A Randomized Trial on the Effect of Phosphate Reduction on Vascular End Points in CKD (IMPROVE-CKD)

Abstract

Visual Abstract

Background 

Hyperphosphatemia is associated with increased fibroblast growth factor 23 (FGF23), arterial calcification, and cardiovascular mortality. Effects of phosphate-lowering medication on vascular calcification and arterial stiffness in CKD remain uncertain.

Methods 

To assess the effects of non–calcium-based phosphate binders on intermediate cardiovascular markers, we conducted a multicenter, double-blind trial, randomizing 278 participants with stage 3b or 4 CKD and serum phosphate >1.00 mmol/L (3.10 mg/dl) to 500 mg lanthanum carbonate or matched placebo thrice daily for 96 weeks. We analyzed the primary outcome, carotid-femoral pulse wave velocity, using a linear mixed effects model for repeated measures. Secondary outcomes included abdominal aortic calcification and serum and urine markers of mineral metabolism.

Results 

A total of 138 participants received lanthanum and 140 received placebo (mean age 63.1 years; 69% male, 64% White). Mean eGFR was 26.6 ml/min per 1.73 m²; 45% of participants had diabetes and 32% had cardiovascular disease. Mean serum phosphate was 1.25 mmol/L (3.87 mg/dl), mean pulse wave velocity was 10.8 m/s, and 81.3% had abdominal aortic calcification at baseline. At 96 weeks, pulse wave velocity did not differ significantly between groups, nor did abdominal aortic calcification, serum phosphate, parathyroid hormone, FGF23, and 24-hour urinary phosphate. Serious adverse events occurred in 63 (46%) participants prescribed lanthanum and 66 (47%) prescribed placebo. Although recruitment to target was not achieved, additional analysis suggested this was unlikely to have significantly affected the principal findings.

Conclusions 

In patients with stage 3b/4 CKD, treatment with lanthanum over 96 weeks did not affect arterial stiffness or aortic calcification compared with placebo. These findings do not support the role of intestinal phosphate binders to reduce cardiovascular risk in patients with CKD who have normophosphatemia.

Clinical Trial registry name and registration number 

Australian Clinical Trials Registry, ACTRN12610000650099

Epidemiologic studies in the general population and in patients with CKD have consistently shown that serum phosphate levels within and above the normal range are independently and positively correlated with cardiovascular disease and all-cause mortality.^1,2 CKD is characterized by abnormal phosphate homeostasis, and hyperphosphatemia is strongly associated with increased arterial stiffness due to calcification of the tunica media of the arterial wall.³ Arterial stiffness and calcification are highly prevalent in CKD, increasing with worsening kidney function, and are associated with increased risks of adverse cardiovascular events and mortality.^4,5

Positive phosphate balance is thought to contribute to rising levels of fibroblast growth factor 23 (FGF23).⁶ FGF23 is the most potent hormone regulating phosphate homeostasis. It increases urinary excretion of phosphate by inhibiting phosphate reabsorption in the renal proximal tubule. Thus, serum phosphate levels remain normal, regardless of dietary variability of phosphate intake, until late stages of CKD. On the other hand, serum FGF23 levels are increased, even in people with modest degrees of kidney impairment, before serum phosphate levels rise.⁷ Increased serum levels of FGF23 are associated with increased risks of cardiovascular events and mortality in patients with CKD.^8–10 Therefore, FGF23 could be a potential therapeutic target to reduce cardiovascular morbidity and mortality in CKD.

Treatment of hyperphosphatemia in patients with CKD involves dietary phosphate restriction and intestinal phosphate binders. Globally, calcium-based phosphate binders are most commonly prescribed, although exogenous calcium has been associated with the development and progression of vascular calcification.¹¹ Current clinical guidelines suggest minimizing exogenous calcium by reducing exposure to calcium-based binders.¹² Use of lanthanum carbonate, a non–calcium-based phosphate binder, has been associated with reduction in vascular calcification compared with calcium-based binders,¹³ although few placebo-controlled trials have shown benefits of lanthanum beyond lowering serum phosphate.^14–17 In the nondialysis CKD population, randomized trials of phosphate-lowering therapy have predominantly focused on changes in serum phosphate and FGF23, but with only modest efficacy over short study periods.^14–16

The IMpact of Phosphate Reduction On Vascular Endpoints in CKD (IMPROVE-CKD) trial was designed to test the hypothesis that treatment with lanthanum carbonate will result in improvements in arterial compliance and vascular calcification in patients with nondialysis CKD, decreasing intestinal phosphate absorption and thus mitigating the rise in serum FGF23 levels.

Methods

Study Aims and Design

The IMPROVE-CKD study was an investigator-initiated, multicenter, international, randomized, placebo-controlled trial including participants with stage 3b–4 CKD (eGFR 15–44 ml/min per 1.73 m²) to assess the intervention of lanthanum carbonate on intermediate cardiovascular end points over 96 weeks. The protocol for the IMPROVE-CKD study has been published.¹⁸ The study was coordinated by the Australasian Kidney Trials Network (AKTN) and was prospectively registered (ACTRN12610000650099). Participants were recruited from 17 sites in Australia, New Zealand, and Malaysia and ethical approval was obtained by each local institutional ethics committee before study commencement.

