Dihydropyridine Calcium Channel Blockers and Kidney Outcomes

Abstract

Background

Early trials of dihydropyridine calcium channel blockers (DCCBs) suggest a detrimental effect on intraglomerular pressure and an association with albuminuria.

Objective

We sought to evaluate the associations of DCCB initiation with albuminuria and kidney failure with replacement therapy (KFRT) and to determine whether renin-angiotensin system (RAS) blockade modified these associations.

Design

We conducted a target trial emulation study using a new user, active comparator design and electronic health record data from Geisinger Health.

Participants

We included patients without severe albuminuria or KFRT who were initiated on a DCCB or thiazide (active comparator) between January 1, 2004, and December 31, 2019.

Main Measures

Using inverse probability of treatment weighting, we performed doubly robust Cox proportional hazards regression to estimate the association of DCCB initiation with incident severe albuminuria (urine albumin to creatinine ratio > 300 mg/g) and KFRT, overall and stratified by RAS blocker use.

Key Results

There were 11,747 and 26,758 eligible patients initiating a DCCB and thiazide, respectively, with a weighted baseline mean age of 60 years, systolic blood pressure of 143 mm Hg, and eGFR of 86 mL/min/1.73 m², and with a mean follow-up of 8 years. Compared with thiazides, DCCBs were significantly associated with the development of severe albuminuria (hazard ratio [HR], 1.29; 95% confidence interval [CI], 1.16–1.43), with attenuation of risk in the presence of RAS blockade (P for interaction < 0.001). The risk of KFRT was increased among patients without RAS blockade (HR, 1.66; 95% CI, 1.19–2.31), but not with RAS blockade (P for interaction = 0.005).

Conclusions

DCCBs were associated with increased risk of albuminuria and, in the absence of RAS blockade, KFRT. These findings suggest coupling DCCB therapy with RAS blockade may mitigate adverse kidney outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11606-024-08762-2.

INTRODUCTION

Dihydropyridine calcium channel blockers (DCCBs), such as amlodipine and nifedipine, are first-line antihypertensive agents alongside thiazide diuretics, angiotensin-converting enzyme (ACE) inhibitors, and angiotensin receptor blockers (ARB).¹ DCCB use is highly prevalent in the United States and globally, including in low-resource settings where they are often preferred.^2–7 While effective blood pressure control yields health benefits,⁸ DCCBs have class-specific effects that may result in kidney harm. Specifically, DCCBs vasodilate the afferent arteriole more than the efferent arteriole.⁹ Consequently, renal plasma flow can increase, resulting in glomerular hyperfiltration, glomerular hypertension, and albuminuria, a risk factor for chronic kidney disease progression and death.^10–13 This raises the question as to whether DCCBs promote kidney injury, as well as the possibility that renin-angiotensin system (RAS) blockers, which dilate efferent arterioles thus reducing intraglomerular pressure, might mitigate this risk.

In animal studies employing a remnant kidney model, in which the majority of kidney mass is ablated, DCCBs impair the myogenic response to hypertension, resulting in glomerulosclerosis.^14,15 In humans, clinical trial evidence suggests that DCCBs may increase albuminuria.^16,17 For example, in the African American Study of Kidney Disease and Hypertension (AASK) trial, the amlodipine arm was stopped early for harm in the setting of increased albuminuria. However, clinical trials linking DCCBs to kidney failure with replacement therapy (KFRT) have yielded mixed findings.^17–20 Few real-world studies have examined the role of DCCB use in the development of albuminuria and KFRT despite their frequent use.⁶ Given the importance of DCCBs in the management of hypertension, the link between DCCBs and kidney disease requires further investigation.

We sought to examine the association of DCCB use with the development of albuminuria and KFRT among patients with hypertension using target trial emulation methods and electronic health records from Geisinger Health. We compared DCCBs to thiazide diuretics, an active comparator not considered to be nephroprotective beyond its blood pressure lowering effect. We also sought to determine if associations were modified by ACE inhibitor or ARB use.

METHODS

Study Design and Data Source

We performed a target trial emulation employing a new user, active comparator, intention-to-treat design in an electronic health records cohort from Geisinger Health, a large, integrated health care system located in northeastern and north central Pennsylvania. The use of deidentified patient information for this study was approved by institutional review boards at Geisinger Health and Johns Hopkins Bloomberg School of Public Health.

