Trough-to-peak ratio, smoothness index and morning-to-evening ratio

In this issue of the journal, Stergiou et al. [1] report on a direct comparison of different assessments of blood pressure response to antihypertensive treatment provided by indices such as trough-to-peak ratio (TPR), morning-to-evening ratio (MER) and smoothness index (SI). This was carried out by analysing the data obtained in patients sequentially receiving two antihypertensive agents, according to a randomized cross-over design. The relative advantages and limitations of different methods for evaluating the efficacy and the duration of action of antihypertensive drugs over 24 h have already been addressed in previous studies [2–4]. However, these methods, initially designed to assess the time–effect profile of individual antihypertensive drugs, are now more and more frequently used to compare the changes in 24-h blood pressure induced by different, and more or less complex, combination treatment regimens. The proper use and the actual value of these analysis procedures when applied in the latter condition still requires further investigation. Indeed, their best possible implementation in a clinical setting is presently the matter of research and is far from being established. Therefore, any inference on the clinical superiority of one particular treatment regimen agent over another based on a higher TPR, MER or SI remains largely speculative in nature. Each of these methods is characterized not only by specific advantages, but also by intrinsic limitations (Table 1), as testified by the difficulties met by a number of studies addressing their correct application. Moreover, important statistical and biological problems related to the evaluation of antihypertensive treatment by means of TPR, SI and MER (i.e. non-normality of distribution, large confidence intervals, dilution of results related to the inclusion of non-responders, random variations of blood pressure, etc.) have only seldom been considered. In most cases, power calculation has been disregarded, resulting in potentially misleading conclusions.Table 1: Comparative characteristics of TPR, MER and SIFinally, the actual clinical relevance of comparing TPR, SI or MER values of different antihypertensive agents or drug combinations is not yet fully understood. Indeed, only SI has demonstrated any relationship with clinical endpoints, such as cardiac hypertrophy or carotid artery intima–media thickness. However, the minimum size of the difference in any of these indices that might be of clinical importance when comparing two antihypertensive regimens, and the target TPR, SI or MER values that should be reached in order to achieve optimal blood pressure control, are still largely undefined. Thus, while TPR, SI and MER are undoubtedly useful in the early stages of development of antihypertensive agents, and more specifically for the determination of the appropriate dose and dosing interval of single compounds, some caution is still needed when considering their clinical application, particularly in the case of multiple drug administration. In this regard, the following observations on the pros and cons of each of these indices may provide a useful background. Trough-to-peak ratio A value of TPR higher than 50% was initially proposed by the US Food and Drug Administration (FDA) as the prerequisite to approve a given antihypertensive drug for once-a-day administration [5,6]. The main aim of this recommendation was to provide a simple tool to assess the duration of action of a given compound, and to evaluate whether its effect is well balanced over 24 h. This was largely aimed at detecting those drugs with a short duration of action where an inappropriate once-a-day use of a high dose may achieve an optimal trough effect, but only at the expense of a large blood pressure drop at peak [7,8]. A pronounced blood pressure fall at peak may generate a number of side-effects and, in particular, may impair organ perfusion. It may also be associated with an increased blood pressure variability, which was shown to contribute to target organ damage and the risk of cardiovascular events [9–14]. For a number of drugs, elimination half-life is significantly different from biological half-life. Therefore, even when measurable, the plasma concentrations of a given compound may not predict its duration of action. On this background, TPR was introduced as a method of assessing the time–effect profile of antihypertensive drugs. The original recommendations for its proper application [7] required repeated conventional blood pressure measurements to be performed in comparable standardized conditions, both at peak and at trough. The ratio between the average values of these multiple trough and peak measurements was suggested to provide a reliable and reproducible index of drug effect–time distribution. Subsequently, however, TPR has been used extensively, but in most cases improperly [15], sometimes serving ill-advised marketing strategies. Indeed, in most cases, TPR was not assessed from standardized clinical blood pressure measurements due to practical reasons, but was quantified through the analysis of blood pressure profiles obtained by 24-h ambulatory blood pressure monitoring (ABPM). This was performed without considering that the accuracy of TPR, calculated as the ratio between the hourly ambulatory blood pressure values obtained at trough and at peak times, may be heavily affected by behaviour-induced blood pressure variations, and the possibility that this approach leaves most of the recorded blood pressure data disregarded. Actually, the FDA (and subsequently the European agency) did not officially indicate any specific method that should be used to guarantee a reliable calculation of TPR, a calculation that at present is thus largely based on ABPM data obtained under uncontrolled behavioural conditions. Thus, although TPR has been accepted by the scientific community as an important parameter for evaluation of the time–effect profile of antihypertensive agents, the lack of methodological standardization in its assessment may lead to a significant bias, as indicated by previous studies dealing with this issue [1–5,7,8,15], and may be responsible for a poor reproducibility of this index, a limitation to which the known occurrence of random ambulatory blood pressure variations may also contribute [8]. Other important reasons responsible of the wide individual scattering of TPR are inclusion of non-responders to active treatment and lack of standardization of activities, as well