Interventions To Attenuate Vascular Calcification Progression in Chronic Kidney Disease: A Systematic Review of Clinical Trials

Significance Statement

Vascular calcification is associated with cardiovascular morbidity and mortality in people with CKD. In this systematic review, we summarized evidence from randomized and nonrandomized clinical trials investigating effects of interventions that might attenuate progression of vascular calcification in CKD; interventions were compared with placebo, other comparators, or standard of care. We reviewed 77 heterogeneous clinical trials (63 randomized) involving 6898 participants. Therapy involving magnesium or sodium thiosulfate appears the most promising, with consistent findings of attenuation of vascular calcification progression, but evaluable studies were small and of short duration. Many other studies had inconclusive or conflicting outcomes. This study highlights the need for more definitive trials to evaluate interventions targeting vascular calcification in people with CKD, preferably in association with patient-centered outcomes.

Abstract

Background

Vascular calcification is associated with cardiovascular morbidity and mortality in people with CKD. Evidence-based interventions that may attenuate its progression in CKD remain uncertain.

Methods

We conducted a systematic review of prospective clinical trials of interventions to attenuate vascular calcification in people with CKD, compared with placebo, another comparator, or standard of care. We included prospective clinical trials (randomized and nonrandomized) involving participants with stage 3–5D CKD or kidney transplant recipients; the outcome was vascular calcification measured using radiologic methods. Quality of evidence was determined by the Cochrane risk of bias assessment tool and the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) method.

Results

There were 77 trials (63 randomized) involving 6898 participants eligible for inclusion (median sample size, 50; median duration, 12 months); 58 involved participants on dialysis, 15 involved individuals with nondialysis CKD, and 4 involved kidney transplant recipients. Risk of bias was moderate over all. Trials involving magnesium and sodium thiosulfate consistently showed attenuation of vascular calcification. Trials involving intestinal phosphate binders, alterations in dialysate calcium concentration, vitamin K therapy, calcimimetics, and antiresorptive agents had conflicting or inconclusive outcomes. Trials involving vitamin D therapy and HMG-CoA reductase inhibitors did not demonstrate attenuation of vascular calcification. Mixed results were reported for single studies of exercise, vitamin E–coated or high-flux hemodialysis membranes, interdialytic sodium bicarbonate, SNF472, spironolactone, sotatercept, nicotinamide, and oral activated charcoal.

Conclusions

Currently, there are insufficient or conflicting data regarding interventions evaluated in clinical trials for mitigation of vascular calcification in people with CKD. Therapy involving magnesium or sodium thiosulfate appears most promising, but evaluable studies were small and of short duration.

Cardiovascular disease (CVD) is the leading cause of mortality in people with CKD. Although the excess cardiovascular burden is, in part, related to an over-representation of conventional cardiovascular risk factors, there is also a significant contribution from nontraditional, CKD-related risk factors, which manifests as accelerated vascular calcification (VC) and valvular calcification.¹ Ectopic extraosseous calcification is accompanied by reduced bone mineral density^2,3 and associated with accelerated vascular stiffening and significantly increased risk of cardiovascular morbidity and mortality. In CKD, VC is driven to a large extent by impaired mineral homeostasis which constitutes part of the condition termed by Kidney Disease Improving Global Outcomes (KDIGO) as CKD mineral and bone disorder (CKD-MBD).^4,5

VC represents an aggregate end point of diverse vascular disease processes and considerable heterogeneity exists with respect to the anatomic site, the layer of the arterial wall affected, and the size of the calcifications present. Two major types of VC can be distinguished histologically, affecting either the intimal or medial arterial layer. Whereas intimal lesions are typically focal and associated with atheromatous lipid-rich plaques and inflammatory infiltrate, medial VC is usually more diffuse and occurs in the absence of lipid deposition or immune cell infiltration. Although considered distinct entities, with intimal VC showing a predilection for large- and medium-sized conduit arteries and medial VC for the same vessels as well as smaller muscular arteries of the periphery, both can coexist, even within the same arterial segment⁶ and with potentially amplificatory deleterious effects on the cardiovascular system.⁷ Cascades initiating and propagating mineralization at the cellular and biochemical levels are complex and not yet fully understood but thought to involve active cell-mediated mechanisms.^8–11 This heterogeneity leads to potentially divergent pathobiologic associations and cardiovascular outcomes for the patient, with atherosclerotic intimal VC relating to luminal narrowing and medial VC being linked to arterial stiffening and increased cardiac afterload. Despite this heterogeneity with respect to anatomy, natural history, molecular pathogenesis, and clinical sequelae, the common end point for VC is the formation and deposition of hydroxyapatite within the extracellular matrix of the arterial wall. CKD represents a perfect storm for VC with dysregulated mineral metabolism leading not only to the accumulation of promoters (such as hyperphosphatemia) but also to the simultaneous suppression and/or consumption of systemic and local inhibitors (e.g., fetuin-A, matrix Gla protein [MGP], pyrophosphate).¹² Consequently, both intimal and medial VC are highly prevalent and typically more severe in patients with CKD compared with age-matched individuals without kidney dysfunction.

VC has not been regarded as a direct therapeutic target in clinical practice as no intervention to date has been shown to reverse VC and there are few interventions to consistently reduce progression of VC.¹³ There is also a lack of evidence that slowing VC progression improves clinical outcomes, although associative data support this notion.^14,15 Thus, VC is considered a surrogate marker of CVD rather than an actionable target. The majority of candidate drug therapies do not target VC specifically and instead rely on addressing the imbalance of mineralization promoters and inhibitors that may engender ectopic mineralization through direct (e.g., phosphate lowering) or indirect (e.g., MGP activation with vitamin K) means. Management of hyperphosphatemia in CKD may address VC and involves dietary phosphate restriction, phosphate-lowering medication (such as intestinal phosphate binders), and dialysis. Vitamin D deficiency is related to endothelial dysfunction and development of VC and use of vitamin D therapy may have a role in influencing VC. For associated secondary hyperparathyroidism (SHPT) in CKD, active vitamin D analogs and calcimimetics can be initiated to reduce the numerous consequences of SHPT including VC. Use of antiresorptive agents (such as bisphosphonates and denosumab) for bone demineralization accompanying ectopic calcification may also help to mitigate VC. Although there is no comparable evidence for treating VC in CKD alone, newer trials have evaluated effects of targeting VC directly with SNF472 (a novel myo-inositol hexaphosphate derivative that inhibits hydroxyapatite formation) and magnesium, or indirectly with sodium thiosulfate and vitamin K.

For future studies to elucidate whether reduction in VC affects clinical outcomes, it is important to determine which therapies have been shown to effectively attenuate VC progression. This systematic review looks at evidence-based strategies to attenuate the progression of VC in people with CKD as reported in prospective clinical trials.

Methods

This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. The protocol is registered in the International Prospective Register of Systematic Reviews (PROSPERO 2020 CRD42021240400) (available at https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=240400).

Inclusion Criteria

This systematic review included randomized controlled trials (RCTs) and prospective nonrandomized clinical trials evaluating interventions to attenuate VC progression in the CKD population. Included studies (1) had participants with stage 3–5D CKD or kidney transplant recipients; (2) compared an intervention with placebo, another comparator, or standard of care; and (3) involved VC as an outcome measured using radiologic methods (computed tomography [CT] or plain x-ray). Interventions involved pharmacological agents, exercise regimes, and changes in dialysis prescription. Observational studies in patients with kidney failure evaluating VC related to changes in dialysis modality or kidney transplantation were excluded.

Preprints, review articles, retrospective studies, cross-sectional observational studies, protocol papers, and conference abstracts were excluded. Only prospective clinical trials in humans were reviewed, with exclusion of in vitro or experimental studies in animals. Surrogate measures of VC including serum calcification propensity (T₅₀) and calciprotein particles (despite being outlined in the registered protocol) were excluded as outcome measures, as well as other surrogate cardiovascular markers such as carotid intima media thickness and pulse wave velocity.