Target Trial Emulation

Eligibility Criteria and Treatment Strategies

Our main cohort included patients aged ≥ 18 years who initiated either a DCCB (excluding nimodipine, a DCCB only prescribed following subarachnoid hemorrhage) or thiazide or thiazide-type diuretic (collectively referred to hereafter as thiazide) from January 1, 2004, through December 31, 2019 (Supplemental Figure 1). We excluded patients with prevalent KFRT and chronic kidney disease category A3 (i.e., having a urine albumin to creatinine ratio [UACR] > 300 mg/g, an equivalent urine protein to creatinine ratio [UPCR] using the conversion equation provided by the Chronic Kidney Disease-Prognosis Consortium,²¹ or a urine protein dipstick of 2+ or greater on at least one occasion), lacking baseline estimated glomerular filtration rate (eGFR), or who were prescribed both a DCCB and thiazide at medication initiation. Patients without a documented baseline albuminuria measure (UACR, UPCR, or urine protein dipstick) were considered not to have albuminuria and were eligible for inclusion. In the primary analysis, given the high proportion of patients lacking an albuminuria measure after baseline, only patients with an albuminuria measure during follow-up were considered. Our cohort consisted of 38,505 patients.

Thiazides were chosen as the active comparator since they are also first-line antihypertensive agents and, unlike ACE inhibitors and ARBs, are not specifically indicated for albuminuria reduction.¹ Medication use was ascertained from prescription records and was considered initial use only if started at least 1 year following entry into the electronic health record system to exclude prevalent use and achieve new user design. Medication groups were handled as intention to treat (i.e., treatment assignment did not change after baseline to account for medication crossover or discontinuation).

Target Trial Emulation by Inverse Probability of Treatment Weighting

We emulated a target trial by employing inverse probability of treatment weighting (IPTW) to ensure characteristics between the two treatment groups were similar. Data were missing for 9%, 3%, and 3% of the population for body mass index, systolic blood pressure, and diastolic blood pressure, respectively, and this was addressed using multiple imputations using chained equations to create 10 datasets. For each imputed dataset, we calculated propensity scores for DCCB initiation using a multivariable logistic regression model. From this, we derived IPTW, which we applied to the cohort to achieve covariate balance. The assigned weight was 1 divided by the propensity score for DCCB initiation among DCCB initiators, and 1 divided by the quantity 1 minus the propensity score for DCCB initiation among thiazide initiators. Weights were stabilized by multiplying them by the proportion of the cohort initiating each treatment. To reduce the influence of extreme outlier weights, the stabilized weights were winsorized at 5 standard deviations. We targeted standardized mean differences of < 10% between groups after IPTW.

In calculating the propensity score, we included demographic and comorbid variables to account for potential confounders. Baseline age, sex, race (categorized as Black, White, or other, which included unknown race), and baseline calendar year were derived from the health record. Baseline body mass index (kg/m²), systolic blood pressure (mm Hg), diastolic blood pressure (mm Hg), and serum creatinine (mg/dL) measurements were taken as the most recent measure within 3 years prior to medication initiation. eGFR was calculated using the 2021 Chronic Kidney Disease Epidemiology Collaboration creatinine equation.²² Smoking history was classified as current, former, or never. History of diabetes mellitus, stroke, heart failure, and coronary heart disease at or before medication initiation was ascertained from inpatient and outpatient diagnostic codes. Prevalent ACE inhibitor, ARB, beta blocker, loop diuretic, sodium-glucose cotransporter 2 inhibitor, glucagon-like peptide-1 receptor agonist, and insulin use were derived from the health record, along with hospitalization within 1 year before medication initiation (yes/no).

Outcomes and Follow-Up

The primary outcomes were incident albuminuria and KFRT. Albuminuria was defined as a UACR > 300 mg/g, an equivalent UPCR using the conversion equation provided by the Chronic Kidney Disease-Prognosis Consortium,²¹ or a urine protein dipstick of 2+ or greater. We required confirmation of albuminuria with repeat testing between 3 months and 2 years for all positive tests. KFRT, including chronic dialysis or kidney transplantation, was ascertained from corresponding diagnostic codes (Supplemental Table 2).²³ Death was also determined through periodic linkage with the Social Security Death Index. Time at risk began with initial use of a DCCB or thiazide.

