Estimated aortic blood pressure based on radial artery tonometry underestimates directly measured aortic blood pressure in patients with advancing chronic kidney disease staging and increasing arterial stiffness

clinical investigation

www.kidney-international.org

Estimated aortic blood pressure based on radial artery tonometry underestimates directly measured aortic blood pressure in patients with advancing chronic kidney disease staging and increasing arterial stiffness Rasmus K. Carlsen1,2, Christian D. Peters1,2, Dinah S. Khatir1,2, Esben Laugesen3, Hans Erik Bøtker4, Simon Winther4 and Niels H. Buus2,5 1

Department of Renal Medicine, Aarhus University Hospital, Aarhus, Denmark; 2Department of Clinical Medicine, Aarhus University, Aarhus, Denmark; 3Department of Endocrinology and Internal Medicine, Aarhus University Hospital, Aarhus, Denmark; 4Department of Cardiology, Aarhus University Hospital, Aarhus, Denmark; and 5Department of Nephrology, Aalborg University Hospital, Aalborg, Denmark

Central blood pressure (BP) can be assessed noninvasively based on radial tonometry and may potentially be a better predictor of clinical outcome than brachial BP. However, the validity of noninvasively obtained estimates has never been examined in patients with chronic kidney disease (CKD). Here we compared invasive aortic systolic BP (SBP) with estimated central SBP obtained by radial artery tonometry and examined the influence of renal function and arterial stiffness on this relationship. We evaluated 83 patients with stage 3 to 5 CKD (mean estimated glomerular filtration rate [eGFR] 30 ml/min/1.73 m2) and 41 controls without renal disease undergoing scheduled coronary angiography. BP in the ascending aorta was measured through the angiography catheter and simultaneously estimated using radial tonometry. The mean difference between estimated central and aortic SBP was –13.2 (95% confidence interval –14.9 to –11.4) mm Hg. Arterial stiffness was evaluated by carotid-femoral pulse wave velocity (cfPWV) and was significantly increased in CKD patients compared with (versus) control patients (mean 10.7 vs. 9.3 m/s). The difference in BP significantly increased 1.0 mm Hg for every 10 ml/min decrease in eGFR and by 1.6 mm Hg per 1 m/s increase in cfPWV. Using multivariate regression analysis including both eGFR and cfPWV, the difference between estimated central and invasive aortic SBP was significantly increased by 0.7 mm Hg. For the entire cohort brachial SBP significantly better reflected invasive SBP than estimated SBP. Thus, tonometry-based estimates of central BP progressively underestimate invasive central SBP with decreasing renal function and increasing arterial stiffness in CKD patients. Kidney International (2016) j.kint.2016.05.014

-, -–-;

http://dx.doi.org/10.1016/

Correspondence: Rasmus K. Carlsen, Department of Renal Medicine, Aarhus University Hospital, Palle Juhl Jensens Boulevard 100, 8200 Aarhus N, Denmark. E-mail: [email protected] Received 11 February 2016; revised 11 April 2016; accepted 5 May 2016 Kidney International (2016) -, -–-

KEYWORDS: blood pressure; central blood pressure; chronic kidney disease; generalized transfer function; invasive blood pressure; pulse wave velocity Copyright ª 2016, International Society of Nephrology. Published by Elsevier Inc. All rights reserved.

A

rterial hypertension is a cause as well as consequence of chronic kidney disease (CKD) and blood pressure (BP) control is essential to attenuate loss of kidney function and the incidence of cardiovascular disease in CKD patients.1 Despite an apparently well-controlled BP, CKD progresses in many patients, suggesting that the measured brachial BP is not the ideal treatment target.2 Systolic BP (SBP) increases as the pulse wave travels from the aorta to the muscular arteries resulting in a higher SBP in the brachial artery, while the diastolic BP (DBP) and the mean arterial BP (MAP) remain fairly constant throughout the larger arteries. This phenomenon is known as pulse pressure amplification and is most pronounced in young individuals.3 With aging, pulse pressure increases more rapidly in the thoracic aorta than in peripheral arteries, causing an attenuation of the physiological increase in pulse pressure from central to peripheral arteries. The pulse pressure amplification varies between individuals and, accordingly, central BP may yield prognostic information superior to brachial BP, because it more accurately reflects the BP close to vulnerable organs such as the heart, brain, and kidneys. Central BP may therefore be a better predictor of cardiovascular events and has also been suggested as a potential risk factor for CKD progression.4–6 Invasive central BP recordings are not feasible in a daily clinical setting and can even be challenging to obtain in research. Consequently, noninvasive estimates of aortic BP are commonly used and several devices are available for assessment of central BP. The SphygmoCor device (AtCor Medical, Itasca, IL) applies a transfer function to the radial artery waveform obtained with applanation tonometry to estimate the aortic BP.7 The calculation requires a calibrating BP, which 1

clinical investigation

RK Carlsen et al.: Estimated central BP underestimates measured BP

usually means the brachial BP. Several studies have used the SphygmoCor in CKD cohorts.8–11 However, the agreement between invasive measurements and noninvasive estimates of central aortic BP has never been assessed in patients with CKD. Validation in this specific group is especially important as declining glomerular filtration rate (GFR) is closely associated with increased arterial stiffness.12 Hypothetically, increasing arterial stiffness could attenuate the pulse pressure amplification. This would change the difference between peripheral and central BP. The primary aim in this study was to assess the agreement between noninvasively obtained estimates of central SBP and invasive SBP measurements in the aorta in CKD patients. First, we tested whether the transfer function was precise in CKD patients compared with control patients. Next, we compared invasive SBP with noninvasive estimates when calibrated with brachial SBP. Finally, we examined whether declining renal function and increasing arterial stiffness affected the difference between estimated central and invasive aortic SBP. RESULTS Study subjects

