Prognostic Role of Ambulatory Blood Pressure Measurement in Patients With Nondialysis Chronic Kidney Disease

Background  Ambulatory blood pressure (BP) measurement allows a better risk stratification in essential hypertension compared with office blood pressure measurement, but its prognostic role in nondialysis chronic kidney disease has been poorly investigated.

Methods  The prognostic role of daytime and nighttime systolic BP (SBP) and diastolic BP (DBP) in comparison with office measurements was evaluated in 436 consecutive patients with chronic kidney disease. Primary end points were time to renal death (end-stage renal disease or death) and time to fatal and nonfatal cardiovascular events. Quintiles of BP were used to classify patients.

Results  The mean (SD) age of the patients was 65.1 (13.6) years, and the glomerular filtration rate was 42.9 (19.7) mL/min/1.73 m²; 41.7% of the participants were women, 36.5% had diabetes, and 30.5% had cardiovascular disease. Office-measured SBP/DBP values were 146 (19)/82 (12) mm Hg; daytime SBP/DBP was 131 (17)/75 (11) mm Hg, and nighttime SBP/DBP was 122 (20)/66 (10) mm Hg. During follow-up (median, 4.2 years), 155 and 103 patients reached the renal and cardiovascular end points, respectively. Compared with a daytime SBP of 126 to 135 mm Hg, patients with an SBP of 136 to 146 mm Hg and those with an SBP higher than 146 mm Hg had an increased adjusted risk of the cardiovascular end point (hazard ratio [HR], 2.23; 95% confidence interval [CI], 1.13-4.41 and 3.07; 1.54-6.09) and renal death (1.72; 1.02-2.89 and 1.85; 1.11-3.08). Nighttime SBPs of 125 to 137 mm Hg and higher than 137 mm Hg also increased the risk of the cardiovascular end point (HR, 2.52; 95% CI, 1.11-5.71 and 4.00; 1.77-9.02) and renal end point (1.87; 1.03-3.43 and 2.54; 1.41-4.57) with respect to the reference SBP value of 106-114 mm Hg. Office measurement of BP did not predict the risk of the renal or cardiovascular end point. Patients who were nondippers and those who were reverse dippers had a greater risk of both end points.

Conclusion  In chronic kidney disease, ambulatory BP measurement and, in particular, nighttime BP measurement, allows more accurate prediction of renal and cardiovascular risk; office measurement of BP does not predict any outcome.

Guidelines^1-3 have repeatedly highlighted the importance of lowering blood pressure (BP) to less than 130/80 mm Hg in all patients with chronic kidney disease (CKD) not requiring dialysis. However, this recommendation is mainly derived from observational data or post hoc analysis^4-7 of large trials showing that achieving such a BP goal improves renal and cardiovascular outcomes, with a greater benefit in patients with proteinuria. However, recent studies have challenged the guidelines' recommendation,^8,9 contributing to the debate on the BP goal in CKD.^10-12

The controversial results of studies on the prognostic role of office BP measurement in patients with CKD might relate, to some extent, to the limited ability of office BP readings to stratify the global risk of this population. This can be ascribed to the high prevalence of white coat hypertension in CKD^13-16 and to the fact that a patient's nocturnal BP profile, without the pressor effect of physical activity, emotional stress, and environmental factors that are usually present during the day, is more representative of the true BP status¹⁷ and a stronger predictor of cardiovascular outcomes.^18,19 Furthermore, an altered circadian pattern of BP has been associated with higher levels of proteinuria.^20,21 Ambulatory blood pressure monitoring (ABPM) constitutes the only tool to identify simultaneously white coat hypertension and nighttime hypertension.

Solid evidence has accumulated over the past decade of the substantial benefits of ABPM measures to use for risk stratification of the general population,^22-24 hypertensive patients,^25,26 and elderly individuals.²⁷ Conversely, very few data are available in patients with CKD.^28,29 In fact, in this high-risk population, Agarwal and Andersen demonstrated that ABPM is superior to clinic BP measurement in predicting end-stage renal disease (ESRD) and/or all-cause mortality²⁸ and cardiovascular events.²⁹ However, the predictive value diminished after adjustment for the main recognized factors of cardiorenal risk; whether this was because of the major role of other risk factors or the inadequate power of those studies remains to be elucidated. In addition, the finding that the predictive role of ABPM was greater than that of office BP measurement was limited to a single cohort of 217 US veterans, mostly men, with CKD.