Statistical Analysis

Baseline characteristics stratified by medication were reported before and, in a randomly selected imputed cohort, after weighting. Standardized mean differences among baseline characteristics were calculated before and after weighting and visualized with a Love plot.

In a weighted cohort, associations of DCCB initiation with incident albuminuria and KFRT were estimated using Cox proportional hazards regression models (IPTW only) and with further adjustment for any unbalanced variables, defined as those with a weighted mean difference > 2.5 (doubly robust model, the primary analysis). Thiazide initiation was considered the reference, and estimates for each imputed cohort were combined using Rubin’s rule.²⁴ We plotted cumulative incidence curves for each outcome by DCCB or thiazide initiation. These plots reflect the associations observed in the IPTW-only analysis without further adjustment for unbalanced variables. Patients were censored upon their last albuminuria assessment for the albuminuria outcome and death for the KFRT outcome. Given the renal protective effects of renin-angiotensin system blockade,^25,26 we also stratified our cohort by baseline RAS blocker (ACE inhibitor or ARB) use and tested for multiplicative interaction with a product term in the regression model.

Sensitivity Analyses

We performed the following sensitivity analyses using the doubly robust model: (1) restricting to patients with baseline eGFR < 60 mL/min/1.73 m², (2) restricting to patients with a baseline albuminuria assessment, (3) excluding patients with A2 albuminuria at baseline (UACR 30–300 mg/g, equivalent converted UPCR, or 1+ urine protein dipstick), (4) including patients without a follow-up albuminuria assessment (such patients were considered free of albuminuria until censoring), (5) adjusting for the number of antihypertensive prescriptions, (6) as-treated analysis with adjustment for time-updated blood pressure and censoring 30 days after the last DCCB or thiazide refill, and (7) replacing thiazides with beta blockers (excluding sotalol, an antiarrhythmic agent) as the active comparator, since beta blockers are antihypertensive agents not known to affect albuminuria. All analyses were conducted using Stata/SE 17.0, and two-tailed P < 0.05 was considered statistically significant.

RESULTS

Baseline Characteristics

We identified 11,747 patients who initiated a DCCB and 26,758 patients who initiated a thiazide. At baseline, the mean age was 60 (standard deviation [SD] 15) years, 56% were female, 23% had diabetes, and 48% took an ACE inhibitor or ARB (Supplemental Table 3). The mean baseline systolic blood pressure was 143 (SD 21) mm Hg. DCCB initiators were older, had a lower mean eGFR, and were more likely to have a history of diabetes, stroke, heart failure, and coronary heart disease compared with thiazide initiators. Missing covariates are listed in Supplemental Table 4. After weighting, baseline characteristics were similar for DCCB and thiazide initiators, and all standardized mean differences fell within ± 10% (Table 1; Supplemental Figure 2).

Table 1.

Baseline characteristics of patients initiating a dihydropyridine calcium channel blocker or thiazide diuretic after inverse probability of treatment weighting

Baseline characteristic	Dihydropyridine calcium channel blocker	Thiazide
Number, n	11,747	26,757*
Age, mean (SD)	60 (16)	60 (15)
Race
Black	269 (2%)	593 (2%)
Other	74 (1%)	186 (1%)
White	11,406 (97%)	25,979 (97%)
Female	6490 (55%)	14,896 (56%)
Baseline year, median (IQR)	2010 (2007–2013)	2010 (2007–2014)
Body mass index, mean (SD)	33 (8)	33 (8)
Systolic blood pressure, mean (SD), mm Hg	142 (22)	142 (21)
Diastolic blood pressure (SD), mm Hg	82 (13)	82 (13)
Estimated glomerular filtration rate, mean (SD), mL/min/1.73 m2	87 (23)	86 (21)
Albumin to creatinine ratio, median (IQR), mg/g	12 (10–25)	12 (9–25)
Protein to creatinine ratio, median (IQR), mg/g	103 (70–210)	100 (60–160)
Urine protein dipstick, median (IQR)	0 (0–1)	0 (0–0)
No baseline albuminuria measure	5823 (50%)	13,321 (50%)
Diabetes mellitus	2781 (24%)	6208 (23%)
Stroke	689 (6%)	1552 (6%)
Heart failure	738 (6%)	1505 (6%)
History of coronary heart disease	2003 (17%)	4441 (17%)
Smoking
Current	2138 (18%)	4836 (18%)
Former	3953 (34%)	8975 (34%)
Never	5656 (48%)	12,946 (48%)
Concurrent medication use
ACE inhibitor or ARB	5840 (50%)	12,852 (48%)
Beta blocker	3977 (34%)	8966 (34%)
Loop diuretic	1033 (9%)	2202 (8%)
Sodium-glucose transport protein 2 inhibitor	16 (0%)	36 (0%)
Glucagon-like peptide-1 receptor agonist	35 (0%)	74 (0%)
Insulin	761 (6%)	1713 (6%)
Number of antihypertensive agents	2 (1)	2 (1)
Hospitalized within prior 12 months	1369 (12%)	3035 (11%)

