Introduction People with chronic kidney disease (CKD) [1] experience mortality in excess of the general population largely because of accelerated cardiovascular disease [2]–[4]. Although improvements in care have occurred, people with advanced CKD treated with dialysis (CKD stage 5D) experience annual mortality of approximately 15% to 20% [5]. Despite intensive efforts, numerous interventions to improve clinical outcomes in adults with CKD have failed to demonstrate beneficial effects on mortality or cardiovascular events, particularly for people treated with dialysis [6]–[10]. CKD leads progressively to phosphorus retention, impaired vitamin D activation, hypocalcemia, and increased parathyroid hormone (PTH) secretion. However, while global clinical practice guidelines suggest that serum PTH and phosphorus concentrations should be kept within a target range [11], therapies that ameliorate abnormal serum PTH and phosphorus levels (vitamin D compounds and phosphorus binders) have not been shown to improve clinical outcomes in randomized trials [12],[13]. In 2004, cinacalcet hydrochloride was approved in the United States to lower elevated serum PTH levels in patients with CKD stage 5D [14]. Cinacalcet mimics the action of calcium on calcium-sensing receptors in the parathyroid glands to suppress PTH secretion [15],[16] and, based on promising data, has been considered a potential intervention to prevent cardiovascular events and mortality in CKD [17],[18]. Within a decade of the first small randomized trials for cinacalcet [19],[20], and despite an earlier meta-analysis showing no evidence for benefit on clinical outcomes [21], cinacalcet prescribing has become the largest single drug cost for dialysis patients in the United States, with an annual expenditure of at least US$260 million, and community prescribing costs for cinacalcet in the United Kingdom increased by 20%–33% from 2010 to 2011 [5],[22]. A pooled analysis of four placebo-controlled randomized trials of cinacalcet in 2005 showing a large reduction in cardiovascular hospitalization with cinacalcet may have contributed to uncertainty in the medical community about the therapeutic benefits of cinacalcet therapy [18]. In light of widespread use and the recent publication of the largest randomized trial of cinacalcet in dialysis patients [23], we have conducted a systematic review and meta-analysis to summarize the available evidence that calcimimetic therapy improves clinical outcomes in adults with CKD. In the context of high cinacalcet prescribing costs despite an earlier meta-analysis reporting no evidence for benefit [21], we have used cumulative meta-analysis to evaluate the evidentiary basis for routine cinacalcet administration in clinical practice over time. Methods We conducted a systematic review and meta-analysis of randomized controlled trials according to methods from a previously published meta-analysis, and followed a published peer-reviewed protocol [21] (Texts S1 and S2). Data Sources and Searches We conducted electronic searches in the Cochrane Renal Group specialized register (through February 7, 2013) and Embase (January 1, 2012 to February 7, 2013) using search terms relevant to this review without language restriction (Table S1). The specialized register contains studies identified from the Cochrane Central Register of Controlled Trials, MEDLINE, handsearches of kidney-related journals and proceedings of major conferences, and searches of trials registries using search strategies based on the scope of the Cochrane Renal Group (http://www.cochrane-renal.org/crgtopics.php). We additionally searched Embase in the previous year (January 1, 2012, through February 7, 2013) for citations that were not automatically included in the specialized register. Study Selection Two authors independently screened the search results by title and abstract, then full text, to identify potentially eligible trials that fulfilled the inclusion criteria. We considered all randomized controlled trials of any calcimimetic agent (cinacalcet HCl, NPS R-467, or NPS R-568) that reported data for adults with CKD (any stage). We defined CKD according to the National Kidney Foundation Kidney Disease Outcomes Quality Initiative, which considers CKD to be present when there are structural kidney and/or urine abnormalities with or without reduced estimated glomerular filtration rate (below 60 ml/min per 1.73 m2) [1]. We have used standard nomenclature, referring to having an estimated glomerular filtration rate below 60 ml/min per 1.73 m2 but not treated with dialysis as CKD stages 3–5, and treated with dialysis as CKD stage 5D. Data Extraction and Quality Assessment Two authors independently extracted data for population characteristics, interventions, nonrandomized cointerventions, and risk of bias according to prespecified criteria from the Cochrane Collaboration's tool for assessing risk of bias [24] into a purpose-built database. Each author