European Review for Medical and Pharmacological Sciences 2019; 23: 2640-2668 Drug efficacies on bone mineral density and fracture rate for the treatment of postmenopausal osteoporosis: a network meta-analysis L. YANG, N. KANG, J.-C. YANG, Q.-J. SU, Y.-Z. LIU, L. GUAN, T. LIU, X.-L. MENG, Y. WANG, Y. HAI Department of Orthopedics, Beijing Chaoyang Hospital, Capital Medical University, Beijing, China Abstract. – OBJECTIVE: Globally, a great Key Words number of elderly suffer from osteoporosis, es- Postmenopausal osteoporosis, Abaloparatide, De- pecially postmenopausal women. Osteoporosis nosumab, Teriparatide, Network meta-analysis. results in low bone mineral density (BMD) and high risk of fragility fracture. However, there is no defined strategy to select the most suit- able anti-osteoporotic drugs for osteoporosis Introduction patients. Therefore, this study aims to select the most effective anti-osteoporotic drug for post- Osteoporosis is a systemic skeletal disease in menopausal women with osteoporosis. which increased bone weakness highlights the risk of MATERIALS AND METHODS: Literature search fracture1. It is associated with a significant social and was conducted in PubMed, EMBASE, and the Co- public health burden. Elderly women are more likely chrane Library. Raw data from the related ran- domized clinical trials were extracted. A pairwise to suffer from this disease since the reduced estrogen and network meta-analysis model was utilized to levels after menopause contributes to a rapid decline assess the efficacy of ten drugs on the percentage in bone mass2,3. According to the National Osteopo- change of BMD in the lumbar spine and total hip rosis Foundation, there are approximately 9.1 million from baseline to one year of treatment. Risks of osteoporosis women and an additional 26 million vertebral fracture and non-vertebral fracture were women with low bone mass in America, which is far evaluated as well. We reported the effect size with a weighted mean difference (WMD) for continuous more than the estimated 2.8 million osteoporosis men outcomes and odds ratio (OR) for dichotomous and 14.4 million men with low bone mass. Several 4,5 outcomes. All the drugs were ranked based on prospective studies suggested that improved bone the surface under the cumulative ranking curve mineral density (BMD) is associated with a reduction (SUCRA) value. Furthermore, the heterogeneity, in the fracture rate. Hence, improving BMD and re- consistency and publication bias of enrolled litera- ducing fracture are the primary therapeutic goals. ture were assessed. There are two main categories of therapies ap- RESULTS: With regard to lumbar spine BMD, the ten selected drugs all showed significant effi- plied to prevent or treat postmenopausal osteoporo- cacy compared with placebo. In regard to total hip sis. One is anti-resorptive agents containing estrogen BMD and vertebral fracture, with the exception of or selective estrogen receptor modulators (bazedox- calcitonin, the remaining nine drugs all showed sig- ifen, raloxifene), calcitonin, bisphosphonates (alen- nificant efficacy compared with placebo. Six drugs dronate, ibandronate, risedronate, zoledronate), – abaloparatide, alendronate, risedronate, strontium denosumab, and odanacatib. The other category ranelate, teriparatide, and zoledronate – were signifi- cantly more effective compared with placebo for the is the drugs which have anabolic effects on bone treatment of non-vertebral fractures. As the SUCRA such as strontium ranelate, PTH1-84, and PTH1-34 6 values indicated, abaloparatide performed the best (teriparatide) . Recently, a new drug abaloparatide, on improving lumbar spine BMD, vertebral fracture which is also a parathyroid hormone-related protein and non-vertebral fracture, while denosumab was analog drug similar to teriparatide, has completed the best choice to improve total hip BMD. the Phase III trial and exerts a marked effect on the CONCLUSIONS: To sum up, abaloparatide, de- improvement of BMD and fracture rate7. nosumab, and teriparatide showed the best effi- cacy for the treatment of postmenopausal oste- There are multiple therapies for postmenopaus- oporosis, especially abaloparatide. al osteoporosis. Previously, randomized clinical 2640 Corresponding Author: Yong Hai, MD; e-mail: [email protected] Comparison of ten drugs for the treatment of postmenopausal osteoporosis trials (RCTs) and traditional pairwise meta-anal- hip, VF and NVF; (5) Sufficient data should be yses have been performed to determine the most provided in the original studies. effective one. However, the conclusions remain Exclusion data were applied: (1) Men or pre- controversial. Therefore, a network meta-analy- menopausal women were included in the study; sis (NMA) is necessary