A previously reported method¹ for non-invasive quantitation of myocardial blood flow (MBF), employing FAIR labelling and a Look-Locker readout, was assessed for reproducibility and variability. These measures are important to consider when planning a clinical study as they indicate the change required for measurement above systematic error, and describe the variation due to factors introduced between and within sessions. It is important in measurements of MBF to observe changes in pathology and under stress.

The LL-FAIR-ASL sequence (Figure 1) employed slice-selective (SS) and globally-selective (GS) HS8 inversion pulses, each followed by 5 bSSFP readouts in mid-diastole on successive heartbeats. There was a gap of 3 heartbeats between the two blocks to allow for some relaxation and the order of the inversion pulses was varied. Six mid-ventricular scans were acquired (3 SS-GS, 3 GS-SS). Sequence parameters are shown in Table 1. Fifteen seconds were allowed between scans to allow for full relaxation. T₁ values were calculated for both the SS and GS-IR experiments by a three-parameter fit (Figure 2). The ratio of these T₁s was related to MBF by the Belle model².

Data was collected in 8 healthy volunteers (7 male, 31±7y) on a 3T scanner (Trio, Siemens) with ethics approval. Volunteers were scanned twice, on separate days, to assess between-session reproducibility, and during 11 scans the protocol was repeated to assess within-session reproducibility. These data were used to produce Bland-Altman plots³ to describe the two measures. The myocardium of the entire left ventricle was treated as a single ROI. The variability of the MBF estimates was assessed by the coefficient of variation (SD/mean) for the whole sample (CV_all), and each subject between-session (CV_BS) and within-session (CV_WS).

The MBF was 1.15±0.47 ml/g/min which compares well with literature values, where resting MBF has been reported as 0.97±0.64 ml/g/min with single-TI ASL⁴, 1.33±0.32 ml/g/min with PET⁵ and 1.02 ± 0.22 ml/g/min using first-pass CMR⁶. The values of MBF for each subject are presented in Figure 3. CV_all was 40%. The between-session and within-session Bland-Altman plots are shown in Figure 4(a) and (b) respectively and show the mean-difference in each case and the value of ±1.96 times SD which represent the upper and lower 95% confidence bounds. When normalised to the mean MBF, these equate to a coefficient of repeatability (CR) of 38% between-session and 39% within-session. The mean CV_BS and CV_WS were calculated as 10% and 9% respectively.

The LL-FAIR-ASL method is attractive as a non-invasive alternative to SPECT, PET and first-pass CMR. While reproducibility and variability of similar techniques have been investigated in mice^7,8, to our knowledge this has not been carried out in human myocardium. Resting MBF values for healthy volunteers have previously been shown to be heterogeneous in studies utilising PET⁵ and first-pass CMR⁹. The high observed CV_all, 40%, shows that our results reflect this heterogeneous nature. The Bland-Altman plots show the mean difference in both the between-session and within-session cases to be close to zero and all bar one of the data points to lie within the ±1.96 SD bounds. The CR_BS of 38% shows the level of reproducibility expected across repeat scans. It indicates the change in MBF required to show a difference over time and is useful to consider if planning longitudinal studies in patient groups. The CR_WS of 39% gives a useful indication of the detectable change in MBF, which is significantly less than the 3-4-fold increase expected under vasodilator stress^4,6.

The variability compares favourably to similar measures reported for preclinical cardiac ASL⁷ and brain ASL in humans¹⁰. The mean values of CV_BS and CV_WS were 10% and 9%, below the 40% observed for all subjects. This shows that the variation in results exhibited by an individual is much less than across the population as a whole. The CV_BS shows the variation due to the method, plus effects such as repositioning, re-localisation, scanner adjustments etc. The CV_WS primarily reflects the methodological effects.

The use of a single ROI to represent the whole myocardium is a limitation of this study. MBF is known to exhibit spatial heterogeneity^5,11. Further investigation of the sensitivity of the method to this is required.

The results presented suggest that a future clinical study applying the LL-FAIR-ASL method at rest and stress will have the sensitivity required to detect the expected change in MBF. The variability of the method was shown to compare favourably with published values in similar techniques. These results should prove useful in the planning of future clinical studies using this method.