Therefore, whether the relationship between ABPM and outcome holds true in a heterogeneous cohort that is more representative of the general CKD population needs to be addressed. Indeed, women have a slower progression of CKD and a lower incidence of mortality compared with men.^30-33 In addition, age-adjusted cardiovascular mortality rates associated with hypertension, diabetes mellitus, and hypercholesterolemia are significantly higher in the United States than in Japan and Mediterranean countries.³⁴

This multicenter, prospective cohort study was designed to compare the prognostic role of ABPM and office BP measurement in fatal and nonfatal cardiovascular events and renal death in a heterogeneous cohort of 436 patients with CKD regularly monitored in 4 Italian nephrology clinics.

In this multicenter, prospective cohort study, consecutive patients were recruited in the nephrology clinics of 4 nephrology units from January 1, 2003, to December 31, 2005. In the 4 participating institutions, ABPM is performed in all patients with hypertension (systolic BP [SBP] ≥130 mm Hg and/or diastolic BP [DBP] ≥80 mm Hg or patients are receiving antihypertensive treatment). Hypertensive patients were included in this study if they had been monitored for at least 6 months in the nephrology clinics and if they had CKD stages 2 to 5, that is, either with an estimated glomerular filtration rate (GFR) by the 4-variable Modified Diet in Renal Disease equation GFR < 60 mL/min/1.73 m² or a GFR of 60 to 90 mL/min/1.73 m² plus proteinuria of more than 0.3 g/24 h, in 2 consecutive visits with an interval 3 months or more. Exclusion criteria were true normotension (BP <130/80 mm Hg without antihypertensive therapy), changes in GFR of more than 30% in the previous 3 months, changes in antihypertensive therapy 2 weeks before ABPM, atrial fibrillation, dialysis treatment or renal transplantation, and inadequate ABPM (No. of SBP/DBP recordings, <14 and <7 during the day and the night, respectively³⁵). Data on medical history, laboratory test results, and current therapy were collected at the initial visit. Institutional review boards of the participating centers approved the protocol, and informed consent was obtained from patients before study enrollment.

During the physician's visit (8 to 11 AM), office BP was measured according to the recommendations of the European Society of Hypertension.³⁵ Systolic and diastolic BP (Korotkoff phase 1 and phase 5) were measured with the patient seated 3 times at 5-minute intervals. The office BP measurement values reported herein are the mean of the 6 values recorded in the 2 consecutive days in which the ABPM device was installed and removed. Measurements were obtained by the same physician, who was not aware of the results of the ABPM recordings.

Similar ABPM protocols and monitors (Spacelabs 90207; Spacelabs Healthcare, Issaquah, Washington) were used in all centers. Specifically, cuff size was chosen according to the patient's arm circumference and fixed to the nondominant arm. Three BP readings were taken in the morning (8-11 AM) concomitantly with sphygmomanometric measurements to ensure a difference of less than 5 mm Hg between the 2 sets of values. The monitor recorded BP every 15 minutes between 7 AM and 11 PM and every 30 minutes between 11 PM and 7 AM. Daytime and nighttime periods were derived from the diaries recorded by the patients during monitoring. The ABPM was always obtained on a workday and while the patient was receiving regular antihypertensive therapy. Patients had no access to the ABPM values. The BP was considered at target when daytime and nighttime values were less than 135/85 mm Hg and less than 120/70 mm Hg, respectively.³⁵ Dipping status was based on the night to day ratio of mean ambulatory BP. We stratified patients as extreme dippers, dippers, nondippers, and reverse dippers when the night to day ratio was less than 0.80, 0.80 to less than 0.90, 0.90 to 1.00, and greater than 1.00, respectively.¹⁹

In each center, patients were always seen by the same nephrologist. All patients were instructed, with use of a personalized written regimen, to restrict dietary salt (<6 g/d of sodium chloride) and protein (≤0.8 g/kg of body weight per day). Antihypertensive medications were administered to achieve office BP values less than 130/80 mm Hg, according to current Kidney Disease Outcomes Quality Initiative guidelines.² Drugs were distributed from 8 AM to 10 PM; doses of furosemide of 50 mg/d or more were divided in 2 administrations (8 AM and 8 PM).

Renal outcomes were defined a priori and included a composite end point of ESRD or death, whichever occurred first. The end point of ESRD was reached on the day of the first dialysis session. Death certificates and autopsy reports were used to establish the underlying cause of death and to adjudicate cardiovascular deaths based on the ninth revision of the International Classification of Diseases. Cardiovascular outcomes included a composite end point of fatal and nonfatal cardiovascular events that required hospitalization (myocardial infarction, congestive heart failure, stroke, revascularization, peripheral vascular disease, and nontraumatic amputation), whichever occurred first. We collected hospital records to identify the diagnosis according to the American College of Cardiology and the European Society of Cardiology criteria.^36,37 Patients were followed up until September 30, 2010, or death or ESRD and censored on the date of the last nephrology clinic visit.

