An outcome-driven threshold for pulse pressure amplification

Pulse pressure amplification (PPA), defined as the brachial-to-aortic pulse pressure ratio, reflects arterial stiffness and wave reflection phenomena. PPA typically decreases with aging and with cardiovascular risk factors such as hypertension, dyslipidemia, diabetes, and chronic kidney disease. Prior population and patient studies have linked lower PPA to higher cardiovascular risk, coronary disease, kidney function decline, and mortality, and cross-sectional data show PPA falling from about 1.72 in teenagers to 1.25 in octogenarians. However, few prospective studies have addressed incident major cardiovascular complications in relation to PPA, and no outcome-driven PPA threshold has been established. This study aimed to define, calibrate, and validate an outcome-driven threshold for PPA associated with increased risk of cardiovascular and coronary events, using individual participant data from the International Database of Central Arterial Properties for Risk Stratification (IDCARS). The goal was to determine whether a simple PPA cut-point adds prognostic information beyond conventional risk factors and supports clinical risk stratification.

Existing literature shows that PPA decreases with age due to central arterial stiffening and earlier wave reflection, which increases central systolic pressure and reduces the brachial-to-aortic PP ratio. Population and clinical studies have linked lower PPA with higher cardiovascular risk, coronary artery disease, kidney function decline, and mortality. In the Anglo-Cardiff Collaborative Trial, PPA decreased from 1.72 in teenagers to 1.25 in octogenarians. Despite these associations, prospective evidence establishing a specific outcome-driven PPA threshold predictive of incident cardiovascular events has been lacking, motivating the current analysis.

Design and data source: Individual-participant meta-analysis using IDCARS longitudinal population cohorts. Eligible cohorts had baseline brachial and central blood pressure, arterial pulse waveform measurements, cardiovascular risk factors, and follow-up for fatal and nonfatal outcomes; participants were representative of general populations. Ethical approvals and informed consent were obtained; data were de-identified. Participants: Of 6650 eligible IDCARS participants, exclusions were applied for age <30 years without endpoints (n=954), peripheral PP >130 mmHg (n=10), central systolic BP <70 (n=1) or >230 mmHg (n=1), central diastolic BP >150 (n=1) or <55 mmHg (n=15), or missing pulse wave analysis (n=60), leaving 5608 participants aged 30–96 years (median 53.6 years) for analysis. Hemodynamic assessment: Brachial BP was measured twice after 5–15 minutes supine rest; the average was used. Radial arterial waveforms were recorded at the dominant arm by applanation tonometry (Millar SPC-301 micromanometer) using SphygmoCor CvMS. Recordings failing quality criteria (variability >5%, amplitude <80 mV, operator index <70%) were discarded. Central (aortic) waveforms were reconstructed via a validated generalized transfer function, and central BP was calibrated using brachial SBP and DBP. PPA was calculated as brachial PP divided by aortic PP. Endpoints: Primary composite cardiovascular endpoint included cardiovascular mortality, sudden death, and nonfatal myocardial infarction, heart failure, stroke, and coronary revascularization. The coronary endpoint included sudden death, fatal/nonfatal myocardial infarction, and coronary revascularization. Only the first event per category counted; events were validated against medical records. Statistical analysis: Cox proportional hazards models with cohort as a random effect adjusted for sex, age, heart rate, body mass index, smoking, alcohol consumption, total-to-HDL cholesterol ratio, estimated glomerular filtration rate (CKD-EPI), antihypertensive treatment, history of cardiovascular disease, and diabetes. Proportional hazards were tested via a Kolmogorov-type supremum test. The overall cohort was randomly split (stratified by sex, median age 53.6 years, and cohort) into a discovery dataset (n=3945) and a replication dataset (n=1663). Threshold derivation in the discovery set used two approaches: (1) plotting multivariable-adjusted HRs for 0.1 PPA increments across the 10th–90th percentiles to identify where the upper 95% CI crossed unity (indicating decreased risk above that level); (2) determining PPA values conferring a 5-year risk equivalent to specified brachial SBP levels (120, 130, 140, 160 mmHg). Risk model calibration was assessed by comparing predicted risks to overoptimism-corrected Kaplan-Meier estimates across PPA quintiles. Predictive performance was evaluated via sensitivity, specificity, ROC AUC, and reclassification metrics (IDI and NRI). Subgroup analyses examined interactions by sex, age (<60 vs ≥60 years), median SBP (<130 vs ≥130 mmHg), and antihypertensive treatment using deviation-from-mean coding. Sensitivity analyses for the coronary endpoint additionally adjusted for diastolic BP.

