Something Old, Something New

Something old, something new

Citation for published version (APA):

Engels, E. B. (2016). Something old, something new: vectorcardiographic loop size and response to cardiac resynchronization therapy. Maastricht University. https://doi.org/10.26481/dis.20161019ee

Document status and date: Published: 01/01/2016

DOI: 10.26481/dis.20161019ee

Document Version: Publisher's PDF, also known as Version of record

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website. • The final author version and the galley proof are versions of the publication after peer review. • The final published version features the final layout of the paper including the volume, issue and page numbers. Link to publication

General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement: www.umlib.nl/taverne-license

Take down policy If you believe that this document breaches copyright please contact us at: [email protected] providing details and we will investigate your claim.

Download date: 05 Oct. 2021

Something old, something new

Vectorcardiographic loop size and response to cardiac resynchronization therapy

Elien Engels ISBN 978-94-6233-413-7

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the author.

Cover design by B. Spronck

Printed by Gildeprint, The Netherlands Something old, something new

Vectorcardiographic loop size and response to cardiac resynchronization therapy

PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Universiteit Maastricht, op gezag van de Rector Magniﬁcus, Prof. dr. Rianne M. Letschert, volgens het besluit van het College van Decanen, in het openbaar te verdedigen op woensdag 19 oktober 2016 om 14:00 uur

door

Elien Berti Engels

geboren op 4 augustus 1989 te Helmond, Nederland Promotor Prof. dr. F.W. Prinzen

Copromotor Dr. K. Vernooy

Beoordelingscommissie Prof. dr. U. Schotten (voorzitter) Prof. dr. A. Auricchio (Fondazione Cardiocentro Ticino, Lugano, Switzerland) Prof. dr. H.J.G.M. Crijns Prof. dr. T. Delhaas Dr. M. Meine (Universitair medisch centrum Utrecht)

Parts of the research described is this thesis were performed within the framework of CTMM, the Center for Translational Molecular Medicine, project COHFAR (Congestive Heart Failure and Arrhythmia; grant 01C-203).

Financial support by the Dutch Heart Foundation for the publication of this thesis is gratefully acknowledged.

Financial support for the publication of this thesis as provided by Maastricht University, Stichting Hartsvrienden RESCAR, St. Jude Medical, and BIOTRONIK is gratefully acknowledged. Contents

1 General introduction 7

2 Why QRS duration should be replaced by better measures of electrical activation to improve patient selection for cardiac resynchronization therapy 21

3 The synthesized vectorcardiogram resembles the measured vectorcardiogram in patients with dyssynchronous heart failure 43

4 Electrical remodeling in patients with iatrogenic left bundle branch block 59

5 T-wave area predicts response to cardiac resynchronization therapy in patients with left bundle branch block 77

6 T-wave area as biomarker of clinical response to cardiac resynchronization therapy 95

7 Prediction of optimal cardiac resynchronization by device-based vectorcardiography in dyssynchronous canine hearts 113

8 Tailoring device settings in cardiac resynchronization therapy by vectorcardiography using the implanted pacing electrodes. 131

9 General discussion 155

Summary 179

Samenvatting 185

Valorization 191

Dankwoord 197

About the author 201

Chapter 1 General introduction Chapter 1

1.1 Introduction

n a normal healthy heart, the electrical activation starts in the sinoatrial (SA) node after which the electrical activation spreads throughout the right and left atria. The electrical I activation then reaches the atrioventricular (AV) node through which the activation is slowed down, to allow optimal ventricular ﬁlling. Hereafter, the right and left ventricle are quickly activated by the specialized His-Purkinje ﬁbers, resulting in a synchronous contraction of the ventricles needed for an optimal pump function.1

1.2 Heart failure with dyssynchrony

In certain conditions there is abnormal ventricular impulse conduction, such as during left bundle branch block (LBBB), resulting in asynchronous ventricular activation and contraction. With LBBB, the left ventricle (LV) is activated through the slowly conducting ventricular myocardium rather than through the rapidly conducting His-Purkinje system. In LBBB, the electrical activation starts in the septum and right ventricle (RV) and proceeds through the slowly conducting septum to activate the LV.2,3 The prevalence of LBBB ranges from ∼1% in a general hospital population4 to as high as 23.8% in a heart failure (HF) population5. LBBB leads to a reduction in contractility and relaxation6,7 as well as to adverse remodelling8, resulting in LV dysfunction, emerging HF9 and eventually cardiac death10,11. Several experimental and clinical studies show that the heart adapts to the presence of ventricular dyssynchrony in a very complicated manner that is yet incompletely understood. The changes induced by altered electrical activation alter blood ﬂow and myocardial strain throughout the heart, triggering myocardial remodeling. This myocardial remodeling in the long run results in progressive LV dilation, decreasing ejection fraction, and asymmetrical hypertrophy.12 Beside these structural changes there are also electrical changes that emerge during the repolarization phase. This remodeling of repolarization occurs via changes in the expression and function of multiple ion channels and gap junctions, such as calcium channels responsible for the contraction and relaxation, and potassium channels responsible for repolarization.13 The electrical remodeling also leads to changes in the electrocardiogram (ECG). Studies, in which a temporary period of altered electrical activation was induced by RV pacing, showed that after this period the T-wave remained in the direction of the paced R-wave when returning to the normal activation, a phenomenon called cardiac memory.14,15 Changes in the ECG have also been demonstrated between acute and chronic RV pacing. These changes manifest as a reduction in T-vector amplitude of the paced beats with increased duration of pacing.14 It is however unknown whether this phenomenon is also present in patients with LBBB. We hypothesize that the presence of LBBB also induces electrical remodeling that leads to

