CT-Derived FFR Predicts Heart Attack and Death Risk Better Than Anatomy Alone
A 7,836-patient nationwide study shows FFR-CT independently forecasts MI, cardiovascular death, and all-cause mortality over 3 years.
Summary
Doctors increasingly use a heart CT scan plus computer modeling (FFR-CT) to estimate whether a narrowed artery is limiting blood flow. This study asked whether that score also predicts future heart attacks and deaths. Researchers followed 7,836 English patients with suspected stable coronary disease for about three years. Lower scores meant higher risk. Scores of 0.70 or below were linked to roughly three times the heart attack risk, even after accounting for age, sex, blood pressure, diabetes, cholesterol and how narrowed the arteries looked. A score of 0.67 was the best risk threshold. Patients with normal scores had only about a 1.5% three-year risk of heart attack or cardiovascular death. The study was observational, so treatment decisions made after the scan may have influenced the results.
Detailed Summary
Coronary CT angiography is now a first-line test for suspected stable coronary artery disease, and FFR-CT software adds a computed estimate of whether a narrowing actually limits blood flow. Its diagnostic accuracy is well established. Whether it predicts future heart attacks and deaths beyond what age, risk factors and the degree of narrowing already reveal has been unclear, because earlier studies were small, used composite endpoints, or followed patients briefly.
Researchers drew on the FISH&CHIPS cohort of 90,553 patients who had CT angiography at 27 English hospitals from 2017 to 2020. Of 8,532 cases sent for HeartFlow FFR-CT analysis, 7,836 completed it (median age 63, 37.4% women, 16.9% diabetes, 52% hypertension). National hospital and death records gave outcomes over a median 3.1 years: 191 heart attacks (2.4%), 74 cardiovascular deaths (0.9%), 261 deaths from any cause (3.3%) and 1,573 revascularizations (20.1%). Patients were grouped by their lowest FFR-CT value: normal (>0.80), borderline (0.71-0.80), reduced (0.51-0.70) or severely reduced (≤0.50).
Lower stenosis-specific FFR-CT meant steadily higher risk. Unadjusted hazard ratios across abnormal groups ranged from 2.08 to 5.73 for heart attack (P<0.001), 1.51 to 3.40 for cardiovascular death (P<0.001), 1.01 to 1.58 for all-cause death (P=0.01) and 4.79 to 24.48 for revascularization. After adjusting for age, sex, hypertension, diabetes, dyslipidemia and CAD-RADS stenosis severity, values ≤0.70 still carried roughly triple the heart attack risk (adjusted HR 2.94-3.39). Values ≤0.50 were linked to cardiovascular death (adjusted HR 2.92, P=0.02) and all-cause death (adjusted HR 1.49, P=0.045).
FFR-CT added measurable predictive information. For nonfatal heart attack or cardiovascular death, the C index rose from 0.758 (risk factors alone) to 0.771 with CAD-RADS, then to 0.783 with FFR-CT (P<0.001). Adding the pressure drop across a lesion (delta FFR-CT) did not help (0.782, P=0.72). Spline analysis identified 0.67 as the optimal cutoff, well below the conventional 0.80 diagnostic threshold and matching a previously reported invasive FFR value. Borderline readings of 0.71-0.80 were only weakly tied to heart attack and not to death.
Distal-vessel FFR-CT performed similarly overall. Stenosis-specific values fit heart attack and revascularization better, while distal values fit all-cause death better. Among patients with a normal stenosis-specific value, 36.4% had an abnormal distal value, nearly all borderline. Heart attack or cardiovascular death occurred in 1.5% of the concordant normal group and 1.3% of the borderline distal group, but in 5.7% when distal values were reduced (P=0.004). The 159 patients with distal-vessel stenoses had the highest mortality, a hypothesis-generating finding.
Practically, FFR-CT looks best interpreted as a continuum of risk rather than a yes/no result. Values ≤0.70 may flag patients for more intensive preventive therapy, while normal results are reassuring. However, this was a retrospective observational study, and FFR-CT results likely steered treatment, particularly revascularization. Heart attacks could not be separated into periprocedural and spontaneous events, and plaque burden, smoking and family history data were missing.
Key Findings
- Cohort of 7,836 patients (median age 63, 37.4% women) at 27 English sites; median follow-up 3.1 years with 191 MIs, 74 CV deaths, 261 all-cause deaths and 1,573 revascularizations
- Stenosis-specific FFR-CT ≤0.70 was independently associated with MI after adjustment for risk factors and CAD-RADS (adjusted HR 2.94-3.39, P<0.001)
- FFR-CT ≤0.50 was independently associated with cardiovascular death (adjusted HR 2.92, P=0.02) and all-cause death (adjusted HR 1.49, P=0.045)
- Adding FFR-CT to risk factors plus CAD-RADS improved discrimination for MI or CV death (C index 0.771 to 0.783, P<0.001); delta FFR-CT added nothing (P=0.72)
- Optimal stenosis-specific risk cutoff was 0.67, lower than the standard 0.80 diagnostic threshold; borderline values (0.71-0.80) were only weakly linked to MI
- With a normal stenosis-specific value, 3-year MI or CV death was about 1.5%; 36.4% had discordant distal values, which signaled higher risk only when reduced (5.7% vs 1.5%, P=0.004)
- 159 patients with distal-vessel stenoses had the highest cardiovascular and all-cause mortality among distal subgroups (hypothesis-generating)
Methodology
This was a retrospective, multicenter, quasi-experimental observational cohort drawn from the FISH&CHIPS study of 90,553 CCTA patients. The 7,836 patients with completed HeartFlow FFR-CT were stratified by their lowest stenosis-specific and distal-vessel values. Outcomes (MI, cardiovascular death, all-cause death, revascularization) came from NHS hospital episode statistics and Office for National Statistics death records. Analyses used Kaplan-Meier curves, Cox and Fine-Gray models adjusted for age, sex, hypertension, diabetes, dyslipidemia and CAD-RADS, Harrell's C index, AIC and Brier scores, and 3-knot restricted cubic splines.
Study Limitations
The observational design means FFR-CT results probably influenced downstream treatment, particularly revascularization, and MI could not be separated into periprocedural and spontaneous events. Data on plaque burden, smoking status and family history were unavailable, so incremental value over total plaque volume is unknown. Follow-up was about 3 years with relatively few cardiovascular deaths, and conflict-of-interest disclosures were not included in the supplied text, though the analysis used HeartFlow software.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: