Influence of Prediabetes on the Effects of Intensive Systolic Blood Pressure Control on Kidney Events

Abstract

BACKGROUND

More than one-third of US adults have prediabetes, which is typically accompanied by hypertension.

METHODS

We examined whether prediabetes modified the effects of intensive systolic blood pressure (SBP) lowering on the incidence of chronic kidney disease (CKD) and acute kidney injury (AKI) events in a post-hoc analysis of the Systolic Blood Pressure Intervention Trial (SPRINT). Diabetes was a SPRINT exclusion criterion. We defined normoglycemia and prediabetes as fasting plasma glucose <100 mg/dl and ≥100 mg/dl, respectively.

RESULTS

Of the 9,323 participants included in this analysis, 3,898 (41.8%) had prediabetes and the rest (5,425) had normoglycemia. In participants with baseline estimated glomerular filtration rate (eGFR) ≥60 ml/min/1.73 m², incident CKD was defined as a ≥30% decline in eGFR to below 60 ml/min/1.73 m² with repeat confirmation. AKI events were identified clinically. In the non-CKD participants (n = 6,678), there were 164 incident CKD events. The hazard ratios (HRs) for incident CKD for intensive SBP goal (<120 mm Hg) vs. standard SBP goal (<140 mm Hg) in the normoglycemia (HR: 3.25, 95% CI: 2.03, 5.19) and prediabetes (HR: 3.90, 95% CI: 2.17, 7.02) groups were similar (interaction P value 0.64). In the entire analytic cohort (N = 9,323), there were 310 AKI events. AKI HRs for intensive vs. standard SBP in the normoglycemia (HR: 1.59, 95% CI: 1.17, 2.15) and prediabetes (HR: 1.74, 95% CI: 1.22, 2.48) groups were also similar (interaction P value 0.71).

CONCLUSIONS

Prediabetes was highly prevalent, but there was no evidence that prediabetes modified the effects of SPRINT intervention on kidney events.

CLINICAL TRIALS REGISTRATION

NCT01206062

About 1.3 billion adults worldwide are considered to have hypertension based on a systolic blood pressure (SBP) ≥140 mm Hg, a diastolic BP (DBP) ≥90 mm Hg or use of antihypertensive medication.¹ Using the same definition, nearly two-thirds of US adults aged 60 or older have hypertension.² High blood pressure (BP) is a well-established risk factor for stroke, heart failure, sudden death, end-stage renal disease, and death.^3–5 Indeed, the Global Burden of Disease Study identified elevated BP as the leading risk factor, among 67 studied, for death and disability-adjusted life-years lost during 2010.⁶ Furthermore, it was estimated that the total annual cost of hypertension in the United States during 2015 was nearly $110 billion.⁷ Thus, hypertension is common and results in substantial morbidity, mortality, and healthcare costs. From a public health perspective, effective and safe strategies to treat hypertension are critically important.

The recent American College of Cardiology/American Heart Association guideline recommends an SBP goal of less than 130 mm Hg in persons with and without diabetes.⁸ The Systolic Blood Pressure Intervention Trial (SPRINT)^9,10 was a large, randomized controlled trial, sponsored by the National Institutes of Health, which compared the effects of intensive (SBP target <120 mm Hg) and standard (SBP target <140 mm Hg) SBP treatment targets on cardiovascular disease (CVD) events in persons without diabetes. In SPRINT, the BP intervention was stopped early because intensive SBP lowering resulted in a substantial reduction in both the primary composite CVD endpoint and all-cause mortality.¹⁰

Nonetheless, concerns about potential adverse effects of intensive SBP lowering remain unresolved. In SPRINT participants with baseline estimated glomerular filtration rate (eGFR) ≥60 ml/min/1.73 m², the intensive SBP arm had a 3.5-fold higher hazard of incident chronic kidney disease (CKD) (defined as a ≥30% decline in eGFR to <60 ml/min/1.73 m², confirmed on repeat testing). In the entire SPRINT cohort, intensive SBP lowering resulted in a 71% increased risk of acute kidney injury (AKI; defined clinically).¹⁰

The Action to Control Cardiovascular Risk in Diabetes (ACCORD) BP trial tested the effects of an SBP intervention similar to the SPRINT intervention in adults with type 2 diabetes mellitus.¹¹ We recently reported¹² that, despite a clinically similar reduction in SBP in both trials, the 3-year absolute risk differences for incident CKD (as defined previously) between the intensive and standard SBP groups in ACCORD BP and SPRINT were 5.9% (95% CI: 4.3%, 7.5%) and 2.5% (95% CI: 1.8%, 3.2%), respectively, with an interaction P value < 0.001. Thus, with intensive SBP lowering, the risk of incident CKD was higher in persons with type 2 diabetes mellitus.

