- Email: [email protected]
Regional frequency variation during human ventricular fibrillation
Medical Engineering & Physics 31 (2009) 964–970
Contents lists available at ScienceDirect
Medical Engineering & Physics journal homepage: www.elsevier.com/locate/medengphy
Regional frequency variation during human ventricular fibrillation K. Umapathy a,b , S. Masse a , E. Sevaptsidis a , J. Asta a , H. Ross a , N. Thavandiran a , K. Nair a , T. Farid a , R. Cusimano a , J. Rogers a , S. Krishnan b,∗,1 , K. Nanthakumar a,1 a b
The Hull Family Cardiac Fibrillation Management Laboratory, Toronto General Hospital, Toronto, ON, Canada Ryerson University, Toronto, ON, Canada
a r t i c l e
i n f o
Article history: Received 18 November 2008 Received in revised form 12 May 2009 Accepted 13 May 2009 Keywords: Ventricular fibrillation Langendorff setup Isolated human hearts Dominant frequency Anatomical substrate
a b s t r a c t Quantifying the regional frequency variation in ventricular fibrillation (VF) may lead to focal strategies in treating human VF. We hypothesized that during human VF there are quantifiable regional frequency variations in the ventricles and they relate to underlying fixed myocardial substrate. In eight myopathic human hearts, we studied 35 VF episodes. The electrograms during VF were acquired simultaneously from the epicardium and endocardium using 2 electrode arrays each consisting of 112 electrodes. Regional characterization was performed using a ratio parameter derived from the dominant frequency analysis of the electrograms. The findings were related to the anatomical substrate using bipolar voltage maps. The results of the analysis indicate that LV had a larger dominant frequency (DF) span than RV (p = 0.0111) while there was no significant difference (p = 0.1488) in the DF span between LV freewall (FW) and septum (SE). Correlation of areas of abnormal myocardium with the dominant frequency feature matched only in 50% of the cases indicating that ion channel heterogeneity and time-varying physiological factors may play an important role in maintaining VF. © 2009 IPEM. Published by Elsevier Ltd. All rights reserved.
1. Introduction The complex electrical processes during ventricular fibrillation (VF) may spatially organize themselves [1–6]. It is essential to study the regional affinities of these processes as they may provide options for targeted therapeutic strategies to modulate human VF [7]. Studies evaluating regional differences in small and large animals have provided mechanistic insight in understanding the organization of VF in animals [8–10]. One of the main analyses done by groups that observe mother rotors in their preparations is the spatial distribution of fibrillation dynamics [3,4]. However, the extrapolation of those findings to humans and the relevance of those data to human VF are yet to be established. A related work by Wu et al. [11] studied a focal area (32 mm × 38 mm) of human data, however, global VF dynamics including the participation of the entire epicardium and endocardium was not studied. A recent study using phase singularity analysis on epicardial mapping data from patient with preserved ventricular function,
∗ Corresponding author at: Department of Electrical and Computer Engineering, Ryerson University, 350 Victoria Street, Toronto, ON M5B 2K3, Canada. Tel.: +1 416 979 5334; fax: +1 416 979 5280. E-mail address: [email protected] (S. Krishnan). 1 The senior authorship is shared between these authors.
suggested that multiple mechanisms are responsible for the organization of electrical activity in VF [12]. That study did not address if there were regions of high–low or low–high frequency transitions throughout the epicardium and endocardium of the human heart during VF. Regional differences in human VF and their contributions in human VF have not been evaluated in experimental data [13]. In addition, there has been a paucity of human heart studies that have simultaneously mapped significant portions of the endocardium and epicardium in VF [13,14]. In our previous work on human data [15], the focus was on studying rotors and there was no repeated VF induction on limited number of VF episodes. The dataset lacked power for quantifying fibrillation dynamics. In [16], we demonstrated the technique of optical mapping in human VF acquired from a single field of view with no regional comparisons. In this work, we studied human VF in 35 VF episodes by mapping VF globally on the epicardium and endocardium in Langendorff-perfused explanted myopathic human hearts. Here we evaluated regional differences in the spatial frequency characteristics and the potential role of underlying fixed myocardial substrate that would lead to better choices of ablation targets in treating human VF. The paper is organized as follows: Section 2 covers the protocol and the approach used to quantify the regional difference during VF, Section 3 provides the results of our analysis, Section 4 elaborates on the results and discusses the previous evidences in the literature in correlation with our findings, and conclusions are provided in Section 6.
1350-4533/$ – see front matter © 2009 IPEM. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.medengphy.2009.05.009
K. Umapathy et al. / Medical Engineering & Physics 31 (2009) 964–970
2. Methodology 2.1. Protocol The hearts used in this study (with informed consent) were explanted from eight cardiomyopathic patients (seven males and one female) who underwent transplantation. Of the eight human hearts used in this study, four were dilated cardiomyopathic (DCMP) and four were ischemic cardiomyopathic (ICMP). The weights of six hearts were 465, 550, 710, 440, 485 and 335 g. We do not have the weight information of two hearts. All the eight hearts had an ejection fraction of less than 30%. All four ICMP hearts had anterior wall myocardial infarction. None of the patients was on anti-arrhythmic drugs at the time of cardiac transplantation except beta-blocker. The method for performing human Langendorff experimentation is described elsewhere [15]. Briefly, immediately after the heart was explanted from the recipient, the heart was placed in cold Tyrode’s solution and transported to an adjacent room less than 3 min away and flushed thoroughly to remove blood particles. The hearts were selectively cannulated in the right and left coronary arteries and fitted to a Langendorff setup. The hearts were then Langendorff perfused with Tyrode’s solution (95% O2 + 5% CO2 ) at a flow rate of 0.9–1.1 ml/g/min. Perfusion pressure was adjusted to maintain a pressure of 60–70 mmHg. The entire isolated human heart was immersed in the temperature controlled Tyrode solution. The temperature was maintained at 37 ◦ C and continuously monitored in the coronary sinus effluent such that it reflected myocardial temperature. During the isolated heart experiments, the epicardium may experience temperature differences compared to the endocardium. Hence, we monitored the temperature to ensure there was never a temperature differential greater than 0.25◦ between the epi- and endocardium. Without blood perfusion, the hearts develop edema within 90 min of initial perfusion. Acquisition was performed only when the preparation did not reveal injury potentials and the protocol was completed within 45 min of initial cannulation. The heart was paced at a cycle length of 600 ms to keep it perfused and functional. Before the induction of VF, pacing was stopped and then VF was induced by briefly touching the heart at the same place with the two poles of a 9 V battery. The above-described protocol was approved by the University Health Network ethics committee and consent was obtained from each of the patients. 2.2. Acquisition of contact epicardial and endocardial electrograms during VF The details of the mapping tools and acquisition system are described elsewhere in detail [15]. VF was allowed to last 30 s before defibrillation. Recordings were made 5 s after VF had stabilized into a chaotic rhythm with no contraction of the myocardium. Twenty seconds of VF data were acquired. On average, five episodes of VF were induced per heart. A minimum of 7 min was allowed between VF episodes. Each 20 s episode was analyzed in 4 s epochs. The epicardial and endocardial electrograms were acquired using two extensible electrode arrays (Sock array covering the whole of epicardium and a Balloon array covering only the LV endocardium) each of which consisted of 112 electrode locations. Each of these locations consisted of two silver beads (2 mm diameter) (a total of 224 beads) separated by 2.1 mm centre to centre facilitating the possibility of acquiring either 112 unipolar or 112 bipolar electrograms. The 112 electrode locations were geometrically organized in 14 rows of 8 columns dispersed radially around the apex. Both arrays were made of flexible material and distance between adjacent electrode locations varied between 1 and 3 cm. Simultaneous unipolar recordings from both the arrays were performed using a mapping system described previously [17]. The filter settings for
965
