Calculation of Trough-to-Peak Ratio in the Research Unit Setting

Transcription

1 A]H 1996; 9:71S-75S Calculation of Trough-to-Peak Ratio in the Research Unit Setting Advantages and Disadvantages Henry L. Elliott and Peter A. Meredith The trough-to-peak ratio for the response to an antihypertensive drug is a clinically meaningful parameter but only when the calculation has been derived from an appropriate and scientifically robust study. Since the methodological details have not been defined by any regulatory authority, several possible approaches have developed. The major apparent advantages of the intensive study of individual patients in the research unit setting are that the conditions of measurement can be standardized and an accurate account can be taken of the circadian variations in the responses to placebo and active drug treatment. The principal disadvantage is that it is an "artificial" environment that may, or may not, ~ uantification of the trough and peak blood pressure responses during steady state antihypertensive treatment is one of several features that should be assessed in the s of clinical drug development as an essential component for the complete understanding of the drug's clinical pharmacological characteristics and profile of activity. Thus, there are requirements for a detailed pharmacokinetic analysis, a full exploration of the duration of action and the trough-to-peak variability in the antihypertensive response, the clear identification of the concentration-effect relationship, and the calculation of the effective dose range and From the Department of Medicine and Therapeutics, Western Infirmary, Glasgow, Scotland. Address correspondence and reprint requests to Dr. H. L. Elliott, Department of Medicine and Therapeutics, Gardiner Institute, Western Infirmary, Glasgow Gll 6NT, Scotland. be directly relevant to routine clinical circumstances. Nevertheless, the values obtained with this approach to date are directly comparable to values obtained by the alternative approaches, such as ambulatory blood pressure measurements (provided that those are also well-conducted studies). Thus, using the trough-to-peak ratio not only appears valid but also permits the detailed study of individual patients and also lends itself to the incorporation of additional and confirmatory clinical pharmacological assessments. Am J Hypertens 1996;9:71S-75S KEY WORDS: Trough-to-peak ratio, antihypertensive drugs, methodology. most appropriate dose frequency. The relevance of the dose- and concentration-effect relationships to trough-to-peak ratio and the duration of antihypertensive effect have already been discussed in detail. 1 Some of the methodological features necessary for characterizing the concentration-effect relationship correspond directly with those necessary for the measurement of the trough and peak blood pressure responses and the calculation of trough-to-peak ratio in the research unit setting. The fundamental principles are: 1. Accurate quantification of the blood pressure changes attributable to circadian variations and placebo effects; 2. Standardized, controlled and reproducible conditions for the repeated measurements of blood pressure; and 3. Intensive studies of individual patients to mini by the American Journal of Hypertension, Ltd /96/$15.00 Published by Elsevier Science, Inc. PII S (96)00266-X

2 72S ELLIOTT AND MEREDITH AJH-OCTOBER 1996-VOL. 9, NO. 10, PART 2 mize confounding influences and to provide information about interindividual variability. PLACEBO AND CIRCADIAN EFFECTS The systolic blood pressure profile following a typical antihypertensive drug is shown in Figure 1 (upper 14o panel) and, since the lowest value for systolic BP occurs at 6 h postdose, it might be assumed that the maximal or peak antihypertensive effect occurred at 120 this time. This would constitute a baseline-adjusted reduction in blood pressure of approximately 40 mm Hg. However, the corresponding response to the ad- 100 ministration of placebo shows that the lowest blood pressure following placebo also occurs at 6 h postdose (Figure 1, middle panel). If the blood pressure profiles are then adjusted according to both the placebo and baseline values (Figure 1, lower panel), it is 16o readily apparent that the blood pressure response attributable to the drug itself is defined by the placebocorrected measurement with a maximal or peak blood 140 pressure reduction of approximately 30 mm Hg occurring at approximately 2 h postdose. 