Journal of (1998) 12, 229–233  1998 Stockton Press. All rights reserved 0950-9240/98 $12.00

ORIGINAL ARTICLE Can rate predict response to anti-hypertensive drug therapy?

PE Owens, S Lyons and E O’Brien Blood Pressure Unit, Beaumont Hospital, Dublin, Republic of Ireland

The use of heart rate in clinical practice is limited by arterial BP from the initial to the subsequent ABPM (ie, its variability under measurement situations. The mean as a result of treatment) was correlated with the mean heart rate on ambulatory provides a more heart rate on the initial ABPM. A moderate association robust statistic for clinical use. We examined the was found for initial daytime heart rate and BP response relationship between mean heart rate on initial referral to beta-blockers (r = 0.24, P = 0.02), and ACE inhibitors ambulatory blood pressure monitoring (ABPM) to the (r = 0.14, P = 0.05). No such association was found for BP-lowering efficacy of the four main groups of anti- calcium antagonists or diuretics. When the groups were hypertensive medications, in a referral hypertensive divided into those with a mean daytime heart rate Ͻ75 population. Patients were retrospectively identified by and у75 beats per min, BP reduction from beta-blocker review of the BP database, and data collected from the and ACE inhibitor therapy was significantly greater in initial referral BP monitor off medication, and the sub- those patients with a higher daytime heart rate. We con- sequent ABPM after treatment with either beta-blockers, clude that average daytime heart rate on pre-treatment diuretics, calcium antagonists or -con- ABPM can be useful as a predictor of BP response to verting enzyme (ACE) inhibitors. The change in mean beta-blockade or ACE inhibition.

