Reassessment of Ambulatory Blood Pressure Improves Renal Risk Stratification in Nondialysis Chronic Kidney DiseaseNovelty and Significance
Long-Term Cohort Study
In nondialysis chronic kidney disease, ambulatory blood pressure (ABP) performs better than clinic BP in predicting outcome, but whether repeated assessment of ABP further refines prognosis remains ill-defined. We recruited 182 consecutive hypertensive patients with nondialysis chronic kidney disease who underwent 2 ABPs 12 months apart to evaluate the enhancement in risk stratification provided by a second ABP obtained 1 year after baseline on the risk (hazard ratio and 95% confidence interval) of composite renal end point (death, chronic dialysis, and estimated glomerular filtration rate decline ≥40%). The difference in daytime and nighttime systolic BP between the 2 ABPs (daytime and nighttime bias) was added to a survival model including baseline ABP. Net reclassification improvement was also calculated. Age was 65.6±13.4 years; 36% had diabetes mellitus and 36% had previous cardiovascular event; estimated glomerular filtration rate was 42.2±19.6 mL/min per 1.73 m2, and clinic BP was 145±18/80±11 mm Hg. Baseline ABP (daytime, 131±16/75±10 and nighttime, 122±18/66±10 mm Hg) and daytime/nighttime BP goals (58.2% and 43.4%) did not change at month 12. Besides baseline ABP values, bias for daytime and nighttime systolic BP linearly associated with renal outcome (1.12, 1.04–1.21 and 1.18, 1.08–1.29 for every 5-mm Hg increase, respectively). Classification of patients at risk improved when considering nighttime systolic level at second ABP (net reclassification improvement, 0.224; 95% confidence interval, 0.005–0.435). Patients with first and second ABPs above target showed greater renal risk (2.15, 1.29–3.59 and 1.71, 1.07–2.72, for daytime and nighttime, respectively). In nondialysis chronic kidney disease, reassessment of ABP at 1 year further refines renal prognosis; such reassessment should specifically be considered in patients with uncontrolled BP at baseline.
- blood pressure monitoring, ambulatory
- diabetes mellitus
- kidney failure, chronic
- polycystic kidney, autosomal dominant
- renal insufficiency, chronic
Ambulatory blood pressure (ABP) monitoring is considered the gold standard to evaluate hypertension in nondialysis chronic kidney disease (ND-CKD)1,2 because it provides a better correlation with target organ damage when compared with office BP.3–6 More important, ABP is superior to clinic BP (CBP) in predicting adverse outcome in ND-CKD population, including mortality, cardiovascular events, and progression of renal disease.7–12
All prognostic studies based on ABP have used a single set of measurements, which represents a potential limitation leading to inaccuracy in properly defining pressor and circadian profiles. As a consequence, incorrect classification of patients with BP at goal for daytime and nighttime ABP may lead to imprecise risk estimation. Noteworthy, although studies have demonstrated that ABP is reproducible in the short and long terms in essential hypertensive patients,13 information is scarce on ND-CKD and limited to small studies carried out in patients with autosomal dominant polycystic kidney disease14,15 or in patients with advanced CKD.16 These studies, moreover, have focused mainly on the persistence of altered circadian profile (nondipping status) over time although not providing data on the prognostic role of repeated ABP monitoring. Therefore, whether detection of altered ABP readings needs to be confirmed by additional ABP session over time in unselected ND-CKD population remains to be defined. This information is critical to reduce economic costs of ABP and to limit the discomfort to the patients.
To this aim, we designed a prospective observational study to evaluate the enhancement of risk stratification provided by a second ABP monitoring performed 1 year apart in a cohort of ND-CKD patients.
This is a prospective cohort study of consecutive hypertensive ND-CKD patients referred to our renal clinic from January 1, 2006 to December 31, 2010. In our institution, ABP monitoring is performed in all patients with hypertension as part of the initial work up in Renal Clinic with a second ABP monitoring being offered to all patients after 1 year.
Patients were included in this study if they had CKD stages II–V, hypertension (systolic CBP, ≥140 mm Hg or diastolic BP ≥90 mm Hg or receiving antihypertensive treatment), ≥6 months of nephrology care, and a second ABP recording obtained 12±2 months after baseline. Exclusion criteria included dialysis or transplant, changes in glomerular filtration rate (eGFR) >30% in the previous 3 months, changes in antihypertensive therapy 2 weeks before ABP monitoring, atrial fibrillation or inadequate ABP readings. eGFR was calculated by using the CKD-EPI (CKD Epidemiology Collaboration) equation accounting for noncalibrated serum creatinine, as previously described.11 Our Institutional Review Board approved the protocol, and written informed consent was obtained from all patients before study enrollment.