double-checked data extraction and data entry independently, and any discrepancies between authors were resolved by discussion. We extracted data for the following outcomes: all-cause mortality, cardiovascular mortality, parathyroidectomy, fracture, and treatment-related adverse events (including hypocalcemia, hypercalcemia, nausea, vomiting, abdominal pain, diarrhea, upper respiratory tract infection, muscle weakness or parasthesia, dyspnea, and headache). We also extracted data for end-of-treatment serum PTH, phosphorus, and calcium concentrations. Two authors independently evaluated the following risk-of-bias items using standardized methods: sequence generation, allocation concealment, blinding of patients and study personnel, blinding of outcome assessment, analysis by intention-to-treat methods, completeness of outcome data, selective reporting of outcomes, and other threats to validity (unequal treatment comparisons, early termination of trial, industry sponsor as author or involved in data handling and analysis) [24]. We also recorded whether trials published after 2005 reported trial registration in the primary trial report. Data Synthesis and Statistical Analysis For dichotomous outcomes, we calculated the relative risk (RR) and 95% confidence interval (CI). For continuous outcomes, we calculated the mean difference together with a 95% CI. Where only proportions of participants experiencing an event were provided in the trial report (instead of raw event data), we estimated the number of participants experiencing one or more events by multiplying the proportion affected by the sample size, and contacted the trial authors or sponsors for additional information. We summarized effect estimates using standard and cumulative random effects meta-analysis. We assessed for heterogeneity in summary effects using the Cochran Q and the I 2 test (with 95% CIs) [25]. We considered a p-value below 0.10 to indicate significant heterogeneity. We analyzed data for all-cause mortality, cardiovascular mortality, parathyroidectomy, hypocalcemia, nausea, and vomiting within subgroups for CKD comprising adults with CKD stage 3–5 and CKD stage 5D. Insufficient data were available to determine if treatment effects differed by stage of CKD, and data were absent for kidney transplant recipients. In cumulative meta-analysis, outcome data for all-cause mortality, parathyroidectomy, hypocalcemia, and nausea from all available trials were included sequentially according to the year in which they first became available. Additional prespecified subgroup analyses and univariate random effects metaregression were performed to explore potential sources of heterogeneity in treatment effects on all-cause mortality, parathyroidectomy, hypocalcemia, and nausea. The potential sources of heterogeneity included mean age of participants in the trial, proportion of male participants, baseline serum PTH concentration, baseline serum calcium concentration, trial duration, allocation concealment (adequate versus unclear), and year of publication. In addition, for the outcome of hypocalcemia we evaluated the serum calcium concentration used to define one or more hypocalcemia events as a source of heterogeneity in treatment effects for this outcome. Insufficient data were available to evaluate whether dialysis modality (hemodialysis versus peritoneal dialysis) modified treatment effect estimates. To assess potential bias from small-study effects, funnel plots of the log risk ratio in individual studies against the standard error of the risk ratio were generated and formally assessed for asymmetry using Egger's regression test [26]. The Duval and Tweedie trim-and-fill procedure [27] was used to quantify the possible effect of any potential publication bias evident in the meta-analyses. For all analyses, a two-tailed p-value 0.4 for all). Subgroup analyses for effects of allocation concealment were not possible for parathyroidectomy. When we excluded the three trials in which randomized cointervention strategies for vitamin D compounds were not comparable between treatment arms [37],[43],[45], we observed similar treatment estimates in dialysis patients (all-cause mortality RR, 0.97 [95% CI, 0.89 to 1.05]; cardiovascular mortality RR, 0.95 [95% CI, 0.84 to 1.08]; hypocalcemia RR, 6.72 [95% CI, 4.88 to 9.25]; nausea RR, 1.89 [95% CI, 1.38 to 2.60]; vomiting RR, 1.98 [95% CI, 1.71 to 2.30]), although risks of hypercalcemia became less certain (RR, 0.88 [95% CI, 0.55 to 1.41]). When we limited the analysis of all-cause mortality to trials that had a follow-up of 6 mo or longer, we found the treatment effect was unchanged (RR, 0.97 [95% CI, 0.90 to 1.05]). We observed asymmetry in the funnel plot for the outcome of all-cause mortality, suggesting that small studies remained unpublished (Egger's regression test, p = 0.01). When we imputed five potentially missing studies, the risk of all-cause mortality remained unchanged (RR, 0.97 [95% CI, 0.89 to 1.05]) (Figure S9). No asymmetry was observed in funnel plots for hypocalcemia or nausea, and data for parathyroidectomy were insufficient to allow for detection of small-study effects. We conducted sensitivity analyses to check the robustness of treatment effect estimates and their precision when trials in which zero events had occurred in one or both arms were available. We used a continuity correction of 0.5 added to all cells for such trials and found no substantive difference in the results. In addition to the cumulative meta-analysis reporting treatment estimates over time for all-cause mortality, parathyroidectomy, hypocalcemia, and nausea, we evaluated the cumulative treatment effect estimates for cardiovascular mortality before and after the inclusion of the large EVOLVE trial [23]. In the absence of EVOLVE, the summary RR for cardiovascular mortality was 0.25 (95% CI, 0.06 to 1.02), while including EVOLVE provided a RR of 0.68 (95% CI, 0.32 to 1.45). Discussion In high- to moderate-quality evidence from 16 randomized controlled trials involving 6,988 patients, routine cinacalcet (30 to 180 mg/d) therapy in people with CKD stage 5D decreases PTH concentrations (281 ng/l [32 pmol/l]), reduces hypercalcemia, and infrequently prevents surgical parathyroidectomy, but has little or no effect on all-cause mortality, has imprecise effects on cardiovascular death, and is associated commonly with adverse effects including hypocalcemia, nausea, vomiting, and diarrhea. On average, routinely treating 1,000 people for 1 y has no effect on mortality, might prevent three patients from experiencing surgical parathyroidectomy, and leads to approximately 60 and 150 patients experiencing hypocalcemia and nausea, respectively. Evidence in people with CKD stages 3–5 is scant and generally low or very low quality. Because of lower absolute risks of parathyroidectomy in earlier stages of CKD, the benefits of cinacalcet identified in dialysis populations are likely to be smaller if generalized to people with CKD stages 3–5. Data for recipients of a kidney transplant and those treated with peritoneal dialysis were largely absent. Although it remains possible that routine cinacalcet prescribing has a beneficial effect on all-cause mortality, consistent treatment effects across all the available studies providing data suggest that, at best, any benefit for mortality is likely to be small. Given that lag censoring analyses for outcomes (where data were censored 6 mo after patients stopped using the study drug) were reported as prespecified secondary analyses in the EVOLVE trial [23] and suggested a potential benefit for cinacalcet on total mortality (hazard ratio, 0.83 [95% CI, 0.73 to 0.96]), it might be argued that additional trials of cinacalcet are now needed or that cinacalcet lowers mortality. However, we suggest that, given that lag censoring approaches were secondary analyses and that overall data for mortality in this meta-analysis are high-quality according to GRADE criteria, additional placebo-controlled trials of cinacalcet are very unlikely to change the confidence in the size and direction of the treatment estimates we observed. By contrast, given the low- to very-low-quality evidence currently available for people with CKD stages 3–5, and the lack of available data to allow analysis of whether treatment effects differ by stage of CKD, additional trial data for this specific group of patients would be informative. Notably, the trials contributing to the analyses all sought to investigate the use of cinacalcet as “routine” or “first line” therapy for elevated PTH levels. Their findings therefore do not exclude the possibility that cinacalcet may afford benefits in the treatment of elevated PTH levels resistant to treatment with vitamin D compounds and phosphate binders. The generally negative findings of existing trials on patient-level end points have resulted in clinical practice guideline recommendations that suggest that cinacalcet should be used when serum parathyroid levels are very high, other treatments have been ineffective, and surgical parathyroidectomy is contraindicated [46]. However, the specific use of cinacalcet in this clinical setting has not been adequately evaluated in randomized trials, and, in particular, outcomes and adverse events after parathyroidectomy versus cinacalcet have not been studied. Before the development of cinacalcet, vitamin D compounds were the mainstay of therapy to normalize perturbed PTH concentrations in CKD, which if left unchecked lead to painful fractures, bone deformity, and generalized osteopenia. In a now familiar sequence of events in nephrology, although vitamin D therapy was effective for improving a surrogate outcome (lower serum PTH levels) and was associated with lower mortality in nonrandomized studies [47], subsequent randomized trials did not clearly demonstrate beneficial effects of vitamin D compounds on cardiovascular