biophysics of radiofrequency ablation using an irrigated ...mirotzni/pdf/biophysics_of... ·...

Journal of Interventional Cardiac Electrophysiology 5, 377–389, 2001

#2001 Kluwer Academic Publishers. Manufactured in The Netherlands.

Biophysics of Radiofrequency Ablation Usingan Irrigated Electrode

Deeptankar Demazumder1, Mark S. Mirotznik2,

and David Schwartzman1

1Electrophysiology Research Laboratory, Allegheny University of

the Health Sciences, Philadelphia, PA2Department of Electrical Engineering, Catholic University of

America, Washington, DC

Abstract. Background: Previous reports have pro-posed that prevention of electrode-endocardial inter-facial boiling is the key mechanism by whichradiofrequency application using an irrigated elec-trode yields a larger ablation lesion than a non-irri-gated electrode. It has been suggested that maximalmyocardial temperature is shifted deep into myocar-dium during irrigated ablation.

Purpose: To examine the biophysics of irrigatedablation by correlating electrode and myocardialtemperatures with ablation circuit impedance andlesion morphology, and to perform a comparison withnon-irrigated ablation modes. To assess the influence ofirrigant rate, composition, temperature and blood flowvelocity.

Methods:I. Ablation with and without electrode irrigation was

performed in vitro utilizing a whole blood-superfusedsystem. Electrode, electrode–endocardial interface, andintramyocardial temperatures were assessed, as wereablation circuit impedance, total delivered energy, andlesion and electrode morphology. Irrigants assessedwere room temperature normal saline, iced normalsaline, and dextrose. Irrigant flow rates assessed were20 and 100 cc=min. Blood flow velocities assessed were 0and 0.26 m=s.

II. Finite element simulations of myocardialtemperature during irrigated ablation were performedto further elucidate irrigation biophysics and provide amore detailed myocardial temperature profile. Twomodels were constructed, each utilizing a differentcore assumption regarding the electrode-tissue bound-ary: 1. electrode temperature measured in vitro; 2. inter-facial temperature measured in vitro. Intramyocardialtemperatures predicted by each model were correlatedwith corresponding temperatures measured in vitro.

Results:I. Ablation during electrode irrigation with normal

saline was associated with greater ablation energydeposition and larger lesion dimensions than non-irri-gated ablation. The mechanism underlying the largerlesion was delay or inhibition of impedance rise; thiswas associated with attenuation or prevention of elec-trode coagulum. Irrigation did not prevent interfacialboiling, which occurred during uninterrupted radio-frequency energy deposition and lesion growth. Irriga-tion using saline at 100 cc=min was associated with noimpedance rise regardless of blood flow velocity,

whereas during irrigation at 20 cc=min impedancerise was blood flow rate-dependent. Iced salineproduced results equivalent to room temperaturesaline. Irrigation with dextrose was associated withcurtailed energy application and relatively smalllesions.

II. The finite element simulation that used elec-trode–endocardial interfacial temperature as the coreassumption predicted a myocardial temperatureprofile which correlated significantly better with invitro than did the simulation which used electrodetemperature as the core assumption. Regardless ofirrigant and blood flow rates, maximal myocardialtemperature was always within 1 mm of the endocar-dial surface.

Conclusions: Radiofrequency energy applicationvia a saline irrigated electrode resulted in a largerlesion due to attenuation or eradication of electrodecoagulum, thus preventing an impedance rise. Irriga-tion did not prevent interfacial boiling, but boiling didnot prevent lesion growth. The site of maximal myocar-dial temperature during irrigated ablation was rela-tively superficial, always within 1 mm of theendocardial surface. Irrigation with iced saline wasno more effective than with room temperature saline;both were far more effective than dextrose. Higherirrigation rates immunized the electrode from theinfluence of blood flow. The biophysical effects ofblood flow and irrigation were similar.

Key Words. catheter ablation, radio frequency

377

Data presented in part at the 18th annual scientific sessions ofthe North American Society of Pacing and Electrophysiology,New Orleans, LA, 1997Supported in part by a grant from Biosense Webster, Inc.,Diamond Bar, CA

Address for correspondence: David Schwartzman, MD, AtrialArrhythmia Center, University of Pittsburgh PresbyterianHospital B535, 200 Lothrop Street, Pittsburgh, PA 15213-2582.E-mail: [email protected]

Received 31 October 2000; accepted 24 April 2001

Previous investigators have documented thatradio frequency energy (RF) applied to myocar-dium via an irrigated electrode yields deeper andmore voluminous ablation lesions than with anon-irrigated electrode [1–5]. This is associatedwith a higher maximum applied RF power, unen-cumbered by impedance rise. It has beenconcluded that the mechanism of prevention ofimpedance rise by electrode irrigation is coolingof the electrode–endocardial interface, prevent-ing boiling [2,5]. It has been suggested thatirrigation displaces the maximum temperaturepoint deep (more than 2 mm) into the myocar-dium [5]. There have been no detailed reports ofmyocardial temperature profiles during irrigatedablation. Herein, we present the results of aninvestigation detailing the biophysics of ‘‘irri-gated’’ ablation. Both in vitro and finite elementmodeling techniques were utilized. For the invitro experiments, we utilized a whole bloodsuperfused system designed to mimic the intra-cardiac environment. Electrode–endocardialinterface (interfacial) and intramyocardial(intramural) temperatures were measured andcorrelated with ablation circuit impedance andlesion=electrode morphology. Continuous echo-cardiographic monitoring of the interface wasalso performed and correlated with these find-ings. Irrigated ablation was compared to non-irrigated ablation modes, including tempera-ture-guided and constant power. The influenceof irrigation variables, including fluid (irrigant)composition, temperature and flow rate, as wellas blood flow velocity were examined. Finiteelement simulation of myocardial temperatureduring irrigated ablation was performed toextend the in vitro data and provide a moredetailed myocardial temperature profile.