Between August 2, 2013 and January 27, 2015, a total of 155 patients undergoing scheduled coronary angiography (CAG) were invited to participate (Figure 1). Fourteen declined participation and 20 were excluded (6 due to occurrence of arrhythmia after inclusion, 5 because of insufficient time during the CAG procedure to perform the measurements, and 9 because of technical issues). In total, data on 124 patients were available for analysis: 83 CKD patients and 41 control patients. Demographic characteristics are presented in Table 1. Overall, the CKD patients and control patients were similar in terms of age, sex, smoking status, and body mass index. CKD stage 5 patients were younger with less diabetes and lower body mass index than CKD stage 3 patients. Diabetes was present in 30% of the CKD patients, but absent in the control patients. Left ventricular ejection fraction was on average well preserved. Most CKD patients had albuminuria that increased with the CKD stage. Eight CKD patients, 4 of whom were anuric, received dialysis. 155 paƟents invited

54 control paƟents invited

104 CKD paƟents invited

11 declined 2 excluded

3 declined 18 excluded

CKD patients received more antihypertensive drugs than control patients did, especially angiotensin receptor blockers and diuretics. CKD and control patients had similar brachial, aortic, and estimated BP (Table 1). The indications for CAG are presented in Table 2 with the CKD patients stratified into stages 3 to 5. The majority of control patients and CKD stage 3 patients were referred because of suspected ischemic heart disease, whereas CKD stage 5 patients were mainly referred as part of a screening program before kidney transplantation. Assessment of the generalized transfer function

To validate the generalized transfer function, central SBP was estimated using the invasively obtained MAP and DBP as the calibration BP (Table 3). The difference between estimated and invasive aortic SBP increased with 0.3 (–0.02 to 0.5) mm Hg per 10 ml/min decrease in estimated glomerular filtration rate (eGFR) (P ¼ 0.07). Using multivariate adjustment for age, sex, ejection fraction, diabetes, current smoking, and number of antihypertensive drugs, the difference was 0.3 (0.1 to 0.6) mm Hg per 10 ml/min decrease in eGFR (P ¼ 0.02, adjusted R2 ¼ 0.19). Adjustment for carotidfemoral pulse wave velocity (cf-PWV) did not change this result (P ¼ 0.046, adjusted R2 ¼ 0.16). Furthermore, adjustment for renal function as an ordinal variable (CKD stage) with the controls patients as reference did not change the results (P ¼ 0.017, adjusted R2 ¼ 0.19). The difference between estimated and invasive aortic SBP was not significantly different between CKD patients and control patients (mean difference 1.6 [–0.5 to 3.6] mm Hg) (P ¼ 0.13). Cf-PWV was not associated with the SBP difference in either unadjusted or multivariate linear regression (P $ 0.12). BP measurements: estimated central versus invasive aortic SBP and brachial versus invasive aortic SBP

The difference between estimated central BP (using brachial BP for calibration) and invasive aortic SBP was compared with Bland-Altman plots for CKD patients (Figure 2a) and control patients (Figure 2b). For CKD patients, the mean difference was –14.9 (–17.1 to –12.6) mm Hg (P < 0.001). For control patients, the mean difference was –9.7 (–12.5 to –7.0) mm Hg (P < 0.001). The difference was 5.1 (1.5–8.8) mm Hg lower in control patients (P ¼ 0.007). The difference between brachial SBP and invasive aortic SBP was –1.4 (–3.9 to 1.0) mm Hg (P ¼ 0.25) and 3.0 (–0.3 to 6.3) mm Hg (P ¼ 0.07) for CKD patients (Figure 3a) and control patients (Figure 3b), respectively. The difference was 4.5 (0.3–8.6) mm Hg higher in control patients (P ¼ 0.04). BP measurements and differences according to CKD stage

41 control paƟents parƟcipated

83 CKD parƟcipated

Figure 1 | Flow chart describing the inclusion process of the 155 patients invited to participate in the study. CKD, chronic kidney disease. 2

Table 4 shows the association between CKD stage and differences between estimated central and invasive aortic SBP and between brachial and invasive aortic SBP using linear regression analysis. The difference between estimated central SBP and invasive aortic SBP increased stepwise from control Kidney International (2016) -, -–-

clinical investigation

RK Carlsen et al.: Estimated central BP underestimates measured BP

Table 1 | Baseline characteristics in control and CKD patients according to stage Characteristics Patients Age (yr) Female BMI (kg/m2) Smoking (n, current/previous/never) Diabetes Ejection fraction (% and range) eGFR (ml/min/1.73 m2) Urine albumin/creatinine (mg/g) 0-29 30–299 $300 Dialysis þ anuria Antihypertensive medication per patient ACE inhibitor ARB Beta-blocker Calcium channel blocker Diuretics Nitrates Glycerol nitrate during CAGe Adenosine during CAGf Brachial SBP (mm Hg) Brachial DBP (mm Hg) Aortic SBP (mm Hg) Aortic DBP (mm Hg) Estimated central SBP (mm Hg) Estimated central DBP (mm Hg) Carotid-femoral PWV (m/s) Carotid-radial PWV (m/s)

Control

CKD 3

CKD 4

CKD 5

All CKD

41 (33) 63  13 12 (29) 27  5 9/22/10 0 (0) 56 (33–65) 95  10

43 (35) 70  9 16 (37) 29  5 2/30/11 13 (30) 51 (15–65) 48  11

15 (12) 67  14 1 (7) 26  4 5/3/7 8 (53) 55 (45–60) 20  3

25 (20) 55  12 10 (40) 25  3 8/11/6 5 (16) 60 (50–65) 8  5d

83 (67) 65  13 27 (33) 27  5 15/44/24 25 (30) 54 (15–65) 30  19

36 (90) 4 (10) 0 (0)

31 (72) 5 (12) 7 (16)

0 5 13 8 4 3

36 11 28 8 4 3

5 (36) 1 (7) 8 (57)

1 (0–4)

3 (0–5)

3 (0–5)

9 (22) 5 (12) 17 (41) 16 (39) 10 (24) 17 (41) 18 (44) 4 (10) 144  18 82  11 141  25 70  11 131  20 83  11 9.3  2.2 8.1  1.1