120 STANDARDIZED, CONTROLLED AND REPRODUCIBLE CONDITIONS To obtain the placebo and the active drug treatment 10o profiles and to minimize intrasubject and day-to-day variability it is recommended that the blood pressure measurements are obtained under standardized, con- 10 trolled, and reproducible conditions. This is most readily achieved in the controlled environment of a 0 clinical research unit in which it is possible to accu- rately time the recordings and standardize the conditions of measurement, including, for example, the use -2o of semiautomated blood pressure recorders to eliminate observer bias. An important practical consider- -3o ation is that the patient should report to the research unit for a minimum period of 1 h prior to the commencement of the study recordings to establish the study conditions (application of blood pressure cuff, insertion of intravenous cannula, etc) and then to permit a period of rest and acclimatization to the research unit conditions. Drug administration and the subsequent measurement of blood pressure can then be timed precisely. It can be argued that this approach creates an "artificial" environment that does not represent the conditions usually present in the hypertensive patient. However, while this may affect the absolute magnitude of the blood pressure reduction, there is no evidence to suggest that it influences the ratio between the trough and peak measurements. Since the magnitude of both the trough and peak responses are directly determined by the drug's pharmacological activity at these different times (and are often directly related to the underlying drug concentrations), 2 the trough-to-peak ratio should be a direct 160-4O SYSTOUC BP (mmhg) I I I I I I ;YSTOLIC BP (mmhg) I I I i I l I I I SYSTOUC BP (mmhg) I I I I I I FIGURE 1. Systolic blood pressure profiles following a typical antihypertensive drug (upper panel), following the administration of placebo (middle panel), and adjusted to demonstrate the differences between placebo and baseline corrections (lower panel). reflection of the ratio of the pharmacological activities present at these two times. The importance of standardized conditions for the reproducible measurement of blood pressure profiles is further illustrated by the discrepancies identified in several studies that have investigated intrasubject variability and the placebo effect during repeat studies with 24 h ambulatory monitoring, where it is often

3 AJH-OCTOBER 1996-VOL. 9, NO. 10, PART 2 CALCULATION OF T:P IN THE RESEARCH UNIT 73S assumed that there is no placebo effect. While it is conceded that the average daytime, nighttime, and 24 h values of repeat recordings are closely similar and not significantly different, it should be recognized that there may be considerable discrepancies if an attempt is made to compare individual time points. For example, a different pattern of patient activity will significantly alter the blood pressure at a particular time. The influence of some types of daily activity on the blood pressure measurements, relative to the relaxed state, are shown in Table 1. Such differences may be of sufficient magnitude to significantly influence the calculation of the trough-to-peak ratio and this illustrates the potential problem for ambulatory monitoring if the patient's activity is not standardized. There is also evidence in the published literature highlighting the practical problems of placebo or acclimatization effects and the relatively poor reproducibility of 24 h blood pressure measurements when specific times or narrow time periods are compared. 3'4 THE STUDY OF INDIVIDUAL PATIENTS The blood pressure responses in two representative subjects are illustrated in Figure 2 by the placebocorrected profiles for the antihypertensive response to steady-state treatment with a modified release formulation of nifedipine. A number of features are readily apparent: in particular, the profile of the blood pressure reduction is significantly different for these two individuals, in both magnitude and duration. With specific respect to trough-to-peak ratio, there is a clear difference in the timing of the peak antihypertensive effect, which occurs at approximately 2 h postdose with patient 6 and at approximately 5 h postdose TABLE 1. CHANGES IN BLOOD PRESSURE Blood Pressure (mm Hg) Systolic Diastolic Meetings Work Transportation Walking Household chores Telephoning Eating Talking Desk work Reading Watching television Relaxing Sleeping Adapted with permission from Clark LA, Denby I, Pregibon D, et ah The effects of activity and time of day on the diurnal variations of blood pressure. 1 Chron Dis 1987;40: rnmhg) PATIENT 6 i f -20 : : : ; : ' TIME (h) mmhg) IB m PATIENT 13 :,, ; ; ; ; TIME (h) FIGURE 2. The placebo adjusted blood pressure response profiles in two representative subjects following steady state treatment with nifedipine (modified release). with patient 13. Thus the study of individual patients avoids the necessity of arbitrary definitions of, or standardized timing of, the peak response and allows the clear identification of the peak antihypertensive response in each individual patient no matter when it occurs. Additionally, information gained from the study of the plasma drug concentration-time profiles and knowledge of the pharmacokinetic-pharmacodynamic relationship can be used to confirm this timing. If necessary, smoothing