Study Participants

The trial included participants with stage 3b–4 CKD and serum phosphate concentration >1.00 mmol/L (3.10 mg/dl) on at least one occasion over the 6-month period before enrolment, aged ≥18 years, and able to give informed consent. Exclusion criteria included medical conditions, other than CKD, affecting calcium and phosphate metabolism, gastrointestinal/malabsorption disorders or liver dysfunction, kidney transplantation, presence of atrial fibrillation, hospitalization or cardiovascular event within 1 month at screening, pregnancy, or breastfeeding. Several minor changes to the initial eligibility criteria were made after trial commencement to enhance recruitment, specifically removal of albuminuria and change from serum phosphate >1.20 mmol/L (3.72 mg/dl).¹⁸

Study Procedures

Patients who met the eligibility criteria and provided written informed consent attended a baseline visit. Those already taking a phosphate binder at screening underwent a 2-week washout period before attending the baseline visit. Eligible participants were randomized 1:1 to 500 mg lanthanum carbonate or matching placebo three times daily, with meals, for 96 weeks. Study medication was up-titrated by local investigators to a total maximum dose of six tablets daily (3000 mg/d lanthanum carbonate or six placebo per day) if serum phosphate remained persistently >1.60 mmol/L (4.95 mg/dl). Randomization was performed using an adaptive allocation algorithm that minimized imbalance across treatment groups for age (<60 years, ≥60 years), presence of diabetes, CKD stage (stage 3b, stage 4), and study site. The minimization algorithm was implemented via a central password-protected, web-based electronic randomization system provided by The George Institute in Sydney, Australia.

Scheduled visits over the 96-week study period involved eight 3-monthly follow-up visits with pulse wave velocity (PWV) measured at every second visit. Computed tomography (CT) scans were performed at baseline and at 96 weeks to determine abdominal aortic calcification (AAC). Pill counts were undertaken at study visits to assess adherence to study medication. Use of concomitant medications was considered standard clinical care by treating nephrologists. Phosphate binders, in addition to the maximal titrated study medication, could be prescribed for persistent hyperphosphatemia at the discretion of the local investigator.

Study Governance

The IMPROVE-CKD study was overseen by a Trial Steering Committee (TSC) and centrally coordinated by the AKTN. The TSC designed the study and took responsibility for the fidelity of the study’s compliance with the protocol. IMPROVE-CKD investigators wrote all drafts of the manuscript, and were responsible for the completeness and accuracy of the data and analysis. Shire International GmbH, a member of the Takeda group of companies, donated lanthanum carbonate and placebo but played no role in design, conduct, or analysis of the study, drafting of the manuscript, or decisions about submission for publication. Safety of participants was overseen by an independent Data and Safety Monitoring Board. Participants, investigators, the AKTN coordinating center staff, and outcome assessors were all blinded to the treatment assignment.

Study Outcomes

Primary Outcome

The primary outcome of the study was large artery compliance at 96 weeks after randomization, as measured by the mean of two carotid-femoral PWV measurements at each of the five time points from a SphygmoCor device (AtCor, PWV Inc., Sydney, Australia). Carotid and femoral waveforms from which PWV measurements are derived were reviewed at a central cardiac laboratory by two trained study investigators blinded to patient details and study medication. Biologically implausible PWV measurements, or measurements with unsatisfactory waveforms, were excluded from the calculation of mean PWV for each participant.

Secondary Outcomes

The extent of AAC was determined on CT datasets at baseline and at 96 weeks using Agatston scores based on density and size of total calcifications in the wall of the abdominal aorta over a length of 10 cm in the z axis, below the level of the upper end plate of the L2 lumbar vertebral body. Calcifications in the visualized renal and visceral branches were not included in the analysis. CT datasets were centrally analyzed on a single workstation (IntelliSpace Portal; Philips Healthcare, Cleveland, Australia) by the same trained senior radiologist and senior medical imaging technologist, who were blinded to patient details and study medication. As a binary outcome, the presence of AAC was determined by an Agatston score of nonzero at any level of the imaged abdominal aorta. Serum phosphate, calcium, and intact parathyroid hormone (PTH) concentrations were assessed at individual sites throughout the 96-week study period. Another prespecified binary outcome was the proportion of participants with hyperparathyroidism, defined by PTH >6.9 pmol/L (65 pg/ml). Standard biochemical measurements were performed in local hospital laboratories with appropriate regulatory accreditation. Samples for FGF23 measurement were collected 6-monthly, stored centrally at −80°C, and then assayed “en bloc.” C-terminal FGF23 measurements were made using the Immutopics ELISA assay, and intact FGF23 measurements using the Kainos (Tokyo, Japan) ELISA assay. Changes in urinary phosphate excretion were also evaluated using 24-hour urine collections.

Statistical Considerations

Sample Size

The study was designed to detect a clinically meaningful difference of 1 m/s in PWV between study groups at 96 weeks (higher in the placebo group). Assuming a within-group SD of 2.9 m/s, a sample size of 356 patients would detect a 1-m/s difference in PWV at the 5% significance level with 90% statistical power. To account for an estimated 10% study withdrawal rate and 10% nonadherence rate, recruitment of 488 participants was anticipated to be required. No provision was made for drop-ins, given there was no provision for lanthanum carbonate to be accessed in the nondialysis CKD population in Australia, New Zealand, and Malaysia.