The primary aim of this study was to compare the prognostic role of 2 ABPs versus single monitoring in predicting renal outcome. During the interval between the 2 sets of ABP measurements, no specific indication for hypertension management was provided to participating nephrologists, who could adjust treatment according to office readings, ABP or both at baseline or office BP during the elapsing visits. Patients were always followed up by the same nephrologist who prescribed antihypertensive medications. After second ABP, information was collected about the renal end point defined as the composite of death, end-stage renal disease (ESRD) or eGFR decline ≥40%, which is now accepted as an adequate surrogate of hard renal end point.17 ESRD included chronic dialysis or kidney transplant. For survival analysis, we considered the time from the day of second ambulatory BP (start date of follow-up) until ESRD, death or eGFR decline ≥40% whichever occurred first with data censored at that point; patients without events were followed up until December 12, 2014 and censored on the date they had the last clinic visit.
Medical history, laboratory, clinical, and therapeutic data were collected at baseline and at visit when the second ABP was performed. CBP was measured in a sitting position 3 times at 5-minute intervals during the visit by the same physician who was not aware of the results of the ABP recordings. CBP was reported as mean of the 6 values recorded in the 2 consecutive days in which the ABP device was installed and removed.
ABP was obtained on a workday and under regular antihypertensive treatment, as previously described9–11; briefly, the monitor (Spacelabs 90207) recorded BP every 15 minutes from 7:00 to 23:00 hours and every 30 minutes from 23:00 to 7:00 hours. Daytime and nighttime periods were derived from the diaries recorded by the patients. Patients had no access to the ABP values.
All patients received detailed instructions on how to limit salt intake (goal, 6 g NaCl per day). Patients were classified at goal for ABP when 24-hour, daytime, and nighttime BP were <130/80, <135/85, and <120/70 mm Hg, respectively18 and at goal for CBP when values were <140/90 mm Hg. Nondipping status was defined as night/day ratio of systolic ABP >0.90.
Continuous variables are expressed as mean and SD or as median and interquartile range (IQR) according to their distribution and compared using paired Student t test or Wilcoxon signed-rank test, respectively. Categorical variables were expressed as percentage and compared with McNemar test. Agreement of daytime and nighttime BP between first and second ABP was analyzed by calculating the mean difference (bias) and limits of agreement (±1.96 SD of bias) with Bland–Altman plots.19
A multivariable Cox proportional hazard model was used to estimate hazard ratio (HR) and 95% confidence interval (CI) for the composite renal end point. We assessed the role of each baseline daytime and nighttime systolic and diastolic ABP separately (model 1). To estimate the contribution of the second ABP, we added, for each component, the difference between second ABP and baseline ABP (model 2). We preferred to consider the difference between 2 ABPs rather than the absolute values of the 2 ABP measurements to avoid collinearity. Model 1 and model 2 were adjusted a priori for the following covariates recorded at month 12 (start of follow-up): age, sex, diabetes mellitus, history of cardiovascular disease, proteinuria, eGFR, number of antihypertensive drugs, and the difference from baseline in the number of antihypertensive drugs; the latter covariate was added to the model to verify whether risk estimates were influenced by changes in therapy. In both models, all first-order interactions were tested. We also evaluated the risk associated with daytime and nighttime BP at goal during the 2 ABP recordings by classifying patients in 4 mutually exclusive categories: both at goal (reference), only baseline ABP at goal, only second ABP at goal, and neither ABP at goal.
From models 1 and 2, we evaluated the continuous net reclassification improvement (NRI) and integrated discrimination improvement (IDI) using methods accounting for censoring.20,21 NRI is the difference in proportions moving up and down risk strata among patients with event versus those without event when using the 2 ABP sets versus baseline ABP, whereas IDI is the difference in discrimination slope between 2 models, where the discrimination slope is the mean difference in predicted probabilities between patients with and without event.22 The difference in c-statistics between models 1 and 2 was also computed using a modified version for censored data.23 CIs of pairwise differences between c-indexes were calculated by bootstrapping with 1000 replicates.
Data were analyzed using SPSS version 12.0 (SPSS Inc, Chicago, IL) and survIDINRI, Hmisc packages of the R software version 3.1.0 (R Foundation for Statistical Computing, Vienna, Austria). Two-tailed P values of <0.05 were considered significant.
