, Lancet 390 (2017) 1585–1594. Effects of teriparatide on bone mineral density and bone turnover markers in [34] D.W. Dempster, P. Roschger, B.M. Misof, et al., Differential effects of teriparatide Japanese subjects with osteoporosis at high risk of fracture in a 24-month clinical and zoledronic acid on bone mineralization density distribution at 6 and 24 months study: 12-month, randomized, placebo-controlled, double-blind and 12-month in the SHOTZ study, J. Bone Miner. Res. 31 (2016) 1527–1535. open-label phases, Bone 47 (2010) 493–502. [35] P. Vestergaard, N.R. Jorgensen, L. Mosekilde, P. Schwarz, Effects of parathyroid [16] E.S. Orwoll, W.H. Scheele, S. Paul, et al., The effect of teriparatide [human para- hormone alone or in combination with antiresorptive therapy on bone mineral thyroid hormone (1-34)] therapy on bone density in men with osteoporosis, J. Bone density and fracture risk – a meta-analysis, Osteoporos. Int. 18 (2007) 45–57. Miner. Res. 18 (2003) 9–17. [36] R.B. Hopkins, R. Goeree, E. Pullenayegum, et al., The relative efficacy of nine os- [17] M.R. McClung, J.S. Martin, P.D. Miller, et al., Opposite bone remodeling effects of teoporosis medications for reducing the rate of fractures in post-menopausal teriparatide and alendronate in increasing bone mass, Arch. Intern. Med. 165 women, BMC Musculoskelet. Disord. 12 (2011) 209 http://www.biomedcentral. (2005) 1762–1768. com/1471-2474/12/209. [18] A. Miyauchi, T. Matsumoto, H. Shigeta, M. Tsujimoto, D. Thiebaud, T. Nakamura, [37] S.-L. Han, S.-L. Wan, Effect of teriparatide on bone mineral density and fracture in Effect of teriparatide on bone mineral density and biochemical markers in Japanese postmenopausal osteoporosis: meta-analysis of randomised controlled trials, Int. J. women with postmenopausal osteoporosis: a 6-month dose-response study, J. Bone Clin. Pract. 66 (2012) 199–209. Miner. Metab. 26 (2008) 624–634. [38] M.H. Murad, M.T. Drake, R.J. Mullan, et al., Comparative effectiveness of drug [19] P. Hadji, J.R. Zanchetta, L. Russo, et al., The effect of teriparatide compared with treatments to prevent fragility fractures: a systematic review and network meta- risedronate on reduction of back pain in postmenopausal women with osteoporotic analysis, J. Clin. Endocrinol. Metab. 97 (2012) 1871–1880. vertebral fractures, Osteoporos. Int. 23 (2012) 2141–2150. [39] T. Saito, J.M. Sterbenz, S. Malay, L.Z. Zhong, M.P. MacEachern, K.C. Chung, [20] M.D. Walker, N.E. Cusano, J. Sliney, et al., Combination therapy with risedronate Effectiveness of anti-osteoporotic drugs to prevent secondary fragility fractures: and teriparatide in male osteoporosis, Endocrine 44 (2013) 237–246. systematic review and meta-analysis, Osteoporos. Int. 28 (2017) 3289–3300. [21] D.L. Kendler, F. Marin, C.A.F. Zerbini, et al., Effects of teriparatide and risedronate [40] K.E.S. Poole, G.M. Treece, G.R. Ridgway, P.M. Mayhew, J. Borggrefe, A.H. Gee, on new fractures in post-menopausal women with severe osteoporosis (VERO): a Targeted regeneration of bone in the osteoporotic human femur, PLoS One 6 (1) multicentre, double-blind, double-dummy, randomised controlled trial, Lancet 391 (2011) e16190, , https://doi.org/10.1371/journal.pone.0016190. (2018) 230–240. [41] J. Borggrefe, C. Graeff, T.N. Nickelsen, F. Marin, C.C. Glüer, Quantitative computed [22] W.C. Kung, E.G. Pasion, M. Sofiyan, et al., A comparison of teriparatide and cal- tomographic assessment of the effects of 24 months of teriparatide treatment on 3D citonin therapy in postmenopausal Asian women with osteoporosis: a 6-month femoral neck bone distribution, geometry, and bone strength: results from the study, Curr. Med. Res. Opin. 22 (2006) 929–937. EUROFORS study, J. Bone Miner. Res. 25 (2010) 472–481. [23] J.S. Hwang, S.T. Tu, T.S. Yang, J.F. Chen, C.J. Wang, K.S. Tsai, Teriparatide vs. [42] R. Lindsay, J.H. Krege, F. Marin, L. Jin, J.J. Stepan, Teriparatide for osteoporosis: calcitonin in the treatment of Asian postmenopausal women with established os- importance of the full course, Osteoporos. Int. 27 (2016) 2395–2410. teoporosis, Osteoporos. Int. 17 (2006) 373–378. [43] J.L. Vahle, M. Sato, G.G. Long, et al., Skeletal changes in rats given daily sub- [24] A.D. Anastasilakis, D.G. Goulis, S.A. Polyzos, et al., Head-to-head comparison of cutaneous injections of recombinant human parathyroid hormone (1-34) for 2 years risedronate vs. teriparatide on bone turnover markers in women with post- and relevance to human safety, Toxicol. Pathol. 