Keywords: ambulatory blood pressure monitoring; heart rate; drug therapy

Introduction mented that a high resting heart rate may in fact be predictive of future hypertension.5 The heart rate is one of the cardinal , and The hypothesis we wished to test therefore was it has been well described in the normal population whether high mean heart rates as documented on and in various pathological states. The primary ABPM were predictive of the likelihood of patients drawback of reliance on heart rate as a disease indi- to respond to anti-hypertensive medications, in cator has been its great variability over short periods particular to direct sympathetic nervous system 1 of time in response to external stresses. The innate blockade with beta-blocker medication. variability of the heart rate may be lessened by fol- lowing the rate over a period of time, where instan- taneous variation is hidden by the more robust stat- Patients and methods istical averaging of heart rate. Similar mechanisms Patient identification have been used to more accurately determine the true blood pressure (BP) with the use of ambulatory Patients were identified from a retrospective search blood pressure monitoring (ABPM). of the Blood Pressure Unit ambulatory BP database. Heart rate and BP are closely correlated, and Specifically, patients were selected if they had had hypertensive patients are known to have higher an ambulatory BP monitor performed on no medi- resting heart rates than normals.2 The increased cations (which is a routine part of the assessment of sympathetic nervous system tone would appear to patients referred for evaluation of BP in our unit), be the causative element in this increase in heart and had had a subsequent ambulatory BP monitor rate along with a general increase in performed within 1 year of the first, after having and the production of a hyperkinetic circulation.3 been commenced on a single anti-hypertensive This pattern would appear to be most marked in medication. The median length of time between younger hypertensives,4 and it has been docu- each ABPM was 6 months (range 2–12). Patients were excluded if there was not a clear identification of medication history in the time period of the two monitors, if they had been commenced on combi- Correspondence: Dr P Owens, Blood Pressure Unit, Beaumont nation anti-hypertensive medication, or had a Hospital, PO Box 1297, Beaumont Road, Dublin 9, Republic of Ireland coincident medical problem interfering with the Received 30 September 1997; revised and accepted 12 Nov- estimation of heart rate (eg, atrial fibrillation, perma- ember 1997 nent pacemaker in situ), or likely to affect heart rate Heart rate and anti-hypertensive therapy PE Owens et al 230 (eg, clinical heart failure). Patients with secondary (IHR), the average of the heart rate in the first 1 h of hypertension, identified by a secondary hyperten- recording; the mean daytime heart rate (DHR), aver- sion screening protocol at the time of the first clinic aged over the daytime hours as defined above; and visit, were not included in the study. Echocardio- the mean night-time heart rate (NHR), averaged over graphic parameters of impaired left ventricular func- the night-time hours as defined above. The BP data tion, namely an of less than 40% was calculated to give the mean daytime mean or a dilated left ventricular cavity were identified arterial pressure (MAP), and the mean night-time from the patients records, and warranted the MAP. The BP response to anti-hypertensive therapy removal of the patients record from the analysis. The was calculated as the difference between mean day- patients were prescribed the anti-hypertensive time MAP and mean night-time MAP between the medication at their initial clinic visit, and the first and second monitors. decision as to which medication to prescribe was clinically based, and was based on the ABPM meas- Statistical analysis ured pressure. Patients were grouped into groups on the basis of the drug they were prescribed at the first The following analyses were made. Firstly, the cor- clinic visit, which they continued to take until the relation between DHR and NHR on the initial ABPM time of the second monitor. Specifically four drug and the BP response to anti-hypertensive medi- classes were identified: diuretics; beta-blockers; cation was investigated using simple bivariate corre- angiotensin-converting enzyme inhibitors (ACEI); lation. Then, in order to determine whether clini- calcium channel blockers. cally identifiable heart rate bands existed where specific drug interventions were most useful, the Ambulatory BP monitoring patient population was divided into two groups according to diurnal heart rate on the initial ABPM; Initial clinic BPs were measured by a trained nurse those with a DHR Ͻ75 bpm, and those with a DHR according to the British Hypertension Society BP у75 bpm. Differences in anti-hypertensive drug measurement protocol.6 ABPM was performed using efficacy between these groups were explored using a validated device. As the database was extended Student’s t-test. Differences between more than two over 15 years, a variety of devices were used, which groups were explored using analysis of variance, included the Remler-M2000, Accutracker I, Space- with probability adjustment using Bonferroni’s Labs 5200, Tapeda TM-2420, SpaceLabs 90202, method. SpaceLabs 90207, NC-Diasys.200R, CH-Druck, Pro- filomat I and II, Nisser DS-240, and Schiller BR102. In practice, 89.4% of the first BP monitors and Results 93.8% of the subsequent monitors were either Spa- A total of 437 suitable patients were identified. ceLabs 90202 or SpaceLabs 90207. The monitors Mean age was 50.7 (CI 49.7–51.8). The population were applied in accordance with the protocol for showed a mean BP at initial assessment (on no application of ambulatory BP monitors in our unit; medication) of 117.3 (CI 116.3–118.3) mm Hg during the monitor was applied to the non-dominant arm the day, and 98.7 (CI 97.7–99.8) mm Hg at night of between 9.00 am and noon, and the patient was which 24.5% were subsequently prescribed diuretic instructed to carry on life as normal between read- monotherapy, 81% of whom received bendrofluaz- ings but to have the arm rested at heart level during ide with potassium supplementation, at a median readings. The monitors were programmed to meas- dose of 5 mg per day. Other diuretics used were ure BP every half-hour, day and night. The data from triamterene/hydrochlorothiazide (median dose 50 the ambulatory BP monitor was downloaded the mg/25 mg), amilioride/hydrochlorothiazide (median next day into a personal computer and uploaded dose 5 mg/50 mg), and frusemide/amilioride into a specialised software package (DABLv). The (median dose 40 mg/5 mg). Beta-blockers were pre- initial, daytime and night-time systolic diastolic and scribed to 22.2%, of whom 67.3% were prescribed mean BP statistics and the mean daytime and night- atenolol (median dose 100 mg), and 25.0% were pre- time heart rate was calculated. The ‘initial’ period scribed metoprolol (median dose 100 mg). Other was defined as the first 1 h of monitoring, and was beta-blockers used were sotalol (median dose 80 mg) considered apart from the daytime readings so as to and celiprolol (median dose 400 mg); 13.3% of exclude interference from white coat effects, ‘day- patients were prescribed calcium channel blockers, time’ was defined as the hours between 9.00 am and 32.6% of whom were given nifedipine slow release 9.00 pm, and night-time as the hours between (SR), at a median dose of 40 mg; 30.4% were given 1.00 am and 6.00 am. These intervals are chosen in amlodipine (median dose 5 mg), 13.0% were given order to remove the influence of retiring and rising nicardipine (median dose 90 mg); 87% of patients time fluctuations in the assessment of day and night- were on dihydropyridine calcium channel blockers, time BP. Monitors were checked and downloaded all the others being on (median dose the same day as the monitor was removed, and 270 mg). However, of the patients taking dihydropy- erroneous measurement values were not included in ridines, 82.5% were on sustained release or long- the statistical processing of the BP data. acting preparations. Forty per cent of patients were prescribed ACE inhibitors, 52.5% of whom were Definitions given lisinopril (median dose 5 mg), 18.2% were For the purposes of comparison, the heart rate was prescribed perindopril (median dose 4 mg), and divided into three subsets; the initial heart rate 14.1% were prescribed enalapril (median dose Heart rate and anti-hypertensive therapy PE Owens et al 231 Table 1 Clinical characteristics of patients at the time of the initial ABPM. P values are from one-way ANOVA testing for differences across groups. Figures are means (95% confidence intervals)