This raises the question of whether prediabetes, as defined by an elevated fasting blood sugar, increases the risk of kidney events during intensive SBP lowering. This question is highly relevant because more than a third of US adults have prediabetes¹³ and because some studies have identified prediabetes as a risk factor for kidney disease.^14–16

Herein, we report the incidence of kidney events in SPRINT participants with and without prediabetes, and a series of analyses examining whether prediabetes modifies the effects of intensive BP lowering on kidney events.

METHODS

This analysis is based on a limited access SPRINT dataset obtained from the National Heart, Lung and Blood Institute Biologic Specimen and Data Repository Information Coordinating Center.¹⁷ SPRINT was a randomized, controlled open-label trial sponsored by the National Institutes of Health to compare an intensive SBP goal of <120 mm Hg (n = 4,678) with a standard SBP goal of <140 mm Hg (n = 4,683)¹⁰ in 102 clinical centers across the United States and Puerto Rico. The design, details, and primary outcomes of the trial have been published.^9,10 In brief, participants had to be older than 50 years, have a SBP ≥130–180 mm Hg and, be at high risk for CVD events (defined by clinical or subclinical CVD, CKD, 10-year risk for CVD events ≥15% as defined by the Framingham risk score, or 75+ years of age). Exclusions included previous history of stroke, dementia, eGFR <20 ml/min/1.73 m², or diabetes. Participants attended monthly visits for the first 3 months, followed by a visit every 3 months. During visits, 3 BP measurements were obtained using an Omron Model 907 device (Omron Healthcare) while the patient was seated after 5 minutes of rest.¹⁸ Lifestyle modification (e.g., regular exercise, reduction in dietary salt intake, attention to levels of stress and quality of sleep) was often included in recommendations to achieve target SBP. Medications and dosages were prescribed and titrated from a menu of widely used, commercially available drugs within nearly all drug classes, and used at the discretion of the investigators to meet target SBP (<120 mm Hg for intensive SBP lowering, 135–139 mm Hg for standard SBP lowering). In cases where participants in the standard group had SBP below target, dosages were decreased and/or medications were discontinued.

Definition of prediabetes

Plasma glucose concentrations were measured at the SPRINT Central Laboratory in blood samples obtained after overnight fasting at the randomization visit. We defined normoglycemia as a fasting blood glucose level < 100 mg/dl and prediabetes as ≥ 100 mg/dl. There were 333 participants in the prediabetes group who had a baseline fasting blood glucose ≥ 125 mg/dl. In sensitivity analyses, we excluded them from our analyses.

Definitions of incident CKD and AKI

The SPRINT protocol pre-specified incident CKD outcome (based on the 4-variable modification of diet in renal disease (MDRD) study equation to estimate GFR) in participants without CKD at their baseline visit (MDRD eGFR ≥ 60 ml/min/1.73 m2), as a >30% decrease in MDRD eGFR from the baseline value with an end value of <60 ml/min/1.73 m², confirmed at the next available SPRINT blood draw.¹⁹ We previously reported that the effects of intensive SBP lowering on incident CKD, defined using either the MDRD equation or the CKD-EPI equation, were similar.¹⁹

AKI was not a pre-specified endpoint in SPRINT. However, occurrences of AKI or “acute renal failure” were noted as adverse events if they were identified during a hospital admission or occurred during a hospitalization and were reported in the hospital discharge summary as a primary or main secondary diagnosis.¹⁰

Statistical methods

We summarized baseline characteristics of participants by CKD status and glycemia groups using means and standard deviations or medians and interquartile ranges for numeric variables and proportions for categorical variables.

We used intention-to-treat principle for all randomized comparisons between the intensive and standard SBP interventions in the normoglycemic and prediabetes groups (Supplementary Figure 1). We censored follow-up time for incident CKD at the time of the final serum creatinine measurement and follow-up times for AKI events at the last study visit. We used Cox proportional hazards regression analyses to provide estimates of the hazard ratios for incident CKD in the intensive and standard SBP group comparisons. We compared the effects of the intensive SBP interventions, expressed as relative reduction in the hazards between the intensive and standard SBP intervention between the treatment groups, by comparing the difference between the estimated log-transformed hazard ratios to the standard error of this difference. We tested Schoenfeld residuals and there was no evidence of nonproportionality.