Analyses

PWV measurements over time were analyzed using a mixed effects model with repeated measurement (MMRM). Fixed effects in the MMRM were treatment group, categoric time, the treatment-by-time interaction, and baseline measurements of PWV. An unstructured variance-covariance matrix was used to model the within-patient correlation structure. Missing baseline measurements were replaced using mean imputation. The result of primary interest was the effect of lanthanum carbonate on PWV at 96 weeks. Prespecified subgroup analyses were performed for CKD stage (3b versus 4), age groups (<65 years, ≥65 years), and presence of diabetes mellitus. Variables for post hoc subgroup analyses were baseline AAC (presence or absence) and serum phosphate (above or below mean). All subgroups were tested by examining treatment-by-subgroup interactions in MMRM. Differences in other continuous variables between the lanthanum and placebo groups were analyzed by analysis of covariance, adjusting for baseline measurements if there was a single follow-up measurement and MMRM otherwise. Non-normally distributed variables were natural log transformed for analysis. Repeatedly measured binary outcomes were analyzed using a generalized version of the MMRM approach to fit the treatment, and other fixed effects by a log binomial model. Differences between treatment groups on other binary outcome variables were analyzed using log binomial regression models. On-treatment analyses were also performed, where “on treatment” was defined as 80% compliant with randomized treatment. Percent treatment compliance was calculated as the proportion of planned study follow-up visits at which participants were taking the allocated study medication and being followed for outcome assessment. No adjustments for multiple comparisons were made. All statistical hypothesis testing was two-sided with P<0.05 considered statistically significant. Analyses were performed using SAS software (version 9.4; SAS Institute Inc., Cary, NC).

Investigational Product

The active and placebo intervention used in the trial was provided by Takeda. Provision of the investigational product labeling and packaging was by Pharmaceutical Professionals Packaging Pty Ltd.

Data Capture

Study data were collected and managed using REDCap electronic data capture tools hosted at Vanderbilt University. REDCap (Research Electronic Data Capture) is a secure, web-based, software platform designed to support data capture for research studies.

Results

Enrolment and Baseline Characteristics

Recruitment occurred between March 2012 and January 2017, with 278 participants enrolled and randomized to lanthanum carbonate (n=138) or placebo (n=140) (Figure 1, Supplemental Table 1). The study was terminated before full recruitment because of slower than anticipated accrual and funding issues. Baseline characteristics of enrolled participants were similar across treatment groups (Table 1, Supplemental Table 2). Mean age (±SD) was 63.1±12.7 years, with 69% male predominance. Mean eGFR for the overall cohort was 26.6±8.3 ml/min per 1.73 m² (Table 2), with the majority of participants classified as stage 4 CKD (67%). Mean serum phosphate at baseline was 1.25±0.20 mmol/L (3.87±0.62 mg/dl), and 91% were normophosphatemic (serum phosphate ≤1.50 mmol/L [4.65 mg/dl]). Mean carotid-femoral PWV of all trial participants was 10.8±3.6 m/s and, of the 235 participants with usable abdominal CT scans to determine AAC scores, 191 (81%) had vascular calcification. There was a greater proportion of AAC in the lanthanum arm compared with placebo at baseline (86% versus 77%, P=0.23). Detailed baseline characteristics have been published, including associations of PWV and AAC.¹⁹

Figure 1.: Study participant flowchart for the IMPROVE-CKD trial. SAE, serious adverse event.

Table 1. - Baseline demographic and clinical characteristics of study participants, by treatment group

Characteristics	Total (n=278)	Placebo (n=140)	Lanthanum (n=138)
Age (yr)	63.1±12.7	62.5±13.1	63.6±12.2
Sex (male), n (%)	193 (69.4)	97 (69.3)	96 (69.6)
Country of study site, n (%)
Australia	206 (74.1)	105 (75.0)	101 (73.2)
Malaysia	41 (14.7)	20 (14.3)	21 (15.2)
New Zealand	31 (11.2)	15 (10.7)	16 (11.6)
Ethnicity, n (%)
White	177 (63.9)	96 (68.6)	81 (59.1)
Aboriginal/TSI	3 (1.1)	2 (1.4)	1 (0.7)
NZ Māori/Pacific islander	6 (2.1)	1 (0.7)	5 (3.6)
Asian	63 (22.7)	29 (20.7)	34 (24.8)
Other	28 (10.1)	12 (8.6)	16 (11.7)
Weight (kg)	84.4±19.1	85.0±18.5	83.8±19.8
Body mass index (kg/m2)	29.9±6.0	30.1±6.1	29.6±5.9
Waist circumference (cm)	102.7±15.5	103.4±15.6	102.0±15.5
CKD stage, n (%)
Stage 3b	91 (32.7)	45 (32.1)	46 (33.3)
Stage 4	187 (67.3)	95 (67.9)	92 (66.7)
Primary cause of renal disease, n (%)
Diabetic nephropathy	84 (30.3)	45 (32.1)	39 (28.5)
Hypertension/vascular	45 (16.2)	21 (15.0)	24 (17.5)
GN	38 (13.7)	22 (15.7)	16 (11.7)
Reflux nephropathy	9 (3.2)	5 (3.6)	4 (2.9)
Polycystic kidney disease	14 (5.1)	6 (4.3)	8 (5.8)
Other	42 (15.2)	19 (13.6)	23 (16.8)
Unknown	45 (16.2)	22 (15.7)	23 (16.8)
Diabetes mellitus, n (%)	125 (45.0)	61 (43.6)	64 (46.4)
Hypertension, n (%)	249 (89.9)	125 (89.3)	124 (90.5)
Dyslipidemia, n (%)	218 (78.7)	111 (79.3)	107 (78.1)
Cardiovascular disease, n (%)	89 (32.1)	43 (30.7)	46 (33.6)
Peripheral vascular disease, n (%)	27 (9.8)	13 (9.3)	14 (10.3)
Cerebrovascular disease, n (%)	18 (6.5)	10 (7.1)	8 (5.9)
Smoking status, n (%)
Never	135 (49.1)	74 (52.9)	61 (45.2)
Former	121 (44.0)	60 (42.9)	61 (45.2)
Current	19 (6.9)	6 (4.3)	13 (9.6)
Systolic BP (mm Hg)	139±19	140±20	139±17
Diastolic BP (mm Hg)	76±10	76±11	76±10
Phosphate binder before study, n (%)	2 (0.7)	0 (0)	2 (1.4)

Results presented as mean±SD or number (percentage). TSI, Torres Strait Islander; NZ, New Zealand.