Search Strategy

MEDLINE and EMBASE (up to October 8, 2021) were searched with no restrictions on date of publication, number of trial participants, nor language. The complete search strategy is shown in Supplemental Table 1. Two authors (C.X. and N.D.T.) independently screened the title and abstracts of each study, discarding studies that clearly did not meet inclusion criteria. The full texts for all remaining studies were retrieved and assessed for eligibility for inclusion. Any disagreement about eligibility of a study was reviewed by a third author (I.R.).

Data Extraction and Quality Assessment

Two authors (C.X. and N.D.T.) assessed each trial, and extracted data on participant characteristics, interventions, comparators, and measurement of VC. Non-English language studies were translated before assessment. Where more than one publication of a study existed, reports were grouped together and the publication with the most complete data was used in the analysis.

Quality of evidence for intervention groups was assessed by the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach, with each being estimated as high, moderate, low, or very low level of evidence.¹⁶ Methodological quality of each RCT was also assessed using the risk of bias assessment tool developed by the Cochrane Bias Methods Group.¹⁷ The following items were assessed: (1) random sequence generation; (2) allocation concealment; (3) blinding of participants, investigators, and outcome assessors; (4) incomplete outcome data; (5) selective outcome reporting; and (6) any other bias (e.g., insufficient rationale and study design). Study selection, data extraction, risk of bias, and GRADE assessments were undertaken independently by two authors (C.X. and N.D.T.).

Results

Search Results

A PRISMA flow diagram of the search results is shown in Figure 1. A total of 10,646 potentially relevant records were identified through database searches. An additional six records were identified through citation searching. After duplicates were removed, 7932 records were screened by title and abstract and a remaining 126 records were identified for full text review. A total of 77 studies met the inclusion criteria and were included in this review.

Figure 1.

PRISMA flow diagram.

Study Characteristics

Tables 1–9 display the main participant, study characteristics, and outcome of included studies. Included studies involved a total of 6898 participants with CKD (median participants 50, range 7–508; median study duration 12 months, range 3–60 months). Of the 77 studies, 63 were RCTs and 14 were nonrandomized interventional trials. VC featured as a primary end point in 53 studies. Fifteen studies involved participants with nondialysis CKD, 58 involved participants on dialysis and 4 studies involved kidney transplant recipients. A significant proportion of studies included phosphate binder therapy as the intervention (n=25). Other interventions with multiple trials included calcimimetics, vitamin D therapy, β-hydroxy-β-methylglutaryl-CoA (HMG-CoA) reductase inhibitors, antiresorptive therapy, sodium thiosulfate, vitamin K, magnesium, and changes in dialysis prescription. There were single trials for each of exercise training, SNF472, sotatercept (a ligand trap for members of the TGF-β superfamily including bone morphogenetic proteins), oral activated charcoal, spironolactone, and nicotinamide.

Table 1.

Summary of phosphate binder study characteristics and VC outcome

Author and Year	Type of Study	Study Duration (months)	Participant	Intervention (Participants at Baseline)	Comparator (Participants at Baseline)	VC as Primary or Secondary End Point	VC Imaging	VC Site	VC Outcomea
Sperschneider 199318	Non-RCT	36	HD	3 months: calcium carbonate 4.5 g + aluminum hydroxide, n=9	Calcium carbonate 9 g, n=8	Secondary	X-ray	Hand	Linear increase in calcification with calcium carbonate
				6 months: calcium carbonate 9 g 24 months: calcium carbonate 4.5–9 g, n=22
Chertow 200219b (Treat to Goal)	RCT	12	HD	Sevelamer, n=99	Calcium-based phosphate binder (acetate or carbonate), n=101	Secondary	EB-CT	Coronary artery, thoracic aorta, mitral valve, aortic valve	Less progression at coronary arteries and aorta with sevelamer
Braun 200420cAsmus 200521d	RCT	12	HD	Sevelamer, n=55	Calcium carbonate, n=59	Secondary	EB-CT	Coronary artery, thoracic aorta, mitral valve, aortic valve	Less progression at all sites with sevelamer
		24
Block 200522	RCT	18	Incident HD	Sevelamer, n=73	Calcium-based phosphate binder (acetate or carbonate), n=75	Primary	EB-CT	Coronary artery	Less progression with sevelamer
Ferramosca 200523	RCT	12	HD	Sevelamer, n=54	Calcium acetate, n=54	Primary	EB-CT	Coronary artery	Less progression with sevelamer
Russo 200724	RCT	24	CKD stage 3–4	Sevelamer + low phosphate diet, n=30	Calcium carbonate + low phosphate diet, n=30 | Low phosphate diet, n=30	Primary	CT	Coronary artery	Less progression with sevelamer than calcium and diet
Barreto 200825 (BRiC)	RCT	12	HD	Sevelamer, n=52	Calcium acetate, n=49	Primary	CT	Coronary artery	No significant VC attenuation
Qunibi 200826 (CARE-2)	RCT	12	HD	Sevelamer, n=100 (+ atorvastatin, n=79)	Calcium acetate, n=103 (+ atorvastatin, n=100)	Primary	EB-CT	Coronary artery	No significant VC attenuation
Takei 200827	Non-RCT	6	HD	Sevelamer, n=22	Calcium carbonate, n=20	Primary	CT	Abdominal aorta	Less progression with sevelamer (lower serum phosphate with sevelamer)
Kakuta 201128	RCT	12	HD	Sevelamer, n=91	Calcium carbonate, n=92	Primary	CT	Coronary artery	Less progression with sevelamer
Toussaint 201129	RCT	18	HD	Lanthanum carbonate, n=22	Calcium carbonate, n=23	Primary: aorta, secondary: femoral	CT	Abdominal aorta, SFA	Less progression at aorta (significant) and SFA (not significant) with lanthanum
Block 201230 (PNT study)	RCT	9	CKD stage 3b–4	Sevelamer, n=30 | Lanthanum carbonate , n=30 | Calcium acetate, n=30	Placebo, n=58	Secondary	EB-CT	Coronary artery, thoracic aorta, abdominal aorta	No significant VC attenuation at all sites (lower serum phosphate with sevelamer)
Di Iorio 201231	RCT	24	CKD stage 3–4	Sevelamer, n=121	Calcium carbonate, n=118	Primary	CT	Coronary artery	Less progression with sevelamer (lower serum phosphate with sevelamer)
Kalil 201232	RCT	12	HD	Lanthanum carbonate, n=7	Remain on same non-lanthanum phosphate binder (calcium-based +/− sevelamer), n=6	Primary	CT	Coronary artery	Less progression with lanthanum
Lemos 201333	RCT	24	CKD stage 3–5	Sevelamer, n=38 | Rosuvastatin, n=38	Control condition undefined, n=41	Primary	CT	Coronary artery	No significant VC attenuation
Ohtake 201334	RCT	12	HD	6 months of calcium carbonate then 6 months of lanthanum carbonate, n=26	6 months of calcium carbonate then another 6 months of calcium carbonate, n=26	Primary	CT	Coronary artery	Less progression with lanthanum
Seifert 201335	RCT	12	CKD stage 3	Lanthanum carbonate, n=19	Placebo, n=19	Secondary	CT	Carotid artery, coronary artery, aortic arch, thoracic aorta	No significant VC attenuation at all sites
Wada 201436	RCT	12	HD and diabetes	Lanthanum carbonate, n=21	Calcium carbonate, n=22	Primary	CT	Abdominal aorta	Less progression with lanthanum, only when baseline VC is not advanced
Russo 201537	RCT	36	CKD stage 3–4	Sevelamer + phosphate restricted diet, n=66	Calcium carbonate + phosphate restricted diet, n=47	Primary	CT	Coronary artery	Less progression with sevelamer
Wang 201538	RCT	3	HD	Lanthanum carbonate, n=28	Standard of care, n=26	Primary	Lateral abdominal x-ray	Abdominal aorta	Less progression with lanthanum
Zhang 201739	RCT	12	HD	Lanthanum carbonate, n=46	Calcium carbonate, n=46	Primary	CT	Coronary artery	Less progression with lanthanum
Fujii 201840e	RCT	18	Incident HD	Lanthanum carbonate, n=53	Calcium carbonate, n=55	Primary	CT	Coronary artery	Less progression with lanthanum
Toussaint 202041 (IMPROVE-CKD)	RCT	22 (96 weeks)	CKD stage 3b–4	Lanthanum carbonate, n=138	Placebo, n=140	Secondary	CT	Abdominal aorta	No significant VC attenuation
Isaka 202142	RCT 2 × 2 factorial	12	HD	Factor 1: lanthanum carbonate, n=78 | sucroferric oxyhydroxide, n=78Factor 2: strict serum phosphate <4.5 mg/dl, or standard serum phosphate 5 to <6 md/dl		Primary	CT	Coronary artery	Less progression with strict phosphate control; no difference between binders

