Blood correction reduces variability and gender differences in native myocardial T1 values at 1.5 T cardiovascular magnetic resonance - a derivation/validation approach

Abstract

Background: Myocardial native T1 measurements are likely influenced by intramyocardial blood. Since blood T1 is both variable and longer compared to myocardial T1, this will degrade the precision of myocardial T1 measurements. Precision could be improved by correction, but the amount of correction and the optimal blood T1 variables to correct with are unknown. We hypothesized that an appropriate correction would reduce the standard deviation (SD) of native myocardial T1. Methods: Consecutive patients (n = 400) referred for CMR with known or suspected heart disease were split into a derivation cohort for model construction (n = 200, age 51 ± 18 years, 50% male) and a validation cohort for assessing model performance (n = 200, age 48 ± 17 years, 50% male). Exclusion criteria included focal septal abnormalities. A Modified Look-Locker inversion recovery sequence (MOLLI, 1.5 T Siemens Aera) was used to acquire T1 and T1* maps. T1 and T1* maps were used to measure native myocardial T1, and blood T1 and T1*. A multivariate linear regression correction model was implemented using blood measurement of R1 (1/T1), R1* (1/T1*) or hematocrit. The correction model from the derivation cohort was applied to the validation cohort, and assessed for reduction in variability with the F-test. Results: Blood [LV + RV] mean R1, mean R1* and hematocrit correlated with myocardial T1 (Pearson’s r, range 0.37 to 0.45, p < 0.05 for all) in both the derivation and validation cohorts respectively, suggesting that myocardial T1 measurements are influenced by intramyocardial blood. Mean myocardial native T1 did not differ between the derivation and validation cohorts (1030 ± 42.6 ms and 1023 ± 45.2 ms respectively, p = 0.07). In the derivation cohort, correction using blood mean R1 and mean R1* yielded a decrease in myocardial T1 SD (45.2 ms to 36.6 ms, p = 0.03). When the model from the derivation cohort was applied to the validation cohort, the SD reduction was maintained (39.3 ms, p = 0.049). This 13% reduction in measurement variability leads to a 23% reduction in sample size to detect a 50 ms difference in native myocardial T1. Conclusions: Correcting native myocardial T1 for R1 and R1* of blood improves the precision of myocardial T1 measurement by ~13%, and could consequently improve disease detection and reduce sample size needs for clinical research.