Patient characteristics do not predict the individual response to antihypertensive medication: a cross-over trial

Patient characteristics do not predict the individual response to antihypertensive medication: a... Abstract Background International guidelines on hypertension management do not agree on whether patient characteristics can be used for the first choice of treatment of uncomplicated essential hypertension. Objective We wanted to identify predictive patient characteristics to the response of two different classes of antihypertensive drugs in patients with newly diagnosed hypertension in primary care. Methods We conducted a prospective, open label, blinded endpoint cross-over trial in 120 patients with a new diagnosis of hypertension from 10 family practices. Patients received 4 weeks of 12.5 mgr hydrochlorothiazide once daily and 4 weeks of 80 mgr valsartan once daily, each followed by a 4-week washout. The sequence of drugs was randomized. Age, sex and menopausal state were recorded at run in and 24 h ambulatory blood pressure, office blood pressure, plasma renin concentration, NT-proBNP, potassium, estimated glomerular filtration rate, urinary albumin, body mass index and waist circumference at each regimen change. The difference in systolic blood pressure response between both study drugs, calculated from mean daytime ambulatory blood pressures, was the main outcome measure. Results Ninety-eight patients (52% female; median age 53 years) were eligible for per-protocol-analysis. None of the studied variables were predictive for the difference in systolic blood pressure response. Individual systolic blood pressure responses ranged from an increase by 18 mmHg to a decrease of 39 mmHg. Conclusion In a relevant group of primary care patients with newly diagnosed hypertension, we were unable to detect predictors of treatment response. This study rather supports the United States and European guidelines than the United Kingdom and Dutch guidelines on hypertension. Cross-over, family medicine, hypertension, prediction, treatment Introduction Several meta-analyses have demonstrated that on average all known classes of antihypertensive medication are equally effective in reducing blood pressure (1,2). However, some studies have confirmed what clinicians experience in daily practice: in individual patients treatment response can differ substantially from one antihypertensive class to the other (3,4). This finding suggests that in ‘essential hypertension’ a more refined diagnostic subtyping may exist and favours the possibility of patient characteristics that could predict which antihypertensive drug works best in the individual patient. Application of a strategy using predictors of hypotensive response, may reduce polypharmacy, enhance treatment adherence and reduce costs (5,6). Research in animals and humans suggests that several patient characteristics may have predictive qualities. Several characteristics have been studied, such as ethnicity (7–9), plasma renin (4,10,11), NT-proBNP (3,12), potassium (13), waistcircumference (14), BMI (15), sex (16,17), age (18,19) and menopause (20). International consensus is lacking on the use of an age cut-off in hypertension management. In the United States and the European guidelines, an age cut-off is lacking. Age as predictor is not mentioned at all in the US guideline; the European guideline specifically states ‘no evidence is available that different choices should be made based on age or gender’ (9,21). In contrast, the UK NICE guideline has advocated the use of an age cut-off in the treatment algorithm (the A/CD rule) (22). In 2012, this rule has been adopted by the Dutch Guideline on Cardiovascular Risk Management. The rule advises in patients younger than 55 years to start treatment with an ACE inhibitor or ARB (‘A’) whereas in patients of 55 years and older CCB’s should be the first choice medication (or diuretic as alternative, ‘CD’). The NICE recommendation is based on the pathophysiological assumption that at least part of the patients with uncomplicated essential hypertension can be divided in low renin (salt sensitive or ‘volume-driven’) and high renin (vasoconstrictive) hypertension (18,23). Core evidence for the A/CD rule comes from a study demonstrating that plasma renin activity declines with age (18) and clinical data from two cross-over trials (3,4) and one randomized controlled trial (24) suggesting that age can be regarded as surrogate for plasma renin and as such as predictor of treatment response. Because hypertension is diagnosed and managed most often in primary care, studies on individual treatment response should ideally be performed in family practice based populations and reflect an age around 55–60 years, in which diagnosis is commonly made (25,26). Selection of predictive characteristics should preferably be based on the actual feasibility of implementation in daily family practice. The clinical data supporting the A/CD rule, however, do not seem to represent this family practice population. In the two cross-over trials (both with less than 40 patients completing the study) (3,4) only patients younger than 55 years were included, predominantly male (65–70%) and blood pressure measurements were in supine position. In both studies the setting of patient selection was not described, patients needed to be young and with untreated essential hypertension. In the RCT (24) only male, veteran patients were selected from 15 outpatient clinics, with a mean age of the total study population of 59 years. We decided to set up a cross-over study to explore whether predictors of treatment response could be identified in a representative family practice population and with a set of potential predictors that are feasible to implement in the family practice setting. A cross-over study design enabled us to study intra-individual treatment responses. The controversy about the use of age as a predictor of treatment response adds to the relevance of this exploration, especially in a population relevant for family practice. Our primary objective was to identify predictors of the difference in blood pressure response between two classes of antihypertensive drugs with different mechanisms of action representing the most widely acknowledged subtyping of essential hypertension in renin driven hypertension or volume driven hypertension. Methods Patients, design and setting Patients with newly diagnosed hypertension, aged 18–65 years and listed in 10 Dutch family practices affiliated with the Radboud University Medical Center (27) were screened for eligibility to participate in a cross-over study. The methodology, strengths and limitations of the type of cross-over design we used (PROBE: PRospective, Open label, Blinded Endpoint) were described by Hansson et al. (28).We started patient inclusion in August 2007 and final measurements in the last included patient were finished in February 2011. The study was registered in clinicaltrials.gov under NCT00457483. http://clinicaltrials.gov/ct2/show/study/NCT00457483?term=hypertension+and+nijmegen&rank=1. Patients in whom the diagnosis of hypertension was confirmed and who gave written informed consent were included in the study. Exclusion criteria were use of antihypertensive medication, blood pressure higher than 210/110 mmHg, inability to speak or understand Dutch and presence of cardiovascular co-morbidity (diabetes mellitus, peripheral arterial disease, ischemic heart disease, stroke, transient ischemic attack and atrial fibrillation).We excluded these co-morbid conditions because hypertension treatment algorithms for these patients differ from those in uncomplicated patients (9,21,22). After a 2-week run-in period patients followed a treatment scheme in which a patient started with either one of the following two drugs: hydrochlorothiazide 12,5 mg once daily or valsartan 80 mg once daily, for the duration of 4 weeks. These 4 weeks of treatment with one antihypertensive medicine was followed by 4 weeks of no treatment (washout period). Then the next 4 weeks of treatment started but now with the antihypertensive that was not used in the first 4 weeks. Again this period was followed by a 4-week washout period. The sequence of the use of the two medications was randomized. Included patients were assigned a medication order by the practice nurse. For this purpose, all practice nurses were given a list with a random sequence of the digits ‘0’ and ‘1’ each standing for one order of medications. This random sequence was generated by SPSS version 16 and at the start of the study each practice was given an unique random sequence. At the run-in, at the start, and at the end of each medication or washout period we measured 24-h ambulatory blood pressure (24-h ABP), office blood pressure (OBP), plasma renin concentration, NT-proBNP, plasma potassium concentration, eGFR, urinary albumin excretion, body mass index (BMI) and waist circumference. At the run-in also age, sex and menopausal state were recorded. The primary outcome measure was the prediction of the difference in mean daytime systolic blood pressure response between both study drugs determined by ambulatory blood pressure monitoring. Blood pressure measurements During the eligibility screening performed by the local staff of the family practices a confirmed diagnosis of hypertension (≥140/90mmHg) was based on the mean of blood pressure measurements taken at three different office visits. All practices were provided with two validated oscillometric office monitors (Stabil-O-Graph, I.E.M. GMBH, Stolberg, Germany) (29). These monitors were exclusively used for study patients and were re-calibrated every 2 years. The office blood pressure measurements during the study were taken by practice nurses who were trained in the methodology of standardized