^aThere was no difference in serum phosphate levels between groups in the majority of studies (apart from those outlined in the table). Most studies had similar reduction in serum phosphate in both groups, and some had no significant reduction in serum phosphate.

^bRaggi 2004¹³⁹ was not included as this study was a post hoc analysis of Chertow 2002¹⁹ and assessed valvular calcification in this cohort.

^cBraun 2004²⁰ had 114 participants; 93 of these participants were in Chertow 2002¹⁹ study. The extra 21 participants in Braun are included in median and range calculations.

^dAsmus 2005²¹ followed patients in Braun 2004²⁰ for an additional 12 months. None of these participants are included in median and range calculations (as same as Braun 2004 participants).

^eWatanabe 2020¹⁴⁰ was not included as this study was a post hoc analysis of Fujii 2018⁴⁰ and assessed valvular calcification in this cohort.

EB, electron beam; HD, hemodialysis; SFA, superficial femoral artery.

Table 9.

Summary of other singleton therapy study characteristics and VC outcome

Author and Year	Type of Study	Study Duration (months)	Participant	Intervention (Participants at Baseline)	Comparator (Participants at Baseline)	VC as Primary or Secondary End Point	VC Imaging	VC Site	VC Outcome
Raggi 2020134 (CaLIPSO)	RCT	12 (52 weeks)	HD	SNF472 600 mg, n=91 | SNF472 300 mg, n=92	Placebo, n=91	Primary	CT	Coronary artery	Less progression with SNF472
Gueiros 2019135	RCT	12	PD	Spironolactone daily, n=16	Treatment as usual, n=17	Primary	CT	Coronary artery	No significant VC attenuation
Coyne 2019136	RCT	7 (225 days)	HD, anemia	Sotatercept 0.3 mg/kg, n=9 | 0.5 mg/kg, n=8 | 0.7 mg/kg, n=9 | 0.7/0.4 mg/kg, n=6	Placebo, n=11	Secondary	CT	Abdominal aorta	Less progression with sotatercept
Liu 2020137	RCT	12	HD	Nicotinamide + calcium-based phosphate binder, n=49	Placebo + calcium-based phosphate binder, n=49	Secondary	X-ray lateral	Abdominal aorta	No significant VC attenuation
Gao 2019138	RCT	24	CKD stage 3–4	Phase 2: oral activated charcoal, n=17	Lanthanum carbonate, n=17 Calcium carbonate, n=16	Primary during the second phase	CT	Coronary artery	VC progressed least with lanthanum followed by charcoal then calcium

HD, hemodialysis; PD, peritoneal dialysis.

Study Interventions

Phosphate Binders

Twenty-five trials (18 dialysis, 7 nondialysis) evaluated effects of phosphate binders on VC and involved 2571 participants (median participants 97, range 13–278; median duration 12 months, range 3–36 months) (Table 1).^18–42 VC was reported as the primary end point in 19 trials, with the coronary artery the most common site evaluated (n=15). There were only three placebo-controlled RCTs of phosphate binders, all in nondialysis-dependent CKD patients, and only one included a significant proportion of patients with hyperphosphatemia, where a modest but significant reduction in serum phosphate and urinary phosphate excretion was seen with active therapy compared with placebo.³⁰ None of the placebo-controlled trials reported any attenuation of progression of VC with noncalcium-based phosphate binders.

The majority of phosphate binder trials (n=17) evaluated noncalcium-based binders compared with calcium-based binders and although results were conflicting, of studies with >100 participants, five of seven RCTs evaluating sevelamer in hemodialysis cohorts reported less coronary artery calcification (CAC) progression with sevelamer. Overall, trials predominantly demonstrated that noncalcium-based phosphate binders (sevelamer and lanthanum carbonate) attenuated VC progression when compared with calcium-based binders (likely as a result of exogenous calcium loading promoting VC with calcium-based binders), but not when compared with placebo or standard of care.

Of trials involving dialysis participants, the largest two trials (≥200 participants each) both evaluated sevelamer compared with calcium-based binders and showed conflicting results. Qunibi et al. (n=203) reported no difference in VC between sevelamer and calcium-based binders (both arms were prescribed concurrent atorvastatin), whereas Chertow et al. (n=200) reported less progression of VC with sevelamer.^19,26 Of note, the study by Qunibi et al. had several limitations which included a greater preexisting cardiovascular risk in participants (higher frequencies of diabetes and smokers at baseline) and a high dropout rate.

Of the seven phosphate binder trials that enrolled nondialysis patients, four of these reported no attenuation of VC progression with noncalcium-based binders (three were placebo-controlled). In the largest two trials in nondialysis CKD (≥200 participants each), Di Iorio et al. (n=239) reported less progression of VC with sevelamer compared with calcium binders, but Toussaint et al. (n=278) showed no difference in VC with lanthanum in a placebo-controlled trial.^31,41

Changes in Dialysis Prescription

Eight trials involved alterations in dialysis prescription and involved 1304 participants (median participants 70, range 50–508; median duration 12 months, range 12–24 months) (Table 2).^43–50 Five trials evaluated effects of changes in dialysate calcium concentration^{44,46,47,49,50} with all but one reporting a lower serum calcium concentration in those randomized to lower dialysate calcium.⁴⁹ Two trials reported that lower dialysate calcium concentration (1.25 mmol/L) attenuated VC progression compared with higher dialysate calcium,^46,47 with the other three reporting no effect on VC.^44,49,50 Two of the larger and longer dialysate calcium studies reported conflicting findings on effects of higher dialysate calcium (1.75 mmol/L) on VC.^44,47 Other solitary studies that involved alterations to dialysis prescription reported attenuation of VC progression with interventions which included vitamin E–coated or high-flux hemodialysis membranes and interdialytic sodium bicarbonate,^43,45,48 but all with <100 participants in each study.

Table 2.