unipolar signals were set to 0.5–200 Hz and the sampling rate was set to 1 K samples/s. In order to quantify the spatial differences, the 112 electrodes of both the electrode arrays were divided between the four regions [Sock array – (1) LV epicardium and (2) RV epicardium, and Balloon array – (3) freewall (FW) endocardium and (4) septum (SE) endocardium)]. The LV (epi) and FW (endo) regions were covered by 80 electrodes and the RV (epi) and SE (endo) regions by 32 electrodes. The four regions will be addressed as LV, RV, FW, and SE in the rest of the manuscript for notational convenience. 2.3. Dominant frequency ratio (DFR) Since this study presents only the spatial organization of VF using frequency analysis, unlike our previous work [18], here we restrict our analysis only to computationally less expensive dominant frequencies (DF) that were extracted from 112 electrograms for each of the VF episodes. The following methodological steps enumerate the DF extraction process. For each of the 112 unipolar electrograms, power spectral density (PSD) was estimated using the Welch averaged modified periodogram [15]. The PSD of the whole electrogram was calculated as the average of the periodograms constructed by segmenting and windowing (Hamming) the electrogram into 256 points segments with 128 points overlap and performing a 2048-point FFT on the windowed segments. Each PSD was then scanned between 1.5 and 12 Hz and the frequency associated with the highest energy component was extracted as the DF [15,16]. We limited the scan for frequency between 1.5 and 12 Hz to avoid low frequency and high frequency artifacts and at the same time to focus on the range of frequencies observed during human VF. Interpolated polar maps, representing epi or endo electrode arrays in two dimensions were generated to study spatial distribution of DFs. The DF values for 112 electrodes for each of the VF episodes were analyzed to identify representative features that quantified regional differences. The DFR was observed to demonstrate the difference between the regions and is defined as the ratio of the maximum and minimum DF in a region and is given by DFRR =
DF
max
DFmin
(1) R
where R indicates one of the four different regions. This feature indirectly measures the range or span of DF within a region. Since only the maximum and minimum DF value of a region is used in computing the DFR, this feature is (statistically) independent of the actual size of different regions, i.e. irrespective of the physical area of the regions analyzed, the DF distribution in these regions will have only one DF maximum and one DF minimum. To ensure electrode locations corresponding to high and low DF points are genuine and not due to noise or other artifacts, the electrograms corresponding to these locations were visually inspected and verified by electrophysiologists. The DFR feature is also immune to overestimation or underestimation of frequency by the Welch’s averaged modified periodogram method as it is derived as a ratio. 2.4. Myocardial substrate As we had mapped globally the entire epicardium and LV endocardium, a global histopathological assessment would interfere with the usual cardiac pathology analysis mandated by the transplant program. Moreover, the main focus of the manuscript is to arrive at targeted therapeutic options in treating human VF. In which case, it is practically impossible to perform biopsies/histopathological analysis during intra-operative procedures to infer about the anatomic substrate. Hence, we used a surrogate way of identifying anatomic substrate. The methodology outlined in our manuscript to infer anatomic substrate using bipolar elec-
966
K. Umapathy et al. / Medical Engineering & Physics 31 (2009) 964–970
Fig. 1. The figure illustrates the regional frequency differences of an isolated human heart during VF using unipolar electrogram. Epicardial and endocardial electrograms were recorded simultaneously.
trogram amplitude is an accepted concept in the literature [19–22]. We also verified that low amplitude of electrograms were located in abnormal tissues from pre-transplant echocardiograms based on landmarks on the heart and echogeneic regions with wall motion abnormality.
In order to relate the frequency transitions to anatomical substrate influences, we constructed abnormal myocardial substrate maps for each of the hearts using an established method [23]. In this established method, during a pacing protocol the bipolar electrode locations that had peak-to-peak amplitude value of less than 0.5 mV
Fig. 2. The figure illustrates the regional frequency differences of an isolated human heart during VF using bipolar electrograms. Epicardial and endocardial electrograms were recorded simultaneously.
K. Umapathy et al. / Medical Engineering & Physics 31 (2009) 964–970
967
Fig. 3. The figure demonstrates the variations in the spatial distribution of DF over 5 VF episodes recorded with the same heart using same induction method. The arrow indicates a DF value in a region with similar DF values.
Fig. 4. The left panel shows the electrode array with 112 electrodes and the LV/FW and RV/SE locations. The top right panel shows the line plot of DF for a sample LV. High and low DF points are indicated in the figure as “A” and “B”. The bottom right panel shows the corresponding polar DF map. The colour bar represents DF from 1 to 12 Hz. DFR is computed as the ratio between the DF at locations “A” and “B”. The left bottom panel shows the electrograms at locations A and B shown in the DF line plot above. A clear difference in the frequency could be observed.
968
K. Umapathy et al. / Medical Engineering & Physics 31 (2009) 964–970
were identified as abnormal myocardium. Since our acquisition system is equipped to measure both unipolar and bipolar electrograms, we were able to use the bipolar electrograms to construct abnormal myocardial maps for each of the hearts. The electrode locations were marked into three (levels) categories: (i) as healthy tissue (>0.5 mV), (ii) as abnormal myocardial tissue (<0.5 mV) and (iii) areas of poor electrode contact. Using the same interpolation procedure as discussed above, finer and smoother abnormal myocardial maps were constructed. These abnormal myocardial maps provided an anatomical substrate mapping that could be related to the DF features. 3. Results We found in these explanted human hearts that during VF different regions had markedly different activation rates, with some regions activating rapidly while other regions show remarkably less activations. In Fig. 1 these differences in activations are illustrated using unipolar recordings. In Fig. 2 bipolar recordings are shown to clearly demonstrate the activation rate changes. Figs. 1 and 2 show the significant intra and interventricular changes in the activation rates between the four regions. 3.1. Repeatability of VF dynamics within the same heart VF dynamics in the same heart for each of the different VF episodes with the same induction method were not similar. To illustrate this, in Fig. 3, we have shown different spatial distribution of DF frequencies over five VF episodes for the same heart. It is evident that the VF dynamics are not entirely reproducible between the different VF episodes induced within the same heart. 3.2. DFR The DFR defined as the ratio of the maximum and minimum DF is illustrated in Fig. 4. The electrograms in the figure corresponding to the electrode locations “A” and “B” show the difference in DF. The DFR for each of the 4 regions was computed for all the 35 VF episodes and its distribution is shown as box plots in Fig. 5. A legend explaining the box plot is shown as an inset in the top right corner of Fig. 5. The mean was found to be 2.10, 1.75, 1.55, and 1.34 for LV, RV, FW, and SE respectively. LV had a larger DFR than RV (p = 0.0111) and FW had a DFR that was not significantly different than SE (p = 0.1488). The p values were computed using mixed linear model (SAS statistical software) taking into account the repeated measures from the same heart. Here it should also be noted since the DFR is computed using maximum and minimum DF of each of the regions, it is independent of the size of the regions with respect to the physical area of the four regions. We also compared the difference in the average DFR (LV–RV vs FW–SE) between the epicardium and endocardium, and found them to be 1.93 and 1.45 respectively. This indicates that on an average the epicardium had larger range of frequencies than the endocardium. We further analyzed the LV–RV relationship observed in the DFR feature. For the DFR to be different between the two regions, either they should differ in their maximum DF or minimum DF or both. We tested this by computing the maximum and minimum DF for these two regions and the results are plotted in Fig. 6. From the box plots in Fig. 6 we could observe that there is no significant difference in their maximum DFs but interestingly there is a significant difference between their minimum DFs. The minimum DFs of the LV were much smaller than RV and well confined in a smaller range (p = 0.0398). The mean value of maximum and minimum DF for LV and RV were 5.03, 4.85, 2.34, and 2.99 respectively. These findings point to chamber specific characteristics to fibrillation dynamics.