Beta-blocker ACEI Calcium antagonist Diuretic P value group group group group n 97 175 58 107 Age 49.2 (46.7–51.6) 50.3 (48.5–52.0) 52.1 (48.9–55.3) 55.8 (53.3–58.4) Sex (m/f) 61/36 104/71 31/27 43/64 BMI 26.4 (25.5–27.3) 27.4 (26.8–28.1) 26.2 (25.2–27.3) 25.8 (25.0–26.6) Smoking % 31.2 38.0 32.1 33.0 Referral SBP 177.5 (171.6–183.4) 171.9 (168.3–175.5) 177.9 (170.8–185.0) 180.3 (175.1–185.4) ns Referral DBP 105.6 (102.6–108.5) 103.5 (101.8–105.2) 106.1 (102.4–109.8) 103.9 (101.0–106.9) ns Creatinine 93.9 (90.2–97.5) 93.1 (89.3–97.0) 91.5 (85.2–97.9) 88.2 (83.9–92.6) ns Cholesterol 5.57 (5.28–5.85) 5.85 (5.58–6.13) 6.07 (5.68–6.45) 5.89 (5.47–6.31) ns Sugar 5.7 (5.0–6.4) 5.4 (5.2–5.7) 5.2 (5.0–5.4) 5.4 (5.1–5.6) ns

SBP = systolic blood pressure; DBP = diastolic blood pressure; BMI = body mass index; ns = non-significant.

