Prognostic Indicators for Coronary Artery Disease: Title and subTitle BreakReady for the Bedside?

Identification of individuals at high risk for coronary events is a growing concern, reflecting increasing confidence in the capability of mechanical and pharmacological interventions to improve the outlook for coronary atherosclerosis. Numerous prognostic indicators have been proposed. Recently, physicians and the lay public have been inundated by advertisements for new tests aimed at identifying or excluding high coronary risk, with an implicit promise of a longer, better, less anxious life as the ultimate payoff.^{1 - 2} Validation of prognostic indices with a full understanding of their appropriate context and their implications for patient management are vitally important in formulating effective and efficient strategies for patient care. Inappropriate application of prognostic indices can lead to excessive use of costly and potentially injurious tests and therapies as well as needless patient anxiety.^{3 - 4}

To be generally useful, prognostic indicators of coronary events must satisfy a number of stringent criteria: (1) high specificity to minimize the worry and expense of false-positive results, which are frequently encountered when prognostic indicators are applied to healthy populations; (2) sufficient sensitivity to yield a meaningful negative result; (3) minimal intrusiveness; (4) modest expense; (5) broad applicability; (6) uniqueness because the indicator must be more than a simple reflection of other, readily available risk factors; and (7) identification of an important precursor process whose modification will lead to patient benefit at an acceptable price, risk, and level of discomfort. The last of these is a significant attribute that should be demonstrated in validating studies. The application of an indicator of coronary prognosis to a healthy population has limited value unless positive findings result in tangibly better patient outcomes. The induction of fear and heightened sense of risk, unwarranted for many individuals, should not be regarded as an unequivocal positive outcome^{5 - 6} and sustaining rationale for testing programs. Everyone exhibiting primary risk factors should undertake effective efforts to modify risk, not just persons made particularly anxious by a positive test result.⁷

Two potential prognostic indicators are evaluated in this issue of THE JOURNAL. Daviglus et al⁸ assessed the predictive ability of minor (nondiagnostic) ST-segment and T-wave changes on annual resting electrocardiograms in 1673 men employed by the Western Electric Corporation. Late coronary mortality was increased among 173 individuals with these electrocardiographic abnormalities, particularly among the small minority (32 subjects) with minor ST-T changes noted on 3 electrocardiograms, each performed at least a year apart. The study is valuable in confirming the transient character of most minor ST-T abnormalities and demonstrating the considerable increase in predictive value with persistent abnormality.

However, certain study features substantially limit general use of minor ST-T changes as a predictive index. Since all deaths and all individuals with overt coronary disease during the first 5 years of observation were excluded, the study necessarily focuses on long-range prediction. The applicability of such long-range forecasting to clinical management is unclear: what is the appropriate, cost-effective diagnostic workup for persons with minor ST-T abnormalities? Would the presence of such abnormalities be a rational basis for treatment to prevent death years or even decades hence? Having observed a minor ST-T abnormality—and probably increased patient and physician anxiety^{5 - 6} —is it reasonable to suspend judgment for 2 more years while awaiting the unlikely but more accurately predictive finding of persistence? These matters need investigation before placing added weight on minor ST-T changes in ordinary clinical practice. Importantly, the data presented by Daviglus et al were derived from an all-male, ethnically restricted cohort from the 1950s with an array of coronary risk factors. Contemporary, more diverse populations might show less linkage between ST-T abnormalities and later coronary death due to better prevention of atherosclerotic progression and a higher frequency of minor ST-T changes unassociated with occlusive coronary disease in major subgroups absent or poorly represented in this study.^{9 - 10} Dependence on specific population characteristics may explain the divergent results in the literature (eg, the lack of predictive value of minor ST-T abnormalities in populations likely more diverse than the individuals followed up by Daviglus et al).¹¹

How do minor ST-T abnormalities rate when assessed by the criteria for prognostic indicators listed above? Measurement of ST-T abnormalities is inexpensive and nonintrusive, and their unique contribution to risk is confirmed by multivariate analysis. Study exclusions and other design features obscure their specificity and sensitivity for significant coronary outcomes. The applicability of this index depends on the definition of predictive abnormality. If at least 3 years are necessary to establish persistence of minor ST-T change, only a small subset (1.9% of screened subjects) is deemed at increased risk. If a single occurrence of ST-T abnormality is viewed as sufficient, the predictive value is modest (risk ratio of 1.67) and the cost, in terms of unneeded anxiety, testing, and even treatment, correspondingly increased.^{3 - 4 ,12} In addition, the patient benefit of pursuing this testing strategy is not established, particularly in a diverse, contemporary patient population.

Lauer et al¹³ take a different approach, examining chronotropic incompetence (reduced peak heart rate relative to predicted values) as a harbinger of mortality during the ensuing 2 years in 2953 low-risk patients referred for symptom-limited exercise thallium²⁰¹ scintigraphy. Chronotropic incompetence was noted in 11% to 26% of individuals even though patients taking β-blockers were excluded from study. Two measures of chronotropic incompetence were each associated with increased total mortality, the study's primary end point, as well as cardiovascular mortality. Multivariate analysis indicated that chronotropic incompetence added independently, and about equally, to mortality prediction provided by scintigraphic defects. This study reinforces the predictive importance of peak heart rate, a physiologic measure of exercise function, in a significant population—potential coronary patients without prior angiography. This physiologic assessment of circulatory limits during effort appears to add to the well-recognized prognostic contribution of exercise duration.¹⁴

The interpretation of the study by Lauer et al is constrained since the connection between cardiovascular mortality and chronotropic incompetence, the linkage of clearest relevance, is based on a small number of cardiovascular deaths (22 of 91 total deaths). The connection between heart rate and total mortality is somewhat suspect because of the preponderance of noncardiovascular deaths. This connection may be spurious, a concern bolstered by the small number of end point events in relation to the number of variables examined.¹⁵ The authors raise the question: should chronotropic incompetence on exercise thallium²⁰¹ testing be used as a basis for advanced testing and therapy? Although data reflecting peak heart rate on exercise may have considerable clinical importance for specific individuals (eg, as a measure of limitation in patients with angina), the findings of Lauer et al have insufficient statistical strength to justify making chronotropic incompetence a preponderant consideration in selecting subsequent clinical management. As in the study of Daviglus et al, the overall benefit of the proposed testing strategy remains uncertain, with no specific course of action clearly shown to enhance clinically useful outcomes. In the study by Lauer et al, there is no reason to believe that cardiovascular testing and intervention would benefit noncardiovascular mortality, a major component of the demonstrated risk of chronotropic incompetence.

The studies by Daviglus et al and Lauer et al raise interesting and important issues, which may be resolved by further examination of the data and prospective validation. However, additional studies confirming these results and providing solid evidence of patient benefit are necessary to justify changing current methods of seeking and managing coronary artery disease.^{7 ,16 - 17}