Aortic arch calcification and risk of cardiovascular or all-cause and mortality in dialysis patients: A meta-analysis

Studies on aortic arch calcification (AAC) and mortality risk in maintenance dialysis patients have yielded conflicting findings. We conducted this meta-analysis to investigate the association between the presence of AAC and cardiovascular or all-cause and mortality risk in maintenance dialysis patients. Observational studies evaluating baseline AAC and cardiovascular or all-cause mortality risk in maintenance dialysis patients were searched through the PubMed and Embase, CNKI, VIP and Wanfang databases until January 2016. A total of 8 studies with 3,256 dialysis patients were identified. Compared with patients without AAC, the presence of AAC was associated with greater risk of cardiovascular mortality (hazard risk [HR] 2.30; 95% confidence intervals [CI] 1.78–2.97) and all-cause mortality (HR 1.44; 95% CI 1.19–1.75). Subgroup analyses indicated that the pooled HR for cardiovascular and all-cause mortality was 2.31 (95% CI 1.57–3.40) and 1.45 (95% CI 1.08–1.96) for the grade 2/3 AAC. Peritoneal dialysis patients with AAC had greater cardiovascular (HR 3.93 vs. HR 2.10) and all-cause mortality (HR 2.36 vs. HR 1.33) than hemodialysis patients. The AAC appears to be independently associated with excessive cardiovascular and all-cause mortality in maintenance dialysis patients. Regular follow-up AAC might be helpful to stratify mortality risk in dialysis patients.

summarizes the baseline characteristics of the included studies. The included studies were published from 2010 to 2015. The follow-up duration ranged from 1.8 to 10 years. Five studies 5,6,8,9,11 were prospective design and three studies 7,10,12 were retrospective design. Among the 8 studies, 2 articles 10,12 were conducted in America and 6 articles [5][6][7][8][9]11 in Asia. All the studies determined the AAC using plain chest X-rays image. The prevalence of AAC in dialysis patients varied from 40.7% to 58%. All the included studies reported all-cause mortality as the outcome and six studies reported cardiovascular mortality. On the basis of NOS for cohort studies, the NOS scores of the included studies ranged from 5 to 8 stars.
All-cause and cardiovascular mortality. A total of 766 all-cause mortality cases were reported in eight studies 5-12 among 3,256 dialysis patients. As shown in Fig. 2, the presence of AAC was associated with 44% greater risk of all-cause mortality (HR 1.44; 95% CI 1.19-1.75; I 2 = 52.9%; P = 0.010) in a random effect model. A total of 287 cardiovascular mortality cases were reported in six studies 5-9,11 among 2,339 dialysis patients. As shown in Fig. 3, the presence of AAC was associated with 1.30 folds greater risk of cardiovascular mortality (HR 2.30; 95% CI 1.78-2.97; I 2 = 0.0%; P = 0.793) in a fixed-effect model.

Subgroup analyses and sensitivity analyses.
Subgroup analyses based on study design, region, patient population sample sizes, grade of AAC, and follow-up duration showed similar results across all the analyses ( Table 2). Sensitivity analyses by excluding any single study at each turn indicated that there were no changes in the direction of pooling risk estimate of all-cause mortality (pooled HR ranges from 1.39 to 1.57) and cardiovascular mortality (pooled HR ranges from 2.19 to 2.63).
Publication bias. Evidences of publication bias for all-cause mortality were not observed based on the funnel plot ( Fig. 4A), Begg's rank correlation test (P = 0.101), and Egger's linear regression test (P = 0.134). There were also no evidences of publication bias for cardiovascular mortality according to the funnel plot (Fig. 4B), Begg's rank correlation test (P = 0.371) and Egger's linear regression test (P = 0.207).