^*Quantity differs from number of thiazide users before inverse probability of treatment weighting due to rounding

Data are presented as number (percent of patients) unless otherwise indicated. ACE angiotensin-converting enzyme, ARB angiotensin receptor blocker, IQR interquartile range, SD standard deviation

Overall Outcomes

Over a mean follow-up of 6 (SD 4) years, 1832 patients developed albuminuria (Table 2). The risk of albuminuria was significantly higher with DCCBs compared to thiazides in both the IPTW-only model (hazard ratio [HR], 1.38; 95% confidence interval [CI], 1.24 to 1.53; Fig. 1A, Table 2) and the doubly robust model (HR, 1.29; 95% CI, 1.16 to 1.43). A total of 391 patients developed KFRT over a mean follow-up of 8 (SD 4) years. Compared to thiazides, DCCBs were significantly associated with incident KFRT in the IPTW-only model (HR, 1.30; 95% CI, 1.04 to 1.63; Fig. 1B) but not the doubly robust model (HR, 1.14; 95% CI, 0.91 to 1.44).

Table 2.

Risks of albuminuria and kidney failure with replacement therapy among patients initiating a dihydropyridine calcium channel blocker compared to those initiating a thiazide in an inverse probability of treatment-weighted cohort with stratification by renin-angiotensin system blocker

Outcome	Number of patients	Mean years of follow-up (SD)	Number of events (% of patients)	HR (95% CI)	P value	Without baseline RAS blocker, HR (95% CI)	With baseline RAS blocker, HR (95% CI)	P for interaction
Albuminuria
IPTW only	38,505	6 (4)	1832 (5%)	1.38 (1.24–1.53)	< 0.001	1.65 (1.41–1.92)	1.16 (1.00–1.34)	0.001
Doubly robust				1.29 (1.16–1.43)	< 0.001	1.61 (1.38–1.88)	1.10 (0.95–1.28)	<0.001
Kidney failure with replacement therapy
IPTW only	38,505	8 (4)	391 (1%)	1.30 (1.04–1.63)	0.02	1.77 (1.28–2.46)	0.99 (0.72–1.36)	0.01
Doubly robust				1.14 (0.91–1.44)	0.3	1.66 (1.19–2.31)	0.86 (0.62–1.19)	0.005

RAS blocker includes angiotensin-converting enzyme or angiotensin receptor blocker. CI confidence interval, HR hazard ratio, IPTW inverse probability of treatment weighting, RAS renin-angiotensin system, SD standard deviation

Figure 1.

Cumulative incidence of a albuminuria (P < 0.001) and b kidney failure with replacement therapy (P = 0.02) among patients initiating a dihydropyridine calcium channel blocker or thiazide in an inverse probability of treatment-weighted cohort. P values reflect the IPTW-only Cox proportional hazard models.

Outcomes Stratified by RAS Blockade

DCCBs were associated with an increased risk of albuminuria among patients without (HR, 1.61; 95% CI, 1.38–1.88) but not with a baseline ACE inhibitor or ARB (HR, 1.10; 95% CI, 0.95–1.28; P for interaction < 0.001). Similarly, DCCBs were associated with an increased risk of KFRT among patients without (HR, 1.66; 95% CI, 1.19 to 2.31) but not with a baseline ACE inhibitor or ARB (HR, 0.86; 95% CI, 0.62 to 1.19; P for interaction = 0.005).