Fig. 5. The box plots show the distribution of the DFR feature over the four regions. The legend at the top right corner of the figure explains the box plot and its relation to the data distribution. The mean of the DFR for each of the four regions is given at the bottom of the figure. The p values for the null hypothesis testing between LV and RV and FW and SE are shown above a double-sided arrow connecting the tested pair. Observing the box plots, the results indicate that the LV has a greater DFR than RV and so does FW than SE.
3.3. Myocardial substrate and DFR The epicardial DFR locations (DFmax and DFmin ) were compared to the abnormal myocardial locations to verify if they co-localize and if they occur at the abnormal myocardial boundaries. Only 33 VF episodes were used in the analysis, as 2 VF episodes did not have the bipolar electrograms. We observed that the DFR electrode locations for LV (45% LV DFmax and 54% LV DFmin ) co-localized with the abnormal myocardial area. In the RV similar to LV only 50% of the DFR electrode locations co-localized, however with different distribution between RV DFmax (66%) and RV DFmin (33%). Table 1 shows the matched and unmatched number of cases for LV–RV DFmax and DFmin locations. Not all the DFR locations co-localized with the anatomically abnormal myocardial locations indicating that the frequency transitions are not completely explained by the anatomical substrate alone. 4. Discussions We have shown here for the first time, that during human VF, there are characteristic regional differences in global fibrillation dynamics. These spatial differences are not entirely explained by underlying fixed intrinsic anatomical and electrophysiological properties of the myocardium. In addition, different VF episodes in the same heart demonstrated frequency changes that do not reproducibly appear at the same location however are chamber specific as seen by characteristic differences between the ventricles. This suggests that during VF, the role of dynamic physiologic Table 1 Correlation of abnormal myocardium with the DFmax and DFmin locations in the epicardium. Region
Matched
Not matched
Total
LV DFmax LV DFmin RV DFmax RV DFmin
15 18 22 11
18 15 11 22
33 33 33 33
K. Umapathy et al. / Medical Engineering & Physics 31 (2009) 964–970
969
Fig. 6. The left panel shows the box plot of the maximum and minimum DF of LV and RV. The mean of the maximum and minimum DF are given at the bottom of the left panel. From the figure it could be observed that the minimum frequency of LV is smaller than that of the RV and well confined to a smaller bandwidth. The p values of the tested regions pairs are shown below (for maximum DF LV–RV comparison—p value not significant) and above (for minimum DF LV–RV comparison). The right panel shows a sample maximum DF of the LV region and the minimum DF of the RV region.
factors such as memory, restitution and region-specific expression of ion channels may be important in regional organization and needs evaluation. The new findings of our study are as follows. In the epicardium, LV exhibits greater DFR than RV. In the endocardium, FW and SE share a similar relationship. Abnormal myocardium as quantified by bipolar amplitude fail to entirely explain the differences in fibrillation dynamics. 4.1. RV–LV epicardial difference The differences between RV and LV indicate that these regions participate in the VF phenomenon in different ways. While the DFR feature differentiated LV from RV, the mean DF did not show strong differences. This could be due to the spatial averaging of DF, which reduces the discriminatory characteristics between the regions. Jalife et al. have shown at least twofold differences in mean DF between LV and RV in guinea pigs during VF [3,4]. Rogers et al. have studied differences between RV and LV during porcine VF [24]. They found that there was a quantifiable difference between the regions with LV showing more wavefronts and activations than RV. However Rogers only mapped the epicardium in a 4 cm × 5 cm area and did not perform global endocardial and/or epicardial mappings. In the guinea pig the rotors they observed were stable spatially and temporally. However, in human VF rotors are not stable temporally and spatially over greater surfaces, thus we believe the averaging over the time domain and spatial domain, dilutes the marked differences present. However, the DFR feature in this work is local in nature and hence does not get diluted by spatial averaging. In fact the figures show that the differences in DF can be as high as 3–4 Hz between independent regions. These sites of maximum and minimum frequencies need further study with regards to VF maintenance, and potential focal treatment strategies. 4.2. Epicardial and endocardial differences Our study is unique in studying the epicardium and endocardium in human VF, which allowed us to compare them at identical time segments during VF. Comparing epicardium and endocardium as a whole, the epicardium seems to be less orga-
nized with higher DFR than endocardium. This finding is consistent with our previous in vivo epicardial-endocardial mapping studies in three humans, where we saw persistence of a defined number of rotors in the endocardium while the epicardium demonstrated disorganization at the same time [9]. Most small animal studies due to the nature of the preparation do not have endocardial mapping information, and other animal studies that have performed endocardial mapping studies have had few electrodes not allowing for global mapping and comparisons between the two mapping surfaces. Though the exact significance of endocardial organization is not explored in our study the potential that the sources may be endocardial with fibrillatory conduction to epicardium may allow for future studies with focal endocardial therapies. 4.3. LV endocardial regional difference In endocardium, the DFR feature did not show much differentiation between FW and SE; however, there were definite regions on the FW that activated much faster than other regions within the FW and SE. These regions seem to coincide to the posterior LV free wall adjacent to the posterior papillary muscle. Indeed Chen’s group has shown that the papillary muscle may help anchor rotors [25]. Our previous optical mapping study in humans has shown greater tendency for wavefronts to block in the septum than the free wall of the LV [16]. Indeed Huang et al. studying porcine VF in search of mother rotor found to their surprise greater conduction block in the septum [26]. Due to the fact that the endocardial freewall had regions with rapid organized activity in VF, it should be targeted for future detailed mapping strategies especially at the base of papillary muscles. 4.4. Myocardial substrate and DFR From our analysis, we were unable to explain our DF findings entirely on the basis of fixed anatomic myocardial substrate. The fact that we were unable to entirely explain regional organization on the basis of fixed anatomic substrate suggests that dynamic physiological factors may determine the regional frequency characteristics. This is further confirmed by the fact that the