Summary of dialysis intervention study characteristics and VC outcome

Author and Year	Type of Study	Study Duration (months)	Participant	Intervention (Participants at Baseline)	Comparator (Participants at Baseline)	VC as Primary or Secondary End Point	VC Imaging	VC Site	VC Outcome
Mune 199943	RCT	24	HD	Vitamin E–coated cellulose membrane dialyser (CL-E), n=25	Unmodified cellulose membrane dialyser, n=25	Secondary	CT	Aorta	Less progression with intervention
Zhang 201444	Non-RCT	24	HD	High DCa 1.75 mmol/L, n=285	Standard DCa 1.5 mmol/L, n=223	Secondary	X-ray, echo	Aortic arch, abdominal aorta, cardiac valve	No VC attenuation. Only aortic arch is reported (increased serum Ca with DCa 1.75)
Fu 201545	RCT	12	HD	High-flux HD, n=25	Low flux HD, n=25	Primary	X-ray lateral	Abdominal aorta	Less progression with intervention
Lu 201646	RCT	12	HD	Low DCa 1.25 mmol/L, n=41	Standard DCa 1.5 mmol/L, n=41	Secondary	X-ray lateral	Abdominal aorta	Less progression with intervention (lower serum Ca with DCa 1.25)
Ok 201647	RCT	24	HD	Low DCa 1.25 mmol/L, n=212	High DCa 1.75 mmol/L, n=213	Primary	CT	Coronary	Less progression with intervention (lower serum Ca with DCa 1.25)
Voiculet 201648	RCT	12	HD	Interdialytic NaHCO3 5 g, n=29. (Reduce HCO3− fluctuations)	No interdialytic NaHCO3, higher intradialytic NaHCO3, n=34	Secondary	X-ray	Coronary	Less progression with intervention
Kim 201749	RCT	12	HD	Low DCa 1.25 mmol/L, n=36	Standard DCa 1.5 mmol/L, n=40	Primary	CT	Coronary	No significant VC attenuation (no difference in serum Ca between groups)
Masterson 201750	RCT	12	Nocturnal HD	Low DCa 1.3 mmol/L, n=24	High DCa 1.6 or 1.75 mmol/L, n=26	Secondary	CT	Abdominal aorta, superficial femoral artery	No significant VC attenuation at all sites (lower serum Ca in DCa 1.3)

DCa, dialysate calcium; HD, hemodialysis.

Calcimimetics

Six trials involved calcimimetics with 562 participants (median participants 41, range 23–360; median duration 12 months, range 8.3–36 months) (Table 3).^51–56 All trials assessed the effect of the calcimimetic cinacalcet on VC and predominantly demonstrated reduction in VC progression; however, studies were small and some were potentially confounded by disparate vitamin D administration between intervention and comparator groups. The largest study, conducted by Raggi et al. (n=360), reported a nonsignificant reduction in progression of CAC determined by the Agatston score with cinacalcet plus low-dose vitamin D analogs compared with vitamin D analogs alone,⁵² but reported a significant attenuation when CAC was measured as a volume score or when aortic valve calcification was assessed. Flexible use of vitamin D in the control group in this study resulted in controls receiving substantially higher doses, making it difficult to isolate effects of cinacalcet on VC. One RCT compared cinacalcet to parathyroidectomy in kidney transplant recipients reporting no difference in VC after 12 months.⁵⁴

Table 3.

Summary of calcimimetic study characteristics and VC outcome

Author and Year	Type of Study	Study Duration (months)	Participant	Intervention (Participants at Baseline)	Comparator (Participants at Baseline)	VC as Primary or Secondary End Point	VC Imaging	VC Site	VC Outcome
Tsurata 200851a	Non-RCT	7–13	HD, SHPT	Cinacalcet, n=8	Control condition undefined, n=60	Primary	CT	Coronary artery	Less progression with cinacalcet
Raggi 201152 (ADVANCE)	RCT	12	HD, SHPT	Cinacalcet + low-dose vitamin D, n=180	Flexible doses of vitamin D, n=180	Primary: % change in Agatston CAC score, secondary: % & absolute change for other sites	CT	Coronary arteries, thoracic aorta, aortic valve, mitral valve	Less progression at aortic valve with cinacalcet. Primary outcome no significant difference
Nakayama 201453	Non-RCT	36	HD, SHPT	Cinacalcet, n=23	No control group	Primary	CT	Abdominal aorta	Less progression with cinacalcet
Cruzado 201654	RCT	12	Kidney transplant recipients	Cinacalcet, n=15	Subtotal parathyroidectomy, n=15	Secondary	CT	Thoracic/abdominal aorta; iliac arteries	No significant VC attenuation at all sites
Susantitaphong 201955	Non-RCT	8.3 (36 weeks)	HD, SHPT	Phase 2: 12 weeks of cinacalcet, n=45	Phase 2: no control group	Secondary	X-ray lateral	Not reported	Less progression with cinacalcet
Eddington 202156	RCT	12	HD, SHPT	Cinacalcet, n=15	Standard treatment, n=21	Primary	CT	Coronary artery, abdominal aorta	No significant difference at both sites

^aTsurata 2008⁵¹ was included in the study participants median and range, but excluded from the study duration median and range calculations due to the duration provided being a range from 7 to 13.

HD, hemodialysis.

Vitamin D Therapy

Six trials involved vitamin D therapy with 289 participants (median participants 44, range 36–76; median duration 15 months, range 6–60 months) (Table 4).^57–62 Half of these studies involved nutritional vitamin D (cholecalciferol) and half evaluated effects on VC of active vitamin D analogs (calcitriol and paricalcitol). All studies showed no difference in VC between groups, regardless of whether the vitamin D intervention was compared with placebo, standard of care, or another vitamin D comparator.

Table 4.

Summary of vitamin D study characteristics and VC outcome

Author and Year	Type of Study	Study Duration (months)	Participant	Intervention (Participants at Baseline)	Comparator (Participants at Baseline)	VC as Primary or Secondary End Point	VC Imaging	VC Site	VC Outcome
Baker 198657	RCT	60	HD	Calcitriol, n=38	Placebo, n=38	Secondary	X-ray	Hands, feet, pelvis, coronary artery	No significant VC attenuation at all sites
Morosetti 200858	Non-RCT	24	HD	High-dose calcitriol, n=18	Low-dose calcitriol + sevelamer, n=18	Primary	CT	Coronary artery	No significant VC attenuation
Delanaye 201359	RCT	12	HD, serum 25(OH)D <30 ng/ml	Cholecalciferol, n=22	Placebo, n=21	Secondary	X-ray lateral	Abdominal aorta	No significant VC attenuation
Kidir 201560	Non-RCT	6	HD or PD	Cholecalciferol, n=26	Standard treatment, n=17	Primary	X-ray hand & pelvis	Radial arteries, digital arteries, iliac arteries, femoral arteries	No significant VC attenuation at all sites
Samaan 201961	RCT	18	CKD stage 3–4, serum 25(OH)D <30 ng/ml	Cholecalciferol, n=23	Placebo, n=24	Primary	CT	Coronary artery	No significant VC attenuation
Anis 202062	RCT	11 (48 weeks)	CKD stage 3–4 with SHPT	Calcitriol, n=22	Paricalcitol, n=22	Primary	CT	Coronary artery, aortic valve, mitral valve	No significant VC attenuation at all sites

HD, hemodialysis; PD, peritoneal dialysis.

HMG-CoA Reductase Therapy

Two trials involved HMG-CoA reductase therapy with 199 participants and involved different CKD populations (nondialysis CKD and kidney transplant recipients) but with similar size and measurement of VC (Table 5).^33,63 Both studies reported that statins did not reduce progression of VC despite benefits on dyslipidemia.

Table 5.

Summary of exercise and statin study characteristics and VC outcome

Author and Year	Type of Study	Study Duration (months)	Participant	Intervention (Participants at Baseline)	Comparator (Participants at Baseline)	VC as Primary or Secondary End Point	VC Imaging	VC Site	VC Outcome
Zhou 2020132 (RENEXC)	RCT	12	CKD stage 3–5	Exercise: endurance training plus: Balance train, n=75	Exercise: endurance training plus: Strength train, n=76	Secondary	X-ray lateral	Abdominal aorta	No significant VC attenuation
Lemos 201333	RCT	24	CKD stage 3–5	Sevelamer, n=38 | Rosuvastatin, n=38	Control condition undefined, n=41	Primary	CT	Coronary artery	No significant VC attenuation
Yazbek 201663	RCT	12	Kidney transplant recipients	Rosuvastatin, n=45 | Atorvastatin, n=16 (due to supply issue)	Control condition undefined, n=59	Primary	CT	Coronary artery	No significant VC attenuation

Antiresorptive Agents

Nine trials involved antiresorptive agents with 346 participants (median participants 31, range 12–101; median duration 12 months, range 6–23 months) (Table 6).^64–72 Of these, seven reported reduced progression of VC, and one study which compared two antiresorptive agents (alendronate versus denosumab) did not show reduced progression of VC.⁷² Of seven trials involving bisphosphonates, four studies examined etidronate in hemodialysis participants and all reported attenuation of VC progression.^64–67 The other three bisphosphonate studies had different study designs, populations, and interventions, with conflicting findings.^68–70 Of two trials involving denosumab, one nonrandomized trial reported attenuation of VC compared with standard of care, and the other, an RCT, showed no difference in VC when compared with alendronate.^71,72

Table 6.