Continuous variables are expressed as mean (SD) or as median (interquartile range), according to their distribution. A multivariable Cox proportional hazards regression model, stratified by center, was used to estimate hazard ratio (HR) and corresponding 95% confidence interval (CI). All Cox models were adjusted for baseline covariates known to be determinants of renal and cardiovascular outcome (age; sex; body mass index, calculated as weight in kilograms divided by height in meters squared; diabetes mellitus; cardiovascular disease; hemoglobin level; proteinuria; and GFR). Office BP measurement and ABPM were analyzed, categorizing patients by means of quintile values. We tested the association of ABPM with risk of renal and cardiovascular end points by stratifying patients according to whether the daytime and nighttime BP targets were achieved. This analysis was performed with each BP target included in the Cox regression model separately or simultaneously (daytime and nighttime). Nested models were compared using the likelihood ratio test. The nonlinear relationship of time and BP was evaluated using restricted cubic splines. Specifically, we placed 5 knots at 0.050, 0.275, 0.500, 0.725, and 0.950 quantiles of the x variable (SBP or DBP).³⁸ The role of nighttime hypertension was further assessed using the defined categories of dipping status; in this Cox model, we further adjusted for 24-hour mean BP, according to the suggestion from the IDACO (International Database on Ambulatory Blood Pressure Monitoring in Relation to Cardiovascular Outcomes) investigators.¹⁹

Data were analyzed using commercially available software (SPSS version 12.0; SPSS Inc, Chicago, Illinois; and R version 2.9.2; R Foundation for Statistical Computing, Vienna, Austria). All statistical tests were 2-tailed, and P < .05 was considered significant.

We included 459 of 530 eligible patients, all of whom were white. Reasons for exclusion were inadequate ABPM recordings (n = 35), change of antihypertensive therapy 2 weeks before the study (n = 24), atrial fibrillation (n = 8), and a GFR change of more than 30% (n = 4). Twenty-three patients were subsequently excluded because they were lost to follow-up after their initial visit. Demographic, clinical, and treatment information on the 436 patients included in the cohort are presented in Table 1 and Table 2.

The median duration of follow-up was 4.2 years (interquartile range, 3.0-5.1 years). During this period, CKD in 86 patients progressed to ESRD and 69 patients died. We recorded 63 nonfatal cardiovascular events and 52 cardiovascular deaths (12 occurring after a first nonfatal cardiovascular event); in particular, we registered 62 acute myocardial infarctions (34 fatal), 22 strokes (12 fatal), 16 peripheral vascular accidents (2 fatal), and 15 acute heart failures (4 fatal). Crude event rates for the composite renal end point (ESRD or death) was 9.01 per 100 patient-years and 6.56 per 100 patient-years for the composite cardiovascular end point (occurrence of fatal/nonfatal cardiovascular episodes).

Table 3 and Table 4 present the results of the predictive role of BP for cardiovascular and renal end points. Patients with daytime SBP in the range of 136 to 146 mm Hg and higher than 146 mm Hg had a 2.23- and 3.07-fold greater risk, respectively, of the cardiovascular end point; these 2 quintiles were also associated with a greater risk of renal death. Similarly, patients in the highest quintile of daytime DBP (Table 4) had a significantly greater risk for both end points. A daytime DBP in the range of 78 to 84 mm Hg was predictive of renal events but not of cardiovascular events. Nighttime readings of SBP higher than 124 mm Hg predicted both end points (Table 3), and DBP values higher than 70 and higher than 75 mm Hg predicted the cardiovascular end point and renal death, respectively (Table 4). Hazard ratios for renal events and cardiovascular events of quintiles of ambulatory SBP/DBP remained unchanged after adjustment for office measurement of SBP/DBP (data not shown). In contrast, office BP measurement (either SBP or DBP) did not predict cardiovascular or renal events.

The adjusted risk associated with ABPM levels is shown in Figure 1 and Figure 2. The risk of renal death increased in a linear fashion with the increment in daytime and nighttime SBP/DBP (Figure 1). A similar pattern was observed for the cardiovascular end point, with the exception of a relationship with the daytime SBP that appeared J-shaped (Figure 2).