970
K. Umapathy et al. / Medical Engineering & Physics 31 (2009) 964–970
frequency characteristics vary from episode to episode within the same heart in spite of the same induction method. To add even more emphasis, interestingly we observed at the same electrode location both DFmax and DFmin frequencies of a region for two different VF episodes. 5. Limitations The hearts were Langendorff perfused in a denervated (cut-off from nerve supply) environment. It is unknown if these factors influence activation patterns. Regional differences of perfusion pressure might play a role in the regional differences of activation rates. However this is unlikely as areas of low bipolar amplitude (infracted/scar) tissues did not explain the differences. In addition, four of the eight hearts studied had normal coronaries and there were no differences that were observed between the non-coronary and the coronary cardiomyopathies with regards to DFR. 6. Conclusions During human VF there are quantifiable regional differences in the DFR. Regional differences in fibrillation dynamics are not entirely explained by fixed anatomic substrate. This suggests that dynamic factors such as ion channel heterogeneity and timevarying physiological factors may play an important role and need to be evaluated during human VF. Acknowledgments This study was supported by the Canadian Institutes of Health Research Grant (NA 777687), Heart & Stroke Foundation Federation Fund, and Ryerson University. Conflict of interest statement None. References [1] Jalife J, Gray R, Chen J. Mechanisms of ventricular fibrillation; drifting scroll waves and phase singularities of electrical activation. In: Cardiac electrophysiology—from cell to bedside. 3rd ed.; 2000. p. 386–404. [2] Guyton A, Hall J. Cardiac arrhythmias and their electrocardiographic interpretation. In: Medical Physiology. 10 ed., 2000. p. 134–42. [3] Chen J, Mandapati R, Berenfeld O, Skanes A, Jalife J. High-frequency periodic sources underlie ventricular fibrillation in the isolated rabbit heart. Circulation Research 2000;86(1):86–93. [4] Samie F, Berenfeld O, Anumonwo J, Mironov S, Udassi S, Beaumont J, et al. Rectification of the background potassium current; a determinant of rotor dynamics in ventricular fibrillation. Circulation Research 2001;89(12):1216–23. [5] Gray R, Pertsov A, Jalife J. Spatial and temporal organization during cardiac fibrillation. Nature 1998;392(6671):75–8. [6] Nanthakumar K, Huang J, Rogers J, Johnson P, Newton J, Walcott G, et al. Regional differences in ventricular fibrillation in the open-chest porcine left ventricle. Circulation Research 2002;91(8):733–40.
[7] Pak H, Oh Y, Liu Y, Wu T, Karagueuzian H, Lin S, et al. Catheter ablation of ventricular fibrillation in rabbit ventricles treated with ˇ-blockers. Circulation 2003;108(25):3149–56. [8] Thomas S, Thiagalingam A, Wallace E, Kovoor P, Ross D. Organization of myocardial activation during ventricular fibrillation after myocardial infarction: evidence for sustained high-frequency sources. Circulation 2005;112(2): 157. [9] Everett T, Wilson E, Foreman S, Olgin IV J. Mechanisms of ventricular fibrillation in canine models of congestive heart failure and ischemia assessed by in vivo noncontact mapping. Circulation 2005;112(11):1532. [10] Zaitsev A, Berenfeld O, Mironov S, Jalife J, Pertsov A. Distribution of excitation frequencies on the epicardial and endocardial surfaces of fibrillating ventricular wall of the sheep heart. Circulation Research 2000;86(4):408–17. [11] Wu T, Ong J, Hwang C, Lee J, Fishbein M, Czer L, et al. Characteristics of wave fronts during ventricular fibrillation in human hearts with dilated cardiomyopathy: role of increased fibrosis in the generation of reentry. Journal of the American College of Cardiology 1998;32(1):187–96. [12] Nash M, Mourad A, Clayton R, Sutton P, Bradley C, Hayward M, et al. Evidence for multiple mechanisms in human ventricular fibrillation. Circulation 2006;114(6):536. [13] ten Tusscher K, Panfilov A. Alternans and spiral breakup in a human ventricular tissue model. American Journal of Physiology—Heart and Circulatory Physiology 2006;291(3):H1088. [14] Nanthakumar K, Walcott G, Melnick S, Rogers J, Kay M, Smith W, et al. Epicardial organization of human ventricular fibrillation. Heart Rhythm 2004;1(1): 14–23. [15] Masse S, Downar E, Chauhan V, Sevaptsidis E, Nanthakumar K. Ventricular fibrillation in myopathic human hearts: mechanistic insights from in vivo global endocardial and epicardial mapping. American Journal of Physiology—Heart and Circulatory Physiology 2007;292(6):H2589. [16] Nanthakumar K, Jalife J, Masse S, Downar E, Pop M, Asta J, et al. Optical mapping of Langendorff-perfused human hearts: establishing a model for the study of ventricular fibrillation in humans. American Journal of Physiology—Heart and Circulatory Physiology 2007;293(1):H875. [17] Sevaptsidis E, Masse S, Parson ID, Downar E, Kimber S. Simultaneous unipolar and bipolar recording of cardiac electrical activity. Pacing and Clinical Electrophysiology 1992;5:45–51. [18] Umapathy K, Masse S, Sevaptsidis E, Asta J, Krishnan S, Nanthakumar K. Spatiotemporal frequency analysis of ventricular fibrillation in explanted human hearts. IEEE Transactions on Biomedical Engineering 2009;56(2):328–35. [19] Soejima K, Stevenson W, Maisel W, Sapp J, Epstein L. Electrically unexcitable scar mapping based on pacing threshold for identification of the reentry circuit isthmus feasibility for guiding ventricular tachycardia ablation. Circulation 2002;106(13):1678–83. [20] Marchlinski F, Zado E, Dixit S, Gerstenfeld E, Callans D, Hsia H, et al. Electroanatomic substrate and outcome of catheter ablative therapy for ventricular tachycardia in setting of right ventricular cardiomyopathy. Circulation 2004;110(16):2293–8. [21] Oza S, Wilber D. Substrate-based endocardial ablation of post infarction ventricular tachycardia. Heart Rhythm 2006;3(5):607–9. [22] Verma A, Kilicaslan F, Schweikert R, Tomassoni G, Rossillo A, Marrouche N, et al. Short- and long-term success of substrate-based mapping and ablation of ventricular tachycardia in arrhythmogenic right ventricular dysplasia. Circulation 2005;111(24):3209–16. [23] Reddy V, Wrobleski D, Houghtaling C, Josephson M, Ruskin J. Combined epicardial and endocardial electroanatomic mapping in a porcine model of healed myocardial infarction. Circulation 2003;107(25):3236–42. [24] Rogers J, Huang J, Pedota R, Walker R, Smith W, Ideker R. Fibrillation is more complex in the left ventricle than in the right ventricle. Journal of Cardiovascular Electrophysiology 2000;11(12):1364–71. [25] Pak H, Kim Y, Lim H, Chou C, Miyauchi Y, Fang Y, et al. Role of the posterior papillary muscle and Purkinje potentials in the mechanism of ventricular fibrillation in open chest dogs and swine: effects of catheter ablation. Journal of Cardiovascular Electrophysiology 2006;17(7):777–83. [26] Huang J, Walcott G, Killingsworth C, Melnick S, Rogers J, Ideker R. Quantification of activation patterns during ventricular fibrillation in open-chest porcine left ventricle and septum. Heart Rhythm 2005;2(7):720–8.
