Most noninvasive measures of diastolic function are made during left ventri
cular (LV) filling and are therefore subject to "pseudonormalization," beca
use variation in left atrial (LA) pressure may confound the estimation of r
elaxation rate. Counterclockwise twist of the LV develops during ejection,
but untwisting occurs rapidly during isovolumic relaxation, before mitral o
pening. We hypothesized that the rate of untwisting might reflect the proce
ss of relaxation independent of LA pressure. Recoil rate (RR), the velocity
of LV untwisting, was measured by tagged magnetic resonance imaging and re
gressed against the relaxation time constant (tau), recorded by catheteriza
tion, in 10 dogs at baseline and after dobutamine, saline, esmolol, and met
hoxamine treatment. RR correlated closely (average r = -0.86) with tau and
was unaffected by elevated LA pressure. Multiple regression showed that tau
, but not LA or aortic pressure, was an independent predictor of RR (P < 0.
0001, P = 0.99, and P = 0.18, respectively). The rate of recoil of torsion,
determined wholly noninvasively, provides an isovolumic phase, preload-ind
ependent assessment of LV relaxation. Use of this novel parameter should al
low the detailed study of diastolic function in states known to affect fill
ing rates, such as aging, hypertension, and congestive heart failure.