13 (30) 16 (37) 28 (65) 16 (37) 24 (56) 19 (44) 19 (44) 7 (16) 146  19 80  11 144  22 67  9 133  20 80  11 11.0  2.6 8.3  1.4

8 (53) 6 (40) 6 (40) 10 (67) 11 (73) 4 (27) 4 (27) 1 (7) 153  21 83  9 155  22 70  9 138 21 84 9 11.8  3.2 8.7  1.3

(0) (28) (72) (31) (16) (0–5)

7 (28) 12 (48) 15 (60) 17 (68) 22 (88) 1 (4) 18 (72) 0 (0) 147  19 88  11 153  23 81  11 133  20 89 11 9.9  2.0 8.7  1.0

P valuea 0.80 0.71b 0.88 <0.001b 0.91c <0.001 <0.001

(48) (15) (37) (10) (5) (0–5)

<0.001c

28 (34) 34 (41) 49 (59) 43 (52) 57 (69) 24 (29) 41 (49) 8 (10) 147  20 83  11 149  23 72  12 134  20 84  11 10.7  2.6 8.5  1.3

0.18 <0.001 0.07 0.18 <0.001 0.17 0.57 0.98 0.39 0.55 0.09 0.60 0.52 0.62 0.004 0.08

ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; BMI, body mass index; cf, carotid-femoral; CKD, chronic kidney disease; cr, carotid-radial; DBP, diastolic blood pressure; PWV, pulse wave velocity; SBP, systolic blood pressure. Data are mean  SD or n (%) unless otherwise indicated. Missing data in groups; cf-PWV: n ¼ 3, 9, 3, 0, respectively; cr-PWV: n ¼ 3, 7, 1, 1, respectively; urine albumin/ creatinine: n ¼ 1, 0, 1, 7, respectively; ejection fraction: n ¼ 1, 0, 3, 0, respectively. Determined by aStudent’s t-test, bchi-square test, or cWilcoxon-Mann-Whitney test comparing the whole CKD group with control patients. d eGFR is set as 5 ml/min/1.73 m2 in dialysis patients with residual renal function and as 0 in patients with anuria (<100 ml/day). e Administered into the coronary arteries. f Administered intravenously.

patients to CKD stage 5 patients. The same was seen for the difference between brachial and estimated central SBP. Moreover, in comparison with estimated central SBP, brachial SBP was a better estimate of the invasively measured aortic SBP in the entire CKD group, in each CKD stage, and in control patients (P < 0.001 for all comparisons). For the entire CKD group, brachial SBP was 13.2 (12.2–14.2) mm Hg closer to invasive aortic SBP than estimated central SBP (P < 0.0001). The difference between estimated and invasive aortic BP increased with 9.9 (5.2–14.6) mm Hg in CKD stage 5 patients compared with control patients (P < 0.001). Table 2 | Indication for coronary angiography Indication

Control patients

CKD 3

CKD 4

CKD 5

Ischemic heart disease Kidney transplantation Heart transplantation control Valve replacement Other reason

37 0 0 3 1

36 0 5 1 1

6 7 1 1 0

1 23 1 0 0

CKD, chronic kidney disease (stages). Data are n (%). Kidney International (2016) -, -–-

(90) (0) (0) (7) (2)

(84) (0) (12) (2) (2)

(40) (47) (7) (7) (0)

(4) (92) (4) (0) (0)

Likewise the difference between brachial and invasive aortic BP increased with 9.5 (4.2–14.9) mm Hg in CKD stage 5 patients compared with control patients (P ¼ 0.001). BP measurements and differences according to eGFR

SBP as a function of eGFR is shown in Supplementary Figure S1A to C. No significant association was found between brachial SBP and eGFR (b ¼ 0.9 [–0.2 to 1.8] mm Hg per 10 ml/min decrease in eGFR, P ¼ 0.10) or estimated central SBP and eGFR (b ¼ 0.8 [–0.3 to 1.7] mm Hg per 10 ml/min decrease in eGFR, P ¼ 0.15), whereas invasive aortic SBP increased with 1.7 (0.5–2.9) mm Hg for every 10 ml/min decrease in eGFR (P ¼ 0.005). The difference between estimated central and invasive aortic SBP as a function of eGFR is shown in Figure 4. The difference increased with 1.0 (0.5–1.5) mm Hg per 10 ml/min decrease in eGFR using unadjusted linear regression (P < 0.001). Multivariate adjustment for age, sex, left ventricle ejection fraction, diabetes, current smoking, and number of antihypertensive drugs did not change this finding substantially, as the difference still increased with 3

clinical investigation

RK Carlsen et al.: Estimated central BP underestimates measured BP

Table 3 | Differences between estimated central SBP and the invasively measured SBP in control and CKD patients using invasively obtained mean arterial BP and diastolic BP for calibration SBP difference

Control

CKD 3

CKD 4

P valuea

CKD 5

Estimated central –3.6  5.0 –4.7  4.8 –7.0  6.3 –4.7  6.3 aortic (mm Hg)

0.21

CKD, chronic kidney disease (stages); SBP, systolic blood pressure. Data are mean  SD. a P for trend (unadjusted linear regression).

