What counts as proof of ischemia: the stressor decides what a negative result is worth

The American Heart Association's 2026 scientific statement on coronary ischemic testing in children describes what follows inadequate coronary flow as an ordered cascade: metabolic changes first, then wall-motion abnormalities, then ECG changes, and only at the end angina and infarction. Different tests capture different links in that chain, which is why their sensitivity, specificity, and appropriateness differ.

The statement also gives the evidence for that order: on dobutamine stress cardiovascular magnetic resonance, myocardial perfusion abnormalities are more common than wall-motion abnormalities both in adults with coronary stenosis and in children with AAOCA. Perfusion imaging is looking at an earlier layer.

AAOCA adds a further wrinkle. The statement notes that AAOCA can create ischemia by blending fixed and dynamic obstruction. The fixed component shows up at rest and under pharmacologic dilation; the dynamic component appears only as aortic pressure, heart rate, and contractility rise. A protocol that covers only the first covers only the first in its conclusions.

It also explains why anatomic studies cannot stand in for functional ones. The CHSS multicentre study of 560 patients (aged 30 or younger at diagnosis) found an intramural course, intramural length, and orifice anomalies associated with evidence of ischemia; yet among those with such evidence, no anatomic feature was shown to be associated with sudden cardiac arrest or death. Anatomy flags risk; it does not measure flow.

The AHA 2026 statement calls CMR the modality of choice for assessing myocardial perfusion in children, on combined diagnostic and safety grounds: no ionising radiation, and perfusion, wall motion, function, and scar in one study. It also names two exceptions — when CMR compatibility or safety is a problem, and when conventional treadmill stress is desirable, with AAOCA given as the example — where nuclear testing may be preferred.

For AAOCA the statement is more specific: stress CMR is a valuable adjunct where CPET does not provide definitive findings congruent with ischemia, precisely because of the test's low sensitivity in this condition. That is the same fact as the 19% and 58% figures above, stated the other way round.

On feasibility, one series of 182 patients under 20 with AAOCA completed 224 dobutamine stress CMR studies, of which 221 were completed and diagnostic, with inducible hypoperfusion in 31 (about 14%). That shows the study can be run routinely at an experienced pediatric coronary centre — and that it needs one: a separate series of 64 patients under 23 with coronary disease undergoing 80 dobutamine stress CMR studies was published specifically on coronary events following the test, which is not a question anyone tracks idly.

Age imposes a further limit. The statement notes that young children often need sedation, and that provocative perfusion imaging under sedation has not been extensively studied; under general anaesthesia, dobutamine stress echocardiography starts from a lowered blood pressure and heart rate, making the 85%-of-predicted target harder to reach. Whether a study can be done and whether its result can be trusted are two separate questions in the younger age range.

A prospective single-centre study enrolled 73 adults with newly diagnosed R-AAOCA and an interarterial/intramural course (mean age 51 ± 13 years), each undergoing dobutamine FFR, adenosine FFR, and resting IVUS. Taking dobutamine FFR ≤0.80 as the reference for haemodynamic relevance, 17 (23%) were abnormal.

The key finding is the disagreement between stressors. Adenosine FFR ≤0.80 occurred in only 5 patients (7%), with 100% specificity and positive predictive value but 29% sensitivity and 82% negative predictive value. A positive adenosine FFR is close to diagnostic; a negative one missed roughly seven in ten of the cases that declared themselves under dobutamine. Resting IVUS behaves in the opposite way: a minimal lumen area ≤5.5 mm² had 100% sensitivity and negative predictive value with 68% specificity and 49% positive predictive value, which makes it a tool for excluding rather than confirming.

A separate dataset of 81 adults with R-AAOCA (mean age 52.3 years) points the same way: adenosine FFR was abnormal in about 6.2% and dobutamine FFR in about 19.8%. The same vessels, and a threefold difference in positive rate depending on the stressor.

How much invasive testing changes decisions has also been quantified. The multicentre prospective MuSCAT study enrolled 76 patients aged 16 or older with an interarterial or intraseptal AAOCA (median age 53). Non-invasive functional testing was positive in 10 (13%) and negative in 59 (78%); invasive functional testing disagreed with the non-invasive result in 21 (28%) and changed the treatment recommendation for 15 patients, 20% of the whole cohort.

All three datasets come from middle-aged adults, and thresholds such as FFR ≤0.80 or an IVUS area of 5.5 mm² do not transfer to children, whose myocardial mass, vessel calibre, and reference flow all differ. What they establish is mechanism: in assessing AAORCA, the stressor has to reproduce exercise physiology, or the exclusionary power of a negative result will be systematically overestimated.

A 58-child AAOCA cohort from Wuhan used a triaged nuclear-perfusion strategy: SPECT-MPI was ordered only for children with ischemia-related symptoms, an abnormal ECG, raised injury biomarkers, or high-risk CTA anatomy. Twenty-seven were imaged, 6 with normal perfusion and 21 with perfusion defects; 7 eventually had surgery, all of them with chest pain, CTA-confirmed ostial stenosis and an intramural course, and a positive perfusion defect.

The methods state that the tracer was 99mTc-MIBI, weight-standardised, and given at rest; no treadmill or pharmacologic stress protocol is described. In the discussion the authors name this as a principal limitation of the study: stress perfusion imaging was absent because caregivers declined the additional testing over concerns about radiation exposure and the risks of pharmacologic stress.

The cohort therefore supports one conclusion — nuclear perfusion imaging does take part in real-world stratification, and positive findings tracked with high-risk CTA anatomy — and not another: that normal resting perfusion means no exercise-induced ischemia. The difference between those two statements is exactly the stress-versus-rest comparison.

That yields a concrete question. When a hospital proposes myocardial perfusion imaging, it is worth establishing first whether it is a resting study or a stress–rest comparison, and if it involves stress, which stressor. The point is not to doubt the test but to know how much weight its result can carry.

The prospective pediatric R-AAOCA cohort at Texas Children's comprised 220 patients under 21, median age 11.4 years; 168 (76%) had no exertional symptoms and 52 (24%) had exertional chest pain or syncope. Stress perfusion imaging was positive in 11 of 120 (9%) without symptoms and 9 of 49 (18%) with them, while exercise testing alone was positive in only 2 of 164 (1.2%) of the asymptomatic group.

Read together, the two figures say two things. First, symptoms are not a reliable filter: about 9% of the asymptomatic group still had inducible ischemia. Second, the modality sets the yield: in the same asymptomatic children, stress perfusion imaging was positive more than seven times as often as exercise testing alone.

The other half of the cohort is its restraint. In the end 56 of 220 (26%) were recommended surgery for high-risk features and 52 were operated, with all alive and returned to exercise at a median follow-up of 4.6 years. Children with demonstrable ischemia are a minority — and that minority is exactly whom this testing strategy exists to find.

Adult data run the same way: a two-centre stress echocardiography study in 46 middle-aged AAOCA patients and a 35-patient study pairing CCTA with exercise SPECT both place functional results alongside anatomy. None of these series is large, and what they share is not a number but a route: anatomy first, function next, and a further step only where the two disagree.