Better Management of Cardiovascular Diseases by Pulse Wave Velocity: Combining Clinical Practice with Clinical Research using Evidence-Based Medicine

Framing the Relevant Clinical Questions

From the above clinical scenarios, the following questions emerged as the basis for this review:

• ▪ In patients with hypertension, diabetes or ESRD, how much risk of mortality is attributable to PWV and how could it be used in risk estimation for individual patients?

• ▪ How much benefit could be expected from reducing PWV in a given patient?

• ▪ Could benefits outweigh potential harms of treatment?

Searching for the Evidence

The MEDLINE and EMBASE databases were searched for relevant studies using “pulse wave velocity” or “arterial distensibility” or “arterial compliance” in combination with “cohort” as study type (yield: 35 articles) or “prognosis” as a MeSH term or a text word (yield: 62 articles). The retrieved papers were screened for their relevance to the framed questions, resulting in 12 papers after exclusion of irrelevant, biased and repeated publications.

Critical Appraisal

For cohort studies, the quality of studies was evaluated with particular focus on the following questions:

• ▪ Was there a representative sample of patients at a well-defined point?

• ▪ Was follow-up sufficiently long and complete?

• ▪ Were all potentially important prognostic factors assessed?

• ▪ What was the effect size (considering the 95% confidence interval [CI])?

Risk Analysis

The relative risk (RR) of mortality in levels of PWV was summarized. Adjusted risk ratios for a 1 standard deviation (SD) change in PWV was either quoted or extracted from other reported indices like odds ratio (OR). Similar risk estimates for systolic blood pressure and age, the main determinants of PWV, were added to the table for better comparison. The 2-by-2 tables were reconstructed according to the information in the text, tables and graphs in order to define the absolute risks and number-needed-to-treat (NNT). While the patient classifications were not based on consistent cutoff points in the studies, they were categorized into low (G1), moderate (G2) and high (G3) risk for analysis. The pooled RR of mortality for the grades of PWV and weighted mean difference of PWV in survived and deceased groups were calculated using the fixed effect model.

PWV Method

PWV is measured by placing pressure sensitive transducers over different arteries (e.g., the right carotid and femoral arteries for carotid-femoral PWV). Dividing the distance separating the sensors by the time corresponding to the transmission time of the pulse wave (defined by the start of the rising phase of the pulse wave) from one site to the other determines the PWV. It is a non-invasive, inexpensive, rapid measurement of arterial stiffness which could be easily performed in an office.

While distance measurement between the probes affects PWV measurement,¹² it is of considerable importance to investigate the method of measurement, or possible adjustments, when studies are compared. However, since the impact is systematic in the measurement, using relative statistics or ratios is not significantly influenced by this factor.

Study Analysis

The 14 eligible studies included:

• ▪ Five reports evaluated ESRD patients. All were performed by a group at F. H. Manhes Hospital, Paris, a gradually increasing study population during progressive research phases.^2,3,13–15 Because these were multiple reports on the same population, the most recent and most complete reports were applied. The average follow-up time was 78 ± 48 months, indicating a relatively short follow-up for some subjects. While in one of their serial reports the proportion of subjects with diabetes was not homogenous across the tertiles of PWV, this factor was not considered in their multivariate analysis. Despite this limitation, other aspects of their study were methodologically acceptable.

• ▪ Five reports evaluated hypertensive patients, all by a group from Broussais Hospital, Paris, who evaluated stroke in one study,⁶ and overall and cardiovascular mortality in three others.^4,16,17 The latest report contained the greatest number of patients and the longest follow-up and was selected for review.¹⁷ The fifth article evaluated coronary events.⁵ Although all these study cohorts were heterogeneous for diabetes, this factor was controlled by multivariate analysis and other aspects of their research methodology were acceptable.

• ▪ Two papers evaluated patients with diabetes, one with ESRD⁸ and one in the general population.⁷ Unfortunately a different PWV method of measurement was used and their results could not be directly applied to the above clinical scenarios. However, their results were used for the risk analysis phase in this review.

• ▪ Two studies evaluated the prognostic value of PWV in the elderly.^9,18 These results could not be extrapolated to our hypothetical cases whose ages are less than 70.

Summarizing and Applying the Evidence

All included studies showed an increasing mortality risk with increased PWV. A summary of the reported risk estimates for each grade of PWV is illustrated in tables 1⇓ and 2⇓. Data from the studies were pooled using the fixed effect model and for one level of increment in PWV grades produced RR of 2.41 (1.81–3.20) and 1.69 (1.35–2.11) for all-cause mortality and cardiovascular events, respectively (P<0.00001) (figure 1⇓). Although pooled RR for two level increment in PWV was 4.59 (2.61–8.05) with a significant P value (0.00001), the tests (I² =77% and P=0.04) demonstrated heterogeneity which made pooling inappropriate.