Methods

I. In VitroAblation Electrode The electrode was 2.3 mm indiameter and 5 mm in length incorporating athermocouple located 2.5 mm from the distalend of the electrode, which provided a measure-ment accuracy of � 2�C (Biosense Webster, Inc.,Diamond Bar, CA, USA). The internal design ofthe electrode permitted the irrigant to travelthrough a center lumen and exit via each of 6irrigation holes, each with a diameter of 0.4 mmlocated radially around the electrode 1 mm fromthe distal tip (‘‘showerhead’’ design). The rate ofelectrode irrigation was regulated by a program-mable motorized pump (Masterflex1, ColeParmer Inc., Vernon Hills, Ill., USA), which deliv-ered the irrigant at a constant volume (cc=min).Two irrigation rates were assessed: 20 and100 cc=min. At an irrigation rate of 20 cc=min,

the maximum irrigant flow velocity (measuredusing Doppler echocardiography at the electrodesurface) was 0.44 m=s. At an irrigation rate of100 cc=min, the maximum velocity was 2.2 m=s.

Ablation System (Fig. 1) This protocol wasapproved by the Institutional Animal Care andUse Committee of the Allegheny University of theHealth Sciences, and was in compliance with theGuide for Care and Use of Laboratory Animalspublished by the National Institutes of Health,publication number 78-83. Bovine myocardiumfrom freshly killed animals was mounted on anisolated ground plate at the base of a tank circu-lating bovine heparinized whole blood (10–12liters) at 37�C. Thermocouples (T-type; 0.3 mmdiameter; measurement accuracy � 2�C; OmegaEngineering Inc., Stamford, CT, USA) wereinserted at depths of 0 (electrode–endocardialinterface; T0), 1 (T1), 3 (T3), 5 (T5) and 7(T7) mm. Each thermocouple was electricallyinsulated and aligned so that its leads wereperpendicular to the applied RF field so as tominimize non-thermal artifacts and prevent fluc-tuations of averaged temperatures >0.5�C root-mean-square. The electrode was mounted on aforce-transducer (AccuForce, Acme Scale Inc.,San Leandro, CA, USA) and locked into positionsuch that it was aligned perpendicular to thetissue, directly above the thermocouples. Approxi-mately 20% of the electrode surface area was incontact with endocardium: this required slightvariations in the applied force (approximately0.1 Newton). Blood flow was directed to the elec-trode–endocardial interface using a pulsatile flowpump (Masterflex, Cole Parmer Inc., VernonHills, Ill., USA) with a pulse rate of 60=min.When activated, the pump delivered blood so asto achieve a peak velocity of 0.26 m=s and a meanvelocity of 0.20 m=s at the electrode–endocardialinterface on the side of the electrode adjacent tothe blood flow source. This is consistent withvelocities commonly measured along the endocar-dial surface in man [6]. When the pump wasinactivated, interfacial flow velocity was negligi-ble. The interface was imaged using a multiplanetransesophageal echocardiographic probemounted into the tank (ATL, Bothell, WA, USA).Two-dimensional imaging was performed at afrequency of 5 MHz. A time calibration wasemployed to permit synchronization of echocar-diography and temperature=impedance measure-ments. Radiofrequency energy was appliedcontinuously in unipolar fashion between theablation electrode and the ground plate utilizinga commercial generator (EP Technologies, Sunny-vale, CA, USA) operating at approximately550 KHz. During each RF application, power,

378 Demazumder, Mirotznik and Schwartzman

voltage, current, ablation circuit impedance andelectrode=myocardial temperatures weremeasured continuously with sampling at 200 Hzusing a commercial data acquisition system(Superscope II, GW Instruments Inc, Somerville,MA, USA). Data were digitized and stored.

Lesion=Electrode Morphology After each RFapplication, the electrode was examined forcoagulum [2,4,5,7]. The endocardial surface wasinspected for disruption (‘‘cratering’’) [2,4,5,7,8]and coagulum. The myocardium was thensectioned coronally and stained with 1% triphe-nyl tetrazolium chloride. This dye stains intra-cellular dehydrogenase, which distinguishesviable and necrotic tissue. Necrosis was observedwhen myocardial temperature exceeded 55�C forat least 10 seconds. The following lesion dimen-sions were measured using a caliper: maximumdepth, maximum width, endocardial surfacediameter, and depth at maximum width (Fig. 2).Lesion volume was calculated by assuming thelesion shape was an oblate ellipsoid and subtract-ing the estimated volume extending above thesurface of the tissue [5].

Experimental Protocol Several different experi-mental groups were established (Table 1). InGroups 1 and 2, the electrode was not irrigatedand RF power titration was guided by interfacialtemperature. In Groups 3 and 4, the electrode

was not irrigated and RF power was fixed at 50watts. In groups 5 through 10, the electrode wasirrigated and RF power was fixed at 50 watts. Ineach group, RF was applied for 60 seconds unlessinterrupted by a significant rise in ablationcircuit impedance, defined as �10 ohms.

Analytical Methods Data are expressed asmean� standard deviation, unless otherwisespecified. The biophysical parameters (power,impedance, current, voltage), temperatures, andlesion dimensions were compared within=betw-between groups using chi-square or student’s t-test. Any significant differences were furtherevaluated using Scheffe’s method for pairwisecomparisons. A P value of < 0.05 was consideredstatistically significant.

The rate of myocardial heating was estimatedby fitting a given temperature–time profile to thefollowing 1-exponential function: T(t)¼TSS þ

(TINITIAL7TSS) � exp(7t=t), where TSS was thesteady-state temperature, TINITIAL was the base-line temperature before RF application, and t wasthe exponential time-constant; (TSS7TINITIAL)=trepresented the linear component of the 1-expo-nential fit or the initial rate of heating (8; �C=s).

II. Finite Element ModelingModeling Methodology Finite element models(FEM) designed to mimic each of the in vitro

Fig. 1. Schematic of in vitro ablation system.