0.8 (0.3–1.4) mm Hg per 10 ml/min decrease in eGFR (P ¼ 0.005, adjusted R2 ¼ 0.13). Albuminuria was not associated with BP differences, neither in unadjusted analysis or multivariate models. Moreover, because albuminuria correlated with eGFR (r2 ¼ 0.24, P < 0.001), we omitted albuminuria from the final multivariate analysis. Sensitivity analysis was performed by excluding patients having CAG before valve replacement as well as those with heart transplants, but omission of these patients had no impact on the described results. Furthermore, adjustment with renal function as an ordinal variable (CKD stage) using control patients as reference did not change the results (Supplementary Table S1). The influence of PWV on BP differences

Cf-PWV was higher in CKD patients and inversely associated with eGFR, increasing with 0.2 (0.03–0.30) m/s per 10 ml/min decrease in eGFR (P ¼ 0.016, r2 ¼ 0.15). This association became stronger when adjusted for age (b ¼ 0.2 m/s per 10 ml/min decrease in eGFR, P ¼ 0.002, adjusted R2 ¼ 0.28). There was also a significant linear association between

a

carotid-radial PWV (cr-PWV) and eGFR (b ¼ 0.1 m/s per 10 ml/min decrease in eGFR, P ¼ 0.006, r2 ¼ 0.07). Figure 5 shows the significant linear association between cf-PWV and the difference between estimated central and invasive aortic SBP. The difference increased with 1.0 (0.3– 1.7) mm Hg for every 1 m/s increase in cf-PWV (P ¼ 0.005, r2 ¼ 0.07). However, the impact of cf-PWV on the difference between estimated central and invasive SBP was different in CKD patients and control patients (P value for interaction ¼ 0.039). The regressions were therefore performed separately for the two groups. In CKD patients the difference between estimated central SBP and invasive SBP increased with 1.1 (0.2–1.9) mm Hg (P ¼ 0.015, r2 ¼ 0.08) and 1.6 (0.5–2.6) mm Hg (P ¼ 0.003, adjusted R2 ¼ 0.10) per 1 m/s increase in cf-PWV in crude and multivariate regression analyses, respectively. In control patients, the association between cf-PWV and the difference between estimated central and invasive aortic SBP was not significant, in both the unadjusted linear regression (P ¼ 0.87) and the multivariate adjusted model (P ¼ 0.64). Cr-PWV was not associated with the difference between estimated central and invasive aortic SBP. Accordingly, the difference increased with 0.8 (–0.7 to 2.4, P ¼ 0.29) for every 1 m/s increase in cr-PWV in unadjusted linear regression. Multivariate regression did not change this result (P ¼ 0.34). PWV and eGFR as independent predictors of BP differences

When both eGFR and cf-PWV were included in the multivariate model, the difference between estimated central and invasive SBP increased with 0.7 (0.1–1.3) mm Hg per 10 ml/ min decrease in eGFR (P ¼ 0.018, adjusted R2 ¼ 0.14) and

b

20

20 Mean + 1.96 SD

Mean + 1.96 SD

0 -10 Mean

-20 -30 Mean – 1.96 SD

-40

Estimated central minus aortic systolic BP (mm Hg)

Estimated central minus aortic systolic BP (mm Hg)

10 0

Mean

-20 Mean – 1.96 SD

-40

-50 -60

-60 80

100

120

140

160

180

Average of estimated central and aortic systolic BP (mm Hg)

200

80

100

120

140

160

180

200

Average of estimated central and aortic systolic BP (mm Hg)

Figure 2 | Bland-Altman plots demonstrating the relation between estimated central and invasive aortic systolic blood pressure (BP) in (a) chronic kidney disease patients and (b) control patients. 4

Kidney International (2016) -, -–-

clinical investigation

RK Carlsen et al.: Estimated central BP underestimates measured BP

a

b

40

Mean + 1.96 SD

20

Brachial minus aortic systolic BP (mm Hg)

0

Brachial minus aortic systolic BP (mm Hg)

Mean + 1.96 SD

20

Mean

-20 Mean – 1.96 SD

-40

-60 80

100

120

140

160

180

200

Mean

0

Mean – 1.96 SD

-20

-40

-60 100

220

120

140

160

180

200

Average of brachial and aortic systolic BP (mm Hg)

Average of brachial and aortic systolic BP (mm Hg)

Figure 3 | Bland-Altman plots demonstrating the relation between brachial and invasive aortic systolic blood pressure (BP) in (a) chronic kidney disease patients and (b) control patients.

with 1.5 (0.5–2.5) mm Hg for every 1 m/s increase in cf-PWV (P ¼ 0.005, adjusted R2 ¼ 0.12). Adjusting for cr-PWV instead of cf-PWV did not modify the association between eGFR and the BP difference (P ¼ 0.014, adjusted R2 ¼ 0.11). Impact of eGFR and PWV on the difference between brachial SBP and invasive aortic SBP

We also analyzed whether substitution of estimated central SBP with brachial SBP yields similar disagreement when compared with invasive aortic SBP. Figure 6a shows the association between brachial and estimated central SBP demonstrating the very close relation between these 2 parameters. Figure 6b shows the relation between eGFR and the difference between brachial and invasive SBP that increased with 0.9 (0.4–1.5) mm Hg per 10 ml/min decrease in eGFR (P ¼ 0.002) in an unadjusted linear regression and with a similar result using multivariate regression (P ¼ 0.015, adjusted R2 ¼ 0.15). In CKD patients, the difference between brachial and invasive SBP increased with 1.2 (95% CI: 0.3–2.2) mm Hg per 1 m/s increase in cf-PWV (P ¼ 0.013) in an unadjusted linear regression (Figure 6c) and with similar results using

multivariate regression (including eGFR) (P ¼ 0.003, adjusted R2 ¼ 0.17). DISCUSSION

Aortic BP can be assessed noninvasively based on radial tonometry, and in the present study, the agreement between invasively measured and estimated aortic BP was assessed in patients with CKD. Our main finding was that the difference between estimated and invasive aortic SBP increased 5 to 10 mm Hg both as renal function deteriorates and as arterial stiffness increases. Brachial SBP seems to be a more accurate estimate of true central SBP compared with the SphygmoCorderived estimate. Our findings warrant caution when using noninvasive assessment of aortic BP in patients with advanced kidney disease. Several studies have compared invasive aortic SBP with noninvasive estimates derived by SphygmoCor (Table 5) and differences with wide limits of agreement have been registrered.13–17 We have previously compared estimated central and invasively obtained aortic BP in type 2 diabetic patients without overt renal disease.18 The results from the control group in the present study are very similar to these previous