- Cohort and events: Among 5608 participants (mean age 54.2 years; 54.1% women), over a median 4.1 years, 255 cardiovascular and 109 coronary events occurred. Standardized incidence rates were 9.59 and 4.05 per 1000 person-years for cardiovascular and coronary endpoints, respectively. - Threshold derivation (discovery set): The upper 95% CI of HRs crossed unity at PPA 1.28 (cardiovascular) and 1.26 (coronary). PPA values yielding 5-year risks equivalent to brachial SBP 140 mmHg were 1.28 (CI: 1.19–1.36) for cardiovascular and 1.29 (CI: 1.22–1.36) for coronary endpoints. Rounding led to a common operational threshold of PPA 1.3. - Risk at threshold: In the discovery set, PPA <1.3 vs ≥1.3 was associated with HR 1.54 (CI: 1.00–2.36) for cardiovascular and HR 2.45 (CI: 1.20–5.01) for coronary endpoints. A 1-SD increase in PPA corresponded to HR 0.74 (CI: 0.26–0.97) for cardiovascular and HR 0.57 (CI: 0.36–0.91) for coronary outcomes. Models were well calibrated. - Replication: Similar HR patterns were observed in the replication dataset; with prior probabilities considered, one-tailed testing was used. Continuous-analysis significance for the coronary endpoint was not achieved (P=0.085), likely due to smaller sample size, but threshold-based findings were consistent. - Predictive performance: Across discovery, replication, and overall datasets, threshold-based classification showed sensitivity ~0.80 and specificity ~0.40; AUC for continuous PPA was ~0.60. NRI for the PPA 1.3 threshold was significant: cardiovascular 16.1%–33.7%, coronary 30.8%–44.9%. Continuous NRI was 9.26%–22.7% (cardiovascular) and 21.5%–30.2% (coronary). IDI was not significant in either threshold or continuous analyses. - Subgroups: Age significantly modified associations (interaction P≤0.041). HRs for PPA <1.3 vs ≥1.3 in participants <60 vs ≥60 years were 3.86 vs 1.19 for cardiovascular and 6.21 vs 1.77 for coronary endpoints. High-risk status (PPA <1.3) was more prevalent in women, particularly aged <60 years (67.7% vs 61.5% in ≥60 years; P<0.001). Younger high-risk individuals also had a higher prevalence of isolated systolic hypertension than younger low-risk peers (12.7% vs 6.43%; P<0.001). - Sensitivity analyses: After additional adjustment for diastolic BP, HRs for coronary endpoints comparing PPA <1.3 vs ≥1.3 were 2.44 (CI: 1.18–5.07) in discovery and 2.79 (CI: 1.10–7.06) in replication; standardized continuous HRs were 0.57 (CI: 0.35–0.92; P=0.022) and 0.64 (CI: 0.34–1.20; P=0.12), respectively.

The study established an outcome-driven pulse pressure amplification threshold of 1.3, below which risk of incident cardiovascular and coronary events is significantly higher, independent of conventional risk factors. This finding directly addresses the prior gap of lacking a clinically actionable PPA cut-point. The two complementary approaches to threshold determination (CI-crossing and risk equivalence) converged at essentially the same value, and results were calibrated and replicated across independent subsets. Although the standalone discrimination of PPA is modest (AUC ~0.60), adding the threshold meaningfully improves risk classification (significant NRI) beyond standard factors. Age emerged as a key effect modifier: low PPA conferred substantially higher relative risks among adults <60 years, suggesting that reduced PPA in midlife signals adverse arterial hemodynamics and vascular aging that presage events. The higher proportion of women classified as high-risk (PPA <1.3), especially between 30 and 60 years, indicates that PPA may capture underrecognized cardiovascular risk in women and that pulse wave analysis could be particularly informative in this group. Sensitivity analyses adjusting for diastolic BP supported the robustness of the coronary findings. Overall, the results support integrating central hemodynamic assessment and PPA into cardiovascular risk stratification strategies.

An operational, outcome-driven PPA threshold of 1.3 (brachial-to-aortic PP ratio) identifies individuals at increased risk of incident cardiovascular and coronary events beyond traditional risk factors. The adverse prognostic impact of low PPA is most pronounced in adults under 60 years and highlights potentially underestimated risk in women aged 30–60 years. These findings support the use of pulse wave analysis to enhance risk stratification in general populations. Future research should validate the threshold across devices and ethnicities, assess its incremental utility within established risk scores, explore longitudinal changes in PPA, and test whether interventions that improve arterial properties and PPA reduce clinical events.