Two hundred thirty-nine patients (54.8%) achieved only the daytime BP target (<135/85 mm Hg) and 189 patients (43.3%) achieved only the nighttime target (<120/70 mm Hg); 166 patients (38.1%) achieved both targets and 174 patients (39.9%) achieved neither target. Patients who did not achieve either target had the greatest risk of cardiovascular (HR, 1.99) and renal (HR, 2.65) end points (Figure 3). Within the group of patients reaching only 1 of the target levels, those achieving only the nighttime target value did not have a greater risk of the renal and cardiovascular end points; conversely, patients achieving exclusively the daytime target had a greater risk of the renal end point. The greater predictive role of the nighttime target BP was confirmed by the likelihood ratio test; indeed, adding the nighttime to the daytime target increased the model fit for renal death (P = .003) but not for the cardiovascular end point (P = .11); the model fit for both end points did not improve when the daytime target was added to the nighttime target (P = .07 for both).

An office BP measurement target lower than 130/80 mm Hg did not influence the outcomes (for renal death, HR 0.91; 95% CI, 0.57-1.44; P = .68 and for fatal/nonfatal cardiovascular events, 1.06; 0.60-1.88; P = .85). This finding persisted after removing 189 patients with white coat hypertension from the analysis (for renal death, HR, 0.71; 95% CI, 0.43-1.16; P = .17 and for fatal/nonfatal cardiovascular events, 0.79; 0.43-1.45; P = .44).

When patient outcomes were stratified according to dipping status, we found that office BP measurement did not change significantly across categories, from extreme dippers to reverse dippers: 146 (17)/86 (12), 146 (18)/83 (11), 146 (20)/81 (11), and 147 (20)/81 (13) mm Hg. Crude event rates for renal death in extreme dippers, dippers, nondippers, and reverse dippers were 3.8, 6.5, 11.0, and 14.6 per 100 patient-years, respectively, and event rates for the cardiovascular end point were 3.9, 3.6, 8.9, and 10.6 per 100 patient-years. Patients who were nondippers and those who were reverse dippers had an increased risk of renal death and cardiovascular events (Table 5).

This study demonstrates that ABPM is a better predictor of renal and cardiovascular end points compared with office BP measurement in patients with CKD. Only 2 studies^28,29 have previously compared the prognostic role of ABPM and office BP measurement in this population. The current study confirms the greater predictive value of ABPM and extends this observation to a more heterogeneous population including women and non-US patients. In contrast to previous studies, our results demonstrate that the predictive role of ABPM is independent of other risk factors, such as diabetes mellitus, cardiovascular disease, proteinuria, hemoglobin level, and GFR. Of interest, the lack of predictive value of office BP measurement, as well as office BP target level, raises concerns regarding the adequacy of recommendations of hypertension guidelines that are derived mainly from expert opinion and post hoc analyses rather than randomized trials.^10-12 We speculate that cardiorenal risk may actually increase in some subgroups of patients (eg, elderly individuals, those with severe atherosclerotic disease, or those with white coat hypertension)³⁹ in which stricter BP control based on office measurements may compromise renal and/or cardiac perfusion.^12,40 However, office measurements alone obviate the possibility of diagnosing and treating masked hypertension, which, when present in patients with CKD, is associated with worse outcomes.²⁹ The ongoing randomized Systolic Blood Pressure Intervention Trial (SPRINT) (http://clinicaltrials.gov/ct2/show/NCT01206062), which is comparing the effect on cardiorenal prognosis of 2 office BP target levels (systolic <120 mm Hg vs <140 mm Hg), will probably provide helpful information for management decisions in this population.

Systolic ABPM was a better predictor of adverse outcomes than was ambulatory DBP values. Among the components of ABPM, nighttime systolic BP was a stronger predictor than daytime systolic BP for both end points, with an increased risk for the renal end point occurring for daytime SBP higher than 135 mm Hg and an increased risk for the cardiovascular end point occurring for nighttime SBP higher than 124 mm Hg. In contrast, DBP had a different association with renal and cardiovascular end points, whereas daytime values allowed a better definition of renal prognosis, nighttime DBP better predicted the risk of fatal/nonfatal cardiovascular events. The analysis with continuous variables confirmed the data obtained with the categorization of BP in quintiles. Interestingly, our data suggest that caution should be exercised with excessively low daytime SBP values because of the potential increase in cardiovascular risk (Figure 2).

