# Limited performance questions retrospective use of quantitative flow ratio in coronary artery bypass grafting

## TL;DR

This study found that using quantitative flow ratio (QFR) to assess coronary arteries before bypass surgery had limited success in predicting graft patency when applied retrospectively.

## Contribution

The study evaluates the feasibility and diagnostic performance of retrospective QFR analysis in predicting arterial graft patency after coronary artery bypass grafting.

## Key findings

- QFR ≤ 0.80 was independently associated with higher arterial graft patency rates.
- Retrospective QFR analysis had poor feasibility and diagnostic performance.
- Visually estimated stenosis from preoperative angiography was not a significant predictor of graft patency.

## Abstract

Hemodynamic assessment of coronary artery stenosis has impact on arterial graft patency. Quantitative flow ratio (QFR) obtains hemodynamic information of coronary artery stenosis.

Patients with history of isolated coronary artery bypass grafting (with ≥1 arterial graft) and at least one postoperative coronary re-assessment were retrospectively investigated. The preoperative angiography was used for retrospective QFR analysis of the native coronary target vessel, to which the arterial bypass graft was anastomosed. Analysis was performed by certified investigators, who were blinded towards postoperative arterial graft patency status. Coronary targets with QFR values of ≤0.80 were defined as hemodynamically relevant, whereas values of >0.80 were defined as hemodynamically irrelevant.

Out of 5,692 patients, 596 patients had a postoperative coronary assessment and were therefore eligible for inclusion. In 196 arterial target vessels QFR analysis was possible. Kaplan–Meier analysis revealed higher graft patency rates for arterial grafts anastomosed to coronary branches with QFR values ≤0.80 (log-rank: p = 0.017). In multivariable Cox regression analysis, QFR ≤ 0.80 remained an independent predictor for arterial graft patency (HR: 0.475, 95% CI: 0.261–0.867; p = 0.015), while visually estimated stenosis from preoperative coronary angiography did not (p = 0.160). With an area under the curve of 0.595, 95% CI (0.503–0.688), the performance of the model was poor to at most moderate. Most target vessels [546 (80.22%)] were not analysable in retrospective fashion.

Though target vessel QFR ≤ 0.80 was associated with higher arterial graft patency, our trial observed low feasibility (high drop out rates) and poor diagnostic performance of QFR used in retrospective fashion. Caution is warranted for retrospective use of QFR in datasets with similar constraints.