5 mg). Other ACE inhibitors used were captopril documented between diurnal mean heart rate and (median dose 50 mg), and quinapril (median dose day (r = 0.13) and night-time (r = 0.14) BP response 15 mg). to ACE inhibition. No relationship was found for The patient characteristics at the time of first mean heart rate on the initial ABPM and BP monitoring for the four groups are shown in Table response to calcium channel blockers or diuretics. 1. Age was broadly similar across the beta-blocker, The patient population was then grouped into calcium antagonist and ACEI groups, but was sig- patients with a mean DHR on the initial ABPM nificantly higher in the diuretic group. This popu- Ͻ75 bpm, and those with a DHR у75 bpm. These lation also was alone in having a primarily female groups were then compared for response to anti- composition. This was felt to be consistent with a hypertensive treatment within the four drug classes, prescription bias in elderly populations toward to determine whether there existed patient sub- diuretic therapy for isolated systolic hypertension, groups identifiable on the basis of resting mean day- and a predominance of females in the more elderly time heart rate, who responded better to specific patient population. Marginal differences were noted drug treatments. The data is summarised in Table in body mass index, this variable being slightly 4. Patients in the higher heart rate group responded lower in the diuretic group. better to beta-blockers than did those in the lower The heart rate and BP values on the first ABPM heart rate group, with respect to the fall in mean are presented in Table 2. Blood pressure values were daytime BP. Similarly, a reduction in day and night- generally comparable across groups. The diurnal time BP was observed in the ACE inhibitor group, and nocturnal systolic BP was significantly lower in with significantly greater falls in BP apparent in the the beta-blocker group when compared with the cal- higher heart rate group. cium antagonist group. Heart rate during daytime, at night and in the initial period was similar. Discussion The change in mean daytime and night-time BP from the first, unmedicated monitor to the second The value of tailored drug therapy for management monitor on medicines is displayed in Table 3, along of disease is that the therapeutic regimen used will with the calculated correlation coefficient and P be efficient with respect to the cost, benefit and side- value for association with the daytime mean heart effect profile of the agent used. The drugs used to rate on the first monitor. The closest association was treat hypertension are particularly pertinent here, as for beta-blockers, where the degree of diurnal BP treatment is likely to last for a number of years, and reduction from treatment was directly proportional compliance with medicines may be less than per- to the mean daytime heart rate on the first ABPM (r fect. The choice of which agent to use is largely a = 0.26). A similar, albeit weaker association was also matter of personal preference, in that most research

Table 2 Baseline ABPM data from the four drug groups, prior to medicating. P values are from one-way ANOVA testing for differences across groups. Figures are means (95% confidence intervals)

Beta-blocker ACEI Calcium antagonist Diuretics P value

Initial SBP 159.2 (155.8–162.6) 163.5 (160.5–166.4) 166.5 (161.2–171.7) 166.0 (162.0–169.9) ns Initial DBP 102.4 (99.8–105.1) 103.2 (101.4–104.9) 100.8 (97.6–104.1) 100.1 (97.6–102.5) ns Initial HR 79.2 (76.1–82.2) 81.0 (79.0–83.1) 79.0 (75.7–82.3) 80.2 (77.8–82.6) ns Day SBP 153.3 (150.6–156.1) 157.3 (154.8–159.7) 160.8 (157.2–164.4) 158.0 (155.1–160.9) 0.02 Day DBP 97.5 (95.2–99.8) 98.3 (96.6–100.0) 98.5 (95.6–101.4) 94.9 (92.8–97.1) ns Day HR 79.7 (77.5–81.8) 80.9 (79.1–82.7) 80.1 (76.9–83.3) 80.0 (77.8–82.2) ns Night SBP 132.2 (128.7–135.6) 133.3 (130.8–135.8) 139.8 (136.4–143.2) 136.1 (132.6–139.7) 0.03 Night DBP 81.4 (78.8–83.9) 80.6 (78.8–82.4) 84.2 (81.8–86.5) 78.9 (76.8–81.1) ns Night HR 66.7 (64.7–68.7) 67.1 (65.5–68.7) 65.9 (63.6–68.1) 65.5 (63.6–67.4) ns Heart rate and anti-hypertensive therapy PE Owens et al 232 Table 3 Change in mean daytime (⌬Day MAP), and night-time (⌬Night MAP) between the first and second ABPM. Correlation coefficients (P value) for association between ⌬Day MAP and ⌬Night MAP associated for daytime heart rate (DHR), and night-time heart rate (NHR) on the first monitor. Figures are means (95% confidence intervals)