Biophysics of Radiofrequency Ablation Using an Irrigated Electrode 379

experimental groups were constructed utilizing acommercial software program (COSMOS1,Structural Research and Analysis Inc., LosAngeles, CA, USA) [9]. Assumptions utilized forelectrical and thermal material properties of theablation electrode, irrigant, myocardium andblood have been validated and reportedpreviously [9,10]. The model was designed topredict myocardial temperature distributionsutilizing a two step approach. In step 1, thedistribution of electric currents produced byapplication of RF at a fixed power was deter-mined by solving Laplace’s equation. Heatdeposition via a resistive mechanism was then

calculated from the predicted current distribu-tion. In step 2, the temperature distributionwithin the myocardium was calculated by solvingthe bioheat transfer equation using the finiteelement method. For these calculations themyocardium and blood temperatures were initi-ally assumed to be 37�C. Myocardial cooling dueto blood flow and irrigation was modeled byimposing a convective boundary condition alongthe electrode-tissue (blood, endocardium)surface. Each model was constructed to reportthe temperature profile predicted to occur at themean duration of RF application observed duringablation in the corresponding in vitro experimen-

Fig. 2. Schematic of ablation lesion, depicted as the light area with viable myocardium dark. MW¼maximum width; SD¼ surfacediameter; DMW¼depth at which lesion width was maximal; MD¼maximal depth.

Table 1. Experimental groups

Group 1 2 3 4 5 6 7 8 9 10

Number of trials (n) 6 6 5 6 6 6 6 6 6 5T0 target (�C) 80 80 – – – – – – – –Irrigant type – – – – RTNS RTNS RTNS RTNS INS DEXIrrigation rate (cc=min) – – – – 20 20 100 100 20 20Blood flow velocity (m=s) 0 0.26 0 0.26 0 0.26 0 0.26 0 0Delivered power (Watts) 10.6�1.0 16.3�1.4 50 50 50 50 50 50 50 50

T0¼ electrode–endocardial interfacial temperature; RTNS¼ room temperature normal saline; INS¼ iced normal saline;DEX¼dextrose.


tal group. Two finite element simulations wereconducted:

FEM 1: for this simulation, electrode tempera-ture was held fixed at the maximum TE whichhad been measured in vitro.

FEM 2: for this simulation, electrode tempera-ture was held fixed at the maximum T0 whichhad been measured in vitro.

Analytical Methods The consistency of the rela-tionship between myocardial temperatures gener-ated by the FEM simulation and those measuredin vitro was assessed using a weighted chi-square method. The reported value representsthe ‘‘goodness of fit’’ between these data sets:

A2¼X FEM � data

SDffiffiffin

p

2664

3775

2

where n¼number of trials, SD¼ standard devia-tion of in vitro data, FEM¼ data from finiteelement simulation, and data¼data mea-sured in vitro. Therefore, a lower chi-squarevalue denotes a better fit. Differences betweenchi-square values were assessed for statisticalsignificance using a variance ratio test. Signifi-cance was defined as a P value < 0.05.

Results

During RF applications in each of the variablepower, T0-guided groups (Groups 1 and 2), T0 wassignificantly greater than all intramural T (Fig.3; Table 2). There were no significant differencesbetween the groups in the initial rates of heating,intramural T or lesion dimensions (Tables 2, 3).In both groups, TE was significantly lower thanT0; the magnitude of the difference was signifi-cantly greater in Group 2 than in Group 1.Impedance rise was not observed, nor was endo-cardial coagulum or disruption, electrode coagu-lum, or echocardiographic bubbling. In bothgroups, the maximal lesion width was at theendocardial surface. The chi-square value forFEM 2 was significantly lower than for FEM 1for Group 1; for Group 2 the comparison hadborderline significance (Fig. 4).

During RF applications in each of the fixed-power, non-irrigated groups (Groups 3 and 4), T0

rapidly approached 100�C, paralleled by TE (Fig.5). In Group 3, as both temperatures approached100�C, two echocardiographic observations weremade consistently: 1. the apparent electrodediameter increased abruptly; 2. there was thesudden appearance of bubbles emanating fromthe interfacial region, associated with an audible

‘‘pop’’ and the initiation of a marked impedancerise, eliciting RF shutoff (Fig. 6). In Group 4, thesudden appearance of bubbles occurred as inGroup 3, but there was a slower increase in theapparent electrode diameter. This was associatedwith a slower initial rate of heating at TE, delayin the timing of impedance rise, permittinggreater total energy deposition, intramuraltemperatures, and lesion dimensions. The initialrates of heating at T0 were not significantlydifferent between Groups 3 and 4. Electrodeand endocardial coagulum were observed consis-tently. The maximal lesion width was at or nearthe endocardial surface. The chi-square valuesfor FEM 1 and FEM 2 were not significantlydifferent in either group (Fig. 4).

During RF application in each of the 20 cc=minroom temperature saline irrigated groups(Groups 5 and 6), T0 rapidly approached 100�C.In Group 5, the initial rate of heating at TE wassimilar to that at T0, but the maximum tempera-ture was significantly lower than T0. In Group 6,the initial rate of heating and maximumtemperature at TE were significantly lower thanthose at T0 (Fig. 7). For both groups, as T0 neared100�C, there was the sudden appearance of echo-cardiographic bubbles without a significantchange in the electrode diameter. This was asso-ciated with an audible pop and the initiation of animpedance rise which was not significant (<10ohms) and which therefore did not elicit RF shut-off. Continued RF application was associatedwith progressive increase in intramural tempera-tures. In some trials, during continued RF appli-cation there was a second sudden increase inbubble density at the interface, usually asso-ciated with another audible pop. In each case,this was correlated with intramural temperaturereaching 100�C. In Group 5, a gradual, signifi-cant increase in the apparent electrode diameterwas observed in all trials. In concert with this,the impedance gradually rose. In 3 of 6 experi-ments, it eventually exceeded 10 ohms, elicitingpremature RF shutoff. In Group 6, there was nosignificant change in the apparent electrodediameter throughout the RF application period.Impedance was stable after the early minor riseand RF was delivered for the planned duration inall experiments. Delay or elimination of a signifi-cant impedance rise in Groups 5 and 6 permittedgreater total RF energy delivery than in Groups 3and 4, which was associated with higher intra-mural temperatures and deeper and more volu-minous lesions. Endocardial coagulum wasobserved after most applications. Electrodecoagulum was observed consistently in Group 5but never in Group 6. The depth at which lesionwidth was maximal was significantly greater inGroups 5 and 6 than in Groups 3 and 4. For both


Groups 5 and 6, the chi-square value for FEM 2was significantly lower than for FEM 1.