Table 2. - Baseline laboratory parameters and cardiovascular intermediate markers of study participants, by treatment group

Parameters	Total (n=278)	Placebo (n=140)	Lanthanum (n=138)
Hemoglobin (g/L)	124.3±16.1	123.8±16.2	124.8±16.0
Albumin (g/L)	38.9±4.3	38.7±4.0	39.1±4.5
Bicarbonate (mmol/L)	23.5±3.5	23.7±3.2	23.3±3.7
Urea (mmol/L)	16.0±5.2	15.8±5.1	16.2±5.4
Creatinine (µmol/L) a	221.4±59.4	219.2±58.5	223.6±60.3
eGFR (ml/min per 1.73 m2) b	26.6±8.3	27.0±8.6	26.1±8.1
Uric acid (mmol/L)	0.46±0.11	0.47±0.11	0.45±0.11
Glucose (mmol/L)	6.3±3.0	6.3±2.8	6.4±3.1
Calcium (mmol/L)	2.32±0.13	2.32±0.12	2.33±0.14
Phosphate (mmol/L)	1.25±0.20	1.26±0.18	1.24±0.22
ALP (U/L)	88.9±30.5	92.0±29.8	85.8±31.1
PTH (pmol/L) c	12.0 (7.2–17.8)	11.9 (7.4–16.8)	12.6 (6.4–19.4)
25-Hydroxyvitamin D (nmol/L)	70.1±30.7	68.9±29.0	71.3±32.6
1,25-Dihydroxyvitamin D (pmol/L)	77.8±26.8	75.5±26.7	80.3±26.9
C-terminal FGF23 (RU/ml)	221.1 (154.3–334.1)	225.1 (160.5–327.1)	220.7 (151.7–353.1)
Intact FGF23 (pg/ml)	133.0 (89.1–202.0)	136.0 (89.4–201.4)	124.8 (88.9–208.5)
Urinary phosphate excretion (mmol/24 h)	24.4±11.4	26.1±12.6	22.7±9.9
Cholesterol (mmol/L)	4.5±1.1	4.4±1.1	4.6±1.1
Triglyceride (mmol/L)	2.1±1.4	2.0±1.2	2.2±1.5
C-reactive protein (mg/L)	4.7±6.3	4.7±6.4	4.8±6.3
Protein excretion (mg/24 h)	718 (182–1710)	823 (200–1720)	535 (120–1710)
Urine ACR (mg/mmol) d	68 (20–139)	74 (23–141)	54 (16–138)
PWV (m/s)	10.8±3.6	10.8±3.6	10.8±3.6
Augmentation index (%)	27.3±9.6	26.9±10.4	27.5±9.0
Aortic calcification e	191 (81.3%)	90 (76.9%)	101 (85.6%)
Agatston score e	1535 (63–5744)	839 (11–3839)	2133 (278–6081)

Results presented as mean±SD, number (percentage) or median (interquartile range). ALP, alkaline phosphatase; ACR, albumin-creatinine ratio.

^aConversion to mg/dl: multiply by 0.0113.

^beGFR calculated using CKD–Epidemiology Collaboration equation.

^cConversion to pg/ml: multiply by 9.4.

^dConversion to mg/g: multiply by 8.85.

^eA total of 235 participants had usable CT scans to determine aortic calcification (118 lanthanum, 117 placebo).

Primary End Point

A total of 30 participants had no PWV follow-up measurements (28 with no measurements and two excluded after blinded central review of PWV measurements; missing data equally distributed between groups, Supplemental Table 3). At 96 weeks, PWV adjusted for baseline values (n=248) did not differ significantly between groups, with a difference of +0.6 (95% CI, −0.3 to 1.5) m/s, P=0.20 (Figure 2). Results were similar after adjustment for the three variables used in the minimization algorithm for treatment allocation, with a difference of +0.6 (95% CI, −0.3 to 1.5) m/s, P=0.19. PWV slope was not significantly different between groups: lanthanum slope was 0.38 m/s, placebo slope was −0.27 m/s (mean slope difference, 0.65; 95% CI, −0.26 to 1.57; P=0.16).

Figure 2.: Change in PWV (with 95% CI) over 96 weeks, by treatment group.

Figure 3 displays PWV results at 96 weeks for three prespecified subgroups (CKD stage, age group, diabetes mellitus) and two post hoc subgroups (AAC, serum phosphate). No subgroup interaction test (treatment×visit×subgroup) was statistically significant (all P>0.30), and all treatment group differences within subgroups favored placebo.

Figure 3.: PWV results at 96 weeks for pre-specified (CKD stage, age, diabetes) and post-hoc (AAC [abdominal aortic calcification], serum phosphate) subgroups showing the effect of lanthanum carbonate relative to placebo.

Secondary End Points

Aortic Calcification

At 96 weeks, the mean AAC Agatston score was not significantly different (+172; 95% CI, −200 to 545; P=0.36) (Table 3). The proportion of participants with AAC seen at 96 weeks was 79% in the placebo arm and 88% in the lanthanum arm (P=0.10), with the difference being similar to that observed at baseline.