• Download figure
• Open in new tab
• Download PowerPoint

Figure 1.

Pooled, weighted RR for one level increment in PWV grade using the fixed effect model in terms of all-cause mortality and cardiovascular events.

Comparison of the reported mean PWV from 6 studies showed a significant difference between PWV in survived and deceased subjects; the overall mean difference was determined to be 1.46 (95% CI, 1.04–1.88) and 2.61 (95% CI, 2.51–2.71) for the low and high risk groups, respectively (table 3⇓). Assuming that reducing PWV by better control of conventional cardiovascular risk factors, as well as non-classic ones, would decrease cardiovascular risk outcomes as much as an increase in PWV would increase risk, we calculated relative and absolute risk reduction and NNT to prevent a death (table 4⇓). From this analysis it can be demonstrated that in a high risk group, such as ESRD, if PWV is reduced from the highest grade (G3, e.g., 13) to the lowest (G1, e.g., 9.2), or even to the moderate grade (G2, e.g., 11.5), approximately 2 patients need to be treated to prevent a death. PWV reduction from G2 to G1 (e.g., 10.5 to 9) for 8 patients for 6 years would also save one life in this population. Although the theoretical benefit in hypertensives is not as profound as the predicted benefit in the ESRD group, it is still convincing. For instance NNTs to save one life by reducing PWV from 13 to 11.5 or 9.5 was calculated at 9 or 27 patients, respectively. Treatment of less than 17 patients to reduce PWV from 11.5 to 9.5 for a similar duration would also prevent a death. Likewise, reducing PWV of 5 diabetic patients to the low level over 5 years of treatment would also save one life. Although the translation from risk to benefit is optimistic, the treatment may be less effective in practice, but it could facilitate clinical decision making. Furthermore, available evidence, although limited, from longitudinal studies is consistent and demonstrates RR of 0.71 (0.6 to 0.86) for all-cause mortality and 0.79 (0.69–0.93) for cardiovascular mortality as a consequence of a decrease in PWV of 1 m/s.¹⁹ Hence, clinical intervention with the intention of reducing PWV by at least one grade seems reasonable.

Potential Risks of Treatment

The possible benefit of treatment should be considered alongside the possible risks. Therapeutic trials based on cardiovascular mortality have shown that systolic blood pressure reduction requires normalization of both large artery stiffness and wave reflections.^20–22 Clinical trials suggest angiotensin converting enzyme inhibitors and angiotensin receptor blockers as the leading candidates for treatment of “arterial pulse hypervelocity” and wave reflection attenuation. They are also very well tolerated with few adverse effects and good long-term compliance.^23,24 Other medications are less effective long term, even with a comparable reduction in blood pressure.^15,25

Relationship of PWV with Systolic Blood Pressure and Age

Comparison of the RRs for these factors after adjustment by multivariate analysis (table 2⇑) shows the risk of a 1 SD increment in PWV is equivalent to 10 years of aging. It is also approximately 1.5 to 2 times the RR of 10 mmHg increase in systolic blood pressure.

Clinical Scenarios Resolution

Case 1. While not clearly hypertensive, assessment by the Coronary Heart Disease (CHD) Event and Stroke Risk Calculator²⁶ demonstrated this patient has a 12.6% 10-year CHD risk which is equal to a 16.75% cardiovascular event risk within 10 years. Without the PWV information there is no treatment indication. However, since his PWV was at least 1 SD more than the low risk tertile (an equivalent of 10 years of aging), reassessment demonstrated a risk of 19.7% for coronary and 26.2% for cardiovascular events. According to recent international guidelines, every patient with a 10-year cardiovascular risk of 20% (equivalent to 15% CHD risk) or greater should be treated.²⁷ Hence, medical treatment (targeting arterial function by reducing blood pressure, blood glucose, lipids, and renal function) is justified for this patient regardless of his blood pressure. Case 2. The patient’s systolic blood pressure is acceptable although her diastolic blood pressure is just above the target point. Since the patient was not diabetic, risk assessment was performed using the CHD Event and Stroke Risk Calculator.²⁶ The 10-year risk of CHD was 12.6% which is compatible with a cardiovascular risk of 16.8% without taking the PWV result into account. However, adding 10 years to her age represents 1 SD difference with the low grade PWV, and the estimated 15.5% CHD risk (20.7% cardiovascular risk) supports treatment. Case 3. Since PWV did not suggest an excess risk and considering this patient’s history of controlled blood pressure, his high blood pressure was likely to be a white coat effect and a 24-hour monitoring for the patient was ordered. The result demonstrated an average daytime blood pressure of 135/85 and a lack of nighttime blood pressure dipping. Hence, no further blood pressure medication is required. Nevertheless, the observed non-dipping in blood pressure represents an expected vascular risk in a patient with ESRD which could possibly be improved by timing adjustment of his current medication.