blood pressure measurement. All 24-h ABP recordings were taken on week days and patients were asked to keep a standardized diary of daily activities. Practices used the same, validated 24-h ABPM device (Mobil-O-Graph, I.E.M. GMBH, Stolberg, Germany) (30). The measurement interval was 20 min from 7 am to 11 pm and 1 h from 11 pm to 7 am. We defined the mean day blood pressure from 9 am to 9 pm and we used this measure as our primary blood pressure outcome. Only recordings with 70% or more valid readings were used. The display of the device was set not to show the blood pressure values. Ambulatory blood pressure values were not communicated to the patient, prescribing physician and researcher until after completion of the study protocol. Study variables and study medication We selected demographic and biochemical variables that family practitioners commonly use or that are easily implemented in daily practice. All blood and urine samples were collected in the morning (before 12.30 pm) and analyzed at one primary care orientated diagnostic centre. Sample analyses were performed batch-wise. The following laboratory tests were used: creatinine (CREA plus, Roche Diagnostics), renin (DSL-25100 ACTIVE Renin IRMA kit, Diagnostic Systems Laboratories), NT-proBNP (Elecsys, Roche Diagnostics), potassium (ISE Indirect, Roche Diagnostics) and urine albumine (COBAS, Roche Diagnostics). As requested by the local medical ethics committee, prescription of the medication was as similar to usual care as possible and thus according to the recommendations of the Dutch multidisciplinary guideline on cardiovascular risk management (31). The principle in this guideline is to start with a low dose and if a patient does not reach hypertension control in 4–8 weeks add a second medication rather than increase doses. Therefore patients were prescribed 80 mg valsartan once daily and 12.5 mg hydrochlorothiazide once daily (31). Following the PROBE design, medication was not blinded for both patients and the prescribing general practitioners. We recorded compliance and adverse events with the use of standardized questionnaires. Statistical analyses and sample size calculation First, we applied descriptive statistics to compare baseline patient characteristics of patients who completed the study with those who were lost to follow-up. We used Pearson’s test for correlation to study the relationship between the individual blood pressure responses of both medications. We used frequency plots to show the treatment response to both studied medications and to show the difference in treatment response. Next, with repeated measures analysis we checked for integrity of our cross-over design by assessing whether the mean drop in blood pressure during the medication periods was followed by a full recovery of blood pressure after the washout period. If this would not occur than it is more likely that results of the first period of the cross-over influence results of the second period of the cross-over trial and thus hampering valid interpretation of results. In addition, using a mixed model analysis (with a compound symmetry correlation matrix) we determined whether relevant carry-over, period and treatment effects occurred. These effects could introduce forms of bias specific to a cross-over design. When these effects prove not to be large and not statistically significant conclusions drawn are (more) reliable. In the final step, according to a per protocol analysis on all patients that completed the second medication period, we performed both univariate and multivariate regression analysis with the difference in mean systolic blood pressure response (blood pressure at start of treatment minus blood pressure after 4 weeks of treatment) between valsartan and hydrochlorothiazide as primary dependent variable. We predefined P = 0.05 as a relevant cut off for inclusion of variables in a multivariable regression model, but we explored for masked associations not using this cut-off in a multivariate backward regression model so that the common effect of the predictors on the treatment response could be determined. Renin and NT-proBNP were modelled after log transformation could be determined. In the evaluation of blood pressure responses to one of the studied medications, a negative outcome implies that blood pressure rose during treatment. A negative result of the primary outcome measure means that hydrochlorothiazide was more effective in blood pressure reduction than valsartan. In all analyses we considered a two sided P-value of < 0.05 to be significant. Statistical test were performed with SPSS version 16. (SPSS Inc, Chicago, IL). With 10 variables that in theory could all be part of the multiple regression model and with a minimum requirement of at least 10 observations per variable (32) a sample size of 100 patients was needed for final analyses. Taking loss to follow-up into account we planned to include 144 patients. Results Of 159 patients formally assessed for eligibility, 120 patients signed informed consent and 98 patients were included in the final analysis of the primary outcome. Patient flow and loss to follow up are visualized in Figure 1. In Table 1, the main patient characteristics at baseline (start of the first medication period) are depicted. Figure 1. View largeDownload slide Patient flow and loss to follow up for 159 eligible patients with hypertension in primary care. Four weeks after the second medication period the study protocol included a final measurement. Primary goal of this T5 measurement (Fig. 3) was to validate our cross over design (with regard to carry over, period and time effects). For this analysis, data from 83 patients were available. Data from this T5 measurement were not required for analysis of the primary outcome. This analysis was based on data of the 98 patients with full data after two medication periods. Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. Figure 1. View largeDownload slide Patient flow and loss to follow up for 159 eligible patients with hypertension in primary care. Four weeks after the second medication period the study protocol included a final measurement. Primary goal of this T5 measurement (Fig. 3) was to validate our cross over design (with regard to carry over, period and time effects). For this analysis, data from 83 patients were available. Data from this T5 measurement were not required for analysis of the primary outcome. This analysis was based on data of the 98 patients with full data after two medication periods. Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. Table 1. Baseline characteristics for patients who completed both medication periods compared to the subgroup of patients lost to follow-up   Completed  Loss to follow-up  Population studied, No.  98  22  Age, median (SD), years  53 (9)  54 (8)  Sex, % females  52  55  Menopause, % of females  49  64  DayABPM SBP, mean (SD)  152 (12)  150 (14)  DayABPM DBP, mean (SD)  95 (8)  97 (11)  OBPM SBP, mean (SD)  155 (16)  155 (14)  OBPM DBP, mean (SD)  97 (11)  96 (10)  Body Mass Index, mean (SD)  28.2 (4)  27.1 (3)  Waist Circumference, mean (SD)  96 (12)  96 (11)  Renin, median (SD)  12.0 (13)  9.3 (14)  NT-proBNP, median (SD)  5.0 (9)  5.0 (7)  Potassium, mean (SD)  4.4 (0.3)  4.5 (0.3)  eGFR in MDRD, mean (SD)  84.3 (14)  85.4 (15)  Urine albumin, mean (SD)  16.7 (25)a  8.1 (6)b    Completed  Loss to follow-up  Population studied, No.  98  22  Age, median (SD), years  53 (9)  54 (8)  Sex, % females  52  55  Menopause, % of females  49  64  DayABPM SBP, mean (SD)  152 (12)  150 (14)  DayABPM DBP, mean (SD)  95 (8)  97 (11)  OBPM SBP, mean (SD)  155 (16)  155 (14)  OBPM DBP, mean (SD)  97 (11)  96 (10)  Body Mass Index, mean (SD)  28.2 (4)  27.1 (3)  Waist Circumference, mean (SD)  96 (12)  96 (11)  Renin, median (SD)  12.0 (13)  9.3 (14)  NT-proBNP, median (SD)  5.0 (9)  5.0 (7)  Potassium, mean (SD)  4.4 (0.3)  4.5 (0.3)  eGFR in MDRD, mean (SD)  84.3 (14)  85.4 (15)  Urine albumin, mean (SD)  16.7 (25)a  8.1 (6)b  BNP, brain natriuretic peptide; DBP, diastolic blood pressure; DayABPM, daytime ambulatory blood pressure measurement; eGFR, estimated glomerular filtration rate; MDRD, modification of diet in renal disease; OBPM, office blood pressure measurement; SBP, systolic blood pressure. Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. an = 89; bn = 14. View Large There was no correlation between the responses to hydrochlorothiazide and valsartan (r = 0.02, P = 0.88). In our study age and renin were not correlated (Pearson correlation coefficient of r = –0.06; P = 0.55). In Figure 2a–c, the frequency plots of the individual responses to both studied drugs as well as the frequency plot of the difference in responses are shown. The differences in systolic blood pressure responses amongst patients ranged from –18 to 34 mmHg and –17 to 39 mmHg for hydrochlorothiazide and valsartan, respectively. Figures 2. View largeDownload slide (A–C) Frequency distributions of the systolic blood pressure response to valsartan (A) and hydrochlorothiazide (B) and of the difference in treatment response between valsartan and hydrochlorothiazide (C). Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. Figures 2. View largeDownload slide (A–C) Frequency distributions of the systolic blood pressure response to valsartan (A) and hydrochlorothiazide (B) and of the difference in treatment response between valsartan and hydrochlorothiazide (C). Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. Repeated measures analyses on systolic blood pressures demonstrated that there was no significant difference between the daytime ambulatory blood pressure values at the start of the first medication period and at the end of each washout period (Fig. 3). There was no difference in carry- over effect between both sequences of study medication (P = 0.44) and there was no period effect (P = 0.16). Valsartan reduced systolic blood pressure on average by 3 mmHg more than hydrochlorothiazide (P = 0.04). Figure 3. View largeDownload slide Study design and time course of mean daytime ambulatory systolic blood pressures. Error bars represent standard error of the mean; there was no statistical significant difference and no clinical relevant difference between mean daytime ABPM systolic blood pressures of T1 and T3 (P = 0.67), T1 and T5 (P = 0.85), T3 and T5 (P = 0.82); the difference in BP between run in and T1 is a known effect, related to getting accustomed to the 24-h measurement. Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. Figure 3. View largeDownload slide Study design and time course of mean daytime ambulatory systolic blood pressures. Error bars represent standard error of the mean; there was no statistical significant difference and no clinical relevant difference between mean daytime ABPM systolic blood pressures of T1 and T3 (P = 0.67), T1 and T5 (P = 0.85), T3 and T5 (P = 0.82); the difference in BP between run in and T1 is a known effect, related to getting accustomed to the 24-h measurement. Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. We did not find one studied variable to be a statistical significant predictor of the difference in systolic blood pressure response between hydrochlorothiazide and valsartan (Table 2). The effect sizes of most variables—including age or renin—were small. There was a trend for menopausal women to respond better to hydrochlorothiazide than to valsartan. Table 2. Results of univariate linear regression analyses with difference between the systolic blood pressure response to valsartan minus hydrochlorothiazide as dependent variable   Β  P  R2  Age  –0.18  (–0.49, 0.14)  0.26  0.01  Sex  –2.33  (–8.24, 3.58)  0.44  0.01  Menopausal state  –6.98  (–15.09, 1.14)  0.10  0.06  BMI  –0.59  (–1.36, 0.17)  0.13  0.02  Waistcircumference  –0.21  (–0.46, 0.05)  0.11  0.03  logRenin  1.53  (–2.41, 5.48)  0.44  0.01  logBNP  0.71  (–2.11, 3.53)  0.62  0.00  Potassium  2.49  (–6.12, 11.09)  0.59  0.00  MDRD  –0.07  (–0.28, 0.15)  0.54  0.00  Urinary albumin  0.01  (–0.12, 0.13)  0.91  0.00    Β  P  R2  Age  –0.18  (–0.49, 0.14)  0.26  0.01  Sex  –2.33  (–8.24, 3.58)  0.44  0.01  Menopausal state  –6.98  (–15.09, 1.14)  0.10  0.06  BMI  –0.59  (–1.36, 0.17)  0.13  0.02  Waistcircumference  –0.21  (–0.46, 0.05)  0.11  0.03  logRenin  1.53  (–2.41, 5.48)  0.44  0.01  logBNP  0.71  (–2.11, 3.53)  0.62  0.00  Potassium  2.49  (–6.12, 11.09)  0.59  0.00  MDRD  –0.07  (–0.28, 0.15)  0.54  0.00  Urinary albumin  0.01  (–0.12, 0.13)  0.91  0.00  Menopause n = 50; A positive β means a greater response to valsartan in comparison with the response to hydrochlorothiazide, e.g. in case of menopause a β of –6.98 means that menopausal women respond better to HCT than to valsartan; in case of BMI for each point BMI gain HCT reduces systolic BP with 0.6 mmHg more than valsartan; R-squared is the percentage of the response variable variation that is explained by the factor (0.01 means 1 %). Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. BMI, body mass index. View Large Discussion In this study with newly diagnosed hypertensive patients in family practice plasma renin concentration, age, NT-proBNP, potassium, eGFR, urinary albumin, waist circumference, sex, BMI did not predict a difference in systolic blood pressure responses between valsartan and hydrochlorothiazide. The only variable with a large effect size was menopausal status and therefore menopausal women may have had more benefit from hydrochlorothiazide than valsartan. Our data support the United States and European hypertension guidelines which do not apply an age cut-off or initiation of treatment (9,21). Conversely, our data do not support the UK NICE or the Dutch guideline (22,31). If age and renin would indeed be potent predictors, one would expect to find a majority of studies demonstrating age and renin as potent predictors of the antihypertensive response. However our study of the literature revealed conflicting results, often with weak associations, demonstrating a lack of predictive power (3,4,11,33–38). One explanation why there is no strong evidence for age as a predictor might be that in the age range in which hypertension is diagnosed most frequently, the pathophysiological mechanism of the development of essential hypertension is multifactorial. The model of high versus low renin hypertension appears to apply to a small minority at the lower and higher ages of the general hypertensive population. Our results show that in some individuals differences in blood pressure response are very large. Unfortunately, our study does not help clinicians to tell beforehand what the predominant mechanism in an individual patient is. The search for predictors of antihypertensive treatment response using a cross-over design is not unique. However, in contrast with previous studies our main objective was to present data which can better generalized to everyday family practice. With our choice to exclude patients older than 65 years and patients with co-morbid cardiovascular conditions our conclusions do not apply to patients aged 65 years and over. We decided to exclude this age group because chances were high that usual care would prohibit us to use the medication regime of our study protocol. Although our conclusions unfortunately do not apply to elderly patients the mean age of the diagnosis of hypertension lies around 55–60 years (25,26). In our study sample these patients are part of the sample while in the previous cross-over trials this was not so, patients were considerable younger. We drew a study sample from a relevant, primary care based population and the sample was relatively diverse. However, our results may not apply to populations in other parts of the world and with other characteristics. Our study has a sample size even bigger than the combined sample sizes of the two cross-over trials used to support the theory of the A/CD rule (3,4). However, as can be seen in Table 2, most effects were small and 95% CI relatively large. With a larger sample some of the effects might have reached significance. Except from menopausal state with an effect size of more than 6 mmHg, most effects would still remain clinically not relevant (<5 mmHg). The relatively large 95% CI’s seem to be most likely due to the heterogeneous mix of included patients and the sample size of the study. Because menopausal state was only applicable to women these results were based on 50 instead of 98 observations which may have been the cause of the lack of significance. Assessment of urinary sodium excretion (10,11,23)may have enabled improved subtyping of patients in different sorts of essential hypertension. However, we decided not to include sodium excretion because we deemed it practically difficult to implement in family practice. We hypothesized that we would find considerable intra-individual differences in blood pressure response to both study drugs, but that the mean blood pressure reduction of valsartan and hydrochlorothiazide would be the same. However, in our study sample valsartan was slightly more effective in reducing blood pressure than hydrochlorothiazide. We aimed to use equipotent doses of both studied drugs (39), but a minor difference in potency may explain the mean difference in response between valsartan and hydrochlorothiazide. As directed by the local medical ethical committee we were obliged to work exactly according to applicable guidelines and as such could not choose the best possible equipotent doses (valsartan 160 mgr and hydrochlorotiazide 25 mgr). Implications and considerations for future research Our current knowledge does not allow pathophysiological subtyping of patients with essential hypertension in order to improve hypertension control in family practice. Nevertheless, some studies have been executed on the use of treatment algorithms based on plasma renin activity; until now results are inconclusive (40).In our view, it makes more sense to focus future research on new possibilities of subtyping of patients with essential hypertension. Physicians starting treatment in patients with essential hypertension without relevant co-morbidity have a choice of three to four classes of antihypertensive agents without a clear preference. Except for black patients in whom thiazides and CCB’s are preferred, the choice of the first drug is to be based on costs, anticipated adverse events, co-morbidity and patient opinion and for now will remain a matter of trial and error. Declarations Funding: this study was funded by the department of Primary and Community Care, Radboud university medical center and by an unconditional grant of Novartis to cover the material costs of the study. Novartis asked and was granted that the ARB used in our study would be valsartan. In no other way, Novartis had any role in design and conduct of the study; collection, management, analysis and interpretation of the data; and preparation, review or approval of the manuscript. Ethical approval: the study was approved by the local Medical Ethics Committee of the RUNMC (Central Committee on Research involving Human Subjects, Arnhem-Nijmegen, The Netherlands) and registered in clinicaltrials.gov under NCT00457483. Conflict of interest: as former head of the Department of Primary and Community Care Chris van Weel supervised research projects some of which (in part) are funded as unconditional research grant by Bayer, NovoNordisk, AstraZeneca, Boehringer Ingelheim, GlaxoSmithKline or Novartis. He is retired as of the first of January 2013. All other authors have no conflict of interest to declare. Acknowledgements We would like to thank all participating family practices (Wijkgezondheidscentrum Lindenholt, Huisartsenpraktijk De Linie, Gezondheidscentrum Frans Huygen, Universitair Gezondheidscentrum Heyendael, Huisartspraktijk De Poort, Huisartspraktijk De Heelhoek, Huisartspraktijk Oosterhout, Huisartspraktijk Meijers en Mazure, Huisartspraktijk Adriaansens en Matser, Huisartsenpraktijk Winssen en Gezondheidscentrum De Kroonsteen); the Stichting Huisartsenlaboratorium Oost (SHO) and all participating patients for their contribution to the NAMI study. One of the founding fathers of the NAMI study Dr. Carel Bakx, FP, PhD died in January 2013. In the Netherlands, he was a well-known researcher in the field of hypertension and without him this study would not have succeeded. References 1. Law MR, Morris JK, Wald NJ. Use of blood pressure lowering drugs in the prevention of cardiovascular disease: meta-analysis of 147 randomised trials in the context of expectations from prospective epidemiological studies. BMJ  2009; 338: b1665. Google Scholar CrossRef Search ADS PubMed  2. Psaty BM, Lumley T, Furberg CDet al.   