Summary of antiresorptive study characteristics and VC outcome

Author and Year	Type of Study	Study Duration (months)	Participant	Intervention (Participants at Baseline)	Comparator (Participants at Baseline)	VC as Primary or Secondary End Point	VC Imaging	VC Site	VC Outcome
Hashiba 200464a	RCT	12	HD	Etidronate for 6 months, n=8	Control group, n=10	Primary	CT	Abdominal aorta	Less progression with etidronate
Nitta 200465	Non-RCT	Approximately 12	HD	Etidronate, n=35	Control group, n=21	Primary	CT	Coronary artery	Less progression with etidronate
Ariyoshi 200666	RCT	12	HD	Etidronate, n=8	No intervention, n=6	Primary	CT	Coronary artery, thoracic and abdominal aorta	Less progression with etidronate at coronary (not significant) and aorta (significant)
Hashiba 200667	RCT	23	HD	Etidronate for 23 months, n=12	No intervention, n=12	Primary	CT	Aorta	Less progression with etidronate
Torregrosa 201068	RCT	12	Kidney transplant recipients	Risedronate, vitamin D, and calcium, n=52	Vitamin D and calcium, n=49	Secondary	X-ray	Iliac arteries, femoral arteries, radial arteries, digital arteries	Less progression with risedronate. No detail on each site
Toussaint 201069	RCT	18	CKD stage 3–4	Alendronate, n=25	Placebo, n=26	Primary: aortic VC, secondary: femoral VC	CT	Abdominal aorta, superficial femoral artery	No significant VC attenuation at both sites
Okamoto 201470	RCT	24	Kidney transplant recipients	Alendronate, n=5	No intervention, n=7	Secondary	CT	Abdominal aorta	Less progression with alendronate
Iseri 201971	RCT	12	HD	Denosumab sc, n=24 (+ calcitriol & calcium for first 2 weeks)	Alendronate iv, n=24 (+ calcitriol & calcium for first 2 weeks)	Secondary	CT	Coronary artery	No significant VC attenuation
Chen 202072	Non-RCT	6	HD, SHPT	Denosumab, n=21	Standard of care, n=21	Primary	CT	Coronary artery	Less progression with denosumab

^aAnother study, by Tamura et al. 2002,¹³³ is thought to have the same participants as Hashiba 2004.⁶⁴.

HD, hemodialysis; sc, subcutaneously.

Magnesium

Three trials involved magnesium with 202 participants and all reported reduction in progression of VC (Table 7).^73–75 These trials, however, were limited by small sample size, and two of the studies, both involving dialysis patients, were not completed. One trial by Sakaguchi et al. was prematurely terminated based on significant efficacy of magnesium on CAC at interim analysis, and one study by Spiegel et al. was discontinued due to concerns about rapid VC progression in the calcium binder control arm.^73,75 Of note, these two studies were conducted using magnesium as a phosphate binder (one with a calcium binder comparator), rather than systemic magnesium supplementation per se. Two studies reporting less progression of VC demonstrated an increase in serum magnesium concentration with magnesium therapy.^74,75

Table 7.

Summary of magnesium and vitamin K study characteristics and VC outcome

Author and Year	Type of Study	Study Duration (months)	Participant	Intervention (Participants at Baseline)	Comparator (Participants at Baseline)	VC as Primary or Secondary End Point	VC Imaging	VC Site	VC Outcome
Spiegel 200973	Non-RCT	18	HD	Mg/CaCO3, n=7	No control group	Primary	EB-CT	Coronary artery	Less progression with magnesium (P=0.07)
Tzanakis 201474	RCT	12	HD	MgCO3 + calcium acetate, n=36	Calcium acetate, n=36	Primary	X-ray	Hands, abdomen, pelvis, femur	Less progression with magnesium. No detail on each site (increased serum Mg with MgCO3)
Sakaguchi 201975 Prematurely terminated due to significant efficacy of MgO	RCT 2 × 2 factorial	24	CKD stage 3–4	MgO + AST-120, n=35 MgO + control, n=28	Control + AST-120, n=35 Control + control, n=25	Primary	CT	Coronary artery, thoracic aorta	MgO but not AST-120 appeared to slow VC progression in coronary artery but not thoracic aorta (increased serum Mg with MgO)
Kurnatowska 201576	RCT	9 (270 days)	CKD stage 3–5	Vitamin K2 + cholecalciferol, n=29	Cholecalciferol 10 μg, n=13	Primary	CT	Coronary artery	No significant VC attenuation
Li 201777	RCT	3	HD	Vitamin K dialysis buffer, n=50	No vitamin K dialysis buffer, n=50	Primary	CT	Coronary artery	Less progression with vitamin K dialysate
Oikonomaki 201978	RCT	12	HD	Vitamin K2 daily, n=44	No vitamin K2, n=58	Primary	CT	Abdominal aorta	No significant VC attenuation
De Vriese 202079 (Valkyrie)	RCT	18	HD with atrial fibrillation	Rivaroxaban + vitamin K2 3×/week postdialysis, n=42 | Rivaroxaban daily alone, n=46	VKA continued or initiated, n=44	Primary	CT	Coronary artery, thoracic aorta, cardiac valve	No significant VC attenuation at all sites
Witham 202080 (K4Kidneys)	RCT	12	CKD stage 3b–4	Vitamin K2 daily, n=80	Placebo, n=79	Secondary	X-ray lateral	Abdominal aorta	No significant VC attenuation
Levy-Schousboe 202181 (RenaKvit)	RCT	24	HD	Vitamin K2, n=24	Placebo, n=24	Secondary	CT for heart, x-ray for aorta	Coronary, aortic valve, mitral valve, abdominal aorta	No significant VC attenuation at all sites

HD, hemodialysis; VKA, vitamin K antagonist.

Vitamin K

Six trials involved vitamin K therapy with 583 participants (median participants 101, range 42–159; median duration 12 months, range 3–24 months) (Table 7).^76–81 One study reported a benefit on CAC of vitamin K in dialysate over a 3-month period in an RCT of 100 hemodialysis participants.⁷⁷ The remaining five studies reported no benefit on attenuation of VC progression with oral vitamin K2 therapy.

Sodium Thiosulfate

Seven trials involved sodium thiosulfate with 251 participants (median participants 32, range 12–60; median duration 6 months, range 3–6 months) (Table 8).^82–88 All trials involved participants on hemodialysis and all reported reduction in progression of VC with sodium thiosulfate. The largest study by Djuric et al. randomized 60 participants to either sodium thiosulfate or saline over 6 months⁸⁷ and reported attenuation of VC progression in iliac arteries and cardiac valves, but not the abdominal aorta which was the prespecified primary outcome.

Table 8.