30 (2002) 312–321. menopausal osteoporosis: a randomised trial, Int. J. Clin. Pract. 62 (2008) 919–924. [44] B.M. Obermayer-Pietsch, F. Marin, E.V. McCloskey, et al., Effects of two years of [25] R.R. Recker, F. Marin, S. Ish-Shalom, et al., Comparative effects of teriparatide and daily teriparatide treatment on BMD in postmenopausal women with severe os- strontium ranelate on bone biopsies and biochemical markers of bone turnover in teoporosis with and without prior antiresorptive treatment, J. Bone Miner. Res. 23 postmenopausal women with osteoporosis, J. Bone Miner. Res. 24 (2009) (2008) 1591–1600. 1358–1368. [45] B.Z. Leder, J.N. Tsai, A.V. Uihlein, et al., Denosumab and teriparatide transitions in [26] A. Panico, G.A. Lupoli, F. Marciello, et al., Teriparatide vs. alendronate as a treat- postmenopausal osteoporosis (the DATA-switch study): extension of a randomised ment for osteoporosis: changes in biochemical markers of bone turnover, BMD and controlled trial, Lancet 386 (2015) 1147–1155. quality of life, Med. Sci. Monit. 17 (2011) CR442–R448. [46] E. Seeman, T.J. Martin, Co-administration of antirsorptive and anabolic agents: a [27] F. Cosman, E.F. Eriksen, C. Recknor, et al., Effects of intravenous zoledronic acid missed opportunity, J. Bone Miner. Res. 30 (2015) 753–764. plus subcutaneous teriparatide [rhPTH(1-34)] in postmenopausal osteoporosis, J. [47] E.F. Eriksen, J.P. Brown, Concurrent administration of PTH and antiresorptives: Bone Miner. Res. 26 (2011) 503–511. additive effects or DXA cosmetics, Bone 86 (2016) 139–142. [28] C.C. Glüer, F. Marin, J.D. Ringe, et al., Comparative effects of teriparatide and ri- [48] C. Crandall, Parathyroid hormone for treatment of osteoporosis, Arch. Intern. Med. sedronate in glucocorticoid-induced osteoporosis in men: 18-month results of the 162 (2002) 2297–2309. EuroGIOPs trial, J. Bone Miner. Res. 28 (2013) 1355–1368. [49] J.N. Tsai, A.V. Uihlein, S.M. Burnett-Bowie, et al., Effects of two years of ter- [29] J.N. Tsai, A.V. Uihlein, H. Lee, et al., Teriparatide and denosumab, alone or com- iparatide, denosumab, or both on bone microarchitecture and strength (DATA- bined, in women with postmenopausal osteoporosis: the DATA study randomised HRpQCT study), J. Clin. Endocrinol. Metab. 101 (2016) 2023–2030. trial, Lancet 382 (9886) (2013) 50–56. [50] D.M. Black, S.R. Cummings, D.B. Karpf, et al., Randomised trial of effect of alen- [30] B.Z. Leder, J.N. Tsai, A.V. Uihlein, et al., Two years of denosumab and teriparatide dronate on risk of fracture in women with existing vertebral fractures. Fracture administration in postmenopausal women with osteoporosis (the DATA extension Intervention Trial Research Group, Lancet 348 (1996) 1535–1541. study): a randomized controlled trial, J. Clin. Endocrinol. Metab. 99 (2014) [51] D.M. Black, P.D. Delmas, R. Eastell, et al., Once-yearly zoledronic acid for treatment 1694–1700. of postmenopausal osteoporosis, N. Engl. J. Med. 356 (2007) 1809–1822. [31] M.R. McClung, A. Grauer, S. Boonen, et al., Romosozumab in postmenopausal [52] S.R. Cummings, J. San Martin, M.R. McClung, et al., Denosumab for prevention of women with low bone mineral density, N. Engl. J. Med. 370 (5) (2014) 412–420. fractures in postmenopausal women with osteoporosis, N. Engl. J. Med. 361 (2009) [32] D.W. Dempster, H. Zhou, R.R. Recker, et al., Differential effects of teriparatide and 756–765. denosumab on intact PTH and bone formation indices: AVA osteoporosis study, J. [53] M.R. McClung, P. Geusens, P.D. Miller, et al., Effect of risedronate on the risk of hip Clin. Endocrinol. Metab. 101 (2016) 1353–1363. fracture in elderly women. Hip Intervention Program Study Group, N. Engl. J. Med. [33] B.L. Langdahl, C. Libanati, D.B. Crittenden, et al., Romosozumab (sclerostin 344 (2001) 333–340. 8