Sensitivity Analyses

As observed in the primary analysis, the risk of albuminuria was higher with DCCBs in sensitivity analyses when restricting to patients with baseline eGFR < 60 mL/min/1.73 m² or baseline albuminuria assessment, excluding baseline A2 albuminuria, including patients without a follow-up albuminuria assessment, adjusting for the number of antihypertensive agents, and performing an as-treated analysis (Supplemental Table 5). Consistent with the primary analysis, there was no difference in KFRT risk with DCCBs in the sensitivity analyses except for the as-treated analysis, in which DCCBs were significantly associated with KFRT.

We identified 10,701 patients initiating a DCCB and 32,010 patients initiating a beta blocker (Supplemental Table 6). After weighting, all standardized mean differences were within ± 10% (Supplemental Table 7 and Supplemental Figure 3). In doubly robust models comparing DCCBs to beta blockers, the risk of albuminuria was significantly higher (HR, 1.24; 95% CI, 1.10 to 1.41; Supplemental Table 8), and there was no difference in the risk of KFRT (HR, 1.20; 95% CI, 0.89 to 1.60). We observed a significant interaction between baseline ACE inhibitor or ARB and beta blocker whereby associations were stronger in the absence of an ACE inhibitor or ARB for KFRT (P for interaction < 0.001) but not albuminuria (P for interaction = 0.05).

DISCUSSION

In a target trial emulation using an electronic health records–based cohort of patients without known albuminuria, DCCBs as compared to thiazides were associated with a higher risk of albuminuria but not KFRT. In the absence of an ACE inhibitor or ARB, DCCBs were associated with increased risks of albuminuria and KFRT, and these risks were negated by concomitant ACE inhibitor or ARB use. Overall, our findings suggest that DCCBs without concurrent ACE inhibitors or ARBs may contribute to the development and progression of kidney disease.

Although blood pressure control benefits kidney health, each class of antihypertensive agents offers unique benefits and risks. In the case of DCCBs, the mechanism of action at the level of the kidney may contribute to injury. DCCBs inhibit L-type calcium channels, which are expressed in the afferent arteriole, producing afferent vasodilation.²⁷ Autoregulation of afferent and efferent arteriolar resistance controls renal plasma flow and maintains glomerular filtration. Excessive renal plasma flow can cause glomerular hypertension that then leads to podocyte injury, albuminuria, and glomerular sclerosis.²⁸ Thus, DCCB-induced afferent vasodilation may increase glomerular pressure, posing a risk for glomerular injury. This is seen in animal models: 5/6 nephrectomy rats administered DCCBs developed glomerulosclerosis beyond those receiving non-dihydropyridine calcium channel blockers²⁹ and ACE inhibitors.¹⁴

An increased risk of albuminuria with DCCBs compared to both thiazides and beta blockers is consistent with evidence from clinical trials featuring distinct comparators. A systematic review of 28 randomized clinical trials of calcium channel blockers reported that DCCB use was associated with a 2% increase in proteinuria compared to a 30% decrease in proteinuria with non-dihydropyridine calcium blockers.¹⁶ A crossover trial comparing chlorthalidone and amlodipine in kidney transplant recipients found more albuminuria with amlodipine.³⁰ In AASK, the amlodipine arm was stopped early due to increased rates of proteinuria compared to the metoprolol and ramipril arms.¹⁷

We observed no link between DCCBs and incident KFRT in the overall cohort, and clinical trials examining this outcome have yielded mixed findings. The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial found no difference in the risk of KFRT between amlodipine and chlorthalidone, although there were few kidney events,^18–20 while in AASK, the risk of KFRT was higher for amlodipine compared to metoprolol.¹⁷ Both trials featured an ACE inhibitor arm; thus, our analysis of participants without a baseline ACE inhibitor or ARB may serve as a more direct comparison to these trials. In the absence of RAS blockade, we observed an increased risk of KFRT with DCCBs compared to thiazides and beta blockers.