Health outcomes associated with various antihypertensive therapies used as first-line agents: a network meta-analysis. JAMA  2003; 289: 2534– 44. Google Scholar CrossRef Search ADS PubMed  3. Deary AJ, Schumann AL, Murfet Het al.   Double-blind, placebo-controlled crossover comparison of five classes of antihypertensive drugs. J Hypertens  2002; 20: 771– 7. Google Scholar CrossRef Search ADS PubMed  4. Dickerson JE, Hingorani AD, Ashby MJet al.   Optimisation of antihypertensive treatment by crossover rotation of four major classes. Lancet  1999; 353: 2008– 13. Google Scholar CrossRef Search ADS PubMed  5. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med  2005; 353: 487– 97. Google Scholar CrossRef Search ADS PubMed  6. Waeber B, Burnier M, Brunner HR. Compliance with antihypertensive therapy. Clin Exp Hypertens  1999; 21: 973– 85. Google Scholar CrossRef Search ADS PubMed  7. Leenen FH, Nwachuku CE, Black HRet al.   Clinical events in high-risk hypertensive patients randomly assigned to calcium channel blocker versus angiotensin-converting enzyme inhibitor in the antihypertensive and lipid lowering treatment to prevent heart attack trial. Hypertension  2006; 48: 374– 84. Google Scholar CrossRef Search ADS PubMed  8. ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group. The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial. JAMA  2002; 288: 2981– 97. CrossRef Search ADS PubMed  9. James PA, Oparil S, Carter BLet al.   2014 Evidence-Based Guideline for the Management of High Blood Pressure in Adults – Report from the panel members of the Eighth Joint National Committee. JAMA  2014; 311: 507– 20. Google Scholar CrossRef Search ADS PubMed  10. Laragh JH, Baer L, Brunner HRet al.   Renin, angiotensin and aldosterone system in pathogenesis and management of hypertensive vascular disease. Am J Med  1972; 52: 633– 52. Google Scholar CrossRef Search ADS PubMed  11. Preston RA, Materson BJ, Reda DJet al.   Age-race subgroup compared with renin profile as predictors of blood pressure response to antihypertensive therapy. Department of Veterans Affairs Cooperative Study Group on Antihypertensive Agents. JAMA  1998; 280: 1168– 72. Google Scholar CrossRef Search ADS PubMed  12. Wada A, Tsutamoto T, Matsuda Yet al.   Cardiorenal and neurohumoral effects of endogenous atrial natriuretic peptide in dogs with severe congestive heart failure using a specific antagonist for guanylate cyclase-coupled receptors. Circulation  1994; 89: 2232– 40. Google Scholar CrossRef Search ADS PubMed  13. Vasan RS, Evans JC, Larson MGet al.   Serum aldosterone and the incidence of hypertension in nonhypertensive persons. N Engl J Med  2004; 351: 33– 41. Google Scholar CrossRef Search ADS PubMed  14. Giacchetti G, Faloia E, Mariniello Bet al.   Overexpression of the renin-angiotensin system in human visceral adipose tissue in normal and overweight subjects. Am J Hypertens  2002; 15: 381– 8. Google Scholar CrossRef Search ADS PubMed  15. Scaglione R, Argano C, di Chiara Tet al.   Central obesity and hypertensive renal disease: association between higher levels of BMI, circulating transforming growth factor beta1 and urinary albumin excretion. Blood Press  2003; 12: 269– 76. Google Scholar CrossRef Search ADS PubMed  16. Chapman AB, Schwartz GL, Boerwinkle Eet al.   Predictors of antihypertensive response to a standard dose of hydrochlorothiazide for essential hypertension. Kidney Int  2002; 61: 1047– 55. Google Scholar CrossRef Search ADS PubMed  17. Safar ME, Smulyan H. Hypertension in women. Am J Hypertens  2004; 17: 82– 7. Google Scholar CrossRef Search ADS PubMed  18. Brown MJ, Cruickshank JK, Dominiczak AFet al.   Better blood pressure control: how to combine drugs. J Hum Hypertens  2003; 17: 81– 86. Google Scholar CrossRef Search ADS PubMed  19. Freis ED. Age and antihypertensive drugs (hydrochlorothiazide, bendroflumethiazide, nadolol and captopril). Am J Cardiol  1988; 61: 117– 21. Google Scholar CrossRef Search ADS PubMed  20. Taddei S. Blood pressure through aging and menopause. Climacteric  2009; 12 Suppl 1: 36– 40. Google Scholar CrossRef Search ADS PubMed  21. Mancia G, Fagard R, Narkiewicz Ket al.   2013ESH/ESC Guidelines for the management of arterial hypertension: The Task Force for the Management of Arterial Hypertension of the European Society of Hypertension (ESH) and of the European Society of Cardiology (ESC). J Hypertens  2013; 31: 1281– 357. Google Scholar CrossRef Search ADS PubMed  22. National Clinical Guideline Centre. Hypertension: The clinical management of primary hypertension in adults - CG 127. 2011. http://www.nice.org.uk/cg127 (accessed on 01 August 2017). 23. Laragh JH, Letcher RL, Pickering TG. Renin profiling for diagnosis and treatment of hypertension. JAMA  1979; 241: 151– 6. Google Scholar CrossRef Search ADS PubMed  24. Materson BJ, Reda DJ, Cushman WCet al.   Single-drug therapy for hypertension in men. A comparison of six antihypertensive agents with placebo. The Department of Veterans Affairs Cooperative Study Group on Antihypertensive Agents. N Engl J Med  1993; 328: 914– 21. Google Scholar CrossRef Search ADS PubMed  25. Egan BM, Zhao Y, Axon RNet al.   Uncontrolled and apparent treatment resistant hypertension in the United States, 1988 to 2008. Circulation  2011; 124: 1046– 58. Google Scholar CrossRef Search ADS PubMed  26. van der Wel M, Bakx C, de Grauw Wet al.   The influence of guideline revisions on the process and outcome of hypertension management in general practice: a descriptive study. Eur J Gen Pract  2008; 14 Suppl 1: 47– 52. Google Scholar CrossRef Search ADS PubMed  27. van Weel C, Smith H, Beasley JW. Family practice research networks. Experiences from 3 countries. J Fam Pract  2000; 49: 938– 43. Google Scholar PubMed  28. Hansson L, Hedner T, Dahlöf B. Prospective randomized open blinded end-point (PROBE) study. A novel design for intervention trials. Prospective Randomized Open Blinded End-Point. Blood Press  1992; 1: 113– 9. Google Scholar CrossRef Search ADS PubMed  29. Westhoff TH, Schmidt S, Zidek Wet al.   Validation of the Stabil-O-Graph blood pressure self-measurement device. J Hum Hypertens  2008; 22: 233– 5. Google Scholar CrossRef Search ADS PubMed  30. Jones CR, Taylor K, Chowienczyk Pet al.   A validation of the Mobil O Graph (version 12) ambulatory blood pressure monitor. Blood Press Monit  2000; 5: 233– 8. Google Scholar CrossRef Search ADS PubMed  31. NHG-Standaard Cardiovasculair Risicomanagement. Huisarts en Wetenschap 2012; 55: 14– 28. https://www.nhg.org/standaarden/samenvatting/cardiovasculair-risicomanagement (accessed on 01 August 2017). 32. Kleinbaum DG, Kupper LL, Muller KE. Applied Regression Analysis and Other Multivariable Methods . 2nd edition ed. PWS-Kent Publishing Company; 1988. 33. Attwood S, Bird R, Burch Ket al.   Within-patient correlation between the antihypertensive effects of atenolol, lisinopril and nifedipine. J Hypertens  1994; 12: 1053– 60. Google Scholar CrossRef Search ADS PubMed  34. Kusaka M, Atarashi K, Matsumoto Ket al.   Plasma reninactivity and norepinephrine as predictors for antihypertensive effects of nifedipine andcaptopril. Am J Hypertens  1991; 4: 735– 39. Google Scholar CrossRef Search ADS PubMed  35. Bidiville J, Nussberger J, Waeber Get al.   Individual responses to converting enzyme inhibitors and calcium antagonists. Hypertension  1988; 11: 166– 73. Google Scholar CrossRef Search ADS PubMed  36. Falconnet C, Bochud M, Bovet Pet al.   Gender difference in the response to an angiotensin-converting enzyme inhibitor and a diuretic in hypertensive patients of African descent. J Hypertens  2004; 22: 1213– 20. Google Scholar CrossRef Search ADS PubMed  37. Case DB, Wallace JM, Keim HJet al.   Possible role of renin in hypertension as suggested by renin-sodium profiling and inhibition of converting enzyme. N Engl J Med  1977; 296: 641– 6. Google Scholar CrossRef Search ADS PubMed  38. Niutta E, Cusi D, Colombo Ret al.   Predicting interindividual variations in antihypertensive therapy: the role of sodium transport systems and renin. J Hypertens Suppl  1990; 8: S53– 8. Google Scholar CrossRef Search ADS PubMed  39. Wellington K, Faulds DM. Valsartan/hydrochlorothiazide: a review of its pharmacology, therapeutic efficacy and place in the management of hypertension. Drugs  2002; 62: 1983– 2005. Google Scholar CrossRef Search ADS PubMed  40. Viera AJ, Furberg CD. Plasma renin testing to guide antihypertensive therapy. Curr Hypertens Rep  2015; 17: 506. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2017. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Family Practice Oxford University Press

Patient characteristics do not predict the individual response to antihypertensive medication: a cross-over trial

Loading next page...