Kaplan–Meier curves depicted the absolute cumulative risk of each outcome by SBP intervention within each group. We estimated the absolute risk reductions in these events at 3 years between the intensive and the standard BP groups using a generalized linear model with a robust variance estimate and pseudo-survival probabilities as the outcome.^20,21 We compared the absolute risk reductions between the normoglycemic and prediabetes groups by contrasting the difference in the estimated risk reductions in the 2 subgroups to the standard error of this difference. We consider 2-tailed P values <0.05 as indicating statistical significance. We conducted all analyses using SAS 9.4 (Cary, NC).

RESULTS

Of the 9,361 participants, 38 had missing baseline fasting plasma glucose concentrations and were excluded from this analysis (Supplementary Figure 1). The final cohort for the incident CKD and AKI outcomes included 6,678 participants without CKD at baseline and 9,323 participants with or without CKD at baseline, respectively (Supplementary Figure 1).

Of the 9,323 participants included in this analysis, 3,898 (41.8%) had prediabetes and the rest (5,425) had normoglycemia. Baseline characteristics for the study by glycemia and CKD status are summarized in Table 1. Compared to the normoglycemia subgroups, the prediabetes subgroups in the non-CKD and CKD populations had a lower percentage of women and African Americans, lower high-density lipoprotein (HDL) cholesterol, and higher body mass index and serum triglycerides.

Table 1.

Baseline characteristics by CKD and glycemia status in the analytic cohort (N = 9,323)

Baseline characteristics	Non-CKD (eGFR ≥ 60)		CKD (eGFR < 60)
	Normoglycemia FPG < 100 n = 3,849	Prediabetes FPG ≥ 100 n = 2,892	Normoglycemia FPG < 100 n = 1,576	Prediabetes FPG ≥ 100 n = 1,069
Fasting plasma glucose (mg/dl)	90.6 ± 6.4	110.5 ± 12.7	90.5 ± 6.6	109.7 ± 12.0
MDRD eGFR (ml/min/1.73 m2)c	81.0 ± 15.7	81.5 ± 15.3	47.4 ± 9.8	48.5 ± 9.0
Age (years)a	66.6 ± 9.3	65.9 ± 8.7	71.7 ± 9.5	72.2 ± 8.9
Female sex (%)b,d	38	28	45	33
African American (%)b,d	37	30	27	20
Intensive SBP control (%)	50	50	51	49
History of CVD (%)	18	19	23	26
No history of smoking (%)a	45	42	47	44
Aspirin use (%)a,c	48	50	53	60
Statin use (%)b,d	38	44	48	57
Antihypertensive medications (N)b,d	1.7 ± 1.0	1.8 ± 1.1	2.0 ± 1.0	2.2 ± 1.0
Systolic blood pressure (mm Hg)b	140 ± 16	139 ± 15	140 ± 16	138 ± 16
Diastolic blood pressure (mm Hg)d	80 ± 12	79 ± 12	76 ± 12	74 ± 12
Body mass index (kg/m2)b,d	29 ± 6	31 ± 6	29 ± 6	30 ± 6
Fasting total cholesterol (mg/dl)b	193 ± 41	190 ± 41	187 ± 41	184 ± 40
Fasting HDL cholesterol (mg/dl)b,d	55 ± 15	50 ± 13	54 ± 15	50 ± 14
Fasting triglycerides (mg/dl)b,d	97 (72, 136)	114 (83, 164)	105 (76, 141)	123 (87, 171)
Urine Alb/CR ratio (mg/g Cr)	9 (5, 17)	9 (5, 17)	13 (6, 46)	13 (6, 41)

CKD, chronic kidney disease; CVD, cardiovascular disease; eGFR, estimated glomerular filtration rate; FPG, fasting plasma glucose; MDRD, modification of diet in renal disease. Data are presented as mean ± SD or median (IQR) for continuous measures, and % for categorical measures.

^a P value < 0.05.

^b P value < 0.001 for differences between FPG groups in non-CKD.

^c P value < 0.05.

