Heterogeneity in Blood Pressure Response to 4 Antihypertensive Drugs: A Randomized Clinical Trial

Abstract

Importance: Hypertension is the leading risk factor for premature death worldwide. Multiple blood pressure-lowering therapies are available but the potential for maximizing benefit by personalized targeting of drug classes is unknown.

Objective: To investigate and quantify the potential for targeting specific drugs to specific individuals to maximize blood pressure effects.

Design, setting, and participants: A randomized, double-blind, repeated crossover trial in men and women with grade 1 hypertension at low risk for cardiovascular events at an outpatient research clinic in Sweden. Mixed-effects models were used to assess the extent to which individuals responded better to one treatment than another and to estimate the additional blood pressure lowering achievable by personalized treatment.

Interventions: Each participant was scheduled for treatment in random order with 4 different classes of blood pressure-lowering drugs (lisinopril [angiotensin-converting enzyme inhibitor], candesartan [angiotensin-receptor blocker], hydrochlorothiazide [thiazide], and amlodipine [calcium channel blocker]), with repeated treatments for 2 classes.

Main outcomes and measures: Ambulatory daytime systolic blood pressure, measured at the end of each treatment period.

Results: There were 1468 completed treatment periods (median length, 56 days) recorded in 270 of the 280 randomized participants (54% men; mean age, 64 years). The blood pressure response to different treatments varied considerably between individuals (P < .001), specifically for the choices of lisinopril vs hydrochlorothiazide, lisinopril vs amlodipine, candesartan vs hydrochlorothiazide, and candesartan vs amlodipine. Large differences were excluded for the choices of lisinopril vs candesartan and hydrochlorothiazide vs amlodipine. On average, personalized treatment had the potential to provide an additional 4.4 mm Hg-lower systolic blood pressure.

Conclusions and relevance: These data reveal substantial heterogeneity in blood pressure response to drug therapy for hypertension, findings that may have implications for personalized therapy.

Trial registration: ClinicalTrials.gov Identifier: NCT02774460.

Figure 1.. Participant Flow and Analyzed Treatment Periods in the Precision Hypertension Care (PHYSIC) Trial

^aEach treatment period consisted of a 1-week placebo washout period, a 2-week dose increase period, and at least 4 weeks at target dose. For each participant, the order of the 6 treatment periods was randomized without restrictions. ^bParticipants who did not complete a treatment period continued with the next period unless withdrawn. ^cTen of the 280 randomized participants contributed to zero periods. A participant-level graphic overview of adherence and missing data per treatment period is given in eTable 2 in Supplement 2.

Figure 2.. Observed and Model Estimates of Mean Daytime Ambulatory Systolic Blood Pressure (SBP): Comparisons Between Pairs of Treatments

Arbitrary colors distinguish participants. The n refers to participants with at least 1 data point with both treatments. A, Dots show observed mean daytime ambulatory SBP at the end-of-treatment periods for 1 participant. For participants with data from 2 periods, individual period values are given by error bars; the dot is at the participant mean. The black line signifies no difference between treatments; participants above the line had higher SBP on the y-axis than on the x-axis treatment. The blue line is shifted by the mean treatment contrast in the trial population. To show all 6 pairwise comparisons, the same BP values for each treatment appear in 3 panels. Hash marks on the axes show marginal distributions. B, Conditional mean SBP from the fitted primary model. Each dot represents best estimates of the unknown real mean SBP for each participant with each treatment underlying the measured SBP values in panel A. Estimates include the mean difference between treatments, person-specific overall SBP, and person-specific treatment contrasts. Dots close to the blue line indicate participants with low estimated gain from personalization.

Figure 3.. Crossover Differences in Mean On-Treatment Daytime Ambulatory Systolic Blood Pressure (SBP): Second Difference by First Difference

Each dot denotes a participant with 2 crossovers for that pair of treatments. In each panel, both axes show a difference in SBP on the first listed drug minus the second listed drug, for example, SBP on amlodipine minus SBP on candesartan. The x-axis shows the difference in the first crossover, and the y-axis the difference in the second crossover, in the same participant. The order of treatments in each crossover was randomized. The regression lines (blue lines) are from the analyses presented rightmost in Table 2, and the shaded areas the pointwise 95% CIs. The dotted diagonal lines represent identical treatment differences at the first and second crossovers. The hash marks on the axes show the marginal distributions. If there is a potential for personalized treatment choice, a better than mean result of one drug vs the other at the first crossover should be associated with the same participant having a better than mean result also at the second crossover. Conversely, absence of such a pattern indicates lack of potential for personalization.