Cardiovascular Risk in Women with Nonspecific Chest Pain

Abstract

Evaluation of: Robinson JG, Wallace R, Limacher M et al.: Cardiovascular risk in women with non-specific chest pain (from the Women's Health Initiative hormone trials). Am. J. Cardiol. 102(6), 693–699 (2008). Chest pain is common, and often a specific cause fails to be diagnosed. There is particular concern that many women with a diagnosis of nonspecific chest pain (NSCP) may have atypical indications of coronary artery disease. Analysis of data from the Women's Health Initiative hormone therapy trials, amongst 24,834 postmenopausal women aged 50–79 years who were initially free of cardiovascular disease, found that NSCP was associated with a 139% (95% CI: 46–292%) increase in the risk of subsequent angina. Similar proportional increases in risk were found for cardiac surgery or interventions and nonfatal myocardial infarction. Thus, a diagnosis of NSCP in a woman, using current technology, does not rule out the possibility of increased coronary risk. It would be prudent to treat women with NSCP as being at a higher than average risk of cardiovascular disease

Nonspecific chest pain (NSCP) is a diagnosis made when a patient presents with chest pain, but the origin cannot be identified. It is a particular concern that women with chest pain who are at high risk of coronary artery disease (CAD) may present with indications of CAD that are not typical of the male model, which has traditionally led the field of cardiology and, thus, their chest pain may incorrectly be labeled as noncardiac in origin. The latest paper on this topic [1] addresses the issue of whether women that were given a diagnosis of NSCP have a greater than expected risk of future CAD.

Chest pain is a common complaint, leading to more than 6 million hospital admissions annually in the USA [2]. A major concern is that the chest pain may be indicative of CAD; indeed, chest pain is the most common symptom of CAD to cause patients to seek medical attention [3]. Of those presenting with chest pain, it has been estimated that 16% will have cardiovascular disease (CVD), including heart failure, but the same percentage will be diagnosed with NSCP, after ruling out the other common causes: musculoskeletal, gastrointestinal and psychiatric [4]. Studies have demonstrated that women with ischemia are at least twice as likely as men to have nonobstructive or normal coronary arteries [5]. Only approximately 50% of women who have chest pain suggestive of ischemia demonstrate obvious indications of CAD on their coronary angiograms [3,5]. Since a negative angiogram will be a major factor in ruling out CAD in those with chest pain, many people, predominantly women, who receive a diagnosis of NSCP may nevertheless, be at relatively high risk for CAD and, perhaps, other types of CVD. This is supported by a report from the Women's Ischemia Syndrome Evaluation study investigating women with chest pain who were undergoing angiograms for suspected myocardial infarction [6]. This study found that, amongst women without angiogram evidence of obstructive CAD, those with persistent (for at least 1 year) chest pain had a relative risk of 1.95 (95% CI: 1.004–3.79) for cardiovascular events, compared with those without persistent chest pains.

Given its inherently vague nature, it is not surprising that the epidemiology of NSCP has been poorly studied. A recent review [7] concluded that the prevalence of noncardiac chest pain (NCCP) is approximately 25% in developed countries. Using other data quoted above [4], the prevalence of NSCP would thus be almost 5%, which represents approximately 15 million people in the USA. The same review suggested a number of possible risk factors for NCCP, including pregnancy and menopause. This raises the possibility that hormone therapy (HT) could be a risk factor for NSCP.

The Women's Health Initiative

The Women's Health Initiative (WHI) is made up of two distinct studies amongst 50–79 year-old postmenopausal American women: a randomized, controlled trial (RCT) that enrolled 68,132 women; and a cohort study that investigated 93,676 women who were unwilling or ineligible to enter the RCT [8]. The RCT studied four interventions: conjugated equine estrogen HT [9], estrogen plus medroxyprogesterone acetate HT [10], a low-fat diet and calcium plus vitamin D supplementation. Follow-up is still ongoing in both studies, although the randomized interventions have ceased (both HT interventions were ceased earlier than planned for safety reasons). The HT components of the RCT were run as parallel double-blind, randomized studies amongst 27,347 women. Those who had had a hysterectomy before the start of the trial (n = 10,739) were allocated, at random, to receive either 0.625 mg/day of estrogen or a placebo. The remaining 16,608 – with a uterus – were allocated to either 0.625 mg/day of estrogen plus 2.5 mg/day of progestin or a placebo. The RCT was clearly designed to answer specific questions about the causal effects of the interventions, and uses the ideal design to answer such questions [11]. The cohort study was designed to observe trends and associations that arise naturally, leading to it being called the WHI Observational Study (WHIOS). This study would, for example, be able to observe the association between HT use and CVD, just as in the RCT, but the evidence for a causal explanation would be weaker in the cohort study because HT was self-selected.

