Comparative efficacy and safety of paricalcitol versus vitamin D receptor activators for dialysis patients with secondary hyperparathyroidism: a meta-analysis of randomized controlled trials

Background Secondary hyperparathyroidism (SHPT) is a severe complication for dialysis patients. Vitamin D receptor activators (VDRAs) are used to treat SHPT, but the comparative efficacy and safety between paricalcitol and other vitamin D receptor activators for management of SHPT in dialysis patients has been unproven. Methods We searched PubMed, Embase, and the Cochrane Library for the time period through June 2017 to identify randomized controlled trials that evaluated paricalcitol compared with other VDRAs for treatment of SHPT. The primary outcome was the percentage of patients with target reduction of intact parathyroid hormone (iPTH) from baseline. Secondary outcomes included incidences of hypercalcemia and hyperphosphatemia. The random-effects model was used to estimate relative risks (RRs) with 95% confidence intervals (CIs). Results Eight studies (N = 759) were eligible for final inclusion. Compared with other VDRAs, no significant differences were found in the percentage of patients with target reduction of intact parathyroid hormone (iPTH) from baseline for paricalcitol treatment of SHPT in dialysis patients (RR, 1.01; 95% CI, 0. 87–1.18; p = 0.85). There were no differences in the incidence of hypercalcemia (RR, 0.95; 95% CI, 0.74–1.21; p = 0. 65) and hyperphosphatemia (RR, 0.94; 95% CI, 0.77–1.16; p = 0.58). Conclusions The presently available evidence is insufficient to draw a conclusion regarding whether paricalcitol therapy has a comparative efficacy and safety over other VDRAs for treating dialysis patients with SHPT. Large-sample, well-conducted, high-quality RCTs with patient-level outcomes (i.e., mortality) are urgently needed. Electronic supplementary material The online version of this article (10.1186/s12882-017-0691-6) contains supplementary material, which is available to authorized users.

To achieve current targets for the key mineral parameters set by KDIGO guidelines, a combination of dietary phosphorus restriction, phosphate binders, vitamin D receptor activators (VDRAs), and adequate dialysis was adopted. Among the VDRAs used for management of SHPT in dialysis patients, paricalcitol is the most potential vitamin D analog. The observational study by Teng et al. indicated a significant survival advantage of paricalcitol over calcitriol for long-term hemodialysis patients [8]. However, whether or not paricalcitol has a comparative efficacy and safety with other VDRAs for treating SHPT in dialysis patients, based on RCTs, is unknown.
Previous meta-analyses focused mainly on comparisons between paricalcitol and placebo [9][10][11] in both dialysis patients [12] and patients not requiring dialysis [9,10,13]. We performed a meta-analysis making a head-to-head comparison to check therapeutic advantages of paricalcitol over other VDRAs in dialysis patients with SHPT.

Methods
We conducted a meta-analysis of randomized controlled trials in compliance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement) guidelines [14]. The present meta-analysis had no registered protocol.

Data sources and Searches
Literature searches for randomized, controlled trials of vitamin D receptor activators in CKD 5D (dialysis) patients with secondary hyperparathyroidism were performed in PubMed, Embase, and the Cochrane Library for the period through January 2016. The Medical Subject Headings (MeSH) terms and corresponding keywords were used for the electronic searches. The search terms used were (MeSH exp. "Kidney Disease," "Kidney Failure," "Chronic Kidney Failure," and the keywords "chronic kidney", "chronic renal", "dialysis" and "hyperparathyroidism"), and (MeSH exp. "Vitamin D" and keywords "vitamin D" and "paricalcitol"). The complete search strategy is outlined in Additional file 1. The cited references of published primary studies and reviews were manually screened to identify relevant trials. To ensure literature saturation, we reran the searches on June 25, 2017. The ClinicalTrials.gov registry (https://clinicaltrials.gov/) was also searched. Trials were identified without language restriction.

Study selection
Two authors (Yifeng Xie and Peiling Su) independently conducted the search, screened the titles and abstracts, and checked full text for eligibility. Discrepancies between authors were resolved by consensus. Published RCTs meeting the following inclusion criteria were included: (1) population: chronic kidney disease 5D adult patients (hemodialysis or peritoneal dialysis) with secondary hyperparathyroidism; (2) intervention: paricalcitol therapy; (3) comparison: vitamin D analogues such as calcitriol, alfacalcidol, maxacalcitol, et al.; and (4) one or more the following outcomes: percentage of participants with target reduction in intact parathyroid hormone (iPTH) from baseline; incidence of hypercalcemia, hyperphosphatemia, and elevation in calcium phosphorus product; all-cause mortality and end-of-treatment serum phosphorus, calcium, and iPTH levels. Studies that did not meet inclusion criteria (i.e., participants were pediatric patients or comparison was placebo.) were excluded.