as inclusion of patients defined as hypertensive only on office blood pressure measurement, a significant proportion of whom are likely to have only ‘white coat’ hypertension, with normal ambulatory blood pressure values [2,15,16]. Because it is not advisable to select a subset of patients for statistical comparison in a randomized study, responders may be preselected through a preliminary evaluation by a dedicated independent protocol. In the case of a comparison of two active drugs, this may be achieved with a cross-over design. However, this is not usually carried out, and the inclusion of non-responders may lead both to high variability and to overestimation of TPR values. The exclusion of non-responders appears to decrease the dispersion of the TPR ratio and normalizes its distribution. Obviously, responders should be selected on ABPM or on self-blood pressure monitoring at home. Indeed, evaluation of response to treatment by office and by ABPM is characterized by pronounced between-method differences [17,18]. Although a good concordance between ABPM responders and responders based on office blood pressure was reported, no concordance may be found between ABPM non- responders and non-responders defined on the basis of office measurements [16]. These discrepancies may depend on the fact that, when using clinical blood pressure measurements alone, investigators may overestimate baseline blood pressure values, as well as the ‘efficacy’ of treatment, due to the possible occurrence of a ‘white-coat’ effect, its subsequent attenuation and the well-known interference with this conventional blood pressure reading by a placebo effect [19]. An additional issue to be considered for an optimal TPR calculation is that ABPM would be best performed in the controlled environment of a research unit where the standardized conditions of measurement can lead to acceptable accuracy and reproducibility [19]. Thus, an ideal protocol for optimal TPR determination would include only those patients defined as hypertensive on ABPM criteria and previously selected as being ABPM responders to the tested treatments. Moreover, ABPM should be performed in a confined ‘research unit’ setting. This ideal protocol still remains to be tested and validated. Thus, at present, TPR calculation appears to be more suited to the analysis of average peak and trough data from large groups of subjects, where the individual variability in TPR values can be dealt with, than to the data obtained in individual patients [2,8]. Smoothness index The SI was introduced with the aim of overcoming a number of limitations affecting TPR [20,21]. In fact, this index, which is derived from the analysis of 24-h blood pressure profiles obtained before and during drug treatment, takes into account all 24-h blood pressure differences between baseline and treatment, allowing more comprehensive information on the homogeneity of 24-h therapeutic coverage to be obtained. In addition, further information on the magnitude of treatment-induced blood pressure changes all over 24 h is provided. At variance from TPR, which can reach high values even after placebo administration, SI demonstrates low values with placebo, and progressively higher values with the increasing efficacy of drug-induced blood pressure reduction [22]. Therefore, in contrast to TPR, SI is related to both pharmacokinetics and pharmacodynamics, and provides a more comprehensive evaluation of the antihypertensive effect of individual agents. It may prove useful to define the optimal therapeutic dose and dosing interval in phase II studies. In comparative studies, after appropriate adjustment for average 24-h blood pressure changes, it may prove suitable for comparing the stability of blood pressure changes during the interdosing interval of specific antihypertensive agents. Because SI is normally distributed and integrates all the information on treatment-induced 24-h blood pressure changes, it may be superior to TPR [1]. Moreover, compared to TPR, the SI is significantly correlated with the degree of blood pressure variability achieved under treatment [21] and has been reported to be associated with changes in left ventricular mass, as well as with changes in carotid artery wall thickness obtained under antihypertensive treatment [21,23]. However, it remains to be investigated whether the relationship between SI and reduction in target organ damage is explained mostly by the magnitude of the average 24-h blood pressure changes induced by treatment, or whether, and to what extent, the reduced blood pressure variability also plays a role [21]. Morning-to-evening ratio As suggested in the study by Stergiou et al. [1], MER results are consistent with TPR results. However, it may not be an alternative to TPR because it shares the same methodological limitations as TPR, with the additional limitation that it may not detect the excessive peak effect, which is the main advantage of TPR. It is also unlikely to have any major clinical relevance in the absence of a clear temporal relationship of morning and evening blood pressure measurements with drug assumption. One potential advantage of MER is the possibility to obtain its quantification through the use of self-blood pressure measurement at home, the adoption of which has been recently addressed by practical guidelines [24]. Compared to ABPM, self-blood pressure measurement is more readily available, less costly, more acceptable to the patient [25] and allows individualized treatment to be monitored closely. Therefore, MER may prove to be a useful tool for antihypertensive treatment monitoring in the individual patient [24,26], but appears to be of little value in clinical pharmacology when investigating the 24-h profile of the effects of antihypertensive drugs. Conclusions At present, the ‘ideal’ approach for evaluating blood pressure coverage offered by antihypertensive drugs over 24 h, and its impact on the cardiovascular risk of patients with high blood pressure, still requires further research. Some ongoing trials, such as the THOP [27] and the APTH [28], are specifically designed to compare the relative merits of self-measured blood pressure and ABPM versus conventional office blood pressure readings in the assessment of antihypertensive treatment. On the background of the data provided by Stergiou et al. [1], ad-hoc designed trials [29,30] should address the clinical important issue of quantifying the 24-h distribution of treatment-induced blood pressure reduction, and the actual ability of SI to provide a better score than MER and TPR.