During RF application in each of the100 cc=min room temperature saline irrigatedgroups (Groups 7 and 8), T0 rapidly approached100�C; the initial rate of heating at TE wassimilar, but the maximum temperature wassignificantly lower than that at T0. As T0

neared 100�, the same echocardiographic obser-vations and impedance pattern as described inGroup 6 ensued. No significant impedance risewas observed. Endocardial coagulum wasobserved in all Group 7 trials and in half ofGroup 8 trials. Electrode coagulum was notobserved in either group. Lesion depths andvolumes were significantly greater than inGroups 3 or 4, and lesion widths and volumeswere significantly greater than in Groups 5 and6. For both groups, the chi-square value for FEM2 was significantly lower than for FEM 1.

During RF application in the 20 cc=min icedsaline irrigated group (Group 9), T0 rapidlyapproached 100�C; the initial rate of heating atTE was greater than that at T0, but the maximumtemperature was significantly lower. As T0

neared 100�C, the same echocardiographic obser-vations and impedance pattern as described inGroup 5 ensued. In all trials, RF duration wascurtailed because of impedance rise. Lesiondimensions were similar to those in Group 5.The chi-square for FEM 2 was significantlylower than for FEM 1.

During RF application in the 20 cc=min roomtemperature dextrose irrigated group (Group 10),T0 rapidly approached 100�C, paralleled by TE.The same echocardiographic observations andimpedance pattern as described for Group 4ensued. Electrode and endocardial coagulumwere observed consistently. Lesion dimensionswere not significantly different from Group 4.

Fig. 4. Finite element simulation results, separated byexperimental group (labeled on top of each graph): FEM 2(assuming T0 predicts intramyocardial temperatures) in red;FEM 1 (assuming TE predicts intramyocardial temperatures)in green. The data points (mean temperature) with error bars(standard deviation) represent in vitro data. The chi-squarevalue, quantifying the agreement between FEM and in vitrodata, is shown in the corresponding color. A variance ratio testwas performed between the chi-square values of FEM 1 andFEM 2; the resulting P value indicates whether FEM 2 agreedwith the in vitro data significantly better than FEM 1.

Fig. 3. Typical trial from experimental Group 2: noirrigation, blood flow velocity¼0.26 m=s, RF titrated to achieveT0¼80�C.


The chi-square values for FEM 1 and FEM 2 werenot significantly different in Group 10.

Discussion

How Does Saline Irrigation Producea Bigger Lesion?In the present study, a key finding was that salineirrigated RF applications resulted in largerlesions by delaying or eliminating impedancerise despite electrode–endocardial interfacialboiling, which was evidenced by direct measure-ment of temperature, intra-ablation echocardio-graphic bubbling, and post-ablation endocardialcoagulum. RF application uninterrupted by inter-facial boiling was associated with higher intra-mural temperatures and deeper and morevoluminous lesions. Electrode coagulum wasconsistently associated with impedance rise; irri-gation eliminated or delayed its formation.

Previous reports in vitro and in vivo havedemonstrated larger lesions using saline irri-gated versus non-irrigated ablation electrodes,associated as in the present study with no ordelayed impedance rise, higher RF energy deposi-tion, and attenuation or elimination of electrode

coagulum [1–5]. In all but two reports [5,11], itwas assumed that impedance rise was preventedby inhibition of electrode–endocardial interfacialboiling, but myocardial temperature measure-ment was either not performed or was performedat a distance from the interface. Wong et al [11]utilized an ablation electrode which had an inter-nal irrigation design. Temperatures weremeasured at the interface and at depths of 1

Fig. 6. A. Echocardiographic image prior to initiation of RFapplication demonstrating myocardium (M), superfusingblood (B), and ablation electrode (E). The dotted linedemarcates a portion of the electrode–endocardial interface.Scale is 1 centimeter. B. Echocardiographic image during RFapplication, at the moment of interfacial bubbling. Theinterfacial bubbles are difficult to appreciate in a static image,except that they are echodense. A ‘‘plume’’ of bubbles (arrows)can be seen escaping to either side of the electrode–endocardialinterface. Scale is 1 centimeter. C. Echocardiographic imageafter RF application, after resolution of all bubbling. Asignificant widening of the electrode was observed. Scale is 1centimeter.