events or death for people with CKD [13]. Similarly, while cinacalcet was shown 10 y ago to markedly improve surrogate outcomes (both serum PTH and calcium by phosphorus product levels) in people with CKD [20], and observational analyses suggest an association between cinacalcet treatment and improved all-cause and cardiovascular mortality [17], until recently, randomized trial evidence systematically evaluating the effect of cinacalcet on clinical outcomes was not available. Despite this vacuum of high-quality evidence for patient-centered end points and cumulative data indicating frequent side effects including hypocalcemia, nausea, or vomiting, cinacalcet has become the most expensive drug cost and the eighth most frequently prescribed drug for Medicare Part D enrolled dialysis patients in the United States, and year-on-year prescribing costs are increasing rapidly in the United Kingdom [5],[22]. This meta-analysis shows that, although its use is widespread and costly, cinacalcet provides small absolute benefits for parathyroidectomy, provides no reductions in mortality, and frequently leads to adverse gastrointestinal effects that may adversely influence nutrition and quality of life in these patients. Importantly, the effects of cinacalcet treatment on all-cause mortality, parathyroidectomy, hypocalcemia, and nausea were all identifiable before the EVOLVE trial [23] was released in late 2012, and the EVOLVE trial has now largely only increased our confidence in treatment effects. The EVOLVE trial has additionally provided us with important data for cardiovascular mortality, showing that benefits of therapy on this outcome are lower than cumulatively estimated by earlier trials. The EVOLVE trial was needed to provide certainty and high-quality data that routine cinacalcet use provides little or no benefit for adults treated with dialysis; monitoring of prescribing data post-EVOLVE may now reveal a fall in the prescribing costs and frequency of routine cinacalcet administration in parallel with the high-quality evidence available, although questions will remain as to why prescribing costs became so high in the context of insufficient cumulated evidence over the last decade. As with vitamin D compounds previously, the pathway from drug development to clinical use for cinacalcet reminds us that relying on surrogate end points and nonrandomized studies to evaluate treatment efficacy for new interventions is likely to result in unreliable estimates of clinical effectiveness. This, in turn, leads to extensive use of interventions that do not improve population outcomes and unnecessarily increase healthcare expenditure. The treatment effect we observed for cinacalcet on fracture (RR, 0.53) was similar in magnitude to, but less certain than, the risk estimate observed in a pooled analysis of four similarly designed randomized, double-blind, placebo-controlled trials of cinacalcet enrolling 1,184 participants with CKD stage 5D and intact PTH concentrations of 300 ng/l or more, in which the RR of fracture was 0.46 (95% CI, 0.22–0.95) [18]. It was unclear in that publication which trials were included in the pooled analysis, from which data for extended treatment in two trials including about half the randomized participants were included. It is possible that cinacalcet lowers the risk of fracture, but at this time, treatment estimates based on published trial data summarized by meta-analysis are imprecise and lower quality. The current evidence for cinacalcet in this systematic review is consistent with the UK National Health Service National Institute for Health and Clinical Excellence guidance recommending that cinacalcet should not be used for the routine treatment of elevated serum PTH levels in people with CKD and should be limited to people with elevated PTH concentrations refractory to standard therapy, with a normal or high serum calcium concentration, and in whom surgical parathyroidectomy is contraindicated because the risks of surgery outweigh the benefits [46]. The data also support the current US Food and Drug Administration approval for cinacalcet, which is restricted to patients with CKD stage 5D who have secondary hyperparathyroidism, although benefits of treatment in this setting are limited to prevention of surgical parathyroidectomy and avoidance of hypercalcemia [14]. At this time, however, the available randomized evidence for cinacalcet does not support the current Kidney Disease: Improving Global Outcomes clinical practice guidelines suggesting that people with CKD treated with dialysis and elevated or rising PTH levels (beyond two to nine times the upper normal limit) receive vitamin D compounds or calcimimetics or a combination to decrease serum PTH levels to within the suggested range [11]. Although based on a peer-reviewed protocol and conducted using methods developed by the Cochrane Collaboration, our review has limitations that should be considered. First, data for