Data extraction and Quality assessment
Data extraction was performed by Yifeng Xie and Peiling Su using a standardized data extraction sheet (Office Excel®, Microsoft® Corporation) and confirmed independently by another author (Hongsheng Zhang). Three authors double-checked data entry independently. Any disagreements between authors were resolved by discussion and consensus. The collected data included the following: first author, year of publication, country, number of patients, follow-up, age, baseline mineral and bone disorders, laboratory values (serum calcium, phosphorus, and iPTH), types of phosphate binders used, dosing schedule, routine of administration, risk of bias data, and outcomes data. Risk of bias data were those that described randomization, double-blinding, and dropouts and withdrawals. Predefined primary outcome was the percentage of patients with target reduction in iPTH from baseline. Secondary outcomes included incidences of hypercalcemia, hyperphosphatemia, and elevation in the calcium-phosphorus product; end-oftreatment serum iPTH, phosphorus, and calcium; and all-cause mortality. For crossover trials, the data from the first period were used [15].
Two authors (Yifeng Xie and Peiling Su) independently assessed the risk of bias of the included studies using a validated Jadad 5-point scale [16,17]. The scale consists of three items describing randomization (0-2 points), masking (0-2 points), and dropouts and withdrawals (0-1 points) in the report of a randomize controlled trial. The scale ranged from 0 to 5, with higher scores showing better reporting. The trials were rated high quality when the score was higher than 2, or low quality when the score was 2 or below, out of a maximum score of 5.

Statistical analysis
We summarized effect estimates as relative risks (RRs) for dichotomous outcomes and weighted mean differences for continuous outcomes with 95% CIs. We assessed heterogeneity of treatment effects across studies with the I 2 statistic and I 2 > 50% significant heterogeneity [18]. We pooled effect estimates using a random-effects model, taking clinical heterogeneity between studies into account. To examine the influence of various factors on the treatment effects of paricalcitol therapy for dialysis patients with secondary hyperparathyroidism, we performed posthoc subgroup analyses according to baseline serum PTH levels (PTH > 68.4 pmol/L versus PTH < 68.4 pmol/L, with the cut-off value set as 9 times the upper normal limit for the assay), types of dialysis (hemodialysis versus peritoneal dialysis), routine of administration (oral versus intravenous), sample size (n > 100 for large sample studies versus n < 100 for small sample studies), and risk of bias (low quality versus high quality). Publication bias was evaluated visually by funnel plot and quantitatively by Egger's test and Begg's test [19,20]. We performed all statistical analyses using RevMan 5.3 (Nordic Cochrane Centre). P < .05 was considered statistically significant.

Trial selection
The PRISMA statement flowchart shows the process of study identification, screening, selection, and reasons for exclusion, as shown in Fig. 1. The combined search identified a total of 640 records, 614 of which were excluded after removing duplicates and screening the titles and abstracts. Full-text assessment of 26 potentially eligible articles identified 8 studies. Finally, 8 RCTs [21][22][23][24][25][26][27][28] were included in the meta-analysis.

Trials characteristics
Characteristics of patients and interventions of the included trials are summarized in Table 1. The sample size ranged from 18 to 263, with a total of 759 patients. Only two trials [21,27] had more than 100 participants. Of the included trials, seven compared paricalcitol therapy with non-selective vitamin D receptor activators, including calcitriol and alfacalcidol, and the remaining one [21] compared paricalcitol therapy with selective vitamin D receptor activator (maxacalcitol).
Five trials [21][22][23][24]27] reported the primary outcome, which was the percentage of patients with target reduction in intact parathyroid hormone (iPTH) from baseline. Of the five trials reporting the primary outcome, two trials [23,24] used a > 30% reduction as target reduction, whereas the remaining three trials [21,22,27] used a > 50% reduction. All included trials used different cut-off values to define hypercalcemia, ranging from 2.63 mmol/L (10.5 mg/dL) to 2.90 mmol/L (11.5 mg/dL). Five trials [21][22][23][24]27] reported incidence of hypercalcemia, one [24] of them using serum ionized calcium and the others serum total calcium. Only one trial [24] presented all-cause mortality data and others did not report patientlevel outcomes (i.e., all-cause mortality, cardiovascular mortality, and fractures). One trial [28] did not present any outcomes, either primary or secondary. All RCTs had a short duration of follow-up, lasting from 5 to 32 weeks.