Fig. 5. Typical trial from experimental Group 3: noirrigation, no blood flow, RF power was fixed at 50 Watts.


and 2 mm. During irrigation with room tempera-ture saline at 30 cc=min, RF application at a fixedpower of only 20 watts resulted in maximumtemperature at the interface. No boiling or impe-dance rises were observed. In regard to tissuetemperature profile, these data are consistentwith the data reported herein. Nakagawa et al[5], utilizing a canine thigh muscle preparation,employed the same electrode, irrigant, and irri-

gation rate as in the present study. A constant RFvoltage of 66 volts (approximately 50 watts) wasapplied in unipolar fashion. With the possibleexception of blood flow velocity and volume (notreported by the investigators), these variables arevery similar to our Group 6. Intramyocardialtemperatures were measured only at depths of3.5 and 7.0 mm. In a subset of applications,temperature was measured at the electrode-thigh muscle interface. No echocardiographicimaging was performed. Maximum electrodetemperature (38.4� 5.1�C) was significantlylower than in Group 6 (49� 2�C), but the magni-tude of the difference was small. The intramuraltemperatures were very similar. However, inmarked contrast to our findings, the peak inter-facial temperature was approximately 69�C, anddid not approach 100�C during any application.No information on endocardial coagulum wasreported. Based on these data, the authorsconcluded that prevention of interfacial boiling

Table 2. Temperature and rate of heating data

Group 1 2 3 4 5 6 7 8 9 10

TE 73�2 53�2 114�7 103� 6 71� 5 49�2 48�2 45� 4 78�7 101�13kE 7.9� 1 12�3 32�2 25�3 29� 1 22�2 20�3 18� 7 27�4 32� 2T0 81�1 81�1 114�3 121� 5 98� 7 100� 2 100�4 101�10 94�12 102�8k0 8.2� 1 11�2 30�2 33�6 31� 3 34�3 26�7 26� 3 19�2 28� 4T1 67�4 69�3 65�12 77�11 98� 6 99�6 98�4 94� 4 98�2 98� 2k1 2.3� 0.5 1.9� 0.4 7.4� 2 8.4� 2 6.5�3 9.9� 4 7.3�3 7.6�2 8.5� 2 11�1T3 57�6 61�3 46�3 57�7 85� 11 92�3 92�4 90� 6 88�5 72� 7k3 1.1� 0.5 0.84�0.2 2.9� 1 2.5� 1 2.2�1 3.2� 1 3.3�1 2.7�0.6 3.0� 0.8 3.6�0.5T5 50�4 51�3 39�1 47�3 73� 12 81�5 80�3 76� 9 70�6 52� 3k5 0.31�0.1 0.35�0.04 1.3� 1 1.2� 0.5 1.3�0.6 1.4� 0.4 1.1�0.2 1.2�0.4 1.0� 0.2 1.5�0.4T7 40�2 40�2 38�1 39�3 54� 11 67�6 62�6 62� 5 56�6 46� 3k7 0.15�0.05 0.13�0.03 0.0� 0.0 0.0� 0.0 0.49� 0.2 0.69�0.1 0.61� 0.2 0.58�0.1 0.64�0.1 0.84� 0.2

Maximum electrode (E), interface (0), and intramural (1, 3, 5, and 7 mm depth) temperatures (T) and initial rates of heating (8),separated by experimental groups (columns). Shaded columns refer to groups with blood flow velocity¼ 0.26 m=s. Multiplestatistical comparisons were performed within and between groups (see text).

b

Fig. 7. Typical trial from experimental Group 6: irrigationwith room temperature normal saline at 20 cc=min, blood flowvelocity¼ 0.26 m=s, RF power was fixed at 50 Watts. Note thatT0 rises rapidly to boiling, at which point it abruptly deflectsdownward. This deflection was coincident with the suddenelaboration of echocardiographic bubbles at the electrode–endocardial interface and an audible ‘‘pop.’’ Trials stopped atthis instant (data not shown) showed displacement of theinterface thermocouple to a site adjacent to the interface. Aftera few seconds, T0 resumes its rise, but at a much slower rate.Note that at the moment that the downward deflection in T0

occurs, despite continued (constant) power application there isalso a downward deflection in TE, and an insignificant (<10ohm) rise in circuit impedance. We hypothesize that the drop inTE was due to temporary displacement of the electrode fromendocardial contact by the elaboration of bubbles, increasingits exposure to blood cooling.


was a key mechanism by which irrigation per-mitted increased RF energy deposition, anddeeper and more voluminous lesions. The discre-pancy in measured interfacial temperature isobviously crucial, and we cannot offer a definitiveexplanation. The difference in electrode tempera-ture, also difficult to explain, may have played arole, but we expect that if so it was very small anddoes not explain the marked difference in inter-facial temperatures. Nakagawa et al demonstra-ted the time–temperature relationship during anirrigated ablation trial, during which a suddendrop in interfacial temperature after an earlyrapid rise is apparent (Fig. 8). We observed asimilar phenomenon in some trials (Fig. 7), butalways as the interface temperature wasapproaching 100�C. The phenomenon was dueto physical displacement of the interface thermo-couple by the sudden release of gas bubbles. Oncedisplaced, the interface thermocouple reportedtemperature near the interface, which was con-siderably lower than peak interfacial tempera-ture. This phenomenon was also observed inanother in vitro study [4]. It is possible, therefore,that the interfacial temperature values reportedin Nakagawa et al were spuriously low.

Insights from Finite ElementModel SimulationsFEM has been validated in previous reports toestimate intramyocardial temperature distribu-tions [9,10]. In the present study, FEM was usedto gain further insight into the intramyocardialheating profile. For each group, we compared thetemperatures measured in vitro with FEM simu-lations performed utilizing one of two differentassumptions: 1. the electrode temperaturepredicts intramyocardial temperatures (FEM 1);2. the electrode–endocardial interfacial tempera-ture predicts intramyocardial temperatures(FEM 2). For most of the experimental groups,the intramural temperatures simulated withFEM 1 correlated poorly with the measured invitro temperatures (Fig. 4). A markedly improvedagreement was observed using simulations basedon the interfacial temperature. In all saline irri-gated groups, the chi-square value for FEM 2 wassignificantly lower than for FEM 1; in most of theremaining groups, there was no significant differ-ence between the values. We conclude from thesefindings that, at least in the saline irrigationgroups, internal electrode temperature markedlyunderestimated outer electrode surface (at theendocardial interface) temperature. We speculatethat the reason for this was cooling of the elec-trode mass by flow, exposing temperature hetero-geneity in various regions of the electrode. WeT

ab

le3.

