Predictability of baroreflex sensitivity induced by phenylephrine injection via frequency domain indices computed from heart rate and systolic blood pressure signals during deep breathing

Diminished baroreflex sensitivity (BRS) is related to increased risk of sudden cardiac death in myocardial infarction patients and can be used as an indicator for risk level. The BRS is traditionally estimated invasively using vasoactive drugs, such as phenylephrine injection. This method has been widely accepted as a standard in clinical research. Due to its clinical importance, alternative BRS assessment methods have been investigated over the years to eliminate the use of drugs. In this study, the BRS obtained by drug-based (pharmacological) assessment was predicted from a subset of non-pharmacological indices computed from heart rate and systolic pressure signals. In the first phase of a two-phase experimental paradigm, 16 subjects were asked to perform two deep breathings with a 2-min delay in between. In the second phase, the BRS was measured by phenylephrine injection. Indices computed from the first phase describing the spectral and time domain properties of the heart rate and systolic pressure signals were used as predictors to estimate the pharmacological BRS of each subject. In addition to individual spectrum of beat-to-beat interval and systolic blood pressure, indices from cross-spectrum were also computed and evaluated as predictors. A leave one out method was employed to estimate the generalization capacity of the system and explore subset of indices, which gives the highest correlation between pharmacological and predicted BRS. Two predictors provided the highest correlation (r=0.87, p=1.16 x 10(-5)) with pharmacological BRS. The algorithm selected consistently normalized cross-power about the Mayer frequency and average magnitude square coherence in the high frequency band as predictors. These results indicate that the pharmacological BRS can be estimated from the combination of non-pharmacological spectral indices computed from beat-to-beat interval and systolic blood pressure signals obtained during deep breathing and therefore may eliminate the use of drugs. (C) 2010 Elsevier Ltd. All rights reserved.