Contents lists available at ScienceDirect
Medical Engineering & Physics journal homepage: www.elsevier.com/locate/medengphy
Regional frequency variation during human ventricular fibrillation K. Umapathy a,b , S. Masse a , E. Sevaptsidis a , J. Asta a , H. Ross a , N. Thavandiran a , K. Nair a , T. Farid a , R. Cusimano a , J. Rogers a , S. Krishnan b,∗,1 , K. Nanthakumar a,1 a b
The Hull Family Cardiac Fibrillation Management Laboratory, Toronto General Hospital, Toronto, ON, Canada Ryerson University, Toronto, ON, Canada
a r t i c l e
i n f o
Article history: Received 18 November 2008 Received in revised form 12 May 2009 Accepted 13 May 2009 Keywords: Ventricular fibrillation Langendorff setup Isolated human hearts Dominant frequency Anatomical substrate
a b s t r a c t Quantifying the regional frequency variation in ventricular fibrillation (VF) may lead to focal strategies in treating human VF. We hypothesized that during human VF there are quantifiable regional frequency variations in the ventricles and they relate to underlying fixed myocardial substrate. In eight myopathic human hearts, we studied 35 VF episodes. The electrograms during VF were acquired simultaneously from the epicardium and endocardium using 2 electrode arrays each consisting of 112 electrodes. Regional characterization was performed using a ratio parameter derived from the dominant frequency analysis of the electrograms. The findings were related to the anatomical substrate using bipolar voltage maps. The results of the analysis indicate that LV had a larger dominant frequency (DF) span than RV (p = 0.0111) while there was no significant difference (p = 0.1488) in the DF span between LV freewall (FW) and septum (SE). Correlation of areas of abnormal myocardium with the dominant frequency feature matched only in 50% of the cases indicating that ion channel heterogeneity and time-varying physiological factors may play an important role in maintaining VF. © 2009 IPEM. Published by Elsevier Ltd. All rights reserved.
1. Introduction The complex electrical processes during ventricular fibrillation (VF) may spatially organize themselves [1–6]. It is essential to study the regional affinities of these processes as they may provide options for targeted therapeutic strategies to modulate human VF [7]. Studies evaluating regional differences in small and large animals have provided mechanistic insight in understanding the organization of VF in animals [8–10]. One of the main analyses done by groups that observe mother rotors in their preparations is the spatial distribution of fibrillation dynamics [3,4]. However, the extrapolation of those findings to humans and the relevance of those data to human VF are yet to be established. A related work by Wu et al. [11] studied a focal area (32 mm × 38 mm) of human data, however, global VF dynamics including the participation of the entire epicardium and endocardium was not studied. A recent study using phase singularity analysis on epicardial mapping data from patient with preserved ventricular function,
∗ Corresponding author at: Department of Electrical and Computer Engineering, Ryerson University, 350 Victoria Street, Toronto, ON M5B 2K3, Canada. Tel.: +1 416 979 5334; fax: +1 416 979 5280. E-mail address: [email protected] (S. Krishnan). 1 The senior authorship is shared between these authors.
suggested that multiple mechanisms are responsible for the organization of electrical activity in VF [12]. That study did not address if there were regions of high–low or low–high frequency transitions throughout the epicardium and endocardium of the human heart during VF. Regional differences in human VF and their contributions in human VF have not been evaluated in experimental data [13]. In addition, there has been a paucity of human heart studies that have simultaneously mapped significant portions of the endocardium and epicardium in VF [13,14]. In our previous work on human data [15], the focus was on studying rotors and there was no repeated VF induction on limited number of VF episodes. The dataset lacked power for quantifying fibrillation dynamics. In [16], we demonstrated the technique of optical mapping in human VF acquired from a single field of view with no regional comparisons. In this work, we studied human VF in 35 VF episodes by mapping VF globally on the epicardium and endocardium in Langendorff-perfused explanted myopathic human hearts. Here we evaluated regional differences in the spatial frequency characteristics and the potential role of underlying fixed myocardial substrate that would lead to better choices of ablation targets in treating human VF. The paper is organized as follows: Section 2 covers the protocol and the approach used to quantify the regional difference during VF, Section 3 provides the results of our analysis, Section 4 elaborates on the results and discusses the previous evidences in the literature in correlation with our findings, and conclusions are provided in Section 6.
1350-4533/$ – see front matter © 2009 IPEM. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.medengphy.2009.05.009
K. Umapathy et al. / Medical Engineering & Physics 31 (2009) 964–970
2. Methodology 2.1. Protocol The hearts used in this study (with informed consent) were explanted from eight cardiomyopathic patients (seven males and one female) who underwent transplantation. Of the eight human hearts used in this study, four were dilated cardiomyopathic (DCMP) and four were ischemic cardiomyopathic (ICMP). The weights of six hearts were 465, 550, 710, 440, 485 and 335 g. We do not have the weight information of two hearts. All the eight hearts had an ejection fraction of less than 30%. All four ICMP hearts had anterior wall myocardial infarction. None of the patients was on anti-arrhythmic drugs at the time of cardiac transplantation except beta-blocker. The method for performing human Langendorff experimentation is described elsewhere [15]. Briefly, immediately after the heart was explanted from the recipient, the heart was placed in cold Tyrode’s solution and transported to an adjacent room less than 3 min away and flushed thoroughly to remove blood particles. The hearts were selectively cannulated in the right and left coronary arteries and fitted to a Langendorff setup. The hearts were then Langendorff perfused with Tyrode’s solution (95% O2 + 5% CO2 ) at a flow rate of 0.9–1.1 ml/g/min. Perfusion pressure was adjusted to maintain a pressure of 60–70 mmHg. The entire isolated human heart was immersed in the temperature controlled Tyrode solution. The temperature was maintained at 37 ◦ C and continuously monitored in the coronary sinus effluent such that it reflected myocardial temperature. During the isolated heart experiments, the epicardium may experience temperature differences compared to the endocardium. Hence, we monitored the temperature to ensure there was never a temperature differential greater than 0.25◦ between the epi- and endocardium. Without blood perfusion, the hearts develop edema within 90 min of initial perfusion. Acquisition was performed only when the preparation did not reveal injury potentials and the protocol was completed within 45 min of initial cannulation. The heart was paced at a cycle length of 600 ms to keep it perfused and functional. Before the induction of VF, pacing was stopped and then VF was induced by briefly touching the heart at the same place with the two poles of a 9 V battery. The above-described protocol was approved by the University Health Network ethics committee and consent was obtained from each of the patients. 2.2. Acquisition of contact epicardial and endocardial electrograms during VF The details of the mapping tools and acquisition system are described elsewhere in detail [15]. VF was allowed to last 30 s before defibrillation. Recordings were made 5 s after VF had stabilized into a chaotic rhythm with no contraction of the myocardium. Twenty seconds of VF data were acquired. On average, five episodes of VF were induced per heart. A minimum of 7 min was allowed between VF episodes. Each 20 s episode was analyzed in 4 s epochs. The epicardial and endocardial electrograms were acquired using two extensible electrode arrays (Sock array covering the whole of epicardium and a Balloon array covering only the LV endocardium) each of which consisted of 112 electrode locations. Each of these locations consisted of two silver beads (2 mm diameter) (a total of 224 beads) separated by 2.1 mm centre to centre facilitating the possibility of acquiring either 112 unipolar or 112 bipolar electrograms. The 112 electrode locations were geometrically organized in 14 rows of 8 columns dispersed radially around the apex. Both arrays were made of flexible material and distance between adjacent electrode locations varied between 1 and 3 cm. Simultaneous unipolar recordings from both the arrays were performed using a mapping system described previously [17]. The filter settings for
965