Table 4 | Differences between estimated central SBP or brachial SBP and the invasively measured SBP in control and CKD patients SBP difference Estimated central aortic (mm Hg) Brachial aortic (mm Hg)

Control

CKD 3

CKD 4

CKD 5

P valuea

–9.7  8.7 3.0  10.4

–11.4  8.6 1.8  9.9

–16.9  12.1b –2.1  14.8

–19.6  9.6c –6.5  9.8c

<0.001 <0.001

CKD, chronic kidney disease (stages); SBP, systolic blood pressure. Data are mean  SD. a P for trend (unadjusted linear regression). b P < 0.05 and cP < 0.001 as compared to control patients (Student’s t-test). Kidney International (2016) -, -–-

5

clinical investigation

RK Carlsen et al.: Estimated central BP underestimates measured BP

20

Estimated central minus aortic systolic BP (mm Hg)

10 0 -10 -20 -30 -40 -50 -60 0

20

40

60

80

100

120

140

2

eGFR (ml/min per 1.73 m )

Figure 4 | Estimated central minus aortic systolic blood pressure (BP) as a function of estimated glomerular filtration rate (eGFR). The line represents linear regression (r2 ¼ 0.11, P < 0.001).

findings. Our study is the first in CKD patients and the SBP difference is markedly larger as compared to findings in subjects without renal disease. Wassertheurer et al.19 also found a decreasing pulse pressure amplification with increasing CKD stages, which predicted decline of renal function and need of renal replacement therapy. However, only CKD stages 2 to 4 patients were included and only noninvasive estimates of aortic BP were used in the analysis, which may explain why we observed a more marked decrease in pulse pressure amplification. The aortic-brachial pulse pressure amplification is typically reduced with aging, and the increase in PWV also seen with increasing age is further enhanced by CKD as confirmed by our results.3,12 Acceleration of vascular aging by CKD may therefore partly explain our results. However, when adjusted for cf-PWV, eGFR was still significantly associated with the difference between estimated and invasive aortic SBP. We have analyzed several factors to explain the association between 20

Estimated central minus aortic systolic BP (mm Hg)

10 0 -10 -20 -30 -40 -50 4

6

8

10

12

14

16

18

20

22

Carotid-femoral PWV (m/s)

Figure 5 | Estimated central minus aortic systolic blood pressure (BP) as function of carotid-femoral pulse wave velocity (PWV). The line represents linear regression (r2 ¼ 0.07, P ¼ 0.005). 6

eGFR and the increasing difference between estimated central and invasively measured aortic SBP. Adjustment for potential confounders did not substantially affect the association, but the r2 level of the described association demonstrates that other factors related to CKD influence pulse pressure amplification. Elevated stiffness of the muscular arteries due to media sclerosis is a predominant feature of CKD and is reflected in the present study by increasing cr-PWV with decreasing eGFR. A less compressible brachial artery would affect the brachial calibrating BP. However, we found no association between cr-PWV and the observed BP differences. Thus, changes in brachial artery stiffness are not an obvious explanation for our findings. However, future studies could include more thorough evaluations of vascular changes encompassing measurements of intima-media thickness, arterial distensibility, and calcification imaging. Increased albumin excretion has been related to cardiovascular disease and outcome in CKD patients, but we did not detect any association between albuminuria and the difference between estimated and invasive aortic SBP.20 Hydration, plasma concentrations of multiple vasoactive factors as well as sympathetic nervous activity are augmented in CKD potentially influencing the relation between brachial and aortic BP.21 Such factors, not measured in this study, could also contribute to our findings. The validity of the generalized transfer factor was assessed using invasively obtained aortic BP for calibration. No significant difference was found between CKD patients and control patients. However, with decreasing eGFR, the difference between estimated and invasive aortic SBP increased. This might indicate that a specific transfer function should be constructed for patients with severe kidney disease. Alternatively, central BP should be estimated based on pressure curves obtained from the common carotid artery as suggested for patients with isolated systolic hypertension, very old patients, or patients with end-stage renal disease.22 Many participants received several antihypertensive medications. The influence of these on aortic and brachial BP is complex and may differ between drug classes, and it was not possible for us to evaluate the role of individual drugs on our data.23 However, using multivariate regression, we found no association between SBP differences and the total number of drugs. Our measurements were performed several minutes after completing the CAG procedure. Consequently, the administration of glycerol nitrate or adenosine, administered at the discretion of the operator, is not suspected to affect the results. CKD patients and control patients differed in terms of CAG indication, which may introduce confounding by indication. Symptoms of ischemic heart disease increase the likelihood of vascular disease and elevated arterial stiffness being present and PWV actually explained part of the difference between estimated central and aortic SBP. Thus, results from patients without vascular disease, and thereby possibly lower PWV, might have yielded different results. In Kidney International (2016) -, -–-

clinical investigation

RK Carlsen et al.: Estimated central BP underestimates measured BP

b

200 180 160 140 120 100 80 80

100

120

140

160

180

200

c 40

Brachial minus aortic systolic BP (mm Hg)

220

Brachial minus aortic systolic BP (mm Hg)

Brachial systolic BP (mm Hg)

a

20

0

-20

-40

-60

220

0

20

40

60

80

100

120

140

40 30 20 10 0 -10 -20 -30 -40 4

6

2

Estimated central systolic BP (mm Hg)

8

10

12

14

16

18

20

22

Carotid-femoral PWV (m/s)

eGFR (ml/min per 1.73 m )

Figure 6 | Brachial blood pressure (BP) compared with estimated central BP, kidney function, and arterial stiffness. (a) The relation between brachial and estimated central systolic blood pressure. The lines represent linear regression with 95% confidence intervals (r2 ¼ 0.92, P < 0.001). (b) Brachial minus aortic systolic BP as function of estimated glomerular filtration rate (eGFR). The line represents linear regression (r2 ¼ 0.08, P ¼ 0.002). (c) Brachial minus aortic systolic BP as function of carotid-femoral pulse wave velocity (PWV). The line represents linear regression (r2 ¼ 0.07, P ¼ 0.007).