Bio

ph

ysic

al

da

ta

Gro

up

12

34

56

78

910

Base

lin

eim

ped

an

ce(O

)78�

378�

283�

480�

384�

781�

188�

878�

269�

382�

1M

ean

vol

tage

(V)

27�

134�

265�

858�

160�

558�

158�

159�

156�

363�

1M

ean

curr

ent

(A)

0.4�

0.0

07

0.5�

0.0

07

1.0�

0.0

30.9�

0.0

03

0.9�

0.0

20.9�

0.0

03

0.9�

0.0

03

0.9�

0.0

03

1.0�

0.0

30.9�

0.0

2R

Fd

ura

tion

(s)

60�

060�

07�

515�

249�

12

60�

060�

060�

050�

10

16�

3T

otal

del

iver

eden

ergy

(J)

600�

58

941�

77

345�

105

746�

123

2473�

585

3000�

03000�

03000�

02485�

503

810�

160

Au

dib

lep

op(n

)0

04

66

66

66

5S

ign

ifica

nt

(>10O

)im

ped

an

ceri

se(n

)0

05

66

00

06

5

Ech

ocard

iogra

ph

icex

plo

sion

(n)

00

56

66

66

65

Ele

ctro

de

coagu

lum

(n)

00

56

60

00

65

En

doc

ard

ial

coagu

lum

(n)

00

36

46

63

45

En

doc

ard

ial

dis

rup

tion

(n)

00

00

10

03

31

Les

ion

dep

th(m

m)

3.9�

0.2

4.1�

0.2

2.6�

0.2

4.2�

0.3

7.0�

1.3

7.6�

0.8

7.3�

0.5

8.0�

0.6

6.8�

0.9

4.4�

0.4

Les

ion

wid

th(m

m)

6.5�

0.4

6.0�

0.0

6.8�

0.8

9.0�

0.0

10�

0.0

10�

0.0

11�

0.0

11�

0.0

10�

18.5�

0.5

Dep

that

max

wid

th(m

m)

0.0�

0.0

0.1�

0.2

0.0�

0.0

0.8�

0.3

2.4�

1.5

2.8�

0.4

3.1�

0.5

2.8�

0.4

2.6�

0.5

1.3�

0.3

Su

rface

dia

met

er(m

m)

5.5�

0.4

5.1�

0.2

5.8�

0.8

6.1�

0.9

5.6�

1.7

6.8�

1.3

4.5�

0.6

5.4�

0.7

6.5�

0.9

5.6�

0.7

Les

ion

vol

um

e(m

m3)

75�

12

65�

0.5

54�

14

141�

11251�

10

263�

36

298�

28

355�

49

273�

115

122�

31

Sh

ad

edco

lum

ns

refe

rto

gro

up

sw

ith

blo

odfl

owvel

ocit

y¼

0.2

6m=s

.


and others have previously characterized thisphenomenon [12–14].

Given the limited spatial resolution of tissuethermometry, precise information regarding sitesof temperature maxima and gradients could notbe culled from in vitro measurements. The FEMsimulations provided far better spatial resolu-tion. Based on FEM, and as previously reported:1. during RF applications using non-irrigatedelectrodes the site of maximum temperaturewas the electrode–endocardial interface; 2.during RF applications using a saline irrigatedelectrode the site of maximum temperature wasnot the interface. However, for all irrigatedmodes, this site was within 1 mm of the interface(Fig. 4). This data varies with the assumptions ofprevious investigators. For example, Nakagawaet al estimated that the maximal temperaturewas located ‘‘several millimeters’’ beneath theinterface [5]. However, these investigators actu-ally measured intramural temperature only atdepths of 3.5 and 7 mm.

FEM also provided insights regarding intra-mural temperature gradients during RF applica-tions. During variable power, interfacialtemperature-guided, non-irrigated ablation(Groups 1 and 2), the mean gradient at a depthof 0–4 mm was 12�C=mm (no difference betweengroups). During fixed power, non-irrigated abla-tion (Groups 3 and 4), a larger gradient (49

�C=mm) was observed (no difference betweengroups). During saline irrigated ablation (Groups5–9), a gradient (8�C=mm) smaller than Groups1–4 was observed (no differences among Groups5–9). Despite the equivalent RF duration forGroups 1, 2 and 5–9, the smaller gradient in thelatter group was due to the greater cumulativeenergy deposition, yielding higher and moreuniform temperatures. Despite the equivalencein RF power between Groups 3, 4 and 5–9, thelarger gradient in the former group was due tothe curtailed RF duration, markedly limitingcumulative energy deposition.

Influence of Irrigant Flow Rate,Temperature, and CompositionWe evaluated the influence of irrigant flow rate inthe absence of blood flow, utilizing saline irriga-tion at 20 cc=min (Group 5) and 100 cc=min(Group 7). As expected, the higher irrigationrate was associated with a significantly lowerelectrode temperature and initial rate of elec-trode heating. However, maximum interfacialtemperatures and initial rates of interfacial heat-ing were not significantly different. The elimina-tion of electrode coagulum and impedance rise inGroup 7 was associated with a longer RF dura-tion, greater cumulative energy deposition,higher intramural temperatures, and largerlesion widths and volumes relative to Group 5.

Fig. 8. Figure 10 from Nakagawa et al [5] demonstrating an irrigated RF application using parameters almost identical to thoseused in Figure 7 (Group 6). Note that, despite continued (constant) RF application, there are sudden downward deflections inboth ‘‘interface temp’’ and electrode temp (arrow), followed by a more gradual rise. The electrode temperature pattern is virtuallyidentical to that in Figure 7. The interfacial temperature pattern is similar in pattern to but different in scale from that in Figure 7.No impedance tracing was supplied with this figure. Reproduced with permission.