Beta-blocker ACEI Calcium antagonist Diuretic

⌬Day MAP 10.1 (7.5–12.6) 11.1 (9.3–13.0) 9.8 (7.0–12.6) 9.3 (7.6–11.1) ⌬Night MAP 6.4 (3.4–9.4) 7.1 (5.2–8.9) 7.0 (4.7–9.2) 8.2 (5.9–10.5) ⌬Day MAP vs DHR 0.24 (0.02) 0.13 (0.05) 0.07 (ns) 0.14 (ns) ⌬Night MAP vs DHR 0.07 (ns) 0.14 (0.05) −0.06 (ns) −0.02 (ns) ⌬Day MAP vs NHR 0.11 (ns) 0.12 (ns) 0.07 (ns) 0.06 (ns) ⌬Night MAP vs NHR 0.00 (ns) 0.04 (ns) −0.11 (ns) 0.02 (ns)

Table 4 Change in blood pressure in response to anti-hypertensive drug therapy, in groups based on mean DHR Ͻ 75 bpm or у75 bpm on the initial ABPM. Figures are means (95% confidence intervals)

Beta-blockers ACEI Calcium antagonists Diuretics

⌬Day MAP DHR у75 11.9 12.7 9.9 9.8 DHR Ͻ75 6.2 7.4 9.4 8.3 Difference (95% CI) 5.7 (0.5–10.8) 5.2 (1.2–9.3) 0.53 (−5.7–6.8) 1.5 (−2.2–5.3) P value 0.03 0.01 ns ns

⌬Night MAP DHR у75 7.8 8.2 6.7 8.4 DHR Ͻ75 3.4 4.1 6.6 7.7 Difference (95% CI) 4.4 (−1.9–10.7) 4.2 (0.2–8.1) 0.1 (−5.4–5.5) 0.75 (−4.3–5.8) P value ns 0.04 ns ns