Summary of sodium thiosulfate study characteristics and VC outcome

Author and Year	Type of Study	Study Duration (months)	CKD Status	Intervention (Participants at Baseline)	Comparator (Participants at Baseline)	VC as Primary or Secondary End Point	VC Imaging	VC Site	VC Outcome
Adirekkiat 201082	Non-RCT	4	HD	STS post-HD twice weekly, n=16	Dialysis as usual, n=16	Primary	CT	Coronary artery	Less progression with STS
Mathews 201183	Non-RCT	5	HD	STS post-HD three times per week, n=22	No control group	Primary	CT	Carotid artery, coronary artery, thoracic arch	Less progression with STS
Yonova 201484	Non-RCT	6	HD	STS intradialytic, n=6	Controls, n=6	Primary	CT	Coronary artery	Less progression with STS
Yu 201685	RCT	3	HD	STS post-HD + calcium carbonate, n=15	Calcium carbonate, n=10	Primary	CT	Coronary artery	Less progression with STS
Saengpanit 201886	RCT	6	HD	STS intradialytic twice weekly, n=24	Standard of care, n=26	Secondary	CT	Coronary artery	Less progression with STS
Djuric 202087	RCT	6	HD	STS post-HD, n=30	0.9% NaCl during dialysis, n=30	Primary: abdominal aorta, secondary: iliac	CT	Abdominal aorta, iliac arteries, cardiac valves	Less progression with STS at iliac arteries
Bian 202188	RCT	6	HD	STS, n=25	Usual treatment, n=25	Primary	CT	Coronary artery	Less progression with STS

HD, hemodialysis; STS, sodium thiosulfate.

Risk of Bias and GRADE Assessment

Supplemental Table 2 summarizes the risk of bias assessment for RCTs. Random sequence generation and allocation concealment were reported with low risks of bias in 45% and 47% of trials, respectively. Blinding of participants and investigators to the allocated intervention and blinding to outcomes were reported with low risk of bias in 22% and 77% of trials, respectively.

Table 10 outlines the GRADE assessment for quality of evidence of studies for each intervention. The majority of analyses were based on low certainty of evidence, primarily due to risk of bias, inconsistency of results, imprecision of reporting, and publication bias. Inconsistency grading was based on heterogeneity of study design and study outcome. Imprecision grading was based on small event numbers.

Table 10.

Summary of GRADE findings

Intervention	Risk of Bias	Consistency	Imprecision	Indirectness	Publication Bias (No. of Studies)	Quality of Evidence (GRADE)	Comments
Phosphate binders	Moderate	Not all RCTs, inconsistent results	None	CKD	25	⊕⊕OO Due to risk of bias and inconsistency	Many studies did not blind participants and investigators
Alterations to dialysis prescription	Moderate	Not all RCTs, inconsistent results	None, x-ray	CKD	8	⊕OOO Due to risk of bias, inconsistency, and imprecision	Half used x-ray to assess VC
Calcimimetics	High	Not all RCTs, consistent results	None	CKD	6	⊕⊕OO Due to risk of bias and inconsistency	Major study was industry sponsored. Half were non-RCTs
Vitamin D	Moderate	Not all RCTs, consistent results	Few participants, x-ray	CKD	6	⊕OOO Due to risk of bias, inconsistency & imprecision	2 out of 6 were non-RCTs. Half used x-ray to assess VC
Exercise	Moderate	RCT	Few participants, x-ray	CKD	1	⊕OOO Due to risk of bias, imprecision, and publication bias	Only 1 study
Statins	Moderate	All RCTs, consistent results	Few participants	CKD	2	⊕OOO Due to risk of bias, imprecision, and publication bias
Antiresorptive agents	Moderate	Not all RCTs, inconsistent results	Few participants	CKD	9	⊕ΟΟΟ Due to risk of bias, inconsistency, and imprecision	4 out of 9 studies were missing information on randomization method and blinding
Magnesium	Moderate	Not all RCTs, consistent results	Few participants, x-ray	CKD	3	⊕ΟΟΟ Due to risk of bias, inconsistency, imprecision, and publication bias
Vitamin K	Moderate	All RCTs	None	CKD	6	⊕⊕⊕O Due to risk of bias
Sodium thiosulfate	Moderate	Not all RCTs, consistent results	Few participants	CKD	7	⊕ΟΟΟ Due to risk of bias, inconsistency, imprecision	3 out of 7 were non-RCTs
SNF472	High	RCT	Few participants	CKD	1	⊕OOO Due to imprecision and publication bias	Only 1 study, which was industry sponsored
Spironolactone	Low	RCT	Few participants	CKD	1	⊕⊕OO Due to imprecision and publication bias	Only 1 study
Sotatercept	High	RCT	Few participants	CKD	1	⊕ΟΟΟ Due to risk of bias, imprecision, and publication bias	Only 1 study
Nicotinamide	Low	RCT	Few participants, x-ray	CKD	1	⊕⊕OO Due to imprecision and publication bias	Only 1 study
Oral activated charcoal	Moderate	RCT	Few participants	CKD	1	⊕ΟΟΟ Due to risk of bias, imprecision, and publication bias	Only 1 study

All interventions were compared with placebo, control, or another comparator, with VC as the outcome. High quality ⊕⊕⊕⊕; moderate quality ⊕⊕⊕O; low quality ⊕⊕OO; very low quality ⊕ΟΟΟ.

Discussion

We present the first systematic review of prospective clinical trials evaluating effects of all reported interventions on VC in people with CKD. Previous systematic reviews published have only appraised the effect of single interventions on VC in the CKD population.^89–92 Our review identified a large number of heterogeneous prospective clinical trials (mostly RCTs) of numerous interventions studied to evaluate their potential effect on VC in people with CKD. Beneficial effects were more consistently seen with magnesium, sodium thiosulfate, and etidronate in hemodialysis participants. Overall, however, trials were of small sample size (median 50 participants) and generally short in study duration (median 12 months).

Heterogeneity in clinical trials was seen with participant characteristics including CKD stage, with most studies involving patients on dialysis. In the nondialysis CKD population, only four studies included in this systematic review reported attenuation of VC progression. Furthermore, the effect on VC of certain interventions has not been evaluated in clinical trials involving specific CKD cohorts, such as HMG-CoA reductase inhibitors and sodium thiosulfate on dialysis patients and nondialysis patients, respectively. Further studies are required to specifically target the nondialysis CKD population, as VC response to therapy may be different in this population compared with patients on dialysis. For example, the safety and efficacy of phosphate binders in nondialysis CKD has been recently called into question,⁹³ in contrast to the potential benefit of noncalcium-based compared with calcium-based phosphate binders in dialysis patients.⁹⁴

Heterogeneity was also seen in trials with regard to baseline VC, study design (randomized and nonrandomized), study duration, VC imaging modality, and VC anatomic site. No major differences were seen in our systematic review of randomized and nonrandomized studies on the effect of various interventions on VC, with similar proportions reporting no change. Five studies in our review were <6 months in duration (three sodium thiosulfate, one phosphate binder, and one vitamin K), and all demonstrated attenuation of VC progression. This suggests changes in VC can be demonstrated over a short duration and studies may not require long-term follow-up to determine the effects of interventions on VC in CKD, unlike hard cardiovascular end points.

