Parallel conductance determination in cardiac volumetry using dilution manoeuvres: theoretical analysis and practical implications

Abstract

Left ventricular volume calibration based on the conductance catheter depends on the correct determination of the parallel conductance (G _p ). Baan's saline manoeuvre procedure leads to G _p by finding the end-systolic (G _es ) and end-diastolic (G _ed ) conductances, for each beat of the dilution curve rising limb. After plotting such values in an xy-system, their linear regression is back-projected to intersect the identity line, so yielding an estimated G_p. The objective is to theoretically analyse all possible lines, G _es =aG _ed +b (Baan's line) and, based on experimental results, to establish their limitations. This was attained by calculating the regression lines using, first G _ed =f₁(G _es ) and thereafter, G_es=f₂(G _ed ), which led to two values, G _p2 and G _p1 , for the parallel conductance. The morphology of the saline curve was also modified to assess its effect on the extrapolation. Multiple dilutions were recorded in eight experimental dogs injecting different concentrations. Each curve was classified according to the maximum change (VAR) reached by the total average conductance. Over 138 manoeuvres, 276 regressions were processed yielding correlations higher than 0.65. Of this total, 92.4% gave positive parallel conductances. The rest produced negative values and, thus, were neglected. If the two (G _ed , G _es ) statistical relationships were ideal, they should yield G _p =G _p1 =G _p2 ; however, there were differences which, when G _p1 was studied against G _p2 , led to: G _p1 =0.97 G _p2 +0.055, with r=0.9476, and n=85. The remaining 53 were discarded because either some G_p values were negative, or the correlation of G _es which G _ed (or vice versa) was <0.85, and/or VAR<15%; the two latter conditions were found necessary for reliable calibration. Baan's line high correlation is not a unique condition to ensure the accuracy and precision of G_p determination because the slope a depends on VAR and, thus, different intersections with the identity line may be obtained. Its recommended that manoeuvres be used with at least eight data points, with VAR>15% and, finally, with (G _es , G _ed ) correlation better than 0.85. Theoretical analysis of Baan's line offers a reference frame, which contains only a limited number of practical possibilities.