Arterial stiffness (from monitoring of Qkd interval) predict the occurrence of cardiovascular events and total mortality

Arterial stiffness, most often assessed with carotido-femoral pulse wave velocity predicts cardiovascular events but its use in clinical practice remains limited. The 24 h ambulatory monitoring of Blood pressure and timing of Korotkoff sounds (QKD interval) allows an automatic assessment of arterial stiffness and is an independent predictor of cardiovascular events in hypertensive patients. The long term follow up of our cohort of hypertensive patients gave us the opportunity to test the consequences of increased arterial stiffness on the incidence of all causes deaths and to define the populations who could benefit of this measurement beyond risk scores. The sample includes 930 patients (502 males, age 53 ± 13 years, baseline risk SCORE2-OP = 6.70 ± 4.97%) with an average follow up of 12.11 ± 7.4 years (0.3–30). In this population 169 cardiovascular events and 155 deaths were recorded. SCORE2-OP, 24 h Systolic Blood Pressure and arterial stiffness (QKDh) as a continuous or discontinuous variable (normal or reduced) were significantly and independently linked to the occurrence of cardiovascular events or all cause deaths in multivariate Cox model. ROC curves analysis show that measuring arterial stiffness with QKD method offers the best predictive value in patients with low or very low risk scores.


INTRODUCTION
Markers of subclinical target organ damage increase cardiovascular risk prediction beyond traditional risk factors.Arterial stiffness (AS) is one of these markers [1].Observational studies have demonstrated the predictive value for future cardiovascular events (CVE) of both the carotido-femoral pulse wave velocity (PWV) (for patients with renal failure [2], hypertension (HTN) [3] and in general population [4]) and the QKD interval (QKDh, for hypertensive patients [5]) independently of traditional risk factors: sex, age, diabetes, smoking, cholesterol and blood pressure(BP), and left ventricular hypertrophy [5].Therefore AS is a powerful marker of cardiovascular risk and is helpful to reclassify patients at intermediate risk.The aims of this study are to test the consequences of reduced QKDh on the incidence of all causes deaths and to define the populations for which this measurement could improve the risk prediction beyond traditional risk factors.

METHODS Population and inclusion criteria
The patients included in this study belong to the Bordeaux cohort of hypertensive subjects.
Briefly, this registry was started in 1984 and includes all patients consecutively referred to our center before administration of antihypertensive treatment and fulfilling the following criteria: Confirmed HTN with Office BP > 140/90 mmHg on at least two occasions without secondary cause of HTN; Previous CVE, pathologies likely to affect the prognosis, and type 1 diabetes were exclusion criteria.For the validity of the ambulatory measurement of AS through QKD measurement, two other selection criteria were employed: absence of bundle branch block on ECG (QRS duration < 120 msec), and absence of known thyroid pathology.
We included in this study patients who benefited from an ambulatory measurement of BP and QKD either on inclusion in the cohort prior to treatment, or during the follow-up under treatment.Therefore baseline for this study was defined as the time of the first QKD monitoring available for each patient.
All the patients gave their consent to participate in this registry, which was approved by our local committee of ethics and protection of the individual (Committee for Protection of Persons in the South-West and Overseas III, February 28th 2007).

Ambulatory measurement of BP and QKD interval
All the patients benefited from an ambulatory measurement of BP coupled with the measurement of QKD interval [6].The equipment (Diasys 200RK then Diasys integra and from 2020 Diasys 3 plus, Novacor, France) is a device based on the auscultatory measurement of BP, which also determined for each measurement the duration of the QKD interval as the time between the QRS wave on the ECG and the detection of the last Korotkoff sound on deflation of the cuff.This time is the sum of the preejection time and time of transmission of the pulse wave between the aortic valves and the microphone placed on the brachial artery.This interval is measured every 15 min along with BP and heart rate over 24 h.From all the measurements obtained, the software automatically determines the value of the QKD 100-60 [7] which is the theoretical value of the QKD for a 100 mmHg SBP and a 60 bpm heart rate based on the bivariate linear regression linking QKD to these two variables.This index reduces the influence of the pre-ejection time and enables comparison of patients at equal systolic BP.The value obtained is expressed as a percentage of a theoretical normal value based on the height predicted value of a population of normal young subjects [8]: QKDh.The lower this value, the higher the AS.A low QKDh was defined as <100% of the normal height predicted value.
The ABPM with QKD device was set up with a suitable size of cuff by a dedicated nurse on the left arm.Good quality criteria were required to take into account the ABPM.Recordings with 70 measurements or more were considered of good quality.
Results of 24 h BP recordings were sent to GP but the results of QKD were not as there was so far no evidence that this result should influence the treatment of these patients.