procedures can be applied to the blood pressure response profiles to eliminate "rogue" values: this is directly equivalent to the averaging techniques that are often applied when analyzing the multiple values obtained via ambulatory BP monitoring. REPRODUCIBILITY With a standardized approach it is obviously possible to conduct repeat studies of the blood pressure response at different times during chronic treatment, the responses to different drug dosages or following acute and steady state drug administration, and for the direct comparison of different drug types. We have tended to use a two stage design with an initial double-blind, random order, cross-over comparison of the responses to the first dose of active drug and placebo and, thereafter, further assessments of the blood pressure responses when steady state conditions have been established after, for example, 1 and 4 weeks of drug treatment. The reproducibility of the trough-to-peak ratio can then be assessed by comparing the values obtained at different times during treatment and this is illustrated in Figure 3 by comparison of the trough-to-peak ratios identified for each individual on two separate occasions (after 1 week and 4 weeks of treatment with nifedipine). It is clear that there is a significant correlation between these values and that there are no individual patients who deviate significantly from the linear regression line. It is also important to note, however, that there is

4 748 ELLIOTT AND MEREDITH AJH-OCTOBER 1996-VOL. 9, NO. 10, PART 2 s weeks (~) 70, Z 1 week (%) I 7O FIGURE 3. The correlation between the trough-to-peak ratio values derived following I week and 4 weeks treatment with nifedipine (modified release). a relatively wide range of interindividual values, between about 30% and 70% for nifedipine in this example. This highlights that there is interindividual variability in trough-to-peak ratio and suggests that the presentation of the values for trough-to-peak ratio should display a range of values, or show confidence intervals, in addition to a summarizing a mean reference value. There are obviously several different ways in which the variability of trough-to-peak ratio can be conveyed but an additional indication of the number of patients achieving the minimum 50% value may be of particular relevance in the clinical setting. CONFIRMATORY STUDIES The robustness of our standardized research unit approach to the calculation of the trough-to-peak ratio is further illustrated in a study of borderline and normotensive individuals where the primary end point was a comparison of the acute and steady state plasma drug concentration time profiles of amlodipine and felodipine, which are shown in Figure 4. Since it is well established that there are direct relationships between the plasma drug concentration profiles s'6 for both of these dihydropyridine calcium antagonists and their respective antihypertensive response profiles it follows that the trough-to-peak ratio for the blood pressure response should reflect the ratio between the minimum (Cmin) and maximum (Cmax) plasma drug concentrations. The ratio for CmJn :Cmax was 36% for felodipine and 70% for amlodipine and the respecfive trough-to-peak ratios for the diastolic blood pressure responses in this study were 29% and 70%. The concordance of the pharmacodynamic and pharmacokinetic ratios within this study, and the additional concordance with published values of 36% and 38% for diastolic BP with 10 and 20 mg extended release felodipine, respectively, provides further confirmation that this is an appropriate and reliable methodology. CONCLUSIONS The results obtained from studies of trough-to-peak ratio in the research unit setting are entirely comparable to, and consistent with, the results obtained from well conducted studies using other methodological approaches, including ambulatory blood pressure monitoring. Although there are practical differences between the clinical research setting and the ambulatory blood pressure monitoring approach in the community, the principles and basic requirements are in close agreement. Multiple blood pressure recordings are obviously necessary to obtain an accurate measurement not only of the placebo and circa- FELODIPINE (ng/ml) 4 AMLODIPINE (ng/ml) I Amlodipine 12 FIGURE 4. Drug concentrationtime profiles under steady-state conditions for amlodipine and felodipine ER ~ Felodipine ER m i 0 I I I I time (h) i 0 24