Table 3. - Comparison of treatment groups (lanthanum versus placebo) at week 96 for intermediate cardiovascular parameters and biochemical parameters

Outcome Variable	N	Mean (SEM) at 96 Wk (lanthanum, placebo)	Difference (95% CI) a	P Value
PWV (m/s) b	248	11.8 (0.33), 11.2 (0.33)	+0.6 (−0.3 to 1.5)	0.20
Serum phosphate (mmol/L)	262	1.34 (0.03), 1.37 (0.03)	−0.03 (−0.12 to 0.05)	0.40
Serum calcium (mmol/L)	261	2.31 (0.01), 2.29 (0.01)	+0.02 (−0.01 to 0.06)	0.14
Serum calcium-phosphate product (mmol2/L2)	261	3.07 (0.07), 3.13 (0.07)	−0.06 (−0.24 to 0.13)	0.54
PTH (pmol/L) c	226	2.76 (0.06), 2.75 (0.06)	+0.009 (−0.16 to 0.18)	0.92
cFGF23 (RU/ml) c	226	5.82 (0.07), 5.85 (0.07)	−0.03 (−0.24 to 0.18)	0.78
iFGF23 (pg/ml) c	237	5.28 (0.09), 5.35 (0.09)	−0.08 (−0.32 to 0.17)	0.53
cFGF23/iFGF23 ratio c	221	0.69 (0.5), 0.73 (0.05)	−0.04 (−0.12 to 0.20)	0.65
Urinary phosphate excretion (mmol/24 h)	236	22.01 (1.10), 23.17 (1.07)	−1.16 (−4.21 to 1.88)	0.45
Urine phosphate-creatinine ratio (mg/mmol)	231	192.24 (8.86), 195.06 (8.61)	−2.82 (−27.20 to 21.56)	0.82
eGFR (ml/min/1.73 m2)	262	21.5 (0.6), 22.2 (0.6)	−0.70 (−2.42 to 1.01)	0.42
Creatinine clearance (ml/min)	233	30.09 (5.50), 40.63 (5.45)	−10.54 (−25.82 to 4.73)	0.18
Agatston score	203	4147 (134), 3975 (133)	+172 (−200 to 545)	0.36

Variables in the models include treatment group, visit, treatment group by visit interaction, and baseline variable. cFGF23, C-terminal FGF23; iFGF23, intact FGF23.

^aLanthanum relative to placebo.

^bA total of 14 participants had imputed values for missing baseline PWV, and 30 participants had no follow-up data and were excluded.

^cLog transformed.

Serum and Urinary Markers of Bone and Mineral Metabolism

There were no differences in serum phosphate or 24-hour urinary phosphate excretion between groups at 96 weeks (Figure 4). There were also no differences in serum PTH and intact FGF23 levels, or plasma C-terminal FGF23, at 96 weeks (Figure 4). The proportion of participants with hyperparathyroidism was also not different between lanthanum and placebo (Table 4).

Figure 4.: Change over 96 weeks (with 95% CI), by treatment group. (A) Serum phosphate, (B) urinary phosphate excretion, (C) serum PTH level, (D) C-terminal FGF23, and (E) intact FGF23.

Table 4. - Comparison of treatment groups (lanthanum versus placebo) at 96 weeks on binary secondary outcomes

Outcome Variable	N	Risk Ratio (95% CI) a	P Value
Presence of AAC	200	1.11 (0.98 to 1.26)	0.10
eGFR decline ≥30%	278	1.03 (0.80 to 1.33)	0.82
Use of nonstudy phosphate binder	278	0.92 (0.40 to 2.10)	0.85
Use of calcium-based binders	278	0.79 (0.30 to 2.06)	0.63
Proteinuria (PCR >30 mg/mmol) b	240	0.98 (0.88 to 1.08)	0.63
Hyperparathyroidism (PTH >6.9 pmol/L) b	226	0.91 (0.81 to 1.03)	0.12
Serum phosphate (>1.5 mmol/L) b	262	0.82 (0.55 to 1.20)	0.30
Serum calcium (>2.6 mmol/L) b	261	2.56 (0.83 to 7.85)	0.10
Serum calcium-phosphate product (>3.9 mmol2/L2) b	261	0.93 (0.53 to 1.65)	0.81

Results not adjusted for baseline measurements. PCR, protein-creatinine ratio.

^aLanthanum relative to placebo.

^bEstimates from generalized estimating equations with robust SEMs.

Kidney Function

There was no difference in eGFR or 24-hour urinary creatinine clearance between groups at 96 weeks (Table 3), nor change in eGFR slope: lanthanum slope was −2.84 ml/min per 1.73 m², placebo slope was −2.34 ml/min per 1.73 m² (mean slope difference, −0.49; 95% CI, −1.41 to 0.42; P=0.29).

Medication

The median number of study medication tablets consumed was 2.69 (95% CI, 2.20 to 2.90) per day for the placebo group and 2.57 (95% CI, 2.00 to 2.88) per day for the lanthanum group. Final maximum doses of study medication in the lanthanum group were as follows: 129 (93.5%) participants were prescribed three tablets per day, seven (5%) participants were prescribed four tablets per day, zero were prescribed five tablets per day, and two (1.4%) were prescribed six tablets per day. The distribution of prescribed doses was similar for the placebo group. Median adherence was 92% (95% CI, 81% to 94%) in the placebo group and 92% (95% CI, 73% to 94%) in the lanthanum group. In addition to study medication, there were no differences between study groups in the use of either nonstudy phosphate binders or calcium-based phosphate binders (Table 4). Eleven (8%) participants randomized to the placebo arm and ten (7%) participants randomized to the lanthanum arm took phosphate binders as concomitant medication during the study.