particular, patients in the CKD stage 5 group may be confounded by selection bias because patients with severe cardiovascular disease are excluded from transplantation. Thus, some of the patients are most likely healthier than the average CKD stage 5 population. This could lead to an underestimation of the difference between estimated central and aortic SBP. Our CKD group was heterogeneous regarding age. The association between SBP differences and eGFR became even more significant in multivariate regressions when adjusting for age, supporting the theory that aging leads to a decreased brachial-aortic amplification. This could also indicate that the increasing difference between estimated central and aortic SBP with decreasing kidney function may even be underestimations. Aortic BP could be a more accurate representative of the pressure exerted on vital organs such as the heart, kidneys, and brain and thus potentially could be superior to brachial BP in terms of predicting cardiovascular outcome. Because invasive central BP is difficult to obtain, no study has yet tested this hypothesis. Instead noninvasive estimates of central BP are utilized, but even in a meta-analysis with large cohorts of patients, the added value of estimated central BP to office brachial BP in predicting long-term cardiovascular morbidity was nonsignificant.24 Our results and those obtained by others show brachial and estimated central SBP to be very closely associated.25 This indicates that estimated central BP is very dependent on the calibrating BP. Indeed,

the need for calibration of the SphygmoCor with brachial BP is the Achilles heel of this methodology, which our study also highlights. A recent meta-analysis also showed brachial BP not to be significantly different from aortic BP, but with SphygmoCor-derived estimates being considerably more inaccurate.26 Several devices are available for measurement of PWV and the results obtained may vary due to different algorithms. However, the variation is mainly due to inconsistences in the distance measurement, whereas transit times are highly comparable.27 Therefore, the use of other devices would likely have yielded similar conclusions although this must be tested in separate studies. Invasive aortic SBP is progressively underestimated with decreasing renal function and increasing arterial stiffness in patients with CKD stages 3 to 5 using radial tonometry and brachial BP for calibration. Current antihypertensive treatment strategies for CKD patients are entirely based on brachial BP with the optimal level in CKD being continuously debated.28 Due to the incremental decrease in aortic-brachial BP amplification, CKD stages 4 and 5 patients should probably have a lower brachial BP as compared to CKD stage 3 patients in order to obtain similar aortic BP levels. Currently, antihypertensive treatment seems more reliable when based on current standard oscillometric measurements of brachial BP than Sphygmocor-based estimates of central BP. It still remains unknown whether antihypertensive treatment guided on central BP improves renal or cardiovascular outcome in

Table 5 | Previous studies comparing invasive and estimated central SBP

Patients (n) Age (yr) Estimated aortic SBP (mm Hg) Brachial aortic SBP (mm Hg)

Cloud13

Davies14

Smulyan15

Zuo16

Weber17

Laugesen18

30 64  13 –13.3  15.1 1.9  15.8

28 60  10 –7.2  10.1 3.4  10.5

50 54  13 –1.5  11.3 10.9  12.7

45 62  12 –4.2  16.7 8.3  12.5

30 59  11 6.2  8.9 8.8  10.4

34 65  11 –10.2  6.5 4.0  7.9

SBP, systolic blood pressure. Data are mean  SD, unless otherwise indicated. All studies used SphygmoCor calibrated with brachial systolic and diastolic blood pressure obtained by oscillometric measurements. Kidney International (2016) -, -–-

7

clinical investigation

CKD. However, if this idea is to be pursued, it is essential to assess central BP as correctly as possible. MATERIALS AND METHODS Inclusion Eligible participants were referred for elective CAG. CKD was defined as eGFR < 60 ml/min/1.73 m2 for at least 3 months corresponding to stages 3 to 5 according to the current guidelines.29 eGFR was determined by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation30 and based on the average of at least 3 measurements within the previous 3 months. Control patients had eGFR $ 75 ml/min/1.73 m2 without signs of structural kidney disease and diabetes (defined as diagnosed diabetes and/or glycosylated hemoglobin > 48 mmol/mol in agreement with current guidelines).31,32 Exclusion criteria in both groups were cardiac arrhythmia (e.g., atrial fibrillation) or diagnosed subclavian or brachial artery stenoses. The study was approved by the Research Ethics Committee of Central Denmark Region and the Danish Data Protection Agency. All patients gave their written informed consent. Biochemistry and echocardiography Blood samples were collected before the CAG procedure and analyzed for creatinine and glycosylated hemoglobin. Urine samples were analyzed for albumin and creatinine. Left ventricular ejection fraction measured by echocardiography was obtained from the medical records if performed within 3 months of the CAG. BP measurements Following the CAG procedure, the BP in the ascending aorta was measured with a fluid-filled 6-F Boston Scientific Expo Angiographic catheter (Natick, MA) or a 6-F Medtronic Launcher Coronary Guide catheter (Minneapolis, MN) attached to a NAMIC transducer (Navilyst Medical, Marlborough, MA). The catheters were 100 cm long with an internal diameter of 1.4 and 1.8 mm, respectively. Pressure transducers were placed at the midaxillary line and calibrated to 0 before each examination. Catheters were inserted through a femoral or radial sheath into the ascending aorta and were flushed every 2 minutes during the investigation. SBP and DBP were determined as the average of the pressure peaks during 10-second recording periods, and MAP was determined as the area under the curve by the recording equipment software (Philips Xper Physiomonitoring 5, Philips, Amsterdam, the Netherlands; or Siemens Axiom Sensis XP, Siemens, Munich, Germany). Estimated central BP was determined based on the radial artery pressure waveform recorded noninvasively using the SphygmoCor equipment and software (version 8.2, AtCor Medical, Sydney, Australia). Recording time was 10 seconds and only high-quality data with a minimum operator index of 80 were accepted (median operator index was 98). The waveforms were calibrated with the brachial SBP and DBP. Brachial BP was measured with a Microlife BP A100 PLUS automatic oscillometric BP monitor (Widnau, Switzerland), which fulfills current recommendations.33 The BP monitor was tested for accuracy and precision twice during the study. Cuff size was selected to fit the circumference of the upper arm. Except for dialysis patients with an arteriovenous fistula, all participants had BP measurements on both arms and if the difference in SBP or DBP exceeded 5 mm Hg, the arm with the higher pressure was used. Otherwise BP measurements were recorded on the left arm or the arm without a fistula. 8