unipolar signals were set to 0.5–200 Hz and the sampling rate was set to 1 K samples/s. In order to quantify the spatial differences, the 112 electrodes of both the electrode arrays were divided between the four regions [Sock array – (1) LV epicardium and (2) RV epicardium, and Balloon array – (3) freewall (FW) endocardium and (4) septum (SE) endocardium)]. The LV (epi) and FW (endo) regions were covered by 80 electrodes and the RV (epi) and SE (endo) regions by 32 electrodes. The four regions will be addressed as LV, RV, FW, and SE in the rest of the manuscript for notational convenience. 2.3. Dominant frequency ratio (DFR) Since this study presents only the spatial organization of VF using frequency analysis, unlike our previous work [18], here we restrict our analysis only to computationally less expensive dominant frequencies (DF) that were extracted from 112 electrograms for each of the VF episodes. The following methodological steps enumerate the DF extraction process. For each of the 112 unipolar electrograms, power spectral density (PSD) was estimated using the Welch averaged modified periodogram [15]. The PSD of the whole electrogram was calculated as the average of the periodograms constructed by segmenting and windowing (Hamming) the electrogram into 256 points segments with 128 points overlap and performing a 2048-point FFT on the windowed segments. Each PSD was then scanned between 1.5 and 12 Hz and the frequency associated with the highest energy component was extracted as the DF [15,16]. We limited the scan for frequency between 1.5 and 12 Hz to avoid low frequency and high frequency artifacts and at the same time to focus on the range of frequencies observed during human VF. Interpolated polar maps, representing epi or endo electrode arrays in two dimensions were generated to study spatial distribution of DFs. The DF values for 112 electrodes for each of the VF episodes were analyzed to identify representative features that quantified regional differences. The DFR was observed to demonstrate the difference between the regions and is defined as the ratio of the maximum and minimum DF in a region and is given by DFRR =
DF
max
DFmin
(1) R
where R indicates one of the four different regions. This feature indirectly measures the range or span of DF within a region. Since only the maximum and minimum DF value of a region is used in computing the DFR, this feature is (statistically) independent of the actual size of different regions, i.e. irrespective of the physical area of the regions analyzed, the DF distribution in these regions will have only one DF maximum and one DF minimum. To ensure electrode locations corresponding to high and low DF points are genuine and not due to noise or other artifacts, the electrograms corresponding to these locations were visually inspected and verified by electrophysiologists. The DFR feature is also immune to overestimation or underestimation of frequency by the Welch’s averaged modified periodogram method as it is derived as a ratio. 2.4. Myocardial substrate As we had mapped globally the entire epicardium and LV endocardium, a global histopathological assessment would interfere with the usual cardiac pathology analysis mandated by the transplant program. Moreover, the main focus of the manuscript is to arrive at targeted therapeutic options in treating human VF. In which case, it is practically impossible to perform biopsies/histopathological analysis during intra-operative procedures to infer about the anatomic substrate. Hence, we used a surrogate way of identifying anatomic substrate. The methodology outlined in our manuscript to infer anatomic substrate using bipolar elec-
966
K. Umapathy et al. / Medical Engineering & Physics 31 (2009) 964–970
Fig. 1. The figure illustrates the regional frequency differences of an isolated human heart during VF using unipolar electrogram. Epicardial and endocardial electrograms were recorded simultaneously.
trogram amplitude is an accepted concept in the literature [19–22]. We also verified that low amplitude of electrograms were located in abnormal tissues from pre-transplant echocardiograms based on landmarks on the heart and echogeneic regions with wall motion abnormality.
In order to relate the frequency transitions to anatomical substrate influences, we constructed abnormal myocardial substrate maps for each of the hearts using an established method [23]. In this established method, during a pacing protocol the bipolar electrode locations that had peak-to-peak amplitude value of less than 0.5 mV
Fig. 2. The figure illustrates the regional frequency differences of an isolated human heart during VF using bipolar electrograms. Epicardial and endocardial electrograms were recorded simultaneously.
K. Umapathy et al. / Medical Engineering & Physics 31 (2009) 964–970
967
Fig. 3. The figure demonstrates the variations in the spatial distribution of DF over 5 VF episodes recorded with the same heart using same induction method. The arrow indicates a DF value in a region with similar DF values.
Fig. 4. The left panel shows the electrode array with 112 electrodes and the LV/FW and RV/SE locations. The top right panel shows the line plot of DF for a sample LV. High and low DF points are indicated in the figure as “A” and “B”. The bottom right panel shows the corresponding polar DF map. The colour bar represents DF from 1 to 12 Hz. DFR is computed as the ratio between the DF at locations “A” and “B”. The left bottom panel shows the electrograms at locations A and B shown in the DF line plot above. A clear difference in the frequency could be observed.
968
K. Umapathy et al. / Medical Engineering & Physics 31 (2009) 964–970
were identified as abnormal myocardium. Since our acquisition system is equipped to measure both unipolar and bipolar electrograms, we were able to use the bipolar electrograms to construct abnormal myocardial maps for each of the hearts. The electrode locations were marked into three (levels) categories: (i) as healthy tissue (>0.5 mV), (ii) as abnormal myocardial tissue (<0.5 mV) and (iii) areas of poor electrode contact. Using the same interpolation procedure as discussed above, finer and smoother abnormal myocardial maps were constructed. These abnormal myocardial maps provided an anatomical substrate mapping that could be related to the DF features. 3. Results We found in these explanted human hearts that during VF different regions had markedly different activation rates, with some regions activating rapidly while other regions show remarkably less activations. In Fig. 1 these differences in activations are illustrated using unipolar recordings. In Fig. 2 bipolar recordings are shown to clearly demonstrate the activation rate changes. Figs. 1 and 2 show the significant intra and interventricular changes in the activation rates between the four regions. 3.1. Repeatability of VF dynamics within the same heart VF dynamics in the same heart for each of the different VF episodes with the same induction method were not similar. To illustrate this, in Fig. 3, we have shown different spatial distribution of DF frequencies over five VF episodes for the same heart. It is evident that the VF dynamics are not entirely reproducible between the different VF episodes induced within the same heart. 3.2. DFR The DFR defined as the ratio of the maximum and minimum DF is illustrated in Fig. 4. The electrograms in the figure corresponding to the electrode locations “A” and “B” show the difference in DF. The DFR for each of the 4 regions was computed for all the 35 VF episodes and its distribution is shown as box plots in Fig. 5. A legend explaining the box plot is shown as an inset in the top right corner of Fig. 5. The mean was found to be 2.10, 1.75, 1.55, and 1.34 for LV, RV, FW, and SE respectively. LV had a larger DFR than RV (p = 0.0111) and FW had a DFR that was not significantly different than SE (p = 0.1488). The p values were computed using mixed linear model (SAS statistical software) taking into account the repeated measures from the same heart. Here it should also be noted since the DFR is computed using maximum and minimum DF of each of the regions, it is independent of the size of the regions with respect to the physical area of the four regions. We also compared the difference in the average DFR (LV–RV vs FW–SE) between the epicardium and endocardium, and found them to be 1.93 and 1.45 respectively. This indicates that on an average the epicardium had larger range of frequencies than the endocardium. We further analyzed the LV–RV relationship observed in the DFR feature. For the DFR to be different between the two regions, either they should differ in their maximum DF or minimum DF or both. We tested this by computing the maximum and minimum DF for these two regions and the results are plotted in Fig. 6. From the box plots in Fig. 6 we could observe that there is no significant difference in their maximum DFs but interestingly there is a significant difference between their minimum DFs. The minimum DFs of the LV were much smaller than RV and well confined in a smaller range (p = 0.0398). The mean value of maximum and minimum DF for LV and RV were 5.03, 4.85, 2.34, and 2.99 respectively. These findings point to chamber specific characteristics to fibrillation dynamics.