On-Treatment Analyses

Overall, 74% of participants (n=206) were at least 80% compliant, with similar proportions in the two groups (lanthanum, 73% [101/138]; placebo, 75% [105/140]). Although the results for PWV and three of the four secondary outcomes were similar to those from the main analyses based on all participants with analyzable data, serum phosphate levels were significantly lower in the lanthanum group compared with placebo (−0.09 mmol/L, 95% CI, −0.18 to −0.01; P=0.03; Supplemental Table 4). These results for the on-treatment subset were replicated when models based on all participants with analyzable data were adjusted for percent compliance.

Adverse Events

Serious adverse events were reported in 63 (46%) and 66 (47%) participants on lanthanum and placebo, respectively (Table 5).

Table 5. - Adverse events by treatment groups

Outcome Variable	n (%)		P Value
	Placebo (n=140)	Lanthanum (n=138)
Any SAE	66 (47)	61 (44)	0.62
Death from any cause	8 (6)	2 (1)	0.10 a
Life-threatening event	0	4 (3)	0.06 a
Initial or prolonged inpatient hospitalization	64 (46)	58 (42)	0.54
Persistent/significant disability	1 (<1)	2 (1)	0.62 a
Important medical event	7 (5)	8 (6)	0.77
Any adverse drug reaction	16 (11)	18 (13)	0.68
Gastrointestinal event b	10 (7)	8 (6)	0.65
Nausea/vomiting	4 (3)	6 (4)	0.54 a
Other event	6 (4)	4 (3)	0.75 a

SAE, serious adverse event.

^aFisher exact test.

^bIncludes diarrhea, constipation, gastroesophageal reflux.

Discussion

This randomized, placebo-controlled, clinical trial showed that, in patients with stage 3b or 4 CKD, treatment with the phosphate binder, lanthanum carbonate, over 96 weeks did not result in any difference in arterial stiffness or aortic vascular calcification compared with placebo. We also found no difference in serum and urinary markers of bone and mineral metabolism between treatment arms. This study is consistent with other clinical trials of phosphate binders in patients with nondialysis CKD,^14–16,20 and relatively normal serum phosphate concentrations, and suggests that phosphate binder therapy may not be effective in reducing intermediate markers of cardiovascular risk in that setting.

Similar to our study, potential benefits of phosphate-lowering therapy in patients with nondialysis CKD are yet to be demonstrated in other placebo-controlled clinical trials. Block et al.¹⁴ reported that, in 148 patients with stage 3–4 CKD, calcium and non–calcium-based phosphate binders led to minimal reductions in serum phosphate, and had no effect on serum FGF23 concentrations, when compared with placebo over 9 months. Similarly, Chue et al.²⁰ found no significant differences between sevelamer and placebo on left ventricular mass or PWV in 109 participants after 40 weeks, although only 56% of subjects took ≥80% of prescribed therapy. Seifert et al.¹⁵ also reported no differences between lanthanum or placebo in 38 patients over 12 months in serum or urinary phosphate, or in surrogate cardiovascular markers. These studies, and our results, do not support the use of phosphate binders in the nondialysis CKD population with normophosphatemia or mild hyperphosphatemia.

More recently, the COMBINE (CKD Optimal Management With BInders and NicotinamidE) study assessed the effects of lowering serum phosphate and FGF23 in 200 patients with stage 3–4 CKD with lanthanum, nicotinamide, and combination therapy of both.¹⁶ Consistent with our study, there were no significant differences in serum phosphate or FGF23 in any of the three active treatment arms compared with the double placebo group. Gastrointestinal adverse events were also common (12% in lanthanum-placebo arm), and the rate of discontinuation of study medications was relatively high (30% in lanthanum-placebo arm, 42% in double active group). We report gastrointestional events and nausea/vomiting in 10% of study participants, with no difference between lanthanum and placebo, and 74% of participants in our study were compliant (≥80% of study medication), with similar proportions between study arms. Higher gastrointestinal adverse effects in the COMBINE trial, compared with our study, may have resulted from a higher dose of lanthanum carbonate (1000 mg three times daily) in that study, and also combination therapy with nicotinamide in participants on both medications.

The lack of effect of phosphate binders on cardiovascular intermediate markers in our study was consistent with an open-labeled clinical trial, which randomized 120 patients with stage 3–4 CKD to either lanthanum, calcium, or dietary phosphate restriction for 12 months, and reported no effect on phosphate, FGF23, or vascular parameters of coronary artery calcification and PWV.²¹ A more recent study did report potential beneficial effects of a different phosphate binder, ferric citrate, in 203 patients with CKD, demonstrating improvement in serum phosphate and FGF23, but also in hospitalization and the composite end point of death and need for kidney replacement therapy after 9 months.²² The study cohort, however, involved participants with more advanced CKD (stages 4 and 5 CKD) with higher mean serum phosphate. Also in contrast to our trial, that study was an open-label, single-center study, and 37% of those in the usual-care arm received phosphate binders, including calcium-based binders.

