You are leaving Medscape Education
Cancel Continue
Log in to save activities Your saved activities will show here so that you can easily access them whenever you're ready. Log in here CME & Education Log in to keep track of your credits.


CAFE - The Conduit Artery Functional Endpoint Study

Authors: Linda Brookes, MScFaculty and Disclosures


Presenter: Bryan Williams, MD (University of Leicester, United Kingdom)

The first prespecified substudy to be reported from the Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT) has shown that amlodipine-based (calcium channel blocker [CCB]) treatment is much more effective at reducing central aortic blood pressure than a conventional atenolol-based (beta-blocker) regimen.[1] The results of the Conduit Artery Functional Endpoint (CAFE) study also suggest that the central aortic blood pressure may be more predictive of cardiovascular events, such as stroke and heart attack, than traditional peripheral (brachial) blood pressure measurements. CAFE was the first study to repeatedly measure central aortic pressure in a major clinical outcomes trial and the first to show that central aortic pressure is a plausible mechanism to explain the better clinical outcomes seen in patients treated with amlodipine-based therapy in ASCOT.

Study Rationale

Brachial blood pressure is a strong predictor of heart attack and stroke and has been assumed to reflect pressure in the central aorta and the heart. This assumption has been challenged on the basis that different classes of blood pressure-lowering drugs have different mechanisms and therefore might have different effects on central aortic blood pressure. CAFE was designed to test this hypothesis as a major substudy of the blood pressure-lowering arm of ASCOT (ASCOT-BPLA), in which over 19,000 hypertensive patients were randomized to 1 of 2 treatments: amlodipine with or without the angiotensin converting enzyme (ACE) inhibitor perindopril, or atenolol with or without bendroflumethiazide (a thiazide-type diuretic).[2] All patients recruited to ASCOT had blood pressure ≥ 160/100 mm Hg if untreated or ≥ 140/90 mm Hg on treatment, were aged 40-79 years (mean age, 63 years), and had 3 additional cardiovascular risk factors (10-year Framingham risk < 20%).[3] The median duration of follow-up was 5.5 years. The results of ASCOT-BPLA, published in September 2005,[4] showed that the amlodipine-based regimen was more effective in reducing the incidence of cardiovascular events and all-cause mortality than the atenolol-based regimen.


The CAFE substudy recruited patients from 5 ASCOT study centers in the United Kingdom and Ireland. Recruitment began 1 year after the start of ASCOT to ensure that patients would be stable on their antihypertensive medication and approaching blood pressure goal when they were recruited. The CAFE population (2073 patients) had similar characteristics to those of the ASCOT population, with similar numbers randomized to each treatment. About 90% had previously been treated for hypertension; 11% were taking lipid-lowering drugs and 25%, aspirin. About 50% of patients in each treatment group were uptitrated to a second drug, and only 3% in the atenolol group and 7% in the amlodipine group remained on monotherapy.

Blood Pressure Measurements

Brachial blood pressure was measured while patients were seated using a HEM-705CP device (Omron Healthcare Europe, Hoofddorp, The Netherlands). Central aortic blood pressure was measured by using applanation tonometry to record radial artery pulse waves, which were calibrated to brachial blood pressure and transformed to derive a central arterial wave form and central aortic mean systolic and diastolic pressures by means of a commercially available computer system linked to the tonometer (SphygmoCor V7; Atcor Medical, West Ryde, New South Wales, Australia). Over 17,0000 central pressure measurements were done over 4 years of follow-up. Approximately 80% of patients had > 1 tonometry measurement. Patients had an average of 3.4 tonometry measurements each. Average follow-up after first tonometry measurement was 3 years.

Brachial and Central Aortic Pressures

Brachial systolic blood pressure (SBP) was well controlled during the study and did not differ significantly between the 2 treatment arms (Table 1). Brachial diastolic blood pressure (DBP) was slightly lower in the patients randomized to amlodipine, with the results that pulse pressure was slightly higher in the amlodipine-based regimen group and heart rate slower in the atenolol-based regimen group.

Table 1. Hemodynamic Data

Parameter Atenolol Amlodipine Mean Difference (95% CI)
(Atenolol - Amlodipine)
P (t-test)
Brachial SBP (mm Hg) 133.9 133.2 0.7 (-0.4, 1,7) .2/.07
Brachial DBP (mm Hg) 78.6 76.9 1.6 (0.9, 2.4) < .0001
Brachial PP (mm Hg) 55.3 56.2 -0.9 (-1.9, 0) .06
Heart rate (beats/min) 58.6 69.3 -10.7 (-11.5, -9.8) < .0001
CI = confidence interval; DBP = diastolic blood pressure; PP = pulse pressure; SBP = systolic blood pressure

Despite similar brachial pressures, however, central aortic SBP and pulse pressure were significantly lower throughout the study in the group on amlodipine-based treatment compared with patients on the atenolol-based regimen (Table 2).

Table 2. Central Aortic Pressures

Parameter Atenolol Amlodipine Mean Difference (95% CI)
(Atenolol - Amlodipine)
P (t-test)
Central SBP (mm Hg) 125.5 121.2 4.3 (3.3, 5.4) < .0001
Central PP (mm Hg) 46.4 43.3 3 (2.1, 3.9) < .0001
CI = confidence interval; PP = pulse pressure; SBP = systolic blood pressure

The difference between brachial and central aortic SBP was significantly greater for the amlodipine-based treatment (12 mm Hg) compared with atenolol-based treatment (8.3 mm Hg), reflecting the lower central SBP on amlodipine throughout the trial.