In our cohort of 219 white patients, only 23 (10.5%) did not have a second ABP (20 patients refused, 2 reached ESRD, and 1 died). Of 196 eligible white patients, 182 were included (Table 1). Reasons for exclusion were inadequate ABP recordings (n=5), recent change of antihypertensive therapy (n=5), atrial fibrillation (n=3), and eGFR change >30% (n=1).
During the interval between the 2 ABP recordings (12±1 months), patients attended a median of 4 visits (IQR, 2–5); in this period, eGFR decreased (P<0.001) by 2.4 mL/min per 1.73 m2 (IQR from −6.9 to +1.8) and proteinuria declined (P=0.004) by 0.02 g/d (IQR from −0.25 to +0.06) without changes in urinary sodium excretion.
BP and antihypertensive therapy at the time of first and second ABP monitoring are presented in Table 2. Systolic CBP at the time of the second ABP was on median 3.5 mm Hg lower (IQR from −15 to +7), whereas no difference was detected for diastolic CBP (median, 0 mm Hg; IQR from −10 to +7). Bland–Altman analysis indicated no evidence of systemic bias and no trend to vary for daytime and nighttime systolic and diastolic BP (Figure S1 in the online-only Data Supplement). The mean change between the 2 ABP sets for daytime systolic BP was −0.5 mm Hg (median, −2 mm Hg; IQR, −9 to +10) and for nighttime systolic BP was −0.3 mm Hg (median, 0 mm Hg; IQR, −9 to +9) with limit of agreement of 29.0 and 26.1 mm Hg, respectively. Bias for daytime diastolic BP was −1.1 mm Hg (median, −1.5 mm Hg; IQR, −6 to +4) and −0.7 mm Hg (median, −1 mm Hg; IQR, −7 to +5) for nighttime diastolic BP with limit of agreement of 16.8 and 16.1 mm Hg, respectively. Overall, these changes did not induce a significant change in BP goal achieved for CBP and ABP with a moderate agreement between measurements for office and daytime BP goals (κ coefficients, 0.450 and 0.499, respectively) and a substantially greater agreement for nighttime BP goal (κ coefficient, 0.753; Table 3). Dipping status was confirmed across 2 ABPs in 74.7% of patients (54.4% nondippers and 20.3% dippers) with a moderate agreement (κ coefficient, 0.429).
On average, antihypertensive treatment did not change in terms of number of drugs and distribution of drug classes (Table 2). The same held true for the dose of diuretic (furosemide: from 63±56 at baseline to 69±53 mg/d at month 12; hydrochlorothiazide: from 16±10 mg/d to 18±9 mg/d), as well as for the adherence to low-sodium diet (25.3% at baseline and 27.5% at the time of second ABP measurement; P=0.665). No change in antihypertensive therapy occurred in 89 patients (48.9%), whereas the number of drugs diminished (from 3.8±1.6–2.7±1.6) in 38 patients (20.9%) and increased (from 2.9±1.3–4.1±1.4) in 55 (30.2%). Daytime and nighttime BP did not change in patients with unchanged antihypertensive treatment; a reduction was observed in those who underwent intensification of therapy and a slight increase in patients receiving less medication (Table S1).
During follow-up (median, 43 months; IQR, 23–66), 104 patients reached the combined end point (33 deaths, 41 ESRD, and 30 eGFR decline ≥40%). The adjusted risk associated with ABP measures is reported in Table 3. Baseline systolic ABP (daytime and nighttime) was significantly associated with renal events, with 8% to 9% increase in risk for each 5-mm Hg higher BP. The second ABP also contributed to risk stratification (model 2). Besides the effect of baseline daytime and nighttime systolic BP, the change in systolic BP between the 2 ABP recordings was also significantly associated with an adverse outcome; a similar weight was observed for changes in nocturnal BP (HR, 1.18) and diurnal BP (HR, 1.12). As depicted in Figure 1, there was a linear association between changes in both daytime and nighttime systolic values during the second ABP recording and the risk of renal end point. No interaction was found between the change in systolic BP and baseline ABP (P=0.880 and P=0.803 for daytime and nighttime values, respectively) or between the change in antihypertensive therapy and the change in systolic BP (P=0.086 and P=0.132 for daytime and nighttime values, respectively). Similar findings were observed for 24-hour systolic BP (data not shown). The role of diastolic BP (daytime and nighttime) was irrelevant with an effect being limited to the change of nighttime value (Table 3). The prognostic effect of achieving target BP on either or both of the ABP is depicted in Figure 2. Only patients having both ABP values above goal had an increased renal risk (HR, 2.15; 95% CI, 1.29–3.59 and HR, 1.71; 95% CI, 1.07–2.72, for daytime and nighttime goals, respectively).