Abnormalities of mineral metabolism in CKD are associated with the development of vascular calcification, increased arterial stiffness, and greater cardiovascular morbidity and mortality.^5,23In vitro studies have reported osteogenic transformation of vascular smooth muscle cells in the presence of hyperphosphatemia, with upregulation of genes that promote matrix mineralization and vascular calcium deposition.²⁴ Studies also report vascular calcification associated with elevated FGF23 values, independent of serum phosphate.^9,25 Even modest serum phosphate elevations within the normal laboratory range are associated with arterial calcification and cardiovascular events.^26,27 Normalizing a positive phosphate balance (and lowering FGF23) is a potential strategy to reduce cardiovascular risk in CKD. In IMPROVE-CKD, we did not target patients who were hyperphosphatemic and the majority of participants were normophosphatemic (91%). This may be one explanation for a lack of difference between groups on the primary end point and on serum phosphate, although an on-treatment analysis did show a modestly lower phosphate level in participants treated with lanthanum. Serum phosphate values are also likely to be inaccurate for quantifying phosphate balance and cardiovascular risk. Subgroup analysis did not show any benefit of lanthanum on PWV in those with higher serum phosphate, but, instead, participants on placebo had a trend toward less arterial stiffness progression.

PWV is an intermediate measure of cardiovascular risk, with higher velocity associated with increased risk.²⁸ PWV increases with aging, increased vascular comorbidity, hypertension, and diabetes.²⁹ This measure has been used as an end point in several phosphate-lowering studies in patients with CKD^14,15 with no significant change in PWV with intervention, although PWV was not the primary end point in these trials. Our trial used PWV as the primary end point, and the study cohort had a high baseline mean PWV level (>10 m/s), highlighting a population at significantly increased cardiovascular risk. The relatively advanced arterial stiffness could explain why intervention was ineffective over 96 weeks.

The majority of participants in our cohort also showed evidence of extensive vascular calcification at baseline, consistent with observational studies of patients with advanced CKD.³⁰ The high prevalence of AAC in our cohort, 81% at baseline, is also an important finding, indicating participants at extremely high cardiovascular risk. Similarly, advanced calcification may explain the lack of effects on intermediate cardiovascular outcomes. A recent short-term pilot study (8 weeks) in 18 patients with CKD who were normophosphatemic reported that treatment with the non–calcium-based phosphate binder sevelamer did not improve PWV overall, but did in the subgroup with no or limited AAC.³¹ However, even in subgroup analysis of participants without AAC (n=44), we did not find any difference in PWV with phosphate binders compared with placebo. The higher proportion of AAC in the lanthanum group may have biased results toward the null, because one of the strongest predictors of progression is the presence of calcification per se.

We report no difference in serum phosphate or urinary phosphate excretion between lanthanum carbonate and placebo. Phosphate binders should lower intestinal phosphate absorption, but may also upregulate the active phosphate transporter, sodium phosphate cotransporter 2b (Npt2b), in the small intestine.³² As such, use of binders may lower phosphate absorption acutely, but may promote relative hyperabsorption through enhanced Npt2b expression when binders are not present in the intestinal lumen, and this may potentially limit efficacy of phosphate binders in CKD.³³ Other possible reasons for why lanthanum did not lower urinary phosphate excretion, as may have been expected, include that compliance may have been lower than reported, the dose of lanthanum may have been too low, or measurement error may have biased toward the null. On-treatment analysis did demonstrate a reduction in serum phosphate in participants, however, suggesting a significant effect of phosphate lowering in those taking active treatment. This finding does raise the question of whether nonadherence or study medication discontinuation were more common than estimated.

FGF23 is an important regulator of both phosphate and calcitriol levels, and trajectories of increasing FGF23 over time are strongly associated with adverse health outcomes, including cardiovascular disease.³⁴ Phosphate binders could possibly reduce serum FGF23 levels by decreasing gastrointestinal phosphate absorption, and interventions to reduce FGF23, even where serum phosphate levels are normal, could potentially be beneficial to reduce cardiovascular morbidity and mortality, although with a lack of evidence to date. A post hoc, per-protocol analysis of patients in the COMBINE study who remained on study medication found no significant changes in serum phosphate, but an 8% reduction in FGF23 in those on lanthanum, compared with a 14% increase on placebo.¹⁶ We report no difference in C-terminal or intact FGF23 with lanthanum compared with placebo in our study, including on-treatment analyses. Surprisingly, there was also no difference in PTH levels between study groups despite the 96-week study duration.

Implications for Clinical Practice

Findings from our study, together with those of previous studies, suggest treating patients with moderate-to-advanced CKD with phosphate binders to reduce cardiovascular risk is not justified, especially if patients are normophosphatemic at the time. Further studies are required to determine whether phosphate binders have a beneficial effect on patient-level outcomes (cardiovascular events, death, quality of life, etc.) that outweighs their harms (pill burden, gastrointestinal effects, etc.), and the patient groups in whom these benefits might be seen.

Strengths and Limitations

The IMPROVE-CKD trial is the largest and longest placebo-controlled study of phosphate binders in patients with nondialysis CKD to date. The use of a non–calcium-based phosphate binder avoided potentially confounding effects of calcium-based binders on vascular calcification. Participant medication adherence rates were monitored and were high (92%), and performance in 17 centers across three countries enhanced generalizability of findings.