Fig. 5. The box plots show the distribution of the DFR feature over the four regions. The legend at the top right corner of the figure explains the box plot and its relation to the data distribution. The mean of the DFR for each of the four regions is given at the bottom of the figure. The p values for the null hypothesis testing between LV and RV and FW and SE are shown above a double-sided arrow connecting the tested pair. Observing the box plots, the results indicate that the LV has a greater DFR than RV and so does FW than SE.
3.3. Myocardial substrate and DFR The epicardial DFR locations (DFmax and DFmin ) were compared to the abnormal myocardial locations to verify if they co-localize and if they occur at the abnormal myocardial boundaries. Only 33 VF episodes were used in the analysis, as 2 VF episodes did not have the bipolar electrograms. We observed that the DFR electrode locations for LV (45% LV DFmax and 54% LV DFmin ) co-localized with the abnormal myocardial area. In the RV similar to LV only 50% of the DFR electrode locations co-localized, however with different distribution between RV DFmax (66%) and RV DFmin (33%). Table 1 shows the matched and unmatched number of cases for LV–RV DFmax and DFmin locations. Not all the DFR locations co-localized with the anatomically abnormal myocardial locations indicating that the frequency transitions are not completely explained by the anatomical substrate alone. 4. Discussions We have shown here for the first time, that during human VF, there are characteristic regional differences in global fibrillation dynamics. These spatial differences are not entirely explained by underlying fixed intrinsic anatomical and electrophysiological properties of the myocardium. In addition, different VF episodes in the same heart demonstrated frequency changes that do not reproducibly appear at the same location however are chamber specific as seen by characteristic differences between the ventricles. This suggests that during VF, the role of dynamic physiologic Table 1 Correlation of abnormal myocardium with the DFmax and DFmin locations in the epicardium. Region
Matched
Not matched
Total
LV DFmax LV DFmin RV DFmax RV DFmin
15 18 22 11
18 15 11 22
33 33 33 33
K. Umapathy et al. / Medical Engineering & Physics 31 (2009) 964–970
969
Fig. 6. The left panel shows the box plot of the maximum and minimum DF of LV and RV. The mean of the maximum and minimum DF are given at the bottom of the left panel. From the figure it could be observed that the minimum frequency of LV is smaller than that of the RV and well confined to a smaller bandwidth. The p values of the tested regions pairs are shown below (for maximum DF LV–RV comparison—p value not significant) and above (for minimum DF LV–RV comparison). The right panel shows a sample maximum DF of the LV region and the minimum DF of the RV region.
factors such as memory, restitution and region-specific expression of ion channels may be important in regional organization and needs evaluation. The new findings of our study are as follows. In the epicardium, LV exhibits greater DFR than RV. In the endocardium, FW and SE share a similar relationship. Abnormal myocardium as quantified by bipolar amplitude fail to entirely explain the differences in fibrillation dynamics. 4.1. RV–LV epicardial difference The differences between RV and LV indicate that these regions participate in the VF phenomenon in different ways. While the DFR feature differentiated LV from RV, the mean DF did not show strong differences. This could be due to the spatial averaging of DF, which reduces the discriminatory characteristics between the regions. Jalife et al. have shown at least twofold differences in mean DF between LV and RV in guinea pigs during VF [3,4]. Rogers et al. have studied differences between RV and LV during porcine VF [24]. They found that there was a quantifiable difference between the regions with LV showing more wavefronts and activations than RV. However Rogers only mapped the epicardium in a 4 cm × 5 cm area and did not perform global endocardial and/or epicardial mappings. In the guinea pig the rotors they observed were stable spatially and temporally. However, in human VF rotors are not stable temporally and spatially over greater surfaces, thus we believe the averaging over the time domain and spatial domain, dilutes the marked differences present. However, the DFR feature in this work is local in nature and hence does not get diluted by spatial averaging. In fact the figures show that the differences in DF can be as high as 3–4 Hz between independent regions. These sites of maximum and minimum frequencies need further study with regards to VF maintenance, and potential focal treatment strategies. 4.2. Epicardial and endocardial differences Our study is unique in studying the epicardium and endocardium in human VF, which allowed us to compare them at identical time segments during VF. Comparing epicardium and endocardium as a whole, the epicardium seems to be less orga-
nized with higher DFR than endocardium. This finding is consistent with our previous in vivo epicardial-endocardial mapping studies in three humans, where we saw persistence of a defined number of rotors in the endocardium while the epicardium demonstrated disorganization at the same time [9]. Most small animal studies due to the nature of the preparation do not have endocardial mapping information, and other animal studies that have performed endocardial mapping studies have had few electrodes not allowing for global mapping and comparisons between the two mapping surfaces. Though the exact significance of endocardial organization is not explored in our study the potential that the sources may be endocardial with fibrillatory conduction to epicardium may allow for future studies with focal endocardial therapies. 4.3. LV endocardial regional difference In endocardium, the DFR feature did not show much differentiation between FW and SE; however, there were definite regions on the FW that activated much faster than other regions within the FW and SE. These regions seem to coincide to the posterior LV free wall adjacent to the posterior papillary muscle. Indeed Chen’s group has shown that the papillary muscle may help anchor rotors [25]. Our previous optical mapping study in humans has shown greater tendency for wavefronts to block in the septum than the free wall of the LV [16]. Indeed Huang et al. studying porcine VF in search of mother rotor found to their surprise greater conduction block in the septum [26]. Due to the fact that the endocardial freewall had regions with rapid organized activity in VF, it should be targeted for future detailed mapping strategies especially at the base of papillary muscles. 4.4. Myocardial substrate and DFR From our analysis, we were unable to explain our DF findings entirely on the basis of fixed anatomic myocardial substrate. The fact that we were unable to entirely explain regional organization on the basis of fixed anatomic substrate suggests that dynamic physiological factors may determine the regional frequency characteristics. This is further confirmed by the fact that the