A major limitation of the clinical trials included in this systematic review is that they all rely on interventions that not only lack specificity for VC but also undoubtedly target both intimal and medial VC, possibly obscuring a potential effect of these therapies and contributing to the inconsistent and inconclusive findings. This is an important issue as these two types of VC are pathobiologically distinct and disproportionately affected by CKD, with the burden of medial VC cumulatively much greater with declining kidney function. The consequences of targeting intimal and medial VC may also be divergent, because there is evidence that accelerated calcification of atheromatous lesions, as seen with high-intensity statin therapy in non-CKD populations,^95,96 might be beneficial due to stabilization of the plaque and reduced risk of rupture. Of note, 48 of the 77 trials reviewed here assessed VC of the coronary arteries, which even in patients on dialysis remains predominantly intimal in nature.^97,98 This issue is further compounded by the fact that conventional imaging techniques lack sufficient spatial resolution to reliably distinguish medial and intimal VC, and have insufficient sensitivity to detect smaller calcifications that are thought to be indicative of active mineral deposition.⁹⁹

The most common intervention evaluated with respect to VC attenuation was oral phosphate binder therapy (34% of studies reviewed), with trials predominantly exploring the potential superiority of noncalcium-based compared with calcium-based binders. It was also notable that a considerable number of these studies involved pharmaceutical industry sponsorship. Our results are consistent with a previous systematic review on phosphate binders that also reported that reliable conclusions cannot be drawn on mortality and cardiovascular end points.⁹⁰ Recently, the efficacy and safety of phosphate binders has been questioned by the updated KDIGO CKD-MBD guidelines,¹⁵ highlighting a lack of trial data for phosphate binder use especially in the nondialysis CKD population. Despite their widespread use for hyperphosphatemia in CKD, the effect of phosphate binders and phosphate lowering per se on VC, cardiovascular risk, or indeed any patient-centered outcome remains uncertain. To be informative, future trials of phosphate binders should only test effects of active therapy compared with no treatment or placebo.

Various alterations in dialysis prescription have been studied to assess the effect on VC, but with contradictory outcomes or insufficient evidence. Some studies suggest that low dialysate calcium concentrations may attenuate VC progression compared with high dialysate calcium concentrations,^46,47 but others report no difference between different dialysate calcium concentrations on VC.^49,50 Potential adverse effects of higher dialysate calcium on VC may result from calcium loading, similar to data assessing calcium-based phosphate binders with excess exogenous calcium. Two of the larger and longer dialysate calcium studies concluded with conflicting findings on the effects of high dialysate calcium,^44,47 and therefore there is no consistent evidence for attenuation of VC progression with low dialysate calcium. These conflicting results may relate to the fact that serum calcium levels in general are likely maintained too high in patients on dialysis, with correction of hypocalcemia perhaps leading to promotion of VC.¹⁰⁰ There is also minimal evidence for any other dialysis-specific interventions, with a paucity of trials evaluating effects on VC of other changes in dialysis prescription.

Clinical trials involving cinacalcet in dialysis patients consistently demonstrated potential reduction in progression of VC; however, studies were small and were potentially confounded by concomitant vitamin D administration. The other confounder with trials involving cinacalcet is that serum calcium levels were not allowed to decrease considerably and therefore interventions to maintain serum calcium in these trials may have limited the ability to see a more definitive benefit of calcimimetics on reducing VC progression.

Trials of vitamin D treatment alone consistently found no benefit on mitigation of VC. Currently, it is recommended that vitamin D therapy be prescribed in people with CKD due to reduced activation of vitamin D precursors by the kidneys, leading to hypocalcemia, SHPT, and bone resorption.¹⁰¹ Studies examining the effect of various forms of vitamin D treatment on VC, across the stages of CKD, do not indicate benefit for reducing progression of VC. Interestingly, even trials involving CKD participants with documented vitamin D deficiency did not demonstrate any beneficial effect of vitamin D supplementation on VC.^59–61 Studies have demonstrated that calcitriol and other vitamin D receptor activators in nondialysis CKD participants may increase the risk of hypercalcemia, without cardiovascular benefit.^102,103 The updated KDIGO CKD-MBD guidelines have consequently recommended reserving the use of calcitriol and other vitamin D receptor analogs for people with CKD who have severe and progressive SHPT.¹⁵

Lipid-lowering and other potential antiatherosclerotic effects of HMG-CoA reductase inhibitors appear to be insufficient in addressing the multifactorial pathogenesis of VC in CKD. Two studies in our systematic review evaluated the effect of statins in different CKD populations but without any benefit with respect to VC.^33,63 These findings were consistent with a recent meta-analysis of statin studies in the non-CKD population, proposing that a multitarget drug approach is required to target both intimal and medial pathways for VC development.¹⁰⁴ Interestingly, in secondary prevention studies of predominantly non-CKD populations, CAC has been shown to increase in those receiving high-intensity statin therapy despite improvement in clinical outcomes.^95,105

Although the majority of trials involving antiresorptive agents in this systematic review reported possible attenuation of VC progression, use of these agents in people with CKD remains contentious. The effect of bisphosphonates on attenuation of VC in CKD is still unclear despite several studies in hemodialysis patients suggesting benefits on VC. Although the initial hypothesis underlying these studies was that increased bone resorption may lead to increased calcium deposition in the vasculature, the potential beneficial mechanism of action of bisphosphonates is that they are nonhydrolyzable analogs of pyrophosphate, which is a potent endogenous and direct inhibitor of hydroxyapatite formation.¹⁰⁶ This is an important consideration when interpreting clinical trials of bisphosphonates, as etidronate, the first bisphosphonate, is a very good analog of pyrophosphate (which also may explain its side effect of osteomalacia) and studies of etidronate show consistent reduction in progression of VC. Subsequent generations of bisphosphonates were developed with more potency in inhibiting osteoclasts and are used at much lower concentrations, thus resulting in less pyrophosphate mimicry. Studies of bisphosphonates were generally small and of short duration, and there are potential concerns about bisphosphonate accumulation in dialysis patients due to its renal excretion and an increased risk of low bone turnover.¹⁰⁶ Denosumab is not renally excreted although numerous studies have reported an increased risk of hypocalcemia,⁷¹ especially during the first 2 weeks of treatment in CKD cohorts,¹⁰⁷ and evidence for a true benefit on attenuation of VC is lacking.

Magnesium supplementation has been proposed to exert beneficial effects on multiple phases of the calcification pathway, including downregulation of promoters of vascular smooth muscle cell osteogenic differentiation (Wnt/β-catenin, osteocalcin), upregulation of calcification inhibitors (osteopontin, MGP, BMP-7), and restoration of activity of TRPM7 channels.¹⁰⁸ Magnesium is also a direct inhibitor of hydroxyapatite crystal formation and this may be the most likely mechanism of action to reduce VC.¹⁰⁹ Magnesium has also been used as an intestinal phosphate binder, mentioned in two studies in our systematic review. Both RCTs involving treatment with magnesium reported reduction in progression of VC.^74,75 Interestingly, Sakaguchi et al. also reported that progression of CAC, but not thoracic aorta VC, was reduced (greater reduction in those with higher CAC scores), prompting speculation that different pathophysiology between these sites requires a different therapeutic target.⁷⁵ Although there is still a paucity of evidence for attenuation of VC progression by magnesium, results from these studies suggest magnesium may be a promising intervention that is relatively cheap and widely available. Concerns about magnesium supplementation contributing to hypermagnesemia in people with advanced CKD due to its renal excretion have been addressed by small clinical trials.¹¹⁰ Data about the possible risk of reduced bone mineral density with magnesium supplementation will be provided from ongoing trials.¹¹¹

Despite the adverse effect of warfarin, a vitamin K antagonist, promoting VC and experimental studies reporting benefits of vitamin K on attenuating VC, clinical trials involving vitamin K (in the form of vitamin K2) predominantly reported no effect on VC progression. Vitamin K is an essential cofactor for the carboxylase enzyme that activates uncarboxylated MGP (uc-MGP) into carboxylated MGP, one of the most potent systemic VC inhibitors, and a decreased level of dephosphorylated uc-MGP (dp-uc-MGP) has been proposed as a marker of vascular vitamin K status.^112,113 Patients with advanced CKD are often vitamin K deficient, in part due to dietary restrictions but also due to possible impaired endogenous recycling of vitamin K. Only one of six trials evaluating effects of vitamin K on VC in CKD reported reduction in progression of VC. Interestingly, VC change in this study occurred after only 3 months of a vitamin K dialysis buffer, and dp-uc-MGP was not measured. The other five trials reported no effect on VC, but did show reduction in dp-uc-MGP to assess the adequacy of oral supplementation.^76,78–81 Notably, the Valkyrie trial by De Vriese et al. reported dp-uc-MGP reduction was not normalized,⁷⁹ with the median level remaining elevated compared with the general population. The authors speculated that uremia might inactivate the carboxylase enzyme, such that supplementation with vitamin K2 does not enable MGP levels to normalize and does not inhibit VC due to a functional deficiency. Thus, it is unclear whether vitamin K2 supplementation can restore MGP levels let alone inhibit VC in CKD. Unless such a functional deficiency can be reversed therapeutically, the relatively high quality of evidence of null efficacy of vitamin K on VC compared with other therapies suggests that there is little virtue in conducting further RCTs of vitamin K in this patient population. In contrast, in the non-CKD population there is evidence that vitamin K1 does slow CAC progression.¹¹⁴