to date has concentrated on treatment of hyperten- For both the beta-blocker group and the ACE sion as a disease state, and identification of specific inhibitor group, the strongest correlation was found drug efficacy has not been extensively studied. for daytime heart rate and degree of BP reduction. This retrospective observational study examines Night-time heart rate was not significantly associa- whether heart rate, averaged over day and night- ted with BP response to medicines. It is known that time, can be used to predict the efficacy of anti- the heart rate slows during the night, and heart rate hypertensive drug therapy of a number of different variability studies have documented a relative pre- drug groups. We have shown that for beta-blockers dominance of vagal (parasympathetic) activity at and ACE inhibitors, the degree of BP reduction this time, with an abeyance of sympathetic input consequent on treatment with these agents is related into heart rate control.12 Therefore, the correlation in a positive linear fashion to the mean day-time between mean night-time heart rate and BP response heart rate on the initial pre-treatment ABPM. The to sympatholytics is unlikely to be close. implication for practice, therefore, is that patients The finding of a modest correlation between heart noted to have an elevated daytime heart rate on rate and BP response to ACE inhibition might again ABPM (Ͼ75 bpm) will benefit in particular from be explained on the basis of the sympatho-inhibitory treatment with these agents. Furthermore, at lower effect of ACE inhibitors in hypertensive popu- heart rates (Ͻ75 bpm) these agents are less effective, lations.13,14 Angiotensin II receptors on pre-synaptic and calcium channel blockers or diuretics may be adrenergic nerve terminals modulate stimulated preferable. release of neurotransmitters—ACE What pathophysiological correlates are there to inhibition has therefore a ‘beta-blocker’ effect, explain this association? The most obvious is the which may in part explain our findings. influence of the sympathetic nervous system in the There is good reason to use a mid-70s level as the aetiology of hypertension. Abnormalities of the sym- cut-off point for ‘low’ and ‘high’ heart rate. The pathetic nervous system have been well docu- mean population rate from a variety of popu- mented in patients with hypertension,7,8 and modi- lation studies lies between 70 and 80 bpm.2 There fication of sympathetic neural outflow can is evidence from epidemiological studies15 that the dramatically reduce BP in even severe hyperten- incidence of total and coronary heart mortality rises sives.9 Elevated heart rate is a marker for sympath- as heart rate rises above 75 bpm. Furthermore, the etic activity,10 and, in studies of heart rate varia- Framingham heart study shows an increase in cor- bility,11 hypertensive patients have been found to onary heart disease for both men and women have increased LF/HF (low frequency/high with heart rate greater than the 70s quintile.16 The frequency) ratios, suggestive predominance of the choice of 75 bpm has, therefore, clinical correlates sympathetic limb of the . with cardiovascular outcome. It would follow therefore, that hypertension associa- The most significant drawback of this study is the ted with a high mean heart rate would respond best fact that, as an observational study, it can only sug- to a sympatholytic anti-hypertensive medication. gest associations, rather than objectively define con- Heart rate and anti-hypertensive therapy PE Owens et al 233 crete relationships between variables. The actual 5 Garrison RJ, Kannel WB, Stokes J. Incidence and pre- mean differences in day-time BP-lowering effect of cursors of hypertension in young adults. Prev Med beta-blockers and ACE inhibitors between the low 1987; 16: 235–251. and high heart rate groups was not large, being 5.7 6 Petrie JC et al. Recommendations on Blood Pressure and 5.2 mm Hg respectively. These differences may Measurement, second ed. London, UK. British Medical Journal Publications, 1992. be sufficient to reduce the total BP burden and so 7 Esler M et al. Mild high-renin : prevent end-organ complications. neurogenic human hypertension? N Eng J Med 1977; In conclusion, we have shown that BP response 296: 405–411. to beta-blockers and ACE inhibitors is optimal in 8 Miyajima E et al. Muscle sympathetic nerve activity in patients who show higher mean daytime heart rate renovascular hypertension and primary aldosteronism. on ABPM, and that at lower heart rates calcium Hypertension 1991; 17: 1057–1062. channel blockers and diuretics may provide a better 9 Smithwick RH. Hypertensive cardiovascular disease: control of BP. Prospective trial data is required to effect of thoracolumbar splanchnicectomy on mortality further explore this relationship. and survival rate. JAMA 1951; 147: 1611. 10 Ganong WF. Review of Medical Physiology. 13th ed. London, England. Prentice Hall International. 1987, Acknowledgements pp 502. 11 Liao D, Cai J, Barnes RW. Association of cardiac auto- The authors wish to acknowledge the continuing nomic function and the development of hypertension. support of the Royal College of Surgeons of Ireland, Am J Hypertens 1996; 9: 1147–1156. the Charitable Infirmary Charitable Trust, and the 12 Sapoznikov D, Luria MH, Mahler Y, Gotsman MS. Day Irish Heart Foundation. vs Night ECG and patterns in patients without obvious heart disease. J Electrocar- diol 1992; 25: 175–184. References 13 Minatoguchi S et al. Enalapril decreases plasma norad- renaline levels during the in human 1 Levy RL, White PD, Stroud WD, Hillman CC. Transient hypertensives. Clin Exp Pharmacol Physiol 1992; 19: -prognostic significance alone and in 279–282. association with transient hypertension. JAMA 1945; 14 Valles Prats M et al. Quinapril ACE-inhibition effects 129: 585–588. on adrenergic parameters in moderate essential hyper- 2 Palatini P, Julius S. Heart rate and the cardiovascular tension. Kidney Int Suppl 1996; 55: S104–S106. risk. J Hypertens 1997; 15: 3–17. 15 Wilhelmsen L et al. The multi-factor primary preven- 3 Julius S, Nesbitt S. Sympathetic overactivity in hyper- tion trial in Goteborg, Sweden. Eur Heart J 1986; 7: tension; a moving target. Am J Hypertens 1996; 9: 279–288. 113S–120S. 16 Kannel WB, Wilson P, Blair SN. Epidemiological 4 Messerli FH et al. Borderline hypertension: relation- assessment of the role of physical activity and fitness ship between age, haemodynamics and circulating cat- in the development of cardiovascular disease. Am echolamines. Circulation 1981; 64: 760–764. Heart J 1985; 109: 876–885.