RK Carlsen et al.: Estimated central BP underestimates measured BP

Invasive aortic and estimated central BP was recorded simultaneously for 10 seconds, and immediately after this, the brachial BP was measured. In this way all BP measurements were performed within <30 seconds. The procedure was performed twice for each patient. Following the examinations, the SphygmoCor was calibrated with (i) invasive diastolic BP and MAP to examine the precision and accuracy of the transfer function per se and (ii) with brachial systolic and diastolic BP to calculate estimates of aortic BP based on noninvasive calibration as recommended by the manufacturer. Measurement of PWV PWV was measured before CAG in a quiet room after at least 10 minutes of supine rest. Cf-PWV and cr-PWV were measured sequentially with applanation tonometry on the carotid, femoral, and radial arteries with the SphygmoCor device with simultaneous recording of electrocardiogram. Distance measurements for calculation of cf-PWV were performed as a straight line from the carotid to the femoral artery multiplied by 0.8 according to current recommandations.34,35 Cr-PWV measurements were obtained subtracting the carotid-suprasternal notch distance from the radialsuprasternal notch distance. Capture time was set to 10 seconds, and transit time was assessed by the intersecting tangent algorithm. Data were only accepted if 2 measurements were recorded. All PWV and radial tonometry measurements were performed by 1 trained investigator. Statistical analysis Data were analyzed with Stata 13 for Windows (StataCorp LP, College State, TX). Distributions were tested by histograms and QQ plots. Baseline data are presented as mean  SD or median (interquartile range) if data were skewed. Categorical variables are presented as numbers (%) and compared using chi-square test. Agreement between SBP values were assessed as described by Bland and Altman.36 Differences between CKD patients and control patients were tested using unpaired Student’s t-test. If data were skewed, even after logarithmic transformation, groups were compared with a Wilcoxon-Mann-Whitney test. The association between estimated central minus invasive aortic SBP and kidney function (CKD stage or eGFR) was assessed by univariate and multivariate linear regression. Likewise the association between brachial minus invasive aortic SBP and kidney function was assessed by univariate and multivariate linear regression. The multivariate regression included age, sex, ejection fraction, diabetes (yes/no), current smoking, and number of antihypertensive drugs. Interactions were investigated between eGFR and both sex and age. Univariate and multivariate linear regression was used to assess associations between PWV and the difference between estimated central SBP and invasive aortic SBP. The multivariate regression included age, sex, brachial SBP at the PWV investigation, ejection fraction, diabetes, current smoking, and number of antihypertensive drugs. Values are presented as means with 95% confidence intervals unless otherwise stated. P < 0.05 was considered statistically significant. DISCLOSURE

SW has received a research grant and consulting fee from Acarix A/S. All the other authors declared no competing interests. ACKNOWLEDGMENTS

The study has been supported by Aarhus Universitets Forskningsfond, Grosserer L.F. Foghts Fond, Familien Hede Nielsens Fond, and the Kidney International (2016) -, -–-

RK Carlsen et al.: Estimated central BP underestimates measured BP

Danish Diabetes Academy, which is supported by the Novo Nordisk Foundation. The staff at the Cardiology Lab at Aarhus University Hospital is thanked for their collaboration. SUPPLEMENTARY MATERIAL Figure S1. Graphs and linear regression lines demonstrating the relation between estimated glomerular filtration rate (eGFR) and (A) brachial systolic blood pressure (BP) (r2 ¼ 0.02, P ¼ 0.10), (B) estimated central systolic BP (r2 ¼ 0.02, P ¼ 0.15), and (C) invasive aortic systolic BP (r2 ¼ 0.06, P ¼ 0.005). Table S1. Univariate and multivariate linear regression on the differences between estimated central systolic blood pressure (SBP) and aortic SBP or brachial SBP and chronic kidney disease (CKD) stage (ordinal variable): 0 for control patients, 1 for CKD stage 3, 2 for CKD stage 4, and 3 for CKD stage 5. *Adjusted for age, sex, diabetes, ejection fraction, number of antihypertensive drugs, and carotidfemoral pulse wave velocity. Supplementary material is linked to the online version of the paper at www.kindey-international.org. REFERENCES 1. Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work Group. KDIGO 2012 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease. Kidney Int Suppl. 2013;3:1–150. 2. Bidani AK, Griffin KA, Epstein M. Hypertension and chronic kidney disease progression: why the suboptimal outcomes? Am J Med. 2012;125: 1057–1062. 3. McEniery CM, Yasmin, McDonnell B, et al. Central pressure: variability and impact of cardiovascular risk factors: the Anglo-Cardiff Collaborative Trial II. Hypertension. 2008;51:1476–1482. 4. Roman MJ, Devereux RB, Kizer JR, et al. Central pressure more strongly relates to vascular disease and outcome than does brachial pressure: the Strong Heart Study. Hypertension. 2007;50:197–203. 5. Cohen DL, Townsend RR. Central blood pressure and chronic kidney disease progression. Int J Nephrol. 2011;2011:407801. 6. Yuan M, Moody WE, Townend JN. Central blood pressure in chronic kidney disease: latest evidence and clinical relevance. Curr Hypertens Rev. 2014;10:99–106. 7. O’Rourke MF, Adji A. Noninvasive studies of central aortic pressure. Curr Hypertens Rep. 2012;14:8–20. 8. Chirinos JA, Khan A, Bansal N, et al., for the CRIC Study Investigators. Arterial stiffness, central pressures, and incident hospitalized heart failure in the chronic renal insufficiency cohort study. Circ Heart Fail. 2014;7: 709–716. 9. Peters CD, Kjaergaard KD, Jensen JD, et al. No significant effect of angiotensin II receptor blockade on intermediate cardiovascular end points in hemodialysis patients. Kidney Int. 2014;86:625–637. 10. Covic A, Goldsmith DJ, Panaghiu L, et al. Analysis of the effect of hemodialysis on peripheral and central arterial pressure waveforms. Kidney Int. 2000;57:2634–2643. 11. Verbeke F, Maréchal C, Van Laecke S, et al. Aortic stiffness and central wave reflections predict outcome in renal transplant recipients. Hypertension. 2011;58:833–838. 12. Sedaghat S, Dawkins Arce FG, Verwoert GC, et al. Association of renal function with vascular stiffness in older adults: the Rotterdam study. Age Ageing. 2014;43:827–833. 13. Cloud GC, Rajkumar C, Kooner J, et al. Estimation of central aortic pressure by SphygmoCor requires intra-arterial peripheral pressures. Clin Sci (Lond). 2003;105:219–225. 14. Davies JI, Band MM, Pringle S, et al. Peripheral blood pressure measurement is as good as applanation tonometry at predicting ascending aortic blood pressure. J Hypertens. 2003;21:571–576. 15. Smulyan H, Siddiqui DS, Carlson RJ, et al. Clinical utility of aortic pulses and pressures calculated from applanated radial-artery pulses. Hypertension. 2003;42:150–155. 16. Zuo JL, Li Y, Yan ZJ, et al. Validation of the central blood pressure estimation by the SphygmoCor system in Chinese. Blood Press Monit. 2010;15:268–274.