 
/lp/ou_press/patient-characteristics-do-not-predict-the-individual-response-to-NECzquPZh1
Publisher
Oxford University Press
Copyright
© The Author(s) 2017. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0263-2136
eISSN
1460-2229
D.O.I.
10.1093/fampra/cmx075
Publisher site
See Article on Publisher Site

Abstract

Abstract Background International guidelines on hypertension management do not agree on whether patient characteristics can be used for the first choice of treatment of uncomplicated essential hypertension. Objective We wanted to identify predictive patient characteristics to the response of two different classes of antihypertensive drugs in patients with newly diagnosed hypertension in primary care. Methods We conducted a prospective, open label, blinded endpoint cross-over trial in 120 patients with a new diagnosis of hypertension from 10 family practices. Patients received 4 weeks of 12.5 mgr hydrochlorothiazide once daily and 4 weeks of 80 mgr valsartan once daily, each followed by a 4-week washout. The sequence of drugs was randomized. Age, sex and menopausal state were recorded at run in and 24 h ambulatory blood pressure, office blood pressure, plasma renin concentration, NT-proBNP, potassium, estimated glomerular filtration rate, urinary albumin, body mass index and waist circumference at each regimen change. The difference in systolic blood pressure response between both study drugs, calculated from mean daytime ambulatory blood pressures, was the main outcome measure. Results Ninety-eight patients (52% female; median age 53 years) were eligible for per-protocol-analysis. None of the studied variables were predictive for the difference in systolic blood pressure response. Individual systolic blood pressure responses ranged from an increase by 18 mmHg to a decrease of 39 mmHg. Conclusion In a relevant group of primary care patients with newly diagnosed hypertension, we were unable to detect predictors of treatment response. This study rather supports the United States and European guidelines than the United Kingdom and Dutch guidelines on hypertension. Cross-over, family medicine, hypertension, prediction, treatment Introduction Several meta-analyses have demonstrated that on average all known classes of antihypertensive medication are equally effective in reducing blood pressure (1,2). However, some studies have confirmed what clinicians experience in daily practice: in individual patients treatment response can differ substantially from one antihypertensive class to the other (3,4). This finding suggests that in ‘essential hypertension’ a more refined diagnostic subtyping may exist and favours the possibility of patient characteristics that could predict which antihypertensive drug works best in the individual patient. Application of a strategy using predictors of hypotensive response, may reduce polypharmacy, enhance treatment adherence and reduce costs (5,6). Research in animals and humans suggests that several patient characteristics may have predictive qualities. Several characteristics have been studied, such as ethnicity (7–9), plasma renin (4,10,11), NT-proBNP (3,12), potassium (13), waistcircumference (14), BMI (15), sex (16,17), age (18,19) and menopause (20). International consensus is lacking on the use of an age cut-off in hypertension management. In the United States and the European guidelines, an age cut-off is lacking. Age as predictor is not mentioned at all in the US guideline; the European guideline specifically states ‘no evidence is available that different choices should be made based on age or gender’ (9,21). In contrast, the UK NICE guideline has advocated the use of an age cut-off in the treatment algorithm (the A/CD rule) (22). In 2012, this rule has been adopted by the Dutch Guideline on Cardiovascular Risk Management. The rule advises in patients younger than 55 years to start treatment with an ACE inhibitor or ARB (‘A’) whereas in patients of 55 years and older CCB’s should be the first choice medication (or diuretic as alternative, ‘CD’). The NICE recommendation is based on the pathophysiological assumption that at least part of the patients with uncomplicated essential hypertension can be divided in low renin (salt sensitive or ‘volume-driven’) and high renin (vasoconstrictive) hypertension (18,23). Core evidence for the A/CD rule comes from a study demonstrating that plasma renin activity declines with age (18) and clinical data from two cross-over trials (3,4) and one randomized controlled trial (24) suggesting that age can be regarded as surrogate for plasma renin and as such as predictor of treatment response. Because hypertension is diagnosed and managed most often in primary care, studies on individual treatment response should ideally be performed in family practice based populations and reflect an age around 55–60 years, in which diagnosis is commonly made (25,26). Selection of predictive characteristics should preferably be based on the actual feasibility of implementation in daily family practice. The clinical data supporting the A/CD rule, however, do not seem to represent this family practice population. In the two cross-over trials (both with less than 40 patients completing the study) (3,4) only patients younger than 55 years were included, predominantly male (65–70%) and blood pressure measurements were in supine position. In both studies the setting of patient selection was not described, patients needed to be young and with untreated essential hypertension. In the RCT (24) only male, veteran patients were selected from 15 outpatient clinics, with a mean age of the total study population of 59 years. We decided to set up a cross-over study to explore whether predictors of treatment response could be identified in a representative family practice population and with a set of potential predictors that are feasible to implement in the family practice setting. A cross-over study design enabled us to study intra-individual treatment responses. The controversy about the use of age as a predictor of treatment response adds to the relevance of this exploration, especially in a population relevant for family practice. Our primary objective was to identify predictors of the difference in blood pressure response between two classes of antihypertensive drugs with different mechanisms of action representing the most widely acknowledged subtyping of essential hypertension in renin driven hypertension or volume driven hypertension. Methods Patients, design and setting Patients with newly diagnosed hypertension, aged 18–65 years and listed in 10 Dutch family practices affiliated with the Radboud University Medical Center (27) were screened for eligibility to participate in a cross-over study. The methodology, strengths and limitations of the type of cross-over design we used (PROBE: PRospective, Open label, Blinded Endpoint) were described by Hansson et al. (28).We started patient inclusion in August 2007 and final measurements in the last included patient were finished in February 2011. The study was registered in clinicaltrials.gov under NCT00457483. http://clinicaltrials.gov/ct2/show/study/NCT00457483?term=hypertension+and+nijmegen&rank=1. Patients in whom the diagnosis of hypertension was confirmed and who gave written informed consent were included in the study. Exclusion criteria were use of antihypertensive medication, blood pressure higher than 210/110 mmHg, inability to speak or understand Dutch and presence of cardiovascular co-morbidity (diabetes mellitus, peripheral arterial disease, ischemic heart disease, stroke, transient ischemic attack and atrial fibrillation).We excluded these co-morbid conditions because hypertension treatment algorithms for these patients differ from those in uncomplicated patients (9,21,22). After a 2-week run-in period patients followed a treatment scheme in which a patient started with either one of the following two drugs: hydrochlorothiazide 12,5 mg once daily or valsartan 80 mg once daily, for the duration of 4 weeks. These 4 weeks of treatment with one antihypertensive medicine was followed by 4 weeks of no treatment (washout period). Then the next 4 weeks of treatment started but now with the antihypertensive that was not used in the first 4 weeks. Again this period was followed by a 4-week washout period. The sequence of the use of the two medications was randomized. Included patients were assigned a medication order by the practice nurse. For this purpose, all practice nurses were given a list with a random sequence of the digits ‘0’ and ‘1’ each standing for one order of medications. This random sequence was generated by SPSS version 16 and at the start of the study each practice was given an unique random sequence. At the run-in, at the start, and at the end of each medication or washout period we measured 24-h ambulatory blood pressure (24-h ABP), office blood pressure (OBP), plasma renin concentration, NT-proBNP, plasma potassium concentration, eGFR, urinary albumin excretion, body mass index (BMI) and waist circumference. At the run-in also age, sex and menopausal state were recorded. The primary outcome measure was the prediction of the difference in mean daytime systolic blood pressure response between both study drugs determined by ambulatory blood pressure monitoring. Blood pressure measurements During the eligibility screening performed by the local staff of the family practices a confirmed diagnosis of hypertension (≥140/90mmHg) was based on the mean of blood pressure measurements taken at three different office visits. All practices were provided with two validated oscillometric office monitors (Stabil-O-Graph, I.E.M. GMBH, Stolberg, Germany) (29). These monitors were exclusively used for study patients and were re-calibrated every 2 years. The office blood pressure measurements during the study were