^d P value < 0.001 for differences between FPG groups in CKD.

Incidence of CKD in the non-CKD population

Overall, there were 164 incident CKD events during 21,155 person-years of follow-up (0.78% per patient-year). The incidence of CKD was higher in the intensive SBP arm compared with the standard SBP arm within both the normoglycemia as well as the prediabetes groups (Figure 1). Absolute risk differences in intensive minus standard SBP arms for the cumulative incidence of CKD at 3 years were 2.5% (95% CI: 1.5%, 3.5%) in the normoglycemia group and 2.5% (95% CI: 1.5%, 3.6%) in the prediabetes group with an interaction P value of 0.91 (Figure 1). Hazard ratios for intensive vs. standard SBP arms in the normoglycemia group (HR: 3.25, 95% CI: 2.03, 5.19) and prediabetes group (HR: 3.90, 95% CI: 2.17, 7.02) were also similar (interaction P value 0.64).

Figure 1.

Incidence of CKD* in the non-CKD Subgroup. (a) Kaplan–Meier failure plots. (b) Absolute risk differences at 3 years. (c) Hazard ratio for entire follow-up. CKD, chronic kidney disease (eGFR < 60 ml/min/1.73 m²); FPG, fasting plasma glucose; non-CKD, no chronic kidney disease (eGFR ≥ 60 ml/min/1.73 m²); Std, standard; Int, intensive.

Incidence of AKI in the entire analytic cohort

There were 310 AKI events during 28,898 person-years of follow-up in the entire cohort. Absolute risk differences in the intensive minus standard SBP arms for the cumulative incidence of AKI during 3 years of follow-up were 1.1% (95% CI: 0.2%, 2.1%) in the normoglycemia group and 1.8% (95% CI: 0.7%, 3.0%) in the prediabetes group, with an interaction P value of 0.37 (Figure 2). The intensive vs. standard SBP hazard ratios for AKI were also similar in the normoglycemia (HR: 1.59, 95% CI: 1.17, 2.15) and prediabetes (HR: 1.74, 95% CI: 1.22, 2.48) groups (interaction P value 0.71).

Figure 2.

Incidence of AKI* in the entire cohort. (a) Kaplan–Meier failure plots. (b) Absolute risk differences at 3 years. (c) Hazard ratio for entire follow-up. AKI, acute kidney injury; FPG, fasting plasma glucose; Int, intensive; Std, standard.

The results were similar in sensitivity analyses where prediabetes was defined as a fasting plasma glucose 100 to < 125 mg/dl (Supplementary Table 1).

DISCUSSION

Results of this analysis show that in a high CVD risk, nondiabetic, cohort of adults with hypertension, prediabetes defined by impaired fasting glucose did not modify the effects of intensive SBP lowering on incident CKD or AKI.

We previously reported that for a similar level of SBP lowering, the absolute risk increase for incident CKD was higher in ACCORD BP participants with type 2 diabetes compared to SPRINT participants without diabetes.¹² We also reported that the early eGFR decline over the first 12 months was steeper in ACCORD BP compared to SPRINT participants assigned to intensive SBP treatment (a mean change of –11.6 vs. –4.8 ml/min/1.73 m2) or to standard treatment (–5.5 vs. –0.4 ml/min/1.73 m2).¹² The reason(s) for the increased risk of incident CKD with intensive SBP lowering with type 2 diabetes but not prediabetes is unclear. One possibility is that autonomic dysfunction might be more common with established type 2 diabetes mellitus than with prediabetes, resulting in impaired renal autoregulation that increases susceptibility for incident CKD with intensive SBP lowering. Although the antihypertensive drugs used in ACCORD BP and SPRINT were similar, they were not identical; it is possible the regimens used to achieve the intensive BP goal were different enough to result in a difference in CKD incidence. There were differences in BP measurement protocols in SPRINT and ACCORD BP, which have been considered to explain potential differences in SPRINT and ACCORD BP findings.²² However, attended vs. unattended BP measurements did not modify the beneficial effects of SPRINT intervention on CVD outcome.¹⁸ Furthermore, it should be noted that the differences in CVD outcome and all-cause death in SPRINT and ACCORD BP is probably because of the interaction of intensive SBP lowering with intensive glycemic control in ACCORD BP; the beneficial effects of intensive SBP lowering on CVD events were similar in SPRINT and ACCORD BP standard glycemia arm.²³