5 AJH-OCTOBER 1996-VOL. 9, NO. 10, PART 2 CALCULATION OF T:P IN THE RESEARCH UNIT 75S dian responses, but also of the timing of the peak response to treatment. Even with ambulatory monitoring there is a need to standardize to some extent the conditions under which the blood pressure measurements are made and a preference for the patient's activity to be similar on the assessment days. There may be an additional requirement to "smooth" or average the data in some way to obtain accurate measurements for different time periods with satisfactory discriminatory power and without spurious calculations as a result of "rogue" values. Thus, while it can be argued that the research unit setting creates an artificial environment in which the starting level of blood pressure and the absolute magnitude of the blood pressure responses may be different from those measured in "real life," the derived values not only have been shown to be genuinely representative but they also provide further detailed information about the interindividual range of responses. There is the obvious and inherent weakness of a sequential design--which ideally should be avoided by a full double-blind, cross-over trial--but randomization of the first dose and the placebo seems to eliminate the "worst" of the placebo effect. Sequential design is a feature of many of the studies of trough-to-peak ratio and, in some instances, is unavoidable because of ethical and practical considerations about protracted periods without antihypertensive treatment. However, the practical advantages of controlled, standardized, and reproducible assessments make it possible to conduct a random order, cross-over comparison with an assessment of placebo and steady-state treatment after two periods each of 4 weeks when the patient might attend the research unit on two occasions for a profile of blood pressure response measurement. Furthermore, where additional information is sought regarding compliance confirmation or for supportive plasma drug concentration measurements, it is obviously easier and more practicable to study individual patients rather than groups. REFERENCES 1. Meredith PA, Elliott HL: Concentration-effect relationships and implications for trough-to-peak ratio. Am J Hypertens 1996;9:66S-70S. 2. Meredith PA, Donnelly R, Elliott HL: Prediction and optimisation of the antihypertensive response to nifedipine. Blood Pressure 1994;3: Mutti E, Trazzi S, Omboni S, et al: Effect of placebo on 24-h non-invasive ambulatory blood pressure. J Hyperten 1991;9: Gerin W, Rosofsky M, Pieper C, Pickering TG: A test of reproducibility of blood pressure and heart rate variability using a controlled ambulatory procedure. J Hypertens 1993;11: Blychert E, Edgar B, Elmfeldt D, Hedner T: Plasma concentration-effect relationships for felodipine. A metaanalysis. Clin Pharmacol Ther 1992;52: Donnelly R, Miller SHK, Elliott HL: Pharmacodynamic modeling of the antihypertensive response to amlodipine. Clin Pharmacol Ther 1993;54: QUESTIONS AND ANSWERS Dr. Myers: Your studies have involved keeping patients in the research unit during the day, having them return home in the evening and then back to the research unit the following morning. Are the patients any different the next morning and would the results be comparable with a continuous 24 h recording of blood pressure? Dr. Meredith: If I measured the placebo blood pressure at time zero it should generally be the same as at 24 h. However, often it is not, and in that sense, the 24 h pressure tends to be lower. Dr. White: I have listened to your concerns about ambulatory blood pressure monitoring with interest and would like to report a somewhat different experience. Last year we did a study with a new antihypertensive agent, ecadotrfl (Bayer Pharmaceuticals, Basel, Switzerland) for which the Food and Drug Administration required that the initial study be done in hypertensive patients in the clinic setting because of concerns about pos~ral hypotension. We found that 40% of the subjects who met the inclusion criteria for blood pressure based upon out-of-hospital blood pressure measurements ended up with readings in the hospital that were so low that they had to be excluded from further participation in the study. It was apparent that staying in the quiet surroundings of the research setting resulted in a fall in blood pressure, presumably by removing all the external stimuli that were contributing to the hypertension seen during the ambulatory recording. Dr. Elliott: I would agree with you that the absolute values we might obtain for the trough and peak reductions in blood pressure will be different in the research unit setting compared to routine clinical practice but I would maintain that the ratio remains exactly the same regardless of where the measurements are performed. In performing these studies, the expectation is to estimate a trough-to-peak ratio for a drug that is applicable for all hypertensive patients. I have no doubt that our approach is achieving this goal. Dr. Lipicky: At this point we should remember that there is nothing magical about the trough-to-peak ratio. It is only a simple method of estimating the time course of a drug's effect on blood pressure. You have described it in a quantitative model that has some predictive value. One should not consider the trough-to-peak ratio as though it is a number with intrinsic value. It only reflects the time course of a drug's action.