taken by practice nurses who were trained in the methodology of standardized blood pressure measurement. All 24-h ABP recordings were taken on week days and patients were asked to keep a standardized diary of daily activities. Practices used the same, validated 24-h ABPM device (Mobil-O-Graph, I.E.M. GMBH, Stolberg, Germany) (30). The measurement interval was 20 min from 7 am to 11 pm and 1 h from 11 pm to 7 am. We defined the mean day blood pressure from 9 am to 9 pm and we used this measure as our primary blood pressure outcome. Only recordings with 70% or more valid readings were used. The display of the device was set not to show the blood pressure values. Ambulatory blood pressure values were not communicated to the patient, prescribing physician and researcher until after completion of the study protocol. Study variables and study medication We selected demographic and biochemical variables that family practitioners commonly use or that are easily implemented in daily practice. All blood and urine samples were collected in the morning (before 12.30 pm) and analyzed at one primary care orientated diagnostic centre. Sample analyses were performed batch-wise. The following laboratory tests were used: creatinine (CREA plus, Roche Diagnostics), renin (DSL-25100 ACTIVE Renin IRMA kit, Diagnostic Systems Laboratories), NT-proBNP (Elecsys, Roche Diagnostics), potassium (ISE Indirect, Roche Diagnostics) and urine albumine (COBAS, Roche Diagnostics). As requested by the local medical ethics committee, prescription of the medication was as similar to usual care as possible and thus according to the recommendations of the Dutch multidisciplinary guideline on cardiovascular risk management (31). The principle in this guideline is to start with a low dose and if a patient does not reach hypertension control in 4–8 weeks add a second medication rather than increase doses. Therefore patients were prescribed 80 mg valsartan once daily and 12.5 mg hydrochlorothiazide once daily (31). Following the PROBE design, medication was not blinded for both patients and the prescribing general practitioners. We recorded compliance and adverse events with the use of standardized questionnaires. Statistical analyses and sample size calculation First, we applied descriptive statistics to compare baseline patient characteristics of patients who completed the study with those who were lost to follow-up. We used Pearson’s test for correlation to study the relationship between the individual blood pressure responses of both medications. We used frequency plots to show the treatment response to both studied medications and to show the difference in treatment response. Next, with repeated measures analysis we checked for integrity of our cross-over design by assessing whether the mean drop in blood pressure during the medication periods was followed by a full recovery of blood pressure after the washout period. If this would not occur than it is more likely that results of the first period of the cross-over influence results of the second period of the cross-over trial and thus hampering valid interpretation of results. In addition, using a mixed model analysis (with a compound symmetry correlation matrix) we determined whether relevant carry-over, period and treatment effects occurred. These effects could introduce forms of bias specific to a cross-over design. When these effects prove not to be large and not statistically significant conclusions drawn are (more) reliable. In the final step, according to a per protocol analysis on all patients that completed the second medication period, we performed both univariate and multivariate regression analysis with the difference in mean systolic blood pressure response (blood pressure at start of treatment minus blood pressure after 4 weeks of treatment) between valsartan and hydrochlorothiazide as primary dependent variable. We predefined P = 0.05 as a relevant cut off for inclusion of variables in a multivariable regression model, but we explored for masked associations not using this cut-off in a multivariate backward regression model so that the common effect of the predictors on the treatment response could be determined. Renin and NT-proBNP were modelled after log transformation could be determined. In the evaluation of blood pressure responses to one of the studied medications, a negative outcome implies that blood pressure rose during treatment. A negative result of the primary outcome measure means that hydrochlorothiazide was more effective in blood pressure reduction than valsartan. In all analyses we considered a two sided P-value of < 0.05 to be significant. Statistical test were performed with SPSS version 16. (SPSS Inc, Chicago, IL). With 10 variables that in theory could all be part of the multiple regression model and with a minimum requirement of at least 10 observations per variable (32) a sample size of 100 patients was needed for final analyses. Taking loss to follow-up into account we planned to include 144 patients. Results Of 159 patients formally assessed for eligibility, 120 patients signed informed consent and 98 patients were included in the final analysis of the primary outcome. Patient flow and loss to follow up are visualized in Figure 1. In Table 1, the main patient characteristics at baseline (start of the first medication period) are depicted. Figure 1. View largeDownload slide Patient flow and loss to follow up for 159 eligible patients with hypertension in primary care. Four weeks after the second medication period the study protocol included a final measurement. Primary goal of this T5 measurement (Fig. 3) was to validate our cross over design (with regard to carry over, period and time effects). For this analysis, data from 83 patients were available. Data from this T5 measurement were not required for analysis of the primary outcome. This analysis was based on data of the 98 patients with full data after two medication periods. Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. Figure 1. View largeDownload slide Patient flow and loss to follow up for 159 eligible patients with hypertension in primary care. Four weeks after the second medication period the study protocol included a final measurement. Primary goal of this T5 measurement (Fig. 3) was to validate our cross over design (with regard to carry over, period and time effects). For this analysis, data from 83 patients were available. Data from this T5 measurement were not required for analysis of the primary outcome. This analysis was based on data of the 98 patients with full data after two medication periods. Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. Table 1. Baseline characteristics for patients who completed both medication periods compared to the subgroup of patients lost to follow-up   Completed  Loss to follow-up  Population studied, No.  98  22  Age, median (SD), years  53 (9)  54 (8)  Sex, % females  52  55  Menopause, % of females  49  64  DayABPM SBP, mean (SD)  152 (12)  150 (14)  DayABPM DBP, mean (SD)  95 (8)  97 (11)  OBPM SBP, mean (SD)  155 (16)  155 (14)  OBPM DBP, mean (SD)  97 (11)  96 (10)  Body Mass Index, mean (SD)  28.2 (4)  27.1 (3)  Waist Circumference, mean (SD)  96 (12)  96 (11)  Renin, median (SD)  12.0 (13)  9.3 (14)  NT-proBNP, median (SD)  5.0 (9)  5.0 (7)  Potassium, mean (SD)  4.4 (0.3)  4.5 (0.3)  eGFR in MDRD, mean (SD)  84.3 (14)  85.4 (15)  Urine albumin, mean (SD)  16.7 (25)a  8.1 (6)b    Completed  Loss to follow-up  Population studied, No.  98  22  Age, median (SD), years  53 (9)  54 (8)  Sex, % females  52  55  Menopause, % of females  49  64  DayABPM SBP, mean (SD)  152 (12)  150 (14)  DayABPM DBP, mean (SD)  95 (8)  97 (11)  OBPM SBP, mean (SD)  155 (16)  155 (14)  OBPM DBP, mean (SD)  97 (11)  96 (10)  Body Mass Index, mean (SD)  28.2 (4)  27.1 (3)  Waist Circumference, mean (SD)  96 (12)  96 (11)  Renin, median (SD)  12.0 (13)  9.3 (14)  NT-proBNP, median (SD)  5.0 (9)  5.0 (7)  Potassium, mean (SD)  4.4 (0.3)  4.5 (0.3)  eGFR in MDRD, mean (SD)  84.3 (14)  85.4 (15)  Urine albumin, mean (SD)  16.7 (25)a  8.1 (6)b  BNP, brain natriuretic peptide; DBP, diastolic blood pressure; DayABPM, daytime ambulatory blood pressure measurement; eGFR, estimated glomerular filtration rate; MDRD, modification of diet in renal disease; OBPM, office blood pressure measurement; SBP, systolic blood pressure. Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. an = 89; bn = 14. View Large There was no correlation between the responses to hydrochlorothiazide and valsartan (r = 0.02, P = 0.88). In our study age and renin were not correlated (Pearson correlation coefficient of r = –0.06; P = 0.55). In Figure 2a–c, the frequency plots of the individual responses to both studied drugs as well as the frequency plot of the difference in responses are shown. The differences in systolic blood pressure responses amongst patients ranged from –18 to 34 mmHg and –17 to 39 mmHg for hydrochlorothiazide and valsartan, respectively. Figures 2. View largeDownload slide (A–C) Frequency distributions of the systolic blood pressure response to valsartan (A) and hydrochlorothiazide (B) and of the difference in treatment response between valsartan and hydrochlorothiazide (C). Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. Figures 2. View largeDownload slide (A–C) Frequency distributions of the systolic blood pressure response to valsartan (A) and hydrochlorothiazide (B) and of the difference in treatment response between valsartan and hydrochlorothiazide (C). Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. Repeated measures analyses on systolic blood pressures demonstrated that there was no significant difference between the daytime ambulatory blood pressure values at the start of the first medication period and at the end of each washout period (Fig. 3). There was no difference in carry- over effect between both sequences of study medication (P = 0.44) and there was no period effect (P = 0.16). Valsartan reduced systolic blood pressure on average by 3 mmHg more than hydrochlorothiazide (P = 0.04). Figure 3. View largeDownload slide Study design and time course of mean daytime ambulatory systolic blood pressures. Error bars represent standard error of the mean; there was no statistical significant difference and no clinical relevant difference between mean daytime ABPM systolic blood pressures of T1 and T3 (P = 0.67), T1 and T5 (P = 0.85), T3 and T5 (P = 0.82); the difference in BP between run in and T1 is a known effect, related to getting accustomed to the 24-h measurement. Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. Figure 3. View largeDownload slide Study design and time course of mean daytime ambulatory systolic blood pressures. Error bars represent standard error of the mean; there was no statistical significant difference and no clinical relevant difference between mean daytime ABPM systolic blood pressures of T1 and T3 (P = 0.67), T1 and T5 (P = 0.85), T3 and T5 (P = 0.82); the difference in BP between run in and T1 is a known effect, related to getting accustomed to the 24-h measurement. Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. We did not find one studied variable to be a statistical significant predictor of the difference in systolic blood pressure response between hydrochlorothiazide and valsartan (Table 2). The effect sizes of most variables—including age or renin—were small. There was a trend for menopausal women to respond better to hydrochlorothiazide than to valsartan. Table 2. Results of univariate linear regression analyses with difference between the systolic blood pressure response to valsartan minus hydrochlorothiazide as dependent variable   Β  P  R2  Age  –0.18  (–0.49, 0.14)  0.26  0.01  Sex  –2.33  (–8.24, 3.58)  0.44  0.01  Menopausal state  –6.98  (–15.09, 1.14)  0.10  0.06  BMI  –0.59  (–1.36, 0.17)  0.13  0.02  Waistcircumference  –0.21  (–0.46, 0.05)  0.11  0.03  logRenin  1.53  (–2.41, 5.48)  0.44  0.01  logBNP  0.71  (–2.11, 3.53)  0.62  0.00  Potassium  2.49  (–6.12, 11.09)  0.59  0.00  MDRD  –0.07  (–0.28, 0.15)  0.54  0.00  Urinary albumin  0.01  (–0.12, 0.13)  0.91  0.00    Β  P  R2  Age  –0.18  (–0.49, 0.14)  0.26  0.01  Sex  –2.33  (–8.24, 3.58)  0.44  0.01  Menopausal state  –6.98  (–15.09, 1.14)  0.10  0.06  BMI  –0.59  (–1.36, 0.17)  0.13  0.02  Waistcircumference  –0.21  (–0.46, 0.05)  0.11  0.03  logRenin  1.53  (–2.41, 5.48)  0.44  0.01  logBNP  0.71  (–2.11, 3.53)  0.62  0.00  Potassium  2.49  (–6.12, 11.09)  0.59  0.00  MDRD  –0.07  (–0.28, 0.15)  0.54  0.00  Urinary albumin  0.01  (–0.12, 0.13)  0.91  0.00  Menopause n = 50; A positive β means a greater response to valsartan in comparison with the response to hydrochlorothiazide, e.g. in case of menopause a β of –6.98 means that menopausal women respond better to HCT than to valsartan; in case of BMI for each point BMI gain HCT reduces systolic BP with 0.6 mmHg more than valsartan; R-squared is the percentage of the response variable variation that is explained by the factor (0.01 means 1 %). Data were collected from 10 family practices in the eastern part of the Netherlands in the years 2007–2011. BMI, body mass index. View Large Discussion In this study with newly diagnosed hypertensive patients in family practice plasma renin concentration, age, NT-proBNP, potassium, eGFR, urinary albumin, waist circumference, sex, BMI did not predict a difference in systolic blood pressure responses between valsartan and hydrochlorothiazide. The only variable with a large effect size was menopausal status and therefore menopausal women may have had more benefit from hydrochlorothiazide than valsartan. Our data support the United States and European hypertension guidelines which do not apply an age cut-off or initiation of treatment (9,21). Conversely, our data do not support the UK NICE or the Dutch guideline (22,31). If age and renin would indeed be potent predictors, one would expect to find a majority of studies demonstrating age and renin as potent predictors of the antihypertensive response. However our study of the literature revealed conflicting results, often with weak associations, demonstrating a lack of predictive power (3,4,11,33–38). One explanation why there is no strong evidence for age as a predictor might be that in the age range in which hypertension is diagnosed most frequently, the pathophysiological mechanism of the development of essential hypertension is multifactorial. The model of high versus low renin hypertension appears to apply to a small minority at the lower and higher ages of the general hypertensive population. Our results show that in some individuals differences in blood pressure response are very large. Unfortunately, our study does not help clinicians to tell beforehand what the predominant mechanism in an individual patient is. The search for predictors of antihypertensive treatment response using a cross-over design is not unique. However, in contrast with previous studies our main objective was to present data which can better generalized to everyday family practice. With our choice to exclude patients older than 65 years and patients with co-morbid cardiovascular conditions our conclusions do not apply to patients aged 65 years and over. We decided to exclude this age group because chances were high that usual care would prohibit us to use the medication regime of our study protocol. Although our conclusions unfortunately do not apply to elderly patients the mean age of the diagnosis of hypertension lies around 55–60 years (25,26). In our study sample these patients are part of the sample while in the previous cross-over trials this was not so, patients were considerable younger. We drew a study sample from a relevant, primary care based population and the sample was relatively diverse. However, our results may not apply to populations in other parts of the world and with other characteristics. Our study has a sample size even bigger than the combined sample sizes of the two cross-over trials used to support the theory of the A/CD rule (3,4). However, as can be seen in Table 2, most effects were small and 95% CI relatively large. With a larger sample some of the effects might have reached significance. Except from menopausal state with an effect size of more than 6 mmHg, most effects would still remain clinically not relevant (<5 mmHg). The relatively large 95% CI’s seem to be most likely due to the heterogeneous mix of included patients and the sample size of the study. Because menopausal state was only applicable to women these results were based on 50 instead of 98 observations which may have been the cause of the lack of significance. Assessment of urinary sodium excretion (10,11,23)may have enabled improved subtyping of patients in different sorts of essential hypertension. However, we decided not to include sodium excretion because we deemed it practically difficult to implement in family practice. We hypothesized that we would find considerable intra-individual differences in blood pressure response to both study drugs, but that the mean blood pressure reduction of valsartan and hydrochlorothiazide would be the same. However, in our study sample valsartan was slightly more effective in reducing blood pressure than hydrochlorothiazide. We aimed to use equipotent doses of both studied drugs (39), but a minor difference in potency may explain the mean difference in response between valsartan and hydrochlorothiazide. As directed by the local medical ethical committee we were obliged to work exactly according to applicable guidelines and as such could not choose the best possible equipotent doses (valsartan 160 mgr and hydrochlorotiazide 25 mgr). Implications and considerations for future research Our current knowledge does not allow pathophysiological subtyping of patients with essential hypertension in order to improve hypertension control in family practice. Nevertheless, some studies have been executed on the use of treatment algorithms based on plasma renin activity; until now results are inconclusive (40).In our view, it makes more sense to focus future research on new possibilities of subtyping of patients with essential hypertension. Physicians starting treatment in patients with essential hypertension without relevant co-morbidity have a choice of three to four classes of antihypertensive agents without a clear preference. Except for black patients in whom thiazides and CCB’s are preferred, the choice of the first drug is to be based on costs, anticipated adverse events, co-morbidity and patient opinion and for now will remain a matter of trial and error. Declarations Funding: this study was funded by the department of Primary and Community Care, Radboud university medical center and by an unconditional grant of Novartis to cover the material costs of the study. Novartis asked and was granted that the ARB used in our study would be valsartan. In no other way, Novartis had any role in design and conduct of the study; collection, management, analysis and interpretation of the data; and preparation, review or approval of the manuscript. Ethical approval: the study was approved by the local Medical Ethics Committee of the RUNMC (Central Committee on Research involving Human Subjects, Arnhem-Nijmegen, The Netherlands) and registered in clinicaltrials.gov under NCT00457483. Conflict of interest: as former head of the Department of Primary and Community Care Chris van Weel supervised research projects some of which (in part) are funded as unconditional research grant by Bayer, NovoNordisk, AstraZeneca, Boehringer Ingelheim, GlaxoSmithKline or Novartis. He is retired as of the first of January 2013. All other authors have no conflict of interest to declare. Acknowledgements We would like to thank all participating family practices (Wijkgezondheidscentrum Lindenholt, Huisartsenpraktijk De Linie, Gezondheidscentrum Frans Huygen, Universitair Gezondheidscentrum Heyendael, Huisartspraktijk De Poort, Huisartspraktijk De Heelhoek, Huisartspraktijk Oosterhout, Huisartspraktijk Meijers en Mazure, Huisartspraktijk Adriaansens en Matser, Huisartsenpraktijk Winssen en Gezondheidscentrum De Kroonsteen); the Stichting Huisartsenlaboratorium Oost (SHO) and all participating patients for their contribution to the NAMI study. One of the founding fathers of the NAMI study Dr. Carel Bakx, FP, PhD died in January 2013. In the Netherlands, he was a well-known researcher in the field of hypertension and without him this study would not have succeeded. References 1. Law MR, Morris JK, Wald NJ. Use of blood pressure lowering drugs in the prevention of cardiovascular disease: meta-analysis of 147 randomised trials in the context of expectations from prospective epidemiological studies. BMJ  2009; 338: b1665. Google Scholar CrossRef Search ADS PubMed  2. Psaty BM, Lumley T, Furberg CDet al.   