The clinical significance of prediabetes by itself is uncertain. In the Australian Diabetes, Obesity, and Lifestyle Study,²⁴ the risk of all-cause death (HR: 1.6, 95% CI: 1.0, 2.4) and cardiovascular mortality (HR: 2.5, 95% CI: 1.2, 5.1) after a median follow-up of 5.2 years in 10,428 participants were higher in participants with impaired fasting glucose compared to those with normoglycemia. Fasting plasma glucose concentrations exceeding 100 mg/dl were associated with a higher risk of death in a meta-analysis of 820,900 adults in 97 prospective studies.²⁵ However, in another meta-analysis including 102 studies,²⁶ persons with fasting glucose concentrations of 100 to 124 mg/dl had only a modest elevated risk for coronary heart disease (HR: 1.11, 95% CI: 1.04, 1.18) compared with persons with baseline fasting glucose concentrations <100 mg/dl. In persons without a history of diabetes, information about fasting glucose concentration did not significantly improve metrics of vascular disease prediction when added to information about several conventional risk factors.²⁶

The role of prediabetes in the genesis of nephropathy is also uncertain. Plantinga et al. noted that 17.7% of adults with prediabetes had CKD.¹⁶ In a meta-analysis of eleven studies (N = 30,146),²⁷ impaired fasting glucose was significantly associated with the development of eGFR <60 ml/min per 1.73 m² (OR: 1.14, 95% CI: 1.03, 1.26). In contrast, prediabetes was not associated with incident CKD in a multivariable logistic regression analysis of the Framingham Heart Study offspring who were followed for an average of more than 7 years (OR: 0.98, 95% CI: 0.67, 1.45).²⁸ Similarly, in the Atherosclerosis Risk in Communities cohort, impaired fasting glucose was not significantly associated with higher risk of incident CKD (OR: 1.11, 95% CI: 0.87, 1.40) during 9 years of follow-up.²⁹ Thus, the causal nature of the relation between prediabetes and kidney disease remains unclear. Hence, it is not surprising that while type 2 diabetes mellitus appeared to augment the risk of incident CKD with intensive SBP lowering in a previous study, we did not observe an interaction between prediabetes and intensive SBP lowering for incident CKD and AKI in this analysis.

Our findings have important public health implications. A National Diabetes Statistics Report of the Centers of Disease Control and Prevention suggests that 48% persons older than 65 have prediabetes and that nearly 84 million adults in the United States have prediabetes.³⁰ Hypertension and prediabetes commonly coexist.³¹ An earlier report from SPRINT noted that prediabetes did not modify the beneficial effects of intensive SBP lowering on cardiovascular events and all-cause mortality..³² In a recent analysis of ACCORD BP data,²³ we noted that intensive SBP lowering appeared to reduce the risk of CVD events and all-cause mortality in the standard glycemia arm, but not in the intensive glycemia arm of the ACCORD BP trial. Thus, intensive SBP lowering lowers cardiovascular events and all-cause mortality in persons with prediabetes or type 2 diabetes mellitus on standard glycemic control.

There are limitations to this study. Prediabetes can be defined based on impaired fasting glucose levels, impaired glucose tolerance after glucose load or hemoglobin A1C. In this study, we used the American Diabetes Association definition of prediabetes based on fasting glucose levels of 100–125 mg/dl.³³ However, several studies have shown a poor correlation between the HbA1c, impaired fasting glucose and impaired glucose tolerance definitions for prediabetes.^34–37 It is unclear whether the effects of intensive SBP lowering on kidney events would have been augmented in prediabetes defined by levels of HbA1c or impaired glucose tolerance. Nonetheless, impaired fasting glucose is a precursor of type 2 diabetes.^38,39 Finally, although nearly 42% of the participants in this large BP trial had prediabetes, the power to detect effect modification in subgroup analyses is low even in large clinical trials. Hence, a modest interaction between prediabetes and intensive SBP lowering on kidney events cannot be definitively ruled out based on the current analysis.

In summary, we noted a high prevalence of prediabetes in a high cardiovascular risk cohort with high BP. However, the presence of prediabetes did not appear to augment the risk for kidney events during intensive SBP lowering. Our findings support the strategy of intensive SBP lowering in adults with hypertension who have prediabetes and a high risk of CVD.

DISCLOSURES

There are no conflicts of interest for any of the authors. All authors are SPRINT investigators.