Risk of bias assessment
The details of risk of bias are listed in Table 2. Overall, four trials [22,23,26,28] were categorized as low methodological quality and four [21,24,25,27] as high methodological quality. Only one trial [24] described appropriate allocation concealment. Three trials [21,25,27] reported double-blinding, and all trials except one [28] reported drop-outs and withdrawals.

Paricalcitol treatment versus vitamin D receptor activator treatment Primary outcome
Five trials [21][22][23][24]27] with a total of 696 patients were included for the outcome of the percentage of patients with target reduction in iPTH from baseline. Compared to either non-selective vitamin D receptor activators (calcitriol and alfacalcidol) or selective vitamin D receptor activator (maxacalcitol), paricalcitol had similar efficacy in suppressing the iPTH of dialysis patients with SHPT (Fig. 2.). The RR was 1.01 (95% CI: 0.87-1.18; p = 0.85). The statistic I 2 was 31%, indicating low heterogeneity.
There was no significant difference in suppressing iPTH with paricalcitol in comparison to other VDRAs for all subgroup analyses. Details of subgroup analyses were presented in Table 3.

Discussion
This meta-analysis demonstrates that there was no significant difference in effectiveness and safety of paricalcitol therapy in comparison to other VDRAs based on the biochemical end points when treating adult dialysis patients with secondary hyperparathyroidism. Comparisons based on patient-centered outcomes are not possible presently because of the lack of RCTs providing data on patient-level outcomes, such as mortality, cardiovascular death, fracture, and quality of life.
KDIGO's recommendations on dialysis patients with SHPT are to lower elevated phosphorus levels toward the normal range while maintaining intact PTH levels in the range of~2-9 times the upper normal limit for the assay and maintaining serum calcium in the normal range [1]. To reach the treatment target, limited dietary intake of phosphorus, phosphate binder, Vitamin D compounds, and adequate dialysis are adopted. A metaanalysis of 14 observational studies demonstrated that vitamin D compound therapies reduced mortality in CKD patients, particularly in those suffering from secondary hyperparathyroidism [29]. A study by Teng et al. showed that paricalcitol had a survival advantage over calcitriol in patients undergoing long-term hemodialysis [8]. However, all the included trials in this meta-analysis did  [30]. Fifth, dose ratios of 4:1 and 3:1, or an initial dose, depending on iPTH level, were used. Thus, the potential dose-response might affect the effect size [31]. Sixth, durations of follow-up were too short for the included studies. Seventh, fibroblast growth factor 23 (FGF 23) plays an important role in the mineral and bone disorders of CKD patients [32][33][34][35]. However, the included studies of this meta-analysis had no information on the serum FGF23 levels of patients enrolled in the studies, except that one study provided incomplete data of FGF23 [36]. Any significant within-group or between-group differences of FGF23 may affect the treatment effects of the included trials.  The negative result of our meta-analysis is consistent with previous meta-analyses [12,37,38]; however, this meta-analysis reveals several differences. First, the population is limited to CKD adult dialysis patients with SHPT. Patients not on dialysis were excluded because of significant differences of the association between serum phosphorus/calcium/iPTH and mortality in these two types of patients [1,39,40]. Second, active VDRA treatments were chosen as controls in this meta-analysis; thus, a direct comparison between paricalcitol and other VDRAs is made to check the superiority. Third, compared with recently published meta-analyses [37], this meta-analysis adds a comparison between selective vitamin D analogues (paricalcitol versus maxacalcitol).
This meta-analysis has the following limitations: First, the number of included studies is small and the studies had small sample sizes and short follow-up lengths. Second, the number of included trials is too small to evaluate publication bias of our meta-analysis, either visually by funnel plot or quantitatively by Egger's test [20] and Begg's test [19]. There may be potential publication bias in this meta-analysis. Third, comparisons between paricalcitol and other VDRAs are based on biochemical end points. Patient-level outcomes are not available in the included studies. Fourth, since the control included both non-selectvie VDRAs and selective VDRAs, this could introduce some heterogeneity for the summarized effect.

Conclusions
The evidence that is presently available is insufficient to draw a conclusion regarding whether paricalcitol therapy has a comparative efficacy and safety over other VDRAs