When replacing in Cox models, systolic ABP with office readings, we found no predictive role of either baseline systolic office BP (HR, 1.00; 95% CI, 0.93–1.07) or the change in systolic office BP (HR, 1.06; 95% CI, 0.99–1.11).
Predictors of Outcomes
The risk estimates for each variable included in complete models (model 2) assessed separately for daytime and nighttime systolic ABP are reported in Table S2. Besides ABP, we found that older age, female gender, and higher eGFR were all associated with lower risk of composite renal end point with a significant prognostic role of 24-hour proteinuria being evident only in the model including nighttime BP.
Renal Risk Reclassification
Systolic nighttime BP showed more evident improvement for renal end point prediction in c-statistics difference, continuous NRI, and IDI when compared with daytime BP (Table 3). Indeed, a significantly positive NRI was observed only for differences in nocturnal systolic BP which reclassified 22.4% of patients. Although daytime systolic BP showed some improvement as demonstrated by the IDI, the estimate of NRI for daytime systolic BP was not significant and smaller than that for nighttime systolic (Table 3). Adding nighttime BP information with a second ABP, the proportion of patients in whom the risk scores with the model 2 were higher than the risk scores with the model 1 was 57.9% among patients with event and 35.5% among patients without event. Changes of daytime and nighttime diastolic BP did not contribute to risk reclassification (Table 3). Similarly, risk reclassification did not improve when considering changes of office systolic BP (△c-statistic: 0.005; 95% CI from −0.004 to 0.015, NRI: 0.211; 95% CI from −0.245 to 0.368, and IDI: 0.007; 95% CI from −0.007 to 0.036).
To our knowledge, this is the first study aimed at evaluating the prognostic role of 2 repeated ABP measurements in hypertensive patients with or without CKD. Previous studies have evaluated the stability of ABP measures in patients with autosomal dominant polycystic kidney disease or in those with advanced CKD (stages 4–5), but neither one of them provided any information about their prognostic value.14–16
The main finding of our study is that a second assessment of ABP monitoring provides incremental estimate of the renal risk beyond the initial evaluation. Patients with increased systolic BP between the first and second ABP were in fact exposed to a greater risk independent of the systolic ABP level measured on the first monitoring, whereas those patients in whom ABP declined were at a lower risk (Figure 1). A repeated measurement of ABP after 1 year, as recommended by guidelines for essential hypertension,18,24 allows to correctly reclassify as at-risk ≈22% of patients based on the nighttime systolic component of ABP (Table 3). Results with daytime systolic BP were less consistent, as demonstrated by increased IDI values but nonsignificant values of NRI, thus confirming the superiority of nocturnal assessment over the diurnal measurements as previously reported in all cohort studies in patients with ND-CKD7–12 and further supports the value of ABP in this population because this technique is the only one currently available to measure BP at night. Of interest, the improvement in the risk reclassification was mainly correlated to the changes in systolic ABP; neither the diastolic values recorded at baseline nor the changes in either daytime or nighttime diastolic ABP predicted the composite end point. The improved risk stratification was not observed when using office BP measurements.
The benefit of repeated ABP recordings in refining prognosis is also evidenced by the results obtained when patients were classified on the basis of achieved BP target (Figure 2). Classifying patients as high-risk based on baseline ABP would miss those individuals who subsequently achieved the daytime or nighttime goal because the renal risk of these patients was not different from those individuals whose BP was at goal on both ABP recordings. In addition, information provided by the second ABP identified patients with poorly controlled BP on both recordings who are truly at higher risk.
During the 12-month interval between the 2 ABPs, antihypertensive treatment was modified in ≈50% of subjects. Because our patients attended several visits during this period, it is difficult to relate changes in the antihypertensive regimen exclusively to their baseline ABP profile because some of these changes could respond to office BP values obtained during the visits attended in between the 2 ABP recordings. To exclude that this approach, inherent to the clinical practice, may have influenced the association with prognosis of the second ABP, we adjusted survival models for the changes of antihypertensive therapy between the 2 ABP sessions; we found that risk estimates for systolic BP change during daytime and nighttime remained unchanged, and there was no interaction between variation of antihypertensive therapy and ambulatory systolic BP changes, thus supporting its prognostic usefulness also in patients undergoing changes of BP-lowering drugs.
