A nonlinear HRV index used as an indirect measure of left ventricular function

D.S. ANDRES, I.M. IRURZUN, M.M. DEFEO, E.E. MOLA, J. MITELMAN, L. GIMENEZ, C. DE LUCA, M. MASCIA, P. VELAZCO

Buenos Aires

Congreso; World Congress of Cardiology 2008; 2008

World Heart Federation

A nonlinear HRV index used as an indirect measure of left ventricular function D. S. Andres1, I. M. Irurzun1, M. M. Defeo2, E. E. Mola1, J. Mitelman34, L. Gimenez34, C De Luca3, M Marcia2, P Velazco2.1, I. M. Irurzun1, M. M. Defeo2, E. E. Mola1, J. Mitelman34, L. Gimenez34, C De Luca3, M Marcia2, P Velazco2. 1Facultad de Ciencias Exactas. Universidad Nacional de La Plata (INIFTA, CONICET). Argentina.Facultad de Ciencias Exactas. Universidad Nacional de La Plata (INIFTA, CONICET). Argentina. 2Hospital Interzonal General de Agudos ´´Prof. R. Rossi´´. La Plata. Argentina.Hospital Interzonal General de Agudos ´´Prof. R. Rossi´´. La Plata. Argentina. 3Hospital General de Agudos "T. Alvarez". Buenos Aires, Argentina.Hospital General de Agudos "T. Alvarez". Buenos Aires, Argentina. 4Instituto Universitario de Ciencias de la Salud, Facultad de Medicina, Fundación Barceló, Buenos Aires, Argentina.Instituto Universitario de Ciencias de la Salud, Facultad de Medicina, Fundación Barceló, Buenos Aires, Argentina. Introduction In the present work we study the nonlinear characteristics of the heart rate variability (HRV) in patients with normal and abnormal left ventricular function, and explore their relation to both the number of ventricular ectopic beats (VEB) in 24 hours and the ejection fraction. We use the false nearest neighbor method (FNN), a wellknown nonlinear analysis, which allows obtaining the embedding dimension (EDIM) and the false nearest neighbor fraction at dimension 10 (FNNF10).In the present work we study the nonlinear characteristics of the heart rate variability (HRV) in patients with normal and abnormal left ventricular function, and explore their relation to both the number of ventricular ectopic beats (VEB) in 24 hours and the ejection fraction. We use the false nearest neighbor method (FNN), a wellknown nonlinear analysis, which allows obtaining the embedding dimension (EDIM) and the false nearest neighbor fraction at dimension 10 (FNNF10). Objective To study the relation of both the EDIM and the FNNF10 to the number of VEBs in 24 hours and to a known measure of left ventricular function (i.e. left ventricular ejection fraction, LVEF). 10 to the number of VEBs in 24 hours and to a known measure of left ventricular function (i.e. left ventricular ejection fraction, LVEF). Methodology We studied Holter recordings of a total of 68 cardiac patients and 59 clinically healthy subjects used as a control group. In a first stage we considered two populations: subjects with more than 100 VEBs in 24 h (n=33) and subjects with congestive heart failure (CHF, n=35). In a second stage, the LVEF of 10 out of the total of patients was obtained: 7 had the LVEF measured with 2D-echocardiogram and 3 with a gamma camera. The Holter recordings were carefully analyzed and all of them showed less than 8% artifacts and did not have any consecutive artifact segments longer than 20 sec. These time series were then used to perform the calculation of both, the EDIM and the FNNF10.10. Results Stage 1: EDIM values were always between 7 and 10 for control subjects, reaching values as high as 15 for the other groups. The EDIM and FNNF10 were higher than normal in all the patients with more than 3000 VEBs/24 h., showing a welldefined correlation between EDIM and frequency of VEBs. The FNNF10 also progressively increased from the control group (0.0047±0.0043) to the VEB (0.0346±0.0417) and the CHF (0.1347±0.1443) groups. (The mean difference between groups was always statistically significant with p<0.0001.) In the control group the FNNF10 remained below a maximum of 0.03 and in the VEB group below 0.18. In the CHF group we obtained FNNF10 values as high as 0.3.10 were higher than normal in all the patients with more than 3000 VEBs/24 h., showing a welldefined correlation between EDIM and frequency of VEBs. The FNNF10 also progressively increased from the control group (0.0047±0.0043) to the VEB (0.0346±0.0417) and the CHF (0.1347±0.1443) groups. (The mean difference between groups was always statistically significant with p<0.0001.) In the control group the FNNF10 remained below a maximum of 0.03 and in the VEB group below 0.18. In the CHF group we obtained FNNF10 values as high as 0.3. Stage 2: Out of a total of 10 LVEFs measured, 6 were diminished (lower than 60%). Two of them had elevated EDIM>10, which was never found in subjects with a normal heart. Therefore, the indexes studied were more specific than sensitive. Conclusion The nonlinear indexes derived from the calculation of FNN are high in patients with VEB and their value correlates with the number of ectopic beats in 24 hours. They are even higher in patients with altered left ventricular function, showing high specificity. In developing countries there exist great difficulties to perform LVEF measures. Therefore, the calculation of these simple noninvasive indexes deserves to be considered as an indirect measure of the left ventricular fraction. These results should be further tested in a larger population.