WHI cohort study analysis of NSCP

Amongst other things, the WHIOS has studied the relationship between NSCP and CVD [12]. After excluding those with CVD at baseline, 83,622 women with NSCP were followed up for an average of 8 years. NSCP was defined as an initial hospital diagnosis of NSCP (International Classification of Diseases [ICD] version 9 [13] codes of 786.50 [chest pain – unspecified], 786.51 [precordial pain], 786.59 [chest pain – other; discomfort, pressure and tightness]), with no prior diagnosis of coronary heart disease. Demographic and CVD risk factors, but not HT use, were compared between women with and without NSCP – those with NSCP had a relatively unfavorable cardiovascular risk profile. Hazard ratios for subsequent CVD events were estimated from Cox proportional hazards models – these demonstrated a clear trend towards excess risk of CAD for women with NSCP ( Table 1 ).

Table 1.

Hazard ratios (95% CI) for cardiovascular events, those with nonspecific chest pain compared with those without, from the Women's Health Initiative.

Outcome	Cohort study [12]	Clinical trial [1]	Pooled
Hospitalized angina	2.18 (1.66–2.86)	2.39 (1.46–3.92)	2.23 (1.76–2.83)
CABG/PTCA	1.67 (1.28–2.20)	1.99 (1.20–3.30)	1.74 (1.37–2.20)
Nonfatal MI	1.59 (1.10–2.31)	2.10 (1.11–3.98)	1.71 (1.24–2.35)
Coronary death	1.19 (0.61–2.31)	1.77 (0.54–5.76)	1.31 (0.73–2.34)
Heart failure	1.75 (1.27–2.41)	1.29 (0.63–2.64)	1.66 (1.24–2.23)
Stroke	1.32 (0.91–1.92)	1.20 (0.56–2.56)	1.30 (0.93–1.81)

Both studies adjusted for age, hypertension, BMI, race/ethnicity, education, physical activity, smoking, alcohol, diabetes, medication for high cholesterol, aspirin use, family history of stroke and/or MI, estrogen use and progestin use and stratified by clinic of origin.

CABG: Coronary artery bypass graft; MI: Myocardial infarction; PCTA: Percutaneous coronary angioplasty.

WHI clinical trial analysis of NSCP

The latest publication from the WHI uses the HT components of their RCT to study NSCP [1]. This study aimed to determine whether those allocated to either of the two HT preparations had any excess risk of incident NSCP during the RCT, and to seek confirmation of the excess cardiovascular risks subsequent to a diagnosis of NSCP that were found in the WHIOS. To answer both questions, Cox models were used on the study population that omitted women with CVD at baseline (9427 in the estrogen vs placebo intervention; 15,407 in the estrogen plus progestin vs placebo intervention). In both analyses, follow-up was terminated when the trial interventions ceased, giving an average of 7.1 years observation in the estrogen versus placebo comparison and 5.6 years in the estrogen plus progestin versus placebo comparison. NSCP was as defined as in the WHIOS study [12].