cinacalcet therapy were largely limited to adults with CKD stage 5D. Insufficient data were available to determine whether treatment effects differed according to severity of CKD. Second, data for recipients of a kidney transplant were absent, although as in other stages of CKD, cinacalcet use may provide benefits outweighing treatment hazards for people requiring parathyroidectomy in whom surgical therapy is contraindicated. Third, due to a relative absence of trials in patients receiving peritoneal dialysis, treatment estimates for this specific group are uncertain. Finally, because of the lack of head-to-head data in available trials, the comparative effectiveness of cinacalcet versus vitamin D compounds for patient-level outcomes remains uncertain. In conclusion, cinacalcet therapy provides small reductions in the risk of surgical parathyroidectomy but has little or no effect on all-cause mortality and uncertain effects on cardiovascular death for people with CKD and is commonly associated with nausea and vomiting. Routine use of cinacalcet therapy in people with CKD does not appear warranted, and benefits may be limited to preventing parathyroidectomy in the small number of patients for whom surgery is contraindicated. Additional trials in patients with CKD stage 5D are unlikely to change the estimates of treatment effects for cinacalcet. Supporting Information Figure S1 Risk of bias in trials of cinacalcet therapy versus conventional treatment in adults with chronic kidney disease. (TIF) Click here for additional data file. Figure S2 Effect of cinacalcet plus conventional therapy versus placebo or no treatment plus conventional therapy on all-cause mortality in adults with chronic kidney disease. (TIF) Click here for additional data file. Figure S3 Effect of cinacalcet plus conventional therapy versus placebo or no treatment plus conventional therapy on cardiovascular mortality in adults with chronic kidney disease treated with dialysis. (TIF) Click here for additional data file. Figure S4 Effect of cinacalcet plus conventional therapy versus placebo or no treatment plus conventional therapy on parathyroidectomy in adults with chronic kidney disease treated with dialysis. (TIF) Click here for additional data file. Figure S5 Effect of cinacalcet plus conventional therapy versus placebo or no treatment plus conventional therapy on fracture in adults with chronic kidney disease treated with dialysis. (TIF) Click here for additional data file. Figure S6 Effect of cinacalcet plus conventional therapy versus placebo or no treatment plus conventional therapy on hypocalcemia in adults with chronic kidney disease. (TIF) Click here for additional data file. Figure S7 Effect of cinacalcet plus conventional therapy versus placebo or no treatment plus conventional therapy on nausea and vomiting in adults with chronic kidney disease. (TIF) Click here for additional data file. Figure S8 Effect of cinacalcet plus conventional therapy versus placebo or no treatment plus conventional therapy on hypercalcemia in adults with chronic kidney disease. (TIF) Click here for additional data file. Figure S9 Funnel plot to assess bias in estimates of all-cause mortality caused by small-study effects. Funnel plot assessing for potential publication bias. Individual studies reporting one or more events (n = 11), together with a diamond denoting the log rate ratio and 95% CI for actual studies, are shown in blue. Imputed hypothetical studies (n = 5) inserted using the Duval and Tweedie trim-and-fill method to account for missing studies with a lower risk for all-cause mortality are shown, together with the associated log rate ratio and its 95% CI, in red. The risk estimate for all-cause mortality adjusted for potentially missing studies is 0.97 (95% CI, 0.90 to 1.05). (TIF) Click here for additional data file. Table S1 Electronic search strategies. (PDF) Click here for additional data file. Table S2 Included studies comparing cinacalcet plus conventional therapy versus placebo or no treatment plus conventional therapy in adults with chronic kidney disease. (PDF) Click here for additional data file. Table S3 Definitions of parathyroid hormone and calcium targets triggering reduction in cinacalcet dose, and definition of hypocalcemia and hypercalcemia end points in included trials. (PDF) Click here for additional data file. Table S4 Effects of cinacalcet plus conventional therapy versus placebo or no treatment plus conventional therapy on end-of-treatment serum parathyroid hormone, phosphorus, and calcium concentrations in adults with chronic kidney disease. (PDF) Click here for additional data file. Table S5 Univariate metaregression exploring the role of patient and trial characteristics in the effect of cinacalcet therapy on clinical outcomes. (PDF) Click here for additional data file. Text S1 PRISMA checklist. (DOC) Click here for additional data file. Text S2 Protocol. (PDF) Click here for additional data file.