Intravenous sodium thiosulfate may be a promising intervention to reduce progression of VC, although its exact mechanism of action remains uncertain. Sodium thiosulfate was initially thought to reduce VC by forming a soluble complex with calcium from tissue deposits, with more recent evidence pointing toward thiosulfate acting as an antioxidant H₂S-donor in vivo; however, these hypotheses were disproved as major effects.^{11,84,115–117} Seven clinical trials demonstrated reduction in VC progression. Notably, all were in participants on hemodialysis and all were 6 months or less in duration.^82–88 Studies have shown that conventional CVD risk factors are more strongly associated with abdominal aortic VC, whereas CKD-related calcium and phosphate disturbances are more strongly associated with CAC.^118,119 This is debatable, however, as CAC is generally considered intimal in nature and in the CKD population it may be that CAC lesions are more calcified (i.e., of greater severity). Sodium thiosulfate may have a greater effect on smaller peripheral arteries compared with larger vessels but further RCTs evaluating effects of sodium thiosulfate on VC should be longer, adequately powered, and ideally placebo-controlled.

Multiple interventions included in our systematic review only had a single clinical trial assessing effects on VC, which was insufficient to draw conclusions. Studies of SNF472, sotatercept, and oral activated charcoal each reported attenuation of VC progression, whereas studies of exercise training, spironolactone, and nicotinamide did not. We did not include in our review other studies which reported effects on VC after changes in kidney replacement therapies, such as several observational studies assessing the effect of kidney transplantation on VC^120–125 and several evaluating changes in VC after commencement of more frequent or longer hours of hemodialysis (such as nocturnal dialysis).^123,126,127 These studies have predominantly reported that VC progression is attenuated by both kidney transplantation and nocturnal hemodialysis when compared with continuation of conventional hemodialysis, but not all studies have consistently shown this.^119,124,125

The strengths of this review include a systematic search of medical databases, with data extraction, analysis, and quality assessment involving two independent reviewers. Inclusion of both randomized and nonrandomized clinical trials across all stages of CKD was comprehensive but heterogeneity in participant populations and methods of quantifying VC prevented meta-analysis. Our systematic review was limited by studies of small sample size. Further, many studies were not adequately designed with sufficient sample size and statistical power to determine differences in VC as this outcome measure was a secondary end point in some trials. VC is also only a surrogate measure of CVD and it is unclear what magnitude of changes in VC (if any) may translate into a clinically meaningful effect on cardiovascular health. Indeed, if it transpires that VC is a secondary phenomenon to antecedent inflammatory processes as suggested by numerous preclinical studies,¹²⁸ then targeting VC per se is unlikely to produce any clinical improvement. Moreover, with the exception of SNF472 (and perhaps etidronate and magnesium) whose effects are specific to mineral growth inhibition, the broad physiologic effects of the other interventions reviewed here imply that, even if effective, these therapies could not provide definitive proof that VC has a causal role in CVD outcomes in patients with CKD.

This systematic review found a large number of heterogeneous clinical trials for all interventions that have been shown to effect VC progression in people with CKD, many with conflicting results and providing a low quality of evidence (Table 11). The studies highlight that there are multiple therapeutic targets to address the complex and multifactorial VC pathogenesis in CKD, and further trials need to be more adequately powered to determine whether any intervention is beneficial to attenuate VC progression in this population. Choice of an effective intervention may need to be better tailored to baseline patient characteristics, such as presence of conventional cardiovascular risk factors, hyperphosphatemia, vitamin D deficiency, bone mineral density, and uremia. Appropriate selection of patients on the basis of more objective functional assessments of systemic calcification propensity such as the T₅₀ test,^129,130 could be considered in future trial protocols. Translation of promising research imaging tools with improved spatial resolution and sensitivity to better distinguish medial and intimal VC and detect earlier microcalcifications (e.g., fused positron emission tomography-CT using the hydroxyapatite-seeking positron-emitting isotope ¹⁸F-fluoride) may also provide more insight into the efficacy of existing and emerging drug candidates.

Table 11.

Summary table of evidence-based interventions reported to attenuate VC progression in CKD

Probably Reduce VC Progression	Possibly Reduce VC Progressiona	Unlikely To Reduce VC Progression
Magnesium	Antiresorptive therapy	Vitamin D therapy
Sodium thiosulfate	Calcimimetics	Vitamin K2
Kidney transplantation	SNF472	HMG-CoA reductase inhibitors
Increased hours hemodialysis (nocturnal dialysis)	Low dialysate calcium concentration	Exercise
Noncalcium-based phosphate binders (when compared with calcium-based binders)	Strict control of phosphate balance	Phosphate binders (compared with placebo/standard of care)
	Sotatercept	Spironolactone
	Oral activated charcoal	Nicotinamide

^aThere is a paucity of evidence or conflicting results.

Based on this systematic review, the most effective evidence-based strategies to reduce progression of VC in CKD appear to be treatment with magnesium, sodium thiosulfate, and etidronate, and reducing exogenous calcium loading, but these findings require confirmation in larger cohorts. Future clinical trials of single interventions to slow VC progression should prioritize inhibitors of calcification (such as sodium thiosulfate, magnesium, SNF472, or etidronate), or administered in conjunction with efforts to restrict calcium and phosphate, as multiple interventions simultaneously tackling both insufficient inhibition and promoter excess may be required to redress the propensity to calcify. However, augmentation of calcification inhibition with either endogenous or synthetic analogs will require prudent evaluation of potentially deleterious effects on bone health given the common mineral target.¹³¹ Osteomalacia is a known side effect of etidronate and reduced bone mineral density has recently been reported with SNF472.¹³¹ Because interventions aimed at modifying advanced VC lesions are unlikely to produce tangible health benefits, future trials should ideally focus on the nondialysis CKD population where long-term patient gains may be cumulatively greater. Critically, however, we await definitive evidence that targeting VC yields the anticipated improvement in patient-centered outcomes and this needs to be embraced as a research priority by public funding entities to lessen the potential risk of bias incurred by pharmaceutical company involvement as has undoubtedly been the case with phosphate binders.

Disclosures

E.R. Smith holds stock in Calciscon; has consultancy agreements with Vifor Pharma; has received research funding from Amgen, Baxter, Sanofi, and Vifor Pharma; has received honoraria from Shire; and is a recipient of a Viertel Charitable Foundation Clinical Investigator award. M. Tiong is a member of the Australia and New Zealand Dialysis and Transplant Registry and is a member of the Aboriginal and Torres Strait Islander Health Working Group. N.D. Toussaint has received honoraria, travel support, and research funding from Amgen, Sanofi, Shire, and Takeda. All remaining authors have nothing to disclose.

Funding

None.

Author Contributions

N. Toussaint and C. Xu conceptualized the study, were responsible for data curation, formal analysis, investigation, methodology, and validation, and wrote the original draft; I. Ruderman, E. Smith, M. Tiong, and N. Toussaint were responsible for providing supervision; and all authors reviewed and edited the manuscript.

Supplemental Material

This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2021101327/-/DCSupplemental.

Supplemental Table 1. Search strategy.

Supplemental Table 2. Risk of bias assessment for RCTs.