Estimation of risk SCORE
An estimation of 10 year risk of fatal and non-fatal cardiovascular disease was calculated from the SCORE2-OP (SCORE2 for patient <70 years, OP for patients >70 years) [9,10] charts for low risk populations using gender, office systolic BP, smoking status and cholesterol levels.As this SCORE is not validated in patients with diabetes, these patients were excluded from this study.

Follow up
After the initial evaluation, the patients were given antihypertensive treatment and followed up by their family physicians who were not informed of arterial stiffness results.These patients were not systematically convened, but could be sent again by their doctors during the follow-up period.However, information from these patients was obtained regularly (every 2 years on average) either directly from the patients or from their physicians.A detailed report was requested in the event of CVE.Events retained for analysis were: myocardial infarction, coronary revascularization, stroke documented by CT scan or MRI, sudden death, arteriopathy of the lower limbs requiring revascularization, aneurism of the abdominal aorta operated or ruptured, carotid stenosis requiring angioplasty, aortic dissection.In 2021, we made a systematic search of all cause deaths for all the patients in the registry from INSEE (Institut National de la Statistique et des Etudes Economiques).However, we were not able to obtain the cause of death for all patients.

Statistics
The data were analysed with SPSS 26 IBM software.We tested the relative interest of raw average 24 h QKD, QKD 100-60 and QKDh to predict CVE and total mortality with ROC curves.The relationship between QKDh and the variables used to calculate the SCORE2-OP was analysed by multivariate regression analysis and the linear regression between QKDh and SCORE2-OP was also tested.First we tested the independence of the QKDh to predict events with a Cox proportional model with forward stepwise conditional method and likelihood ratio.The analysis were carried out separately for (1) the first outcome (CVE or death), (2) CVE and (3) all cause deaths.QKDh was tested as a continuous or a discontinuous variable (normal or low).The other variables included for analyses were: risk SCORE2-OP and 24 h average SBP.
Then we tested the factors that may influence the predictive value of QKDh.We analysed the link between basal data and events with ROC curves analysis for QKDh (a low value is linked to the event), age, 24 h SBP,   SCORE2-OP (a high value is linked to the event) in the whole population then in subgroups (SCORE2-OP ≤ 5%, ≤10%, males or females, treated or untreated patients).Differences in AUC were tested with paired tests (paired sample design).

RESULTS
In end 2021, 1105 patients with 24 h monitoring of QKD were included in the registry.For this study, 96 patients with diabetes, 35 patients with QRS duration exceeding 119 ms and 44 patients with thyroid diseases were excluded (Fig. 1) so the sample for the present analysis included 930 patients whose main characteristics are summarised in Table 1.The average follow-up duration for this sample was 12.11 ± 7.4 years (0.3-33).In this population 169 CVE were recorded (Table 2), and 155 deaths (37 of proven cardiovascular origin, 50 of proven non cardiovascular origin and 68 of undetermined cause).Cardiovascular event or death as the first outcome was recorded in 254 patients.Table 2 describes the relative value of the different indices of QKD to predict CVE or deaths with ROC curves analysis.All indices are significantly predictive of events but as expected QKDh provide the best results.Figure 2 shows the ROC curve of QKDh to predict CVE and Fig. 3 shows variations of sensitivity and specificity to predict CVE according to QKDh values.All events were significantly more frequent in patients with reduced QKDh at baseline (Table 3).QKD as a marker of arterial stiffness was correlated to age (r = −0.32,p < 0.001), 24 h SBP (r = −0.196,p < 0.001) and to cholesterol (r = −0.124,p = 0.001) but in multivariate regression analysis these 3 variables explain only 15% of its variance.SCORE2-OP alone is correlated to QKDh (r = −0.346,p < 0.001) and explain only 12% of its variance.In this population (Table 4), SCORE2-OP, 24 h SBP and QKDh as a continuous or discontinuous variable were significantly and independently linked to the occurrence of first outcome (CVE or death), CVE and all cause deaths in multivariate Cox models.Figure 4 shows the first outcome free survival curves from the Cox Model for QKDh as a dichotomic variable.As a sensitivity analysis, the same analysis was performed in the whole cohort (n = 1105, 192 deaths) with similar results (results not shown).
ROC curve analysis shows that baseline SCORE2-OP, age, 24 h SBP and QKDh are significantly linked to first outcome (Table 5) and CVE (Table 6).In the population with very low (≤5%) baseline   risk, there is a trend to a greater AUC observed with QKDh as compared to SCORE2-OP (p = 0.155) for the prediction of CVE.A value of 95% offer the best compromise between sensitivity and specificity (Youden indice = 1.38) with 63% sensitivity and 67% specificity.The same analysis performed separately in males and females showed no significant difference with the limitation of the much smaller number of events in females (Supplementary table 1).When taking into account the presence or absence of antihypertensive treatment at baseline (Supplementary Table 2), the AUC for all predictors were higher in untreated patients.But the lower AUC in treated patients was observed as well for variables dependent of treatment (24 h SBP and QKD) or not dependent (age and SCORE).