Our data suggest that when DCCBs are prescribed, they should be accompanied by an ACE inhibitor or ARB to reduce the risk of albuminuria. ACE inhibitors and ARBs vasodilate the efferent arteriole, thereby decreasing glomerular pressure and reducing hyperfiltration, potentially offsetting the adverse effects of DCCBs.³¹ In line with this mechanistic understanding, we found that concomitant RAS blockade mitigated the DCCB-associated risk of albuminuria. Likewise, in a clinical trial, the combination of amlodipine and fosinopril reduced albuminuria more than amlodipine alone, though this may have been due to better blood pressure control with the additional antihypertensive agent.³² Among baseline ACE inhibitor or ARB users, DCCBs were not associated with KFRT. The benefit of combining a DCCB with RAS blockade was also observed in the Avoiding Cardiovascular Events through Combination Therapy in Patients Living with Systolic Hypertension trial, where amlodipine with benazepril resulted in a lower risk of chronic kidney disease progression than hydrochlorothiazide and benazepril.³³ While some uncertainty remains regarding the optimal regimen, administering an ACE inhibitor or ARB with a DCCB may limit kidney injury.

Our findings may have significant public health implications, especially in low-resource settings. Compared to ACE inhibitors, ARBs, and thiazides, which can cause electrolyte abnormalities and creatinine fluctuations, DCCBs can be administered with little laboratory monitoring and therefore are favored where barriers to laboratory monitoring exist.^2,3 However, the elevated risk of albuminuria and KFRT with DCCB initiation in the absence of an ACE inhibitor or ARB may negate this potential benefit. Validation of our findings in diverse economic settings is warranted.

Study strengths include a new user study design, an active comparator with similar baseline characteristics after IPTW, and the use of a large community cohort with comprehensive prescription records and long-term clinical follow-up. We also acknowledge some limitations. First, low rates of UACR testing necessitated the conversion of UPCR and urine protein dipstick to UACR. Second, it is possible that the indications for DCCBs and thiazides differed in ways that we were unable to capture with IPTW. However, similar results were observed when beta blockers were considered the active comparator. Third, we were unable to gauge medication adherence, and medication assignment reflects prescription of medication and not actual use. Fourth, it is possible that not all incident KFRT incidences were captured. Fifth, the study population included few non-White patients, which may limit the generalizability of our findings.

In summary, we found that DCCBs were associated with the development of albuminuria and, in the absence of RAS blockade, KFRT. These risks were reduced with concurrent ACE inhibitor or ARB use. This highlights a possible medication-induced instance of glomerular hyperfiltration and suggests that coupling DCCB therapy with an ACE inhibitor or ARB may limit adverse kidney outcomes.

Supplementary Information

Below is the link to the electronic supplementary material.

Data Availability:

The datasets analyzed during the current study are not publicly available because they contain protected health information.

Declarations:

Conflict of Interest:

Dr. Blum reports grants or contracts: National Kidney Foundation and Johns Hopkins University; meeting or travel support: American Society of Nephrology. Dr. Chang reports grants or contracts: Bayer, Novartis, Novo Nordisk; consulting fees: Novartis, Reata, Amgen; data safety monitoring board or advisory board: Occurrence of Dyskalemia with Treatment for Hypertension Study. Dr. Inker reports grants or contracts: National Kidney Foundation, National Institutes of Health, Omeros, Dialysis Clinics Inc., Reata; consulting fees: Tufts Medical Center; data safety monitoring board or advisor board: Dimerix, TMC, Tricida, Healthlogistics. Dr. Chen reports grants or contracts: National Institutes of Health, Yale University; payment or honoraria: American Society of Nephrology; meeting or travel support: CKD Biomarkers Consortium Meeting. Dr. Grams reports grants or contracts: National Institutes of Health and National Kidney Foundation; payment or honoraria: University of Pennsylvania, Columbia University Medical Center, American Society of Nephrology; meeting/travel support: Kidney Disease Improving Global Outcomes, European Renal Association, University of Pennsylvania, and Korean Society of Nephrology; leadership or fiduciary role: American Society of Nephrology, Kidney Disease: Improving Global Outcomes, National Kidney Foundation, United States Renal Data System, American Society of Clinical Investigation, Clinical Journal of American Society of Nephrology, American Journal of Kidney Diseases, and Journal of American Society of Nephrology. The remaining authors declare that they have no relevant financial interests.