The analysis of the estrogen versus placebo randomized comparison estimated the hazard ratio for NSCP to be 1.04 (95% CI: 0.81–1.32), after adjustment for CVD risk factors and HT use at baseline (the same adjustments were made as are listed in the footnote to Table 1 ). From the parallel RCT, the equivalent hazard ratio, comparing estrogen plus progestin to placebo, was 0.78 (95% CI: 0.59–1.02). This suggests that estrogen alone has no effect on the incidence of NSCP, but when added to progestin it reduces the risk of NSCP by an almost significant 22%, so that further data are needed to test the hypothesis that the combined HT may protect against NSCP. However, the authors of this paper caution that their finding for the compound HT may be an artifact of their method of analysis, which gave a shorter follow-up to women with CAD events resulting from estrogen plus progestin. However, this conjecture is not tested. More importantly, treatment with the combined HT cannot be recommended to avoid NSCP since a previous publication from the WHI [9] and a paper from the Heart and Estrogen/Progestin Replacement Study research group [14] has demonstrated that it increases the risk of CAD.

To address the other aim of this study, the authors used the same methods as in the WHIOS [12] to evaluate the association between NSCP and subsequent CVD events. In this, they chose to combine data from the two parallel HT intervention trials. Thus, they used the combined RCT as if it were a cohort study – that is, making a nonrandomized comparison. The hazard ratios reported are shown in Table 1 (note that, in this case, the adjustment includes the RCT interventions). These demonstrate strong evidence of an effect of NSCP on angina, coronary surgery or intervention, non-fatal myocardial infarction and heart failure. There was no strong evidence of an effect of NSCP status on stroke. Two drawbacks to the methodology are that the Cox models used were not stratified to allow for the differing background risks in the two parallel RCTs, consequent to the 18-month difference in trial durations and, as in the WHIOS [12], no combined CAD end point is analyzed. Nevertheless, the paper clearly demonstrates that NSCP is significantly associated with CAD events in this study. Broadly, the results do, indeed, confirm those from the WHIOS, as given in Table 1 . Given the similarities in design, conduct and results of the two studies, it appears sensible to pool the results to produce overall estimates of the effects of NSCP. This new analysis has been added to Table 1 , using inverse variance-weighted meta-analysis [11].

A further analysis in the paper demonstrates the effects of a range of different factors, including sociodemographic and CVD risk factors, on the risk of all CVD events (taken together) amongst those women who had NSCP [1]. This shows that women with NSCP had significantly (p < 0.05, not corrected for multiple testing) higher risk if they also had either diabetes or hypertension. However, the power for the tests made will be low since only a maximum of 571 women were included in these analyses. Therefore, many of the other factors that the authors considered may still have a role to play in the risk of CVD subsequent to NSCP, as, indeed, was concluded in the WHIOS [12].

Unfortunately, the study was also under-powered to investigate another relevant issue in the epidemiology of NSCP in women: whether or not HT and NSCP status interact in their effects on CAD events. The paper demonstrates that NSCP is still a risk factor after controlling for the effects of HT, but not whether HT modifies the detrimental effects of NSCP.

Conclusion & future perspective

The new study [1] fits in logically with the former work of the same research group, who are responsible for the impressive WHI. It adds to this former work to demonstrate that NSCP is likely to be an independent risk factor for CAD events amongst middle-aged and older women. Further studies are needed to confirm this in other settings and in younger women. Much larger data sets are required to investigate the role of HT in both NSCP as an outcome and as a risk factor for CAD (and, potentially, for stroke). The WHI will have a role to play both in providing further follow-up data from their studies and contributing data to meta-analyses.

The current paper makes it abundantly clear that NSCP, up to now, has been an inadequate diagnosis. Future improvements in diagnostic testing, for example, MRI, will surely lead to better identification of coronary risk in women with chest pain. A crucial goal will be to ensure that such new technology is both accurate and cost effective. In the meantime, the authors rightly conclude that women who are given a diagnosis of NSCP deserve aggressive CVD risk-factor management. Discharge from hospital with NSCP cannot be considered as a clean bill of health.

Executive summary

• Chest pain without a known cause is a common medical diagnosis.

• Traditional ways of ruling out cardiac causes of chest pain may be biased against women.

• Nonspecific chest pain in women appears to be associated with more than doubling of the risk of coronary artery disease.

• The role of hormone therapy in the diagnosis of, and subsequent coronary risks from, nonspecific chest pain remains to be determined.

• Women diagnosed with nonspecific chest pain should be considered for aggressive cardiovascular risk management.