DISCUSSION
This study confirms that AS evaluated from 24 h ambulatory monitoring of BP and QKD is an independent predictor of CVE and mortality.
The ambulatory measurement of QKD interval enables determination of an index of arterial stiffness coupled with measurements of BP.This time delay is related to the velocity of the pulse wave (PWV) and the length of the arterial route.The QKD 100-60 provides an isobaric index of arterial stiffness, making it possible to compare patients at equal BP.We have also shown with an invasive study [11] that the QKD interval was significantly correlated to aortic systolic BP and could be used for an assessment of central BP by a non-invasive method [12].We have already shown in our cohort [5,13,14] that QKDh is significantly related to the occurrence of the CVE independently of age, mean 24 h BP and other traditional risk factors including left ventricular hypertrophy.This new analysis of our cohort is based on a larger number of CVE and deaths with a systematic search of deaths from INSEE registry performed in 2021.It confirms our previous results showing a relation between QKDh and CVE or all cause deaths independently of risk SCORE2-OP and 24 h SBP demonstrating clearly that QKDh add predictive information to the European SCORE as other measurements of PWV [15].The sensitivity analysis based on the whole registry confirms the robustness of the analysis but as duration of QRS and thyroid diseases strongly influence the QKD via the pre-ejection time, monitoring of QKD cannot be considered as reliable in patients with large QRS, thyroid disease but also cardiac failure.
The present analysis affords new data.The added predictive value of arterial stiffness as compared to SCORE2-OP is greater in patients with low or very low risk.Previous studies on the link between PWV and cardiovascular or total mortality were often performed in populations with higher baseline risk than ours (average age >60 years, renal failure, diabetes, previous CVE) [16].But AS measurement is probably even more informative in low risk patients and could contribute to an earlier and more efficient control of all risk factors.However the difference between the AUC of SCORE2-OP and QKD did not reach statistical significance in our study and these results deserve confirmation with a greater sample or more events.Some important limitations stem from the structure this study.It is a prospective registry.Initially the antihypertensive treatments were proposed by our team, but later choices and modifications were often carried out by the family doctors and were based mostly on the office BP.It was impossible to study the influence of the different classes of antihypertensive agent in this cohort.As follow up 24 h monitoring of BP and QKD were not systematically performed in this cohort, we have so far not enough data to evaluate the evolution of these parameters with time and treatments but the survey of these patients will continue with more systematic evaluation of BP and QKD with 24 h monitoring.Also there remains potential confounding in models that contain age, systolic BP, and large artery stiffness because of collinearity of these markers.
In conclusion AS evaluated by the ambulatory measurement of QKD interval has an independent prognostic value in the hypertensive, independent of 24 h SBP measured simultaneously and add significant predictive information to the risk SCORE2-OP.The potential addition of AS evaluation seems to be even more important in a very low risk population.

Fig.
Fig. ROC curve QKDh and cardiovascular events.

Fig. 3
Fig.3Variations of sensitivity and specificity to predict cardiovascular events according to QKDh values.

Fig. 4
Fig. 4 Event free survival curves for patents with normal or reduced QKDh (mulyivariate Cox model).

Table 1 .
Main characteristics of population studied.
BMI body mass index, BP blood pressure, D diastolic, S systolic, M mean, QKDh measured QKD 100-60 /height predicted value in normal subjects.

Table 2 .
Comparison of the predictive values of the different QKD indices with ROC curves.
AUC Area under the curve, CI confidence interval.

Table 3 .
Cardiovascular events and deaths observed as a function of normal or reduced QKDh at baseline.

Table 4 .
Main results of multivariate Cox analysis in the whole population.