Discussion
The present meta-analysis provided evidences that the presence of AAC significantly increased the risk of all-cause mortality by 44% and cardiovascular mortality by 130% in dialysis patients. To the best of our knowledge, this is the first meta-analysis to investigate the relationship between the presence and severity of AAC and risk of cardiovascular and all-cause mortality in dialysis patients. Given AAC is easily determined by chest X-ray in clinical practice, regular follow-up AAC might be a simple and helpful method to stratify the mortality risk in dialysis patients.
Subgroup analysis revealed that the statistical significance of an association with mortality was more obvious in patients with grade 2 and 3 AAC. This finding supports a higher degree of AAC corresponds to a greater mortality risk. In addition, the presence of AAC in peritoneal dialysis patients appeared to have a greater mortality risk than those undergoing hemodialysis patients. Moreover, progression of AAC over one year was also an independent predictor of cardiovascular and all-cause mortality in incident peritoneal dialysis patients 8 .
Approximately 20-30% of people older than 65 years had calcification in the aorta 2 . In dialysis patients, the prevalence of AAC ranged from 37.29% to 58% based on the chest X-ray findings 7,12 . The high prevalence of AAC in dialysis patients indicates the importance to early detect the presence and progression of AAC. Several potential explanations may explain the presence and progression of AAC and mortality risk. AAC represented the magnitude of whole aortic calcification in the general population and dialysis patients 13,14 . The extent of AAC may be correlated to the degree of atherosclerosis. Calcification can increase stiffness and reduce elasticity of large arteries, resulting in substantial mortality 15 .
Our meta-analysis had several limitations. First, the most important concern is the sensitivity for detecting AAC in chest X-ray. Compared with plain X-ray, multi-detector computed tomography or electron beam-computed tomography are the gold standard for evaluating AAC, with the power of detecting small amounts of calcification. However, these examinations are too expensive to perform in all the dialysis patients. AAC assessed by a chest X-ray may underestimate the true calcium deposition in the aortic wall. Second, most dialysis patients in our analysis were adult and elder with a trend of acceleration of vascular calcification. Thus, predictive values of AAC on mortality risk cannot be extrapolated to relatively younger dialysis patients. Third, the included studies did not adjust covariates in a consistent way, lacking adjustment for these covariates may have led to a slight overestimation of the risk estimate. Finally, Despite we made a comprehensive literature search, there were very few studies included in this meta-analysis.The conclusion based on the limited number of study may be not robust, particularly in the subgroup analyses.
In conclusion, this meta-analysis indicates that AAC appears to be independently associated with greater risk of cardiovascular and all-cause mortality, and higher grade of AAC corresponds to a greater risk in dialysis patients. Our finding support incorporation of AAC into the existing risk factors for dialysis patients may improve the prognostic stratification. However, more well-designed prospective studies are need to confirm our findings because there were very few studies being included in the meta-analysis.

Methods
Search strategy. This meta-analysis was performed in accordance with the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-analyses Statement 16 An extensive electronic database search was conducted in PubMed, Embase, China National Knowledge Infrastructure, VIP and Wanfang databases up to January 2016. The following search terms were used: 'hemodialysis' OR 'haemodialysis' OR 'peritoneal dialysis' OR 'end stage renal disease' AND 'aortic calcification' OR 'aortic arch' AND 'calcification' OR 'calcium' AND 'mortality' OR 'death' AND 'follow-up' OR 'longitudinal' . Additionally, the reference lists of the selected papers were manually searched for additional possible studies.
Selection criteria. Studies were considered eligible for the present meta-analysis if: 1) original observational studies; 2) participants in the end-stage kidney disease who are undergoing maintenance dialysis; 3) investigating the relationship between the presence and extent of AAC at baseline and subsequent cardiovascular or all-cause mortality risk; and 4) reporting risk estimate of cardiovascular or all-cause mortality events. The severity of calcification was classified as grade 0 to 3 in accordance with previous studies 6,13,14 . For the multiple articles from the same research group, we only selected the most recent comprehensive publication. Studies were excluded if they were cross-sectional design, reviews or duplicated publication.     Data extraction and quality assessment. Two authors (A Zhang and SJ Wang) independently collected data from included studies using a structured form. Extracted information included first author' s name, publication year, study design, geographical region of study, baseline characteristics of patients, detection methods, prevalence of AAC, event numbers, fully adjusted risk ratio (RR) or hazard ratio (HR) and 95% confidence intervals (CI), duration of follow-up, and adjustment for covariates. Any discrepancies during the data extraction were resolved by discussion. We applied the Newcastle-Ottawa Scale (NOS) for cohort studies to evaluate the methodological quality of each study 10 . The NOS ranges from zero to nine stars. Studies achieving a rating of more than 6 stars were considered to be of higher quality.
Statistical analysis. The overall risk estimates were pooled using the most fully adjusted RR or HR with their 95% CI comparing with and without AAC. Heterogeneity between studies evaluated by the Cochran's Q (heterogeneity was set at a value of p < 0.10) and I 2 tests (I 2 > 50%). Random effect model was used for meta-analysis when there was significant heterogeneity; otherwise, a fixed-effect model was applied 17 . Subgroup analyses were performed by study design (prospective vs. retrospective), region (America vs. Asia), population (hemodialysis vs. peritoneal dialysis), sample sizes (> 500 vs. < 500), grade of AAC, follow-up duration (≥ 4 years vs. < 4 years), and NOS scores (≥ 6 stars vs. < 6 stars). The possibility of publication bias was tested by the Begg's 18 and Egger's 19 tests with significant publication bias considered as a p-value < 0.1.We performed a sensitivity analysis by excluding any single study at each turn to test the robustness of the pooled results. All statistical analyses were conducted using Stata 12.0 software.