Regression of left ventricular hypertrophy as a surrogate end-point for morbid events in hypertension treatment trials.

OBJECTIVE

To assess whether regression of left ventricular hypertrophy (LVH) can be used as a surrogate end-point for morbid events in hypertension treatment trials.

DESIGN AND METHODS

Statistical, epidemiologic and treatment trial literature was reviewed to identify the criteria that should be met by a surrogate end-point and to determine whether these criteria are met by existing data on the regression of LVH.

RESULTS

Relevant criteria include: (1) a consistent relationship between LVH and subsequent morbid events; (2) prediction of lower or higher complication rates by LVH regression or progression; (3) evidence that the relationship between LVH regression/progression and morbidity/mortality is consistent in different populations and with different treatments; and (4) demonstration of a quantitative relationship that allows prediction of a change in clinical risk from a measured change in LVH. The results of seven electrocardiographic and 10 echocardiographic studies with a total of about 20000 subjects have shown consistently higher risks of morbid events in subjects with than without LVH (odds ratios 1.4-5.4). The available data (four studies, 1145 subjects) suggest that morbid events will occur in a higher proportion of subjects in whom LVH progresses (13-59%) rather than regresses (7-12%). However, the latter data are derived almost entirely from white subjects, predominantly male, with incomplete knowledge of interim treatment and blood pressure in most instances; no information on the mathematical relationship between change in LVH and subsequent morbidity and mortality is yet available.

CONCLUSIONS

A strict definition of the information required to establish a fully adequate surrogate end-point for morbid events in antihypertensive has been partially but not completely satisfied. The consistent relationship between baseline LVH and subsequent morbid events and the initial evidence of a parallelism between LVH change and prognosis need to be supplemented by additional studies that examine the latter relationship in diverse populations under varied treatments, and which examine the quantitative relationship between measured change in LVH and the subsequent rates of morbid events. Additional data will come from ongoing treatment trials (approximately 12 000 subjects) and observational studies (approximately 8000 subjects) with serial assessments of LVH.