Few studies have evaluated the persistence of altered ABP across ≥2 ABPs in patients with CKD, and none of them related the ABP findings with outcome. Only 2 small studies have focused on the change in prevalence of dipping status across ABP recorded 1 year apart. In 30 autosomal dominant polycystic kidney disease patients with mild renal impairment, Covic et al14 reported that dipping status (nocturnal fall above or below the median value of 10.6%) was consistently maintained in 63.3% of patients. Similarly, when ABP monitoring was performed serially in 4-month intervals in 34 patients with CKD stages 4 to 5, the dipping status did not change in 62% of paired examinations.16 In those studies, the prevalence of patients persisting in the same dipping category was lower than the one we found in our study (75%) possibly because of the smaller sample size, and marked differences in the cause of CKD, age, and renal function.
Our work has some limitations; first, all patients were whites, and it is not known if the prognostic values of repeated ABP is generalizable to other races. Second, participating nephrologists were free to modify antihypertensive therapy on the basis of CBP, ABP, or both and as such, results of this study cannot be applied to patients where changes in the antihypertensive regimen are driven exclusively by either office or ABP results.
The results of this study have important practical clinical implications. The major reason to perform an ABP recording in patients with CKD is to identify patients with nocturnal hypertension, which constitutes a major predictor of cardiovascular events and progression to ESRD.1,2 Performing a second ABP monitoring provides better prognostic information and allows reclassification of patients at-risk based on nighttime systolic BP. Therefore, in routine clinical practice, physicians need to weight the benefit of refining prognosis with a second ABP against the burden for the patients and the economic costs associated with ABP.
R. Minutolo, L. De Nicola, P. Chiodini, F.B. Gabbai, and G. Conte have done substantial contributions to conception and design. P. Chiodini, C. Garofalo, G. Stanzione, M.E. Liberti, M. Pacilio, S. Borrelli, and M. Provenzano contributed to acquisition of data. R. Minutolo, S. Borrelli, L.D. Nicola, P. Chiodini, and F.B. Gabbai contributed to analysis and interpretation of data. All authors contributed to drafting of the article and revising it critically for important intellectual content, final approval of the version to be published, and agreement to be accountable for all aspects of the work. We certify that the article represents valid work and that neither this article nor one with substantially similar content under their authorship has been published or is being considered for publication elsewhere. R. 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.
The online-only Data Supplement is available with this article at http://hyper.ahajournals.org/lookup/suppl/doi:10.1161/HYPERTENSIONAHA.115.05820/-/DC1.
- Received May 9, 2015.
- Revision received May 24, 2015.
- Accepted June 15, 2015.
- © 2015 American Heart Association, Inc.
- Agarwal R,
- Andersen MJ.
- Peterson GE,
- de Backer T,
- Gabriel A,
- Ilic V,
- Vagaonescu T,
- Appel LJ,
- Contreras G,
- Kendrick C,
- Rostand S,
- Phillips RA
- Iimuro S,
- Imai E,
- Watanabe T,
- Nitta K,
- Akizawa T,
- Matsuo S,
- Makino H,
- Ohashi Y,
- Hishida A
- Gorostidi M,
- Sarafidis PA,
- de la Sierra A,
- Segura J,
- de la Cruz JJ,
- Banegas JR,
- Ruilope LM
- Minutolo R,
- Agarwal R,
- Borrelli S,
- Chiodini P,
- Bellizzi V,
- Nappi F,
- Cianciaruso B,
- Zamboli P,
- Conte G,
- Gabbai FB,
- De Nicola L.
- Minutolo R,
- Gabbai FB,
- Agarwal R,
- Chiodini P,
- Borrelli S,
- Bellizzi V,
- Nappi F,
- Stanzione G,
- Conte G,
- De Nicola L.
- Gabbai FB,
- Rahman M,
- Hu B,
- et al
- Elung-Jensen T,
- Strandgaard S,
- Kamper AL.
- Levey AS,
- Inker LA,
- Matsushita K,
- Greene T,
- Willis K,
- Lewis E,
- de Zeeuw D,
- Cheung AK,
- Coresh J.
Novelty and Significance
What Is New?
We have evaluated for the first time the prognostic role of 2 repeated ambulatory blood pressure (ABP) measurements in a heterogeneous population of patients with chronic kidney disease.
What Is Relevant?
Second assessment of ABP provides incremental estimate of the renal risk beyond the initial evaluation and allows to correctly reclassify as at-risk >22% of patients based on the nighttime systolic component of ABP.
In patients not achieving BP goals at both first and second ABPs, the risk of adverse renal outcome was almost doubled.
Reassessment of ABP at 1 year further refines renal prognosis especially in patients with uncontrolled BP at baseline. In routine clinical practice, physicians need to weight the benefit of refining prognosis with a second ABP against the burden for the patients and the economic costs associated with ABP.