for identifying the most (2) Participants were treated by combined thera- effective therapy8. However, one previously-pub- py or sequential therapy; (3) The duration of fol- lished NMA9 had some drawbacks because it in- low-up was less than 12 months. cluded some sub-standard trials reporting oste- oporosis induced by glucocorticoid treatment or Data Extraction and Quality male osteoporosis. Furthermore, some other arti- Assessment cles10,11 only assessed a few drugs and the sample Data extracted from the included studies were size was small. assessed independently by two reviewers (San Therefore, we conducted this NMA in order Zhang, Si Li), with discussion to the third review- to evaluate the comparative efficacy of ten prima- er (Er Wang) to resolve any discrepancies. Infor- ry drugs in postmenopausal women with osteo- mation including the name of first author, year of porosis, including abaloparatide (ALE), alendro- publication, sample size, comparators, drug dos- nate (ALE), calcitonin (CT), denosumab (DEN), age, mean age of patients, blinding condition, out- ibandronate (IBA), risedronate (RIS), raloxifene comes of the study and maximum follow-up time (RLX), strontium ranelate (STR), teriparatide were extracted. The quality of the included studies (TPD), and zoledronate (ZOL). The efficacy of was assessed according to the modified JADAD the ten drugs on percentage change of BMD from score (out of 7). The studies gaining scores of 4 to baseline to one-year treatment in the lumbar spine 7 were considered as high-quality and regarded as and total hip, vertebral fracture (VF) rate and low-quality if they gained a score of 1 to 312. non-vertebral fracture (NVF) rate was assessed. Statistical Analysis Four outcomes (lumbar spine BMD, total hip Materials and Methods BMD, VF, and NVF) were analyzed. Pairwise me- ta-analysis of studies that directly compared different Search Strategy treatments was conducted using STATA 14.0 (Stata Literature search and identification process were Corp., College Station, TX, USA) software. The per- conducted in PubMed, EMBASE, and the Cochrane centage change of BMD in the lumbar spine and total Library from inception until April 18, 2018 (date hip was reported using the weighted mean differ- of final search). Articles published in both English ence (WMD) and the 95% confidence interval (95% and Chinese languages were searched using the CI). The risk of vertebral fracture and non-vertebral following medical subject headings: “Osteoporo- fracture was reported using the odds ratio (OR) and sis, Postmenopausal, Abaloparatide, Alendronate, the 95% confidence interval (95% CI). Cochran’s Q Calcitonin, Denosumab, Ibandronate, Risedronate, test and Higgins’ I-squared test were used to test the Raloxifene, Strontium Ranelate, Teriparatide, Zole- heterogeneity of enrolled studies. A p-value of the dronate, randomized controlled trial”, and their syn- Cochran’s Q test statistic less than 0.05 or I-square onyms. The search procedures were performed by larger than 50% indicated significant heterogeneity two independent reviewers. Reference lists of ac- among included studies for each pairwise compari- quired articles were manually searched. son. The fixed-effect model was applied for studies without significant heterogeneity and otherwise, the Inclusion and Exclusion Criteria random-effects model was applied. Studies meeting the following criteria were in- Next, the network meta-analysis was performed cluded: (1) A randomized controlled trial; (2) Sub- by Bayesian analysis methods using R software jects were postmenopausal women with osteopo- (Version 3.2.3; R Foundation for Statistical Com- rosis; (3) The study was designed to compare the puting, Vienna, Austria)13. The effect sizes of WMD effects of the following drugs with placebo (PLA) and OR and their corresponding 95% confidence in- or between each other: ABL, ALE, CT, DEN, tervals (95% CI) were calculated with a random-ef- IBA, RIS, RLX, STR, TPD, ZOL; (4) At least one fects model. In order to evaluate the consistency of of the following outcomes was assessed in each the network meta-analysis model, the node-splitting study: percentage change of BMD from baseline method was utilized to assess the difference between to one-year treatment of the lumbar spine or total direct and indirect comparisons. Furthermore, the 2641 L. Yang, N. Kang, J.-C. Yang, Q.-J. Su, Y.-Z. Liu, L. Guan, T. Liu, X.-L. Meng, Y. Wang, Y. Hai effects of drugs were ranked by the surface under articles were removed for other reasons (e.g., com- the cumulative ranking (SUCRA) curve values14. bined therapy or sequential therapy, unrelated drugs Higher SUCRA value indicated the pronounced or diseases other than osteoporosis, re-analysis or efficacy of the drug. Funnel