We evaluated the influence of irriganttemperature in the absence of blood flow, utilizingroom temperature (Group 5) and iced (Group 9)saline. There were no significant differences inthe incidence of electrode coagulum or impedancerise, intramural temperatures or lesion dimen-sions. Maximal interfacial temperatures were notsignificantly different, but the initial rate ofinterfacial heating was markedly lower inGroup 9. The latter observation suggests thatthere was superior interfacial cooling by icedsaline irrigation but that the magnitude of theeffect was relatively small.

We evaluated the influence of irrigant composi-tion in the absence of blood flow, utilizing 0.9%saline (Group 5) and dextrose (Group 10). In Group10, impedance rise severely curtailed RF duration,which was associated with lower intramuraltemperatures and smaller lesions relative toGroup 5. Electrode temperature was significantlyhigher in Group 10. The behavior of this group wasvery similar to Group 4. We hypothesize thatdextrose, a non-conductive fluid, acted as an elec-trical insulator. By diminishing conductivity at theelectrode–blood interface, it effectively diminishedthe electrode surface area, resulting in a highelectrode temperature and coagulum formation.Consistent with this, Shake et al reported a directrelationship between lesion dimensions and irri-gant electrolyte concentration [15].

Influence of Blood Flow VelocityAt saline irrigation rates of 20 cc=min (Groups 5and 6) and 100 cc=min (Groups 7 and 8), weevaluated the influence of blood flow velocity.During irrigation at 20 cc=min, the introductionof blood flow was associated with lower electrodetemperature, lower initial rate of electrode heat-ing, eradication of electrode coagulum, longer RFduration, greater total delivered energy andhigher intramural temperatures. The data fromGroups 6 and 7 are very similar, suggesting thatblood flow and irrigation have the same biophy-sical effect. In contrast, during irrigation at100 cc=min, the introduction of blood flow hadlittle effect on biophysical parameters, intra-mural temperatures, or lesion dimensions. Wehypothesize that, at this rate of irrigation, bloodwas effectively excluded from the electrode,rendering it without influence on events occur-ring at the electrode surface.

Lesion MorphologyConsistent with previous reports, we observedthat the maximum width of an irrigated lesionwas well beneath the endocardial surface (Table 3)[2–5]. This was in contrast to non-irrigated abla-tion, in which the maximum width was at or very

close to the endocardial surface. Displacement ofmaximum lesion depth during irrigated ablationhas been used as evidence that the maximumintramural temperature is similarly displaced,which we have demonstrated not to be the case.The depth at which lesion width is maximal is notsimply a manifestation of the heating patterndirectly below the electrode. Radial myocardialtemperature is the result of competition betweenthermal conduction and convection. Despite thehigh temperatures which we have demonstratedat the electrode–endocardial interface, it is likelythat there was a rapid radial decline in tempera-ture. This may have been exaggerated for theshowerhead irrigation electrode design testedherein, given that the main direction of irrigantflow was radial to the electrode [16]. Low surfacetemperature would act as a heat sink, preventingradial myocardium at and adjacent to the endo-cardial surface from reaching ablative tempera-ture. Deeper regions would be relatively resistantto this effect.

As in other reports, we observed the phenom-enon of endocardial disruption (‘‘lesion cratering’’)in several of the irrigated lesions [2–4]. Thisphenomenon has also been used as evidence thatthe maximum intramural temperature isdisplaced deep into myocardium. In the presentstudy, cratering was seen after trials during whichintramural temperature reached 100�C, asso-ciated with a second sudden increase in echocar-diographic bubble density at the electrode–endocardial interface and a second audible pop.The endocardial aspect of the lesion demonstrateda small tear, presumably the area where steamwas suddenly released from a site of subendocar-dial accumulation. Note that despite multipletrials with identical RF parameters, crateringwas not seen consistently. Even though intra-mural temperatures commonly approached theboiling point, if the second sudden increase inechocardiographic bubble density at the elec-trode–endocardial interface was not observedthen cratering was not observed. The variabilityin the incidence of lesion cratering could have beendue to small differences in electrode–endocardialcontact, the rate of intramural temperature rise,and=or in the arrangement of the myocytes andconnective tissue in the heating zone.

Echocardiographic ObservationsThis is the first report in which measured elec-trode and myocardial temperatures, impedanceand echocardiographic observations have beensynchronized. Two main echocardiographic obser-vations were made which increased insight intothe biophysics of irrigated ablation. First, wedemonstrated the phenomenon of interfacial gas


bubble liberation during RF application. As theelectrode–endocardial interface approached boil-ing, bubbles were observed emanating rapidlyfrom the interface into the surrounding blood.This was associated with the initial audible popand a small (<10 ohm) rise in impedance. Second,in some trials we observed an increase in theechocardiographic image of the electrodediameter during RF application, which was corre-lated with electrode coagulum. During lesionswhich involved the formation of electrode coagu-lum, bubbles were observed to be trapped at theelectrode surface. A marked impedance riseoccurred only when bubble trapping wasobserved. We hypothesize that gas liberated inthe absence of electrode coagulum transientlyencountered the electrode surface, whereas gasliberated in the presence of coagulum had asustained presence at the electrode surface. Gaswould function as an electrical insulator, precipi-tating impedance rise.

LimitationsOur data have several limitations. First, weutilized an in vitro system in which the targetedmyocardium was not perfused. Previous studieshave demonstrated significant changes in myocar-dial properties soon after excision without perfu-sion [17]. Similarly, the ex vivo whole blood mayhave had properties, including rheologic andbiophysical, which might be substantially differ-ent in vivo. Second, our observations may be speci-fic to the RF power, duration and irrigation ratesutilized herein. Third, we evaluated only onemanner of stable electrode–endocardial angula-tion and contact, utilizing smooth endocardium.Finally, only one irrigation electrode design(showerhead) was tested in this study. It is concei-vable that other designs may produce disparateresults. However, the same in vitro observations asin this study were also reported for two otherelectrode designs in another study [18].

