female/male differences in noninvasive cardiac testing

Subgroup analysis of the PROMISE trial found a significant difference in prognostic information in women vs men by the type of cardiac test done (see cad testing men vs women jacc2016​ indropbox, or DOI: 10.1016/j.jacc.2016.03.523).  for my review of the PROMISE trial overall, see http://gmodestmedblogs.blogspot.com/2015/03/coronary-angiography-or-exercise_17.html  , which includes my concerns about radiation exposure as well as what what defines the gold standard to evaluate sensitivity/specificity. details:

--in the original PROMISE trial, 10,003 outpatients with stable symptoms suggestive of CAD were randomized to a functional test (a stress test, as chosen by the clinician: exercise ECG, stress echo, or stress nuclear) vs anatomic test (computed tomographic angiography - CTA), followed 25 months, and investigators found no difference in outcomes. the current analysis looked at the prognostic capabilities of the different non-invasive tests by sex.

--8966 patients (53% women, mean age 60, 22% ethnic minority, 65% hypertension, 21% diabetes, 37% metabolic syndrome, 68% dyslipidemia, 50% current/former smoker, BMI 30, Framingham risk score 15% in women/29% in men, 75% presented with chest pain though 78% of them were felt to be "atypical") who received a noninvasive test with interpretable results

--CTA was performed in 4500 (52% of the women) and stress testing in 4466 (53% of the women)

--a positive CTA was if there were  ≥70% stenosis in at least one epicardial artery or ≥50% stenosis in the left main

--results:

    --456 women had a positive result (9.7%) but a significantly smaller proportion of positive CTAs vs stress tests: adjusted OR of 0.67 (0.55-0.82, p<0.001), and the CTA had many fewer positive results vs the exercise ECG (OR 0.39), less so for nuclear stress test (OR=0.66), and was not significant for stress echo (OR=0.90)

    --640 men (15.1%) had a positive test, with marginally greater proportion of CTAs being positive (16% vs 14%, p=0.047). men were more likely to have a positive CTA vs stress test with adjusted OR 1.23 (1.04-1.47, p=0.019); CTA had more positive results than an exercise ECG (OR=1.79), or stress echo (OR=2.10) but was not different from a nuclear stress test (OR=1.03)

    --in terms of clinical outcomes over 25 months (primary endpoint being: ​a composite of all-cause death/MI/hospitalization for unstable angina)

        ​--overall a positive noninvasive did strongly predict a primary endpoint

        ​--112 women (2.4%) had primary endpoint: a positive CTA was more than 2x as predictive as a positive stress test (adjusted HR of 5.86 vs 2.27)

        ​--153 men (3.6%) ​had primary endpoint: a positive CTA was nonsignificantly weaker than a positive stress test (adjusted HR of 2.80 vs 4.42).

so, a few points:

--it has been known for a long time that women have more false positive stress tests. this has been attributed to smaller coronary vessel size, higher prevalence in microvascular coronary dysfunction in women, baseline differences in resting ECG, breast attenuation for some of the tests.

--in terms of risk stratification, CTA yields better prognostication over stress tests (ie women with a positive CTA tended to do worse than if they had a positive stress test); men get pretty much the same by either modality (trend to doing better with a positive stress test). 

        --but the difference in women could be because women do have more microvascular heart disease which is missed on a CTA, and microvascular coronary dysfunction has a better prognosis and may not lead to many events in the short 25 month followup of this study; ie, CTA may be prognostically better than functional tests, at least in the short-term, since it picks up more imminent clinical events. but if the treatment is the same for any positive test (aggressive risk factor reduction), one might imagine that a normal CTA could have the very negative impact of leading to less aggressive treatment by the clinician/less follow-through by women ("after all, it was a negative test, so i must be fine"). ie, the higher positivity rate of the functional stress tests ("false positives") could well lead to better longterm outcomes [it is not really clear what the gold standard is. is it really a false positive if a woman has very real microvascular heart disease but it takes longer to manifest itself as clinical cardiac events?]

--there are, of course, several concerns in interpreting the above. this is a secondary analysis of a large pragmatic trial, so there can always be unexpected confounders. also, as a pragmatic trial, there was no systematic randomization to different types of stress tests, and no control over therapies after a positive test (eg, did those women with a positive CTA get different treatments, pharmacologic or invasive, which might explain the different prognostic value/future events?).  (i should also note there was no cardiac cath confirmation of CAD, though i would argue that this is not an uncontested gold standard, as noted in  http://blogs.bmj.com/ebm/2015/03/22/primary-care-corner-with-geoffrey-modest-md-coronary-angiography-or-exercise-testing-for-chronic-angina/ ).​

this study reinforces that there are significant differences in heart disease between men and women, as manifested by the differences in imaging sensitivities as above.  i do not think this trial shows conclusively which test should be done and, importantly, there was no information about how those with positive tests were treated. but i bring this up since it does reinforce that there are significant differences in the pathophysiology and testing for men and women. i will add the following from a more EBM focus (though 3 years old now):

http://www.ncbi.nlm.nih.gov/books/NBK153207/pdf/Bookshelf_NBK153207.pdf is the AHRQ evaluation of noninvasive technologies for the diagnosis of CAD in women, released feb 2013, which notes that:

--exercise EKG: 41 studies, high strength of evidence, sensitivity 62%, specificity 68%

--exercise/stress echo: 22 studies, high strength of evidence, sensitivity 79%, specificity 83%

--exercise/stress radionucleide: 30 studies, high strength of evidence, sensitivity 81%, specificity 78%

--CTA: 8 studies, low strength of evidence, sensitivity 93%, specificity 77%

--and they noted overall that in women, stress ECG and CTA were both less sensitive and less specific than in men, with stress ECG being statistically significantly less specific than the other noninvasive modalities

--their bottom line:

    --in women with no known CAD, the specificity of stress EKG was less than stress echo

    ​--though there is higher radiation exposure levels in women over men with CTA​ (3 of 4 studies), there is not enough info in the literature about relative radiation exposures, or other safety concerns

--so, as per usual, this detailed analysis raised more questions than it answered, including: the risks of noninvasive testing (including radiation), the comparative accuracy in real-world setting, the best sequential order of these tests (including by differing pre-test probabilities), or if there are other population differences besides male/female (race/ethnicity, for example), etc

so, my bottom line at this point is pretty much unchanged: i usually get a stress echo in women, especially since there are so many "false positives" with regular exercise testing, it is a functional vs anatomic test (which, it seems to me, is more likely to reflect the real effect of the cardiac disease on the individual patient), and since there is no radiation exposure. (echo also gives collateral information about cardiac functioning, LVH, valves, etc, which might be useful clinically). and the above large but perhaps only marginally useful PROMISE study does seem to reinforce that stress echos seem closest to their strongest candidate of CTA (though, unfortunately, their clinical outcome data did not allow statistically for breakdown of outcomes by the different types of stress testing by modality)