Pulse wave analysis showed that compared with amlodipine-based therapy, the central aortic outgoing pressure wave (P1 height) was significantly lower with atenolol-based therapy and so could not account for the central augmentation of SBP seen with atenolol. However, augmentation index, a measure of wave reflection, was markedly greater for atenolol-based therapy, suggesting a relation to the increase in central pulse pressure seen with atenolol.

Clinical Outcomes

The composite clinical outcome for CAFE, which was defined posthoc when the ASCOT endpoint rate was known, included all cardiovascular events and procedures plus development of renal impairment (ASCOT-prespecified and -validated endpoints). A significant 16% reduction with amlodipine-based treatment was seen in the CAFE composite outcome in the ASCOT cohort (HR = 0.84 [0.79-0.90]; P <.0001), although since only 305 events occurred in CAFE, the study had limited statistical power for this endpoint. Cox proportional analysis identified 4 significant factors for this endpoint - peripheral pulse pressure, central pulse pressure, augmentation index, and P1 height - but after adjustment for baseline variables, only central pulse pressure remained significant (P = .048; HR 1.11; CI 1.00-1.21)


AHA-designated discussant Joseph L. Izzo, Jr, MD (SUNY-Buffalo School of Medicine, Buffalo, New York), said physicians should be looking beyond blood pressure cuff measurements. SBP is not constant throughout the arterial tree, so brachial cuff blood pressure cannot detect clinically relevant changes caused by reflected waves in central SBP, as was shown in the CAFE study, or the impact of forward pressure wave amplification on organ damage, he said, calling for more studies.

Dr. Izzo also recommended reconsidering some drug actions. For instance, drugs such as beta-blockers, without favorable effects on arterial biomechanics, appear to confer less protection for target organs. Non-blood-pressure benefits of blood pressure-lowering drugs should not be inferred unless SBP is measured throughout the circulation, he stressed, noting that the ASCOT investigators themselves had earlier concluded that the differences they found between the effects of the 2 treatment regimens "might not be entirely explained by better control of blood pressure."[3] It now appears that brachial blood pressure measurements may have overestimated the hemodynamic effects of the atenolol-based regimen and underestimated the beneficial effect of the amlodipine-based regimen on central aortic pressure and hemodynamics.

Speaking outside the session, Dr. Williams provided more insight into the implications of the CAFE results. He believes that they provide insights into the design of the most appropriate types of treatment to most effectively lower central blood pressure. "We need to know what different types of drugs that we use to treat blood pressure actually do to the central circulation. It is quite remarkable that we talk about the genes and about molecular approaches to treatment, and we do not even understand what the drugs that we use every day are doing to the circulation," he emphasized. There are common regimens that are used on a routine basis that should be examined for the effect they have on central arterial pressure and whether there are more effective regimens, Dr. Williams said. "If we can identify consistently that there are, then we should be thinking about using those regimens because we are likely to get better outcomes with them." Dr. Williams expects that the results of CAFE, as well as those of ASCOT, will lead to changes in hypertension management guidelines.

The CAFE investigators plan to calculate central pressures for all of the ASCOT patients at every time point at which blood pressure was measured to specifically determine the impact that central pressure had on the overall outcome of the trial, including endpoints. Dr. Williams suggested that if the difference between treatment arms is found to be 3-4 mm Hg, this could explain most of the difference in the outcome of stroke seen in ASCOT. Dr. Williams and his colleagues also plan to investigate the effect of lipid-lowering therapy on central and brachial pressures, since by the ASCOT protocol half the CAFE population had also been randomized to atorvastatin.

The CAFE study was funded by a medical school research grant from Pfizer UK.


  1. Williams B. Differential impact of blood pressure-lowering drugs on central arterial pressure influences clinical outcomes - principal results of the Conduit Artery Function Evaluation (CAFE) study in ASCOT. Program and abstracts of the American Heart Association Scientific Sessions 2005; November 13-16, 2005; Dallas, Texas. Late Breaking Clinical Trials I.
  2. Williams B, O'Rourke M; Anglo-Scandinavian Cardiac Outcomes Trial. The Conduit Artery Functional Endpoint (CAFE) study in ASCOT. J Hum Hypertens. 2001;15(Suppl 1):S69-73. 3. Sever PS, Dahlof B, Poulter NR, et al; ASCOT investigators. Rationale, design, methods and baseline demography of participants of the Anglo-Scandinavian Cardiac Outcomes Trial. J Hypertens. 2001;19:1139-1147.
  3. Dahlof B, Sever PS, Poulter NR, et al; the ASCOT investigators. Prevention of cardiovascular events with an antihypertensive regimen of amlodipine adding perindopril as required versus atenolol adding bendroflumethiazide as required, in the Anglo-Scandinavian Cardiac Outcomes Trial-Blood Pressure Lowering Arm (ASCOT-BPLA): a multicentre randomised controlled trial. Lancet. 2005;366:895-906.
  • Print