Clinical ImplicationsAssuming appropriate titration of voltage, RFapplication via an irrigated electrode has clearlybeen demonstrated to produce lesions havinggreater depth and volume than those resultingfrom RF application via non-irrigated electrodes.This has, in part, led to its use in targetingentities such as ventricular tachycardia occur-ring in the setting of chronic myocardial infarc-tion [19]. Our data suggest that irrigation mightalso be useful for ablation of relatively superficialtargets, such as smooth atrium. Advantages ofirrigated ablation which might have clinicalutility for such a target would include preventionof coagulum and minimization of endocardiallesion dimensions. We have presented supportivedata in a preliminary form [20,21].

Acknowledgements

The authors gratefully acknowledge the support andassistance of Dr. Stephen M. Dillon, John J. Michele,and Dr. James P. Dilger.

References

1. Mittleman RS, Huang SKS, DeGuzman WT, CuenoudH, Wagshal AB, Pires LA. Use of the saline infusionelectrode catheter for improved energy delivery andincreased lesion size in radiofrequency catheter abla-tion. Pacing Clin Electrophysiol 1995;18:1022–1027.

2. Skrumeda LL, Mehra R. Comparison of standardand irrigated radiofrequency ablation in the canineventricle. J Cardiovasc Electrophysiol 1998;9:1196–1205.

3. Ruffy R, Imran MA, Santel DJ, Wharton JM. Radio-frequency delivery through a cooled catheter tipallows the creation of larger endomyocardial lesionsin the ovine heart. J Cardiovasc Electrophysiol1995;6:1089–1096.

4. Peterson HH, Chen X, Pietersen A, Svendsen JH,Haunso S. Tissue temperatures and lesion sizeduring irrigated tip catheter radiofrequency ablation.Pacing Clin Electrophysiol 2000;23:8–17.

5. Nakagawa H, Yamanashi WS, Pitha JV, Arruda M,Wang X, Ohtomo K, Beckman KJ, McClelland JH,Lazzara R, Jackman WM. Comparison of in vivotissue temperature profile and lesion geometry forradiofrequency ablation with a saline-irrigated elec-trode versus temperature control in a canine thighmuscle preparation. Circulation 1995;91:2264–2273.

6. Ren J-F, Schwartzman D, Michele JJ, Li KS, HoffmanJ, Brode SE, Lighty GW Jr, Dillon SM, Chaudhry FA.Lower frequency (5 MHz) intracardiac echocardiogra-phy in a large swine model: imaging views andresearch applications. Ultrasound Med Biol 1997;23(6):871–877.

7. Petersen HH, Chen X, Pietersen A, Svendsen JH,Haunso S. Temperature-controlled irrigated tip radio-frequency catheter ablation. J Cardiovasc Electrophy-siol 1998;9:409–414.

8. Eick OJ, Gerritse B, Schumacher B. Poppingphenomena in temperature-controlled radiofrequencyablation. Pacing Clin Electrophysiol 2000; 23:253–258.

9. Labonte S. Numerical model for radiofrequency abla-tion of the endocardium and its experimental valida-tion. IEEE Trans Biomed Eng 1994;41:108–115.

10. Panescu D, Whayne JG, Fleischman SD, MirotznikMS, Swanson DK, Webster JG. Three-dimensionalfinite element analysis of current density andtemperature distributions during radiofrequencyablation. IEEE Trans Biomed Eng 1995;42:879–890.

11. Wong WS, VanderBrink BA, Riley RE, et al. Effect ofsaline irrigation flow rate on temperature profileduring cooled radiofrequency ablation. J InterventCardiac Electrophysiol 2000;4:321–326.

12. Mirotznik MS, Schwartzman D. Inhomogenous heat-ing patterns of commercial electrodes for radiofre-quency catheter ablation. J Cardiovasc Electrophysiol1996;7:1058–1062.


13. Smith S, Demazumder D, Mannino SF, Dillon SM,Marchlinski FE, Schwartzman D. Thermometry-guided ablation using commercial systems. J AmColl Cardiol 1996;27(2):123A.

14. McRury ID, Panescu D, Mitchell MA, Haines DE.Nonuniform heating during radiofrequency catheterablation with long electrodes: monitoring the edgeeffect. Circulation 1997;96:4057–4064.

15. Shake JG, Larson DW, Salerno CT, Bianco RW,Bolman RM III. The role of electrolyte in lesion sizeusing an irrigated radio frequency electrode. J InvestSurg 1997;10:339–348

16. Demazumder D, Kallash H, Schwartzman D. Compar-ison of different electrodes for radiofrequency ablationof myocardium using saline irrigation. Pacing ClinElectrophysiol 1997;20(4):1076.

17. Schwartzman D, Chang I, Michele JJ, Mirotznik MS,Foster KR. Electrical impedance properties of normaland chronically infarcted left ventricular myocardium.J Intervent Cardiac Electrophysiol 1999;3(3):213–224.

18. Demazumder D, Mirotznik MS, Schwartzman D.Comparison of irrigated electrode designs for radio-frequency ablation of myocardium. J InterventCardiac Electrophysiol 2001;5(4):391–400.

19. Calkins H, Epstein A, Packer D, Arria AM, HummelJ, Gilligan DM, Trusso J, Carlson M, Luceri R,Kopelman H, Wilber D, Wharton JM, Stevenson W.Catheter ablation of ventricular tachycardia inpatients with structural heart disease using cooledradiofrequency energy. J Am Coll Cardiol 2000;35:1905–1914.

20. Schwartzman D, Fischer WD, Spencer EP, Pariszh-kaya M, Devine W. Linear radiofrequencyatrial lesions deployed using an irrigated electrode:electrical and histologic evolution. Circulation 1998:I-567.

21. Schwartzman D, Fischer WD, Spencer EP.Linear radiofrequency atrial ablation: comparison ofirrigated and non-irrigated electrodes. Circulation1998:I-643.


biophysics of radiofrequency ablation using an irrigated ...mirotzni/pdf/biophysics_of... ·...

Documents