Limitations include achievement of only 57% of target recruitment and, therefore, the study was underpowered for the primary outcome, such that a type 2 statistical error could not be excluded. The study was terminated before full recruitment because of slower than anticipated accrual and funding issues. However, the 95% CI for the primary end point of PWV showed that a 1-m/s difference in favor of lanthanum carbonate was likely not plausible for this proposed outcome. A post hoc futility analysis also revealed that the conditional power to detect the prespecified clinically meaningful difference of a 1-m/s lower PWV in the lanthanum group relative to placebo at 96 weeks was 0.5%. The futility index was >99%. Therefore, it is highly unlikely that, had the trial continued to its recruitment target, there would have been a statistically significant result, and it was not a lack of power to detect the difference of 1 m/s but more that an effect of this magnitude in favor of lanthanum in this CKD population, with a high cardiovascular risk burden, may not exist. A further limitation is that PWV and AAC are intermediate cardiovascular measures, and there is some degree of operator dependency with PWV measurements, although the objective measurement of AAC is a strength because it may precede the presence of coronary artery calcification in patients with CKD and is more strongly correlated with cardiovascular risk factors.^35,36 Other limitations included biochemical parameters measured at local laboratories, with potential for different PTH assays and associated variability, and recruitment occurring over a 5-year period, with potential for changes in some aspects of clinical practice over that time course. However, randomization should evenly distribute any change in background clinical practice.

The IMPROVE-CKD study found no beneficial effect of the phosphate binder, lanthanum carbonate, on intermediate cardiovascular markers and mineral metabolism parameters in patients with stage 3b and 4 CKD and predominantly normal serum phosphate values. This study adds to the current clinical trial data raising further concern about the paucity of proven benefits of phosphate binder therapy, in the nondialysis CKD population with normophosphatemia, to lower serum phosphate and FGF23 levels in an attempt to reduce the burden of cardiovascular disease. Further studies should be adequately powered, and perhaps targeted to those with positive phosphate balance, to assess the utility of phosphate-lowering strategies on patient-level outcomes.

DISCLOSURES

S. Badve reports receiving grants from National Health and Medical Research Council of Australia, personal fees from Bayer AG and Amgen Australia, and nonfinancial support from Bayer AG, unrelated to the study. G. Block reports having current equity ownership of Ardelyx and Reata. G. Block also reports being an employee of Reata Pharmaceuticals; past research funding from Keryx; and past consulting with Kirin, Amgen, Akebia, Keryx, and OPKO. N. Boudville reports personal fees from Baxter; travel grants from Amgen and Roche; and grants from Amgen and Baxter, outside the submitted work. K.L. Campbell reports consultancy fees from Nestle Health Sciences. G.J. Elder reports receiving research funding from Amgen; and travel support and honoraria from Roche and Takeda, outside the submitted work. C.M. Hawley reports research grants from National Health and Medical Research Council of Australia; research funding to her institution from Baxter Healthcare, Fresenius Medical Care, and Shire; and consultancy fees from GlaxoSmithKline, Janssen, and Otsuka paid to her institution; and personal fees from Otsuka and grants from Shire, outside the submitted work. S.G. Holt reports receiving nonfinancial support from Amgen; personal fees from AstraZeneca; other from Baxter; personal fees from Otsuka; grants from Sanofi; and honoraria, travel support, and research funding from Amgen, AstraZeneca, Baxter, and Sanofi, outside the submitted work. M. Jardine reports having served on advisory boards sponsored by Akebia, Baxter, Boehringer Ingelheim, and Vifor; serves on the steering committee for a trial sponsored by CSL and Janssen; and speaking at scientific meetings sponsored by Amgen, Janssen, Roche, and Vifor, with any consultancy, honoraria, or travel support paid to her institution; a Medical Research Future Fund Next Generation Clinical Researchers Program Career Development Fellowship; and unrestricted funding from Gambro, Baxter, CSL, Amgen, Eli Lilly, and MSD. D. Johnson reports grants from Baxter Healthcare, Fresenius Medical Care, and the National Health and Medical Research Council of Australia; personal fees from Baxter Healthcare, Fresenius Medical Care, AWAK, Ono, and AstraZeneca; and other support from Amgen, unrelated to the study. N.M. Lioufas reports support for research from the Australian Commonwealth with an RTP scholarship, outside the submitted work. E. Pedagogos reports honoraria, travel support, and research funding from Amgen, Shire, and Sanofi, outside the submitted work. V. Perkovic reports grants or fees from National Health and Medical Research Council of Australia, Retrophin, Janssen, Merck, Servier, AbbVie, Astellas, AstraZeneca, Bayer, Baxter, BMS, Boehringer Ingelheim, Dimerix, DURECT, Eli Lilly, Gilead, GSK, Mitsubishi Tanabe, Novartis, Novo Nordisk, Pfizer, PharmaLink, Relypsa, Sanofi, Vifor Pharma, and Tricida, outside the submitted work. K.R. Polkinghorne reports grants from National Health and Medical Research Council of Australia; personal fees from Medtronic and AstraZeneca; and travel support from Amgen, outside the submitted work. C. Pollock reports being a speaker for AstraZeneca, Janssen Cilag, Sanofi, Novartis, Vifor, and Otsuka; and reports being an advisory board member for AstraZeneca, Merck Sharp and Dohme, Eli Lilly, Novartis, Vifor, and Otsuka. E. Smith owns stock in Calciscon AG, which commercializes the T50 test. E.R. Smith reports research funding from Amgen and Sanofi, outside the submitted work. A. Yee Moon Wang reports speaker honorarium from Fresinius Kabi; and grants from Sanofi, outside the submitted work. All remaining authors have nothing to disclose.

Funding

This investigator-initiated research work was supported by National Health and Medical Research Council of Australia research grants APP1044302, APP1092957, and ID 631731, and by Shire (a member of the Takeda group of companies) grant IST-AUS-000108. National Health and Medical Research Council of Australia and Shire International did not have any role in study design, collection, analysis and interpretation of data, writing the report, and the decision to submit the report for publication. The Vanderbilt Institute for Clinical and Translational Research received support from National Center for Advancing Translational Sciences grant UL1 TR000445.