Health outcomes associated with various antihypertensive therapies used as first-line agents: a network meta-analysis. JAMA  2003; 289: 2534– 44. Google Scholar CrossRef Search ADS PubMed  3. Deary AJ, Schumann AL, Murfet Het al.   Double-blind, placebo-controlled crossover comparison of five classes of antihypertensive drugs. J Hypertens  2002; 20: 771– 7. Google Scholar CrossRef Search ADS PubMed  4. Dickerson JE, Hingorani AD, Ashby MJet al.   Optimisation of antihypertensive treatment by crossover rotation of four major classes. Lancet  1999; 353: 2008– 13. Google Scholar CrossRef Search ADS PubMed  5. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med  2005; 353: 487– 97. Google Scholar CrossRef Search ADS PubMed  6. Waeber B, Burnier M, Brunner HR. Compliance with antihypertensive therapy. Clin Exp Hypertens  1999; 21: 973– 85. Google Scholar CrossRef Search ADS PubMed  7. Leenen FH, Nwachuku CE, Black HRet al.   Clinical events in high-risk hypertensive patients randomly assigned to calcium channel blocker versus angiotensin-converting enzyme inhibitor in the antihypertensive and lipid lowering treatment to prevent heart attack trial. Hypertension  2006; 48: 374– 84. Google Scholar CrossRef Search ADS PubMed  8. ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group. The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial. JAMA  2002; 288: 2981– 97. CrossRef Search ADS PubMed  9. James PA, Oparil S, Carter BLet al.   2014 Evidence-Based Guideline for the Management of High Blood Pressure in Adults – Report from the panel members of the Eighth Joint National Committee. JAMA  2014; 311: 507– 20. Google Scholar CrossRef Search ADS PubMed  10. Laragh JH, Baer L, Brunner HRet al.   Renin, angiotensin and aldosterone system in pathogenesis and management of hypertensive vascular disease. Am J Med  1972; 52: 633– 52. Google Scholar CrossRef Search ADS PubMed  11. Preston RA, Materson BJ, Reda DJet al.   Age-race subgroup compared with renin profile as predictors of blood pressure response to antihypertensive therapy. Department of Veterans Affairs Cooperative Study Group on Antihypertensive Agents. JAMA  1998; 280: 1168– 72. Google Scholar CrossRef Search ADS PubMed  12. Wada A, Tsutamoto T, Matsuda Yet al.   Cardiorenal and neurohumoral effects of endogenous atrial natriuretic peptide in dogs with severe congestive heart failure using a specific antagonist for guanylate cyclase-coupled receptors. Circulation  1994; 89: 2232– 40. Google Scholar CrossRef Search ADS PubMed  13. Vasan RS, Evans JC, Larson MGet al.   Serum aldosterone and the incidence of hypertension in nonhypertensive persons. N Engl J Med  2004; 351: 33– 41. Google Scholar CrossRef Search ADS PubMed  14. Giacchetti G, Faloia E, Mariniello Bet al.   Overexpression of the renin-angiotensin system in human visceral adipose tissue in normal and overweight subjects. Am J Hypertens  2002; 15: 381– 8. Google Scholar CrossRef Search ADS PubMed  15. Scaglione R, Argano C, di Chiara Tet al.   Central obesity and hypertensive renal disease: association between higher levels of BMI, circulating transforming growth factor beta1 and urinary albumin excretion. Blood Press  2003; 12: 269– 76. Google Scholar CrossRef Search ADS PubMed  16. Chapman AB, Schwartz GL, Boerwinkle Eet al.   Predictors of antihypertensive response to a standard dose of hydrochlorothiazide for essential hypertension. Kidney Int  2002; 61: 1047– 55. Google Scholar CrossRef Search ADS PubMed  17. Safar ME, Smulyan H. Hypertension in women. Am J Hypertens  2004; 17: 82– 7. Google Scholar CrossRef Search ADS PubMed  18. Brown MJ, Cruickshank JK, Dominiczak AFet al.   Better blood pressure control: how to combine drugs. J Hum Hypertens  2003; 17: 81– 86. Google Scholar CrossRef Search ADS PubMed  19. Freis ED. Age and antihypertensive drugs (hydrochlorothiazide, bendroflumethiazide, nadolol and captopril). Am J Cardiol  1988; 61: 117– 21. Google Scholar CrossRef Search ADS PubMed  20. Taddei S. Blood pressure through aging and menopause. Climacteric  2009; 12 Suppl 1: 36– 40. Google Scholar CrossRef Search ADS PubMed  21. Mancia G, Fagard R, Narkiewicz Ket al.   2013ESH/ESC Guidelines for the management of arterial hypertension: The Task Force for the Management of Arterial Hypertension of the European Society of Hypertension (ESH) and of the European Society of Cardiology (ESC). J Hypertens  2013; 31: 1281– 357. Google Scholar CrossRef Search ADS PubMed  22. National Clinical Guideline Centre. Hypertension: The clinical management of primary hypertension in adults - CG 127. 2011. http://www.nice.org.uk/cg127 (accessed on 01 August 2017). 23. Laragh JH, Letcher RL, Pickering TG. Renin profiling for diagnosis and treatment of hypertension. JAMA  1979; 241: 151– 6. Google Scholar CrossRef Search ADS PubMed  24. Materson BJ, Reda DJ, Cushman WCet al.   Single-drug therapy for hypertension in men. A comparison of six antihypertensive agents with placebo. The Department of Veterans Affairs Cooperative Study Group on Antihypertensive Agents. N Engl J Med  1993; 328: 914– 21. Google Scholar CrossRef Search ADS PubMed  25. Egan BM, Zhao Y, Axon RNet al.   Uncontrolled and apparent treatment resistant hypertension in the United States, 1988 to 2008. Circulation  2011; 124: 1046– 58. Google Scholar CrossRef Search ADS PubMed  26. van der Wel M, Bakx C, de Grauw Wet al.   The influence of guideline revisions on the process and outcome of hypertension management in general practice: a descriptive study. Eur J Gen Pract  2008; 14 Suppl 1: 47– 52. Google Scholar CrossRef Search ADS PubMed  27. van Weel C, Smith H, Beasley JW. Family practice research networks. Experiences from 3 countries. J Fam Pract  2000; 49: 938– 43. Google Scholar PubMed  28. Hansson L, Hedner T, Dahlöf B. Prospective randomized open blinded end-point (PROBE) study. A novel design for intervention trials. Prospective Randomized Open Blinded End-Point. Blood Press  1992; 1: 113– 9. Google Scholar CrossRef Search ADS PubMed  29. Westhoff TH, Schmidt S, Zidek Wet al.   Validation of the Stabil-O-Graph blood pressure self-measurement device. J Hum Hypertens  2008; 22: 233– 5. Google Scholar CrossRef Search ADS PubMed  30. Jones CR, Taylor K, Chowienczyk Pet al.   A validation of the Mobil O Graph (version 12) ambulatory blood pressure monitor. Blood Press Monit  2000; 5: 233– 8. Google Scholar CrossRef Search ADS PubMed  31. NHG-Standaard Cardiovasculair Risicomanagement. Huisarts en Wetenschap 2012; 55: 14– 28. https://www.nhg.org/standaarden/samenvatting/cardiovasculair-risicomanagement (accessed on 01 August 2017). 32. Kleinbaum DG, Kupper LL, Muller KE. Applied Regression Analysis and Other Multivariable Methods . 2nd edition ed. PWS-Kent Publishing Company; 1988. 33. Attwood S, Bird R, Burch Ket al.   Within-patient correlation between the antihypertensive effects of atenolol, lisinopril and nifedipine. J Hypertens  1994; 12: 1053– 60. Google Scholar CrossRef Search ADS PubMed  34. Kusaka M, Atarashi K, Matsumoto Ket al.   Plasma reninactivity and norepinephrine as predictors for antihypertensive effects of nifedipine andcaptopril. Am J Hypertens  1991; 4: 735– 39. Google Scholar CrossRef Search ADS PubMed  35. Bidiville J, Nussberger J, Waeber Get al.   Individual responses to converting enzyme inhibitors and calcium antagonists. Hypertension  1988; 11: 166– 73. Google Scholar CrossRef Search ADS PubMed  36. Falconnet C, Bochud M, Bovet Pet al.   Gender difference in the response to an angiotensin-converting enzyme inhibitor and a diuretic in hypertensive patients of African descent. J Hypertens  2004; 22: 1213– 20. Google Scholar CrossRef Search ADS PubMed  37. Case DB, Wallace JM, Keim HJet al.   Possible role of renin in hypertension as suggested by renin-sodium profiling and inhibition of converting enzyme. N Engl J Med  1977; 296: 641– 6. Google Scholar CrossRef Search ADS PubMed  38. Niutta E, Cusi D, Colombo Ret al.   Predicting interindividual variations in antihypertensive therapy: the role of sodium transport systems and renin. J Hypertens Suppl  1990; 8: S53– 8. Google Scholar CrossRef Search ADS PubMed  39. Wellington K, Faulds DM. Valsartan/hydrochlorothiazide: a review of its pharmacology, therapeutic efficacy and place in the management of hypertension. Drugs  2002; 62: 1983– 2005. Google Scholar CrossRef Search ADS PubMed  40. Viera AJ, Furberg CD. Plasma renin testing to guide antihypertensive therapy. Curr Hypertens Rep  2015; 17: 506. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2017. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.

Journal

Family PracticeOxford University Press

Published: Feb 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off