Kidney International (2016) -, -–-

clinical investigation

17. Weber T, Wassertheurer S, Rammer M, et al. Validation of a brachial cuffbased method for estimating central systolic blood pressure. Hypertension. 2011;58:825–832. 18. Laugesen E, Rossen NB, Peters CD, et al. Assessment of central blood pressure in patients with type 2 diabetes: a comparison between SphygmoCor and invasively measured values. Am J Hypertens. 2014;27: 169–176. 19. Wassertheurer S, Burkhardt K, Heemann U, Baumann M. Aortic to brachial pulse pressure amplification as functional marker and predictor of renal function loss in chronic kidney disease. J Clin Hypertens (Greenwich). 2014;16:401–405. 20. Matsushita K, van der Velde M, Astor BC, et al., for the Chronic Kidney Disease Prognosis Consortium. Association of estimated glomerular filtration rate and albuminuria with all-cause and cardiovascular mortality in general population cohorts: a collaborative meta-analysis. Lancet. 2010;375:2073–2081. 21. Rossignol P, Massy ZA, Azizi M, et al., for the ERA-EDTA EURECA-m Working Group. REDINREN Network, F-CRIN INI-CRCT Network. The double challenge of resistant hypertension and chronic kidney disease. Lancet. 2015;386:1588–1598. 22. Safar ME, Blacher J, Pannier B, et al. Central pulse pressure and mortality in end-stage renal disease. Hypertension. 2002;39:735–738. 23. Morgan T, Lauri J, Bertram D, Anderson A. Effect of different antihypertensive drug classes on central aortic pressure. Am J Hypertens. 2004;17:118–123. 24. Vlachopoulos C, Aznaouridis K, O’Rourke MF, et al. Prediction of cardiovascular events and all-cause mortality with central haemodynamics: a systematic review and meta-analysis. Eur Heart J. 2010;31:1865–1871. 25. O’Rourke MF, Pauca A, Jiang XJ. Pulse wave analysis. Br J Clin Pharmacol. 2001;51:507–522. 26. Narayan O, Casan J, Szarski M, et al. Estimation of central aortic blood pressure: a systematic meta-analysis of available techniques. J Hypertens. 2014;32:1727–1740. 27. Rajzer MW, Wojciechowska W, Klocek M, et al. Comparison of aortic pulse wave velocity measured by three techniques: Complior, SphygmoCor and Arteriograph. J Hypertens. 2008;26:2001–2007. 28. Wheeler DC, Becker GJ. Summary of KDIGO guideline: what do we really know about management of blood pressure in patients with chronic kidney disease? Kidney Int. 2013;83:377–383. 29. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis. 2002;39(suppl 1):S1–266. 30. Levey AS, Stevens LA, Schmid CH, et al., for the CKD-EPI Collaboration. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150:604–612. 31. World Health Organization. Use of Glycated Haemoglobin (HbA1c) in the Diagnosis of Diabetes Mellitus: Abbreviated Report of a WHO Consultation 2011. Geneva, Switzerland: World Health Organization; 2013. 32. Rydén L, Grant PJ, Anker SD, et al. ESC guidelines on diabetes, prediabetes, and cardiovascular diseases developed in collaboration with the EASD: the Task Force on diabetes, pre-diabetes, and cardiovascular diseases of the European Society of Cardiology (ESC) and developed in collaboration with the European Association for the Study of Diabetes (EASD). Eur Heart J. 2013;34:3035–3087. 33. Belghazi J, El Feghali RN, Moussalem T, et al. Validation of four automatic devices for self-measurement of blood pressure according to the International Protocol of the European Society of Hypertension. Vasc Health Risk Manag. 2007;3:389–400. 34. Huybrechts SA, Devos DG, Vermeersch SJ, et al. Carotid to femoral pulse wave velocity: a comparison of real travelled aortic path lengths determined by MRI and superficial measurements. J Hypertens. 2011;29: 1577–1582. 35. Van Bortel LM, Laurent S, Boutouyrie P, et al., for the Artery Society, European Society of Hypertension Working Group on Vascular Structure and Function, European Network for Noninvasive Investigation of Large Arteries. Expert consensus document on the measurement of aortic stiffness in daily practice using carotid-femoral pulse wave velocity. J Hypertens. 2012;30:445–448. 36. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1: 307–310.

9