The analysis of daytime and nighttime BP target levels provides another important finding of our study. International guidelines^1-3,35 do not provide specific thresholds for daytime and nighttime BP for patients with CKD. We found that achieving daytime and nighttime BP goals lower than 135/85 mm Hg and lower than 120/70 mm Hg, respectively, was associated with better outcomes. To determine whether achieving these goals implies a favorable risk factor profile or is causally associated with good outcomes will require randomized controlled trials. Our results therefore suggest that the ABPM target levels recommended for patients with essential hypertension can be reasonably used in patients with CKD.

This study also provides evidence that the lack of physiological BP decline during nighttime predicts important clinical outcomes in CKD, as described in a non-CKD population.^18,19 Our results demonstrate a 2-fold greater risk of cardiovascular events in nondipper and reverse dipper subgroups; in these patients, the risk of renal death increased by 62% and 72% in comparison with dippers. Interestingly, this analysis also confirmed the poor predictive role of office BP measurement, as it was ineffective in discriminating risk profile across dipping categories. Previous small, retrospective studies reported^41-44 the relationship between altered circadian pattern of BP and more rapid decline in renal function in specific subgroups of patients with CKD; however, none of these studies used important clinical end points such as ESRD, death, or cardiovascular events. More recently, Redon et al⁴⁵ evaluated the risk of renal death in a small number of patients with CKD stages 3 and 4. They found a greater risk of progression to ESRD in patients with nighttime SBP higher than 130 mm Hg, independent of their dipping status. However, the population studied was scarcely representative of a typical CKD population because diabetic patients, those with previous cardiovascular disease, and those with more advanced CKD were excluded. This selection bias led to an event rate in these patients much lower than that in the current study and in the cohort evaluated by Agarwal and Andersen,²⁸ therefore making the study by Redon et al somewhat underpowered.

Only 1 study has investigated the effect of nondipping on cardiovascular outcome. Analyzing a cohort of US veterans, Agarwal and Andersen²⁹ found that nondipping was associated with increased cardiovascular risk, but its predictive role disappeared after adjusting for other risk factors with the use of propensity scores. In contrast, our study found that an abnormal dipping pattern was a strong predictor of renal and cardiovascular outcomes, even after adjusting for the main demographic and clinical factors influencing dipping status.¹⁷ Our larger sample size may account for this difference.

The current study has some limitations. First, our study included few diabetic patients and was racially homogenous. A larger study with a more diverse population would allow generalization of our findings to patients with diabetes and nonwhite individuals. Second, ABPM data are based on a single set of measurements, which can lead to some imprecision, in particular, for dipping status. Some studies^46,47 have reported a low reproducibility of this phenomenon in the CKD population, with a change in dipping status involving 24% to 38% of patients. The third limitation relates to treatment of hypertension. The IDACO investigators demonstrated that use of antihypertensive drugs removed the significant association between cardiovascular events and daytime BP.¹⁹ However, at variance with the IDACO results, a larger number of patients in our study received antihypertensive therapy (90% vs 22%). Finally, patients with true normotension were excluded, which may explain the difference with the IDACO study.¹⁹

In conclusion, our results show that ABPM is prognostically superior to office BP measurement, even among patients with CKD. The prediction of renal and cardiovascular risk among patients with CKD treated according to BP measured in physicians' offices was better with ABPM, in particular, nighttime BP and an abnormal dipping pattern. In contrast, office BP measurement was prognostically less informative. Interventional studies based on ABPM rather than office BP measurement are urgently required in this high-risk population.

Correspondence: Roberto Minutolo, MD, PhD, Department of Nephrology, Second University of Naples, Via Tiberio 90 I-80125, Naples, Italy (roberto.minutolo@unina2.it).

Accepted for Publication: March 10, 2011.

Author Contributions: Dr Minutolo had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Minutolo, Chiodini, Nappi, Zamboli, Conte, and De Nicola. Acquisition of data: Borrelli, Bellizzi, Nappi, Cianciaruso, and Zamboli. Analysis and interpretation of data: Minutolo, Agarwal, Borrelli, Chiodini, Bellizzi, Gabbai, and De Nicola. Drafting of the manuscript: Minutolo, Chiodini, Bellizzi, Cianciaruso, Conte, Gabbai, and De Nicola. Critical revision of the manuscript for important intellectual content: Minutolo, Agarwal, Borrelli, Chiodini, Bellizzi, Nappi, Cianciaruso, Zamboli, Conte, Gabbai, and De Nicola. Statistical analysis: Agarwal, Borrelli, and Chiodini. Administrative, technical, and material support: Zamboli and De Nicola. Study supervision: Minutolo, Agarwal, Conte, and De Nicola.

Financial Disclosure: None reported.