970
K. Umapathy et al. / Medical Engineering & Physics 31 (2009) 964–970
frequency characteristics vary from episode to episode within the same heart in spite of the same induction method. To add even more emphasis, interestingly we observed at the same electrode location both DFmax and DFmin frequencies of a region for two different VF episodes. 5. Limitations The hearts were Langendorff perfused in a denervated (cut-off from nerve supply) environment. It is unknown if these factors influence activation patterns. Regional differences of perfusion pressure might play a role in the regional differences of activation rates. However this is unlikely as areas of low bipolar amplitude (infracted/scar) tissues did not explain the differences. In addition, four of the eight hearts studied had normal coronaries and there were no differences that were observed between the non-coronary and the coronary cardiomyopathies with regards to DFR. 6. Conclusions During human VF there are quantifiable regional differences in the DFR. Regional differences in fibrillation dynamics are not entirely explained by fixed anatomic substrate. This suggests that dynamic factors such as ion channel heterogeneity and timevarying physiological factors may play an important role and need to be evaluated during human VF. Acknowledgments This study was supported by the Canadian Institutes of Health Research Grant (NA 777687), Heart & Stroke Foundation Federation Fund, and Ryerson University. Conflict of interest statement None. References [1] Jalife J, Gray R, Chen J. Mechanisms of ventricular fibrillation; drifting scroll waves and phase singularities of electrical activation. In: Cardiac electrophysiology—from cell to bedside. 3rd ed.; 2000. p. 386–404. [2] Guyton A, Hall J. Cardiac arrhythmias and their electrocardiographic interpretation. In: Medical Physiology. 10 ed., 2000. p. 134–42. [3] Chen J, Mandapati R, Berenfeld O, Skanes A, Jalife J. High-frequency periodic sources underlie ventricular fibrillation in the isolated rabbit heart. Circulation Research 2000;86(1):86–93. [4] Samie F, Berenfeld O, Anumonwo J, Mironov S, Udassi S, Beaumont J, et al. Rectification of the background potassium current; a determinant of rotor dynamics in ventricular fibrillation. Circulation Research 2001;89(12):1216–23. [5] Gray R, Pertsov A, Jalife J. Spatial and temporal organization during cardiac fibrillation. Nature 1998;392(6671):75–8. [6] Nanthakumar K, Huang J, Rogers J, Johnson P, Newton J, Walcott G, et al. Regional differences in ventricular fibrillation in the open-chest porcine left ventricle. Circulation Research 2002;91(8):733–40.
[7] Pak H, Oh Y, Liu Y, Wu T, Karagueuzian H, Lin S, et al. Catheter ablation of ventricular fibrillation in rabbit ventricles treated with ˇ-blockers. Circulation 2003;108(25):3149–56. [8] Thomas S, Thiagalingam A, Wallace E, Kovoor P, Ross D. Organization of myocardial activation during ventricular fibrillation after myocardial infarction: evidence for sustained high-frequency sources. Circulation 2005;112(2): 157. [9] Everett T, Wilson E, Foreman S, Olgin IV J. Mechanisms of ventricular fibrillation in canine models of congestive heart failure and ischemia assessed by in vivo noncontact mapping. Circulation 2005;112(11):1532. [10] Zaitsev A, Berenfeld O, Mironov S, Jalife J, Pertsov A. Distribution of excitation frequencies on the epicardial and endocardial surfaces of fibrillating ventricular wall of the sheep heart. Circulation Research 2000;86(4):408–17. [11] Wu T, Ong J, Hwang C, Lee J, Fishbein M, Czer L, et al. Characteristics of wave fronts during ventricular fibrillation in human hearts with dilated cardiomyopathy: role of increased fibrosis in the generation of reentry. Journal of the American College of Cardiology 1998;32(1):187–96. [12] Nash M, Mourad A, Clayton R, Sutton P, Bradley C, Hayward M, et al. Evidence for multiple mechanisms in human ventricular fibrillation. Circulation 2006;114(6):536. [13] ten Tusscher K, Panfilov A. Alternans and spiral breakup in a human ventricular tissue model. American Journal of Physiology—Heart and Circulatory Physiology 2006;291(3):H1088. [14] Nanthakumar K, Walcott G, Melnick S, Rogers J, Kay M, Smith W, et al. Epicardial organization of human ventricular fibrillation. Heart Rhythm 2004;1(1): 14–23. [15] Masse S, Downar E, Chauhan V, Sevaptsidis E, Nanthakumar K. Ventricular fibrillation in myopathic human hearts: mechanistic insights from in vivo global endocardial and epicardial mapping. American Journal of Physiology—Heart and Circulatory Physiology 2007;292(6):H2589. [16] Nanthakumar K, Jalife J, Masse S, Downar E, Pop M, Asta J, et al. Optical mapping of Langendorff-perfused human hearts: establishing a model for the study of ventricular fibrillation in humans. American Journal of Physiology—Heart and Circulatory Physiology 2007;293(1):H875. [17] Sevaptsidis E, Masse S, Parson ID, Downar E, Kimber S. Simultaneous unipolar and bipolar recording of cardiac electrical activity. Pacing and Clinical Electrophysiology 1992;5:45–51. [18] Umapathy K, Masse S, Sevaptsidis E, Asta J, Krishnan S, Nanthakumar K. Spatiotemporal frequency analysis of ventricular fibrillation in explanted human hearts. IEEE Transactions on Biomedical Engineering 2009;56(2):328–35. [19] Soejima K, Stevenson W, Maisel W, Sapp J, Epstein L. Electrically unexcitable scar mapping based on pacing threshold for identification of the reentry circuit isthmus feasibility for guiding ventricular tachycardia ablation. Circulation 2002;106(13):1678–83. [20] Marchlinski F, Zado E, Dixit S, Gerstenfeld E, Callans D, Hsia H, et al. Electroanatomic substrate and outcome of catheter ablative therapy for ventricular tachycardia in setting of right ventricular cardiomyopathy. Circulation 2004;110(16):2293–8. [21] Oza S, Wilber D. Substrate-based endocardial ablation of post infarction ventricular tachycardia. Heart Rhythm 2006;3(5):607–9. [22] Verma A, Kilicaslan F, Schweikert R, Tomassoni G, Rossillo A, Marrouche N, et al. Short- and long-term success of substrate-based mapping and ablation of ventricular tachycardia in arrhythmogenic right ventricular dysplasia. Circulation 2005;111(24):3209–16. [23] Reddy V, Wrobleski D, Houghtaling C, Josephson M, Ruskin J. Combined epicardial and endocardial electroanatomic mapping in a porcine model of healed myocardial infarction. Circulation 2003;107(25):3236–42. [24] Rogers J, Huang J, Pedota R, Walker R, Smith W, Ideker R. Fibrillation is more complex in the left ventricle than in the right ventricle. Journal of Cardiovascular Electrophysiology 2000;11(12):1364–71. [25] Pak H, Kim Y, Lim H, Chou C, Miyauchi Y, Fang Y, et al. Role of the posterior papillary muscle and Purkinje potentials in the mechanism of ventricular fibrillation in open chest dogs and swine: effects of catheter ablation. Journal of Cardiovascular Electrophysiology 2006;17(7):777–83. [26] Huang J, Walcott G, Killingsworth C, Melnick S, Rogers J, Ideker R. Quantification of activation patterns during ventricular fibrillation in open-chest porcine left ventricle and septum. Heart Rhythm 2005;2(7):720–8.