8 ls fI-V nadto,teesol eeiec feetia etiua ysnhoy in- and dyssynchrony, duration ventricular functional QRS electrical NYHA prolonged of a by evidence dicated be and LVEF should CRT reduced there on addition, a guidelines In by II-IV. Current evidenced of important. class as HF, very have is should CRT patients from who that benefit patients state not those will of improve who prediction potentially those reliable can and Therefore, that mortality. will therapy reduce a and from life withheld of quality be hand, their not other to should the likely On benefit avoided. to not be likely for are should patients risks procedure patients invasive corresponding this with When after device and CRT during expensive selection. complications an patient of proper implantation therapy, with from starts benefit CRT of success The CRT for selection patient Improving 1.4 investigated. are factors location. two lead first non- LV and are settings, patients device CRT the suboptimal of 30-50% ‘CRT-response’ cardiac of mortality. improve responders. definition and to hospitalization, the known symptoms, on HF is reduce dependent activation CRT and However, intrinsic life simultaneously. the of ventricles with quality both and collide function LV pacing pump paced by a or of RV fronts the wave of the letting either by traction alone. in(VF.CThsbe hw ob oee more be to shown frac- been ejection LV has reduced CRT (LVEF). a tion by treat- accompanied established HF dyssynchrony-induced an with become patients has for (CRT) ment therapy resynchronization cardiac decade, past the In therapy resynchronization Cardiac 1.3 ECG. the on T-wave the by reflected as phase repolarization the during changes fti side h ae tmyb otwiet losuyteiprac frepolarization- of importance the study also to the worthwhile repolarization. be that the may in it hypothesized case, changes we the to indeed 1.2 leads is that this section If remodeling In electrical induces LBBB depolarization. of of presence sequence in abnormalities the a be might further. VCG even The selection patient 1.1). optimize (Fig. to (VCG) the tool vectorcardiogram over of valuable representation the forces 3D heart, electrical a the in the of results of pattern vectors activation hereby-formed direction the and of magnitude heads the the Connecting Vectorcardio- records CRT. time. that for technique selection a patient simplify is complete and graphy of improve definition to the vectorcardiography that use uniform. is to not is problem ECG a 12-lead the but CRT, from LBBB from most benefit morphology QRS ni o h vrhligmjrt frsac nteae fCTdaswt studying with deals CRT of area the in research of majority overwhelming the now Until 16 sisnm ugss R ist eycrnz h ih n etvnrclrcon- ventricular left and right the resynchronize to aims CRT suggests, name its As 18 h anfcoslaigt o-epneaennotmlptetselection, patient non-optimal are non-response to leading factors main The / rLB R morphology. QRS LBBB or 22,23 ff cieta pia hraooia therapy pharmacological optimal than ective nti hssw netgt h possibilities the investigate we thesis this In akrudadgnrlamo h thesis the of aim general and Background 19 nti hss mrvmnsi the in improvements thesis, this In 20,21 ainswt LBBB with Patients 17 9

1 Chapter 1

Figure 1.1: Representation of the 3D vectorcardiogram construction. The vectors change rapidly in direction and length (indicated by the solid black arrows). The QRS loop is indicated in blue and the T loop in red and green. Every color alteration represents 10 ms. The dashed black arrow indicates the direction of the QRS loop. based VCG metrics for the selection of CRT candidates. Therefore, in this thesis, we will compare the eﬀectiveness of diﬀerent QRS variables (depolarization) and T-wave variables (repolarization) as predictors of CRT response.

1.5 Optimizing CRT device settings

Once the CRT device is implanted, one can program its settings non-invasively and possibly adapt them to optimize application of CRT in the individual patient. These settings are the atrioventricular (AV) interval, the interval between atrial and ventricular activation, and the interventricular (VV) interval, the interval between RV and LV stimulation. The AV-interval is responsible for ventricular ﬁlling and timing of contraction as well as for the extent of fusion of the pacing generated activation wave front with the intrinsic activation. The VV- interval is responsible for improved resynchronization between the right and left ventricle to improve systolic performance of the two ventricles. Currently, CRT devices are commonly

10 nlss oti n h olwn eea isaeformulated: vectorcardiographic are using aims CRT general to following response the the end improve this to To is analysis. thesis this of goal ultimate The thesis the of Aims 1.6 settings. could device CRT EGM-VCG the This optimize (EGM-VCG). continuously and VCG individually EGM-based that to the leads used RV be pacing: and then LV for implanted used the on being electrodes not the are from calculated be could VCG similar V n Vitra nbt animals both the in as VV-interval and qualify AV- not do that patients for bias a introduce cer- immediately might patient. to a ‘average’ This settings using (BiV)-paced VV-interval settings. biventricular or optimal or AV- LV- the baseline use optimal ﬁnd The not the do itself. calculate thus individual to they the equation; used on tain is not pacing and atrial population average or the activation on based are they that iecnuigades ouse. to easy and less are consuming methods new time these However, methods. optimization echocardiographic to superiority optimization. continuous almost ADAPTIVECRT an and only provides individually algorithms, intervals automated the mentioned optimize previously to the able Of be continuously. to like would one ideally Therefore, patient. omddrn et nspn oiinadol ne oee,i scnevbeta h op- the that conceivable is and it activity However, patient with once. change only may settings and timal position supine in rest, during formed settings. device the of optimization automated perform to devices CRT in embedded Inc.) eoe:tesmartDelay the settings. veloped: ‘out-of-the-box’ the default in their at that left explains are devices methods CRT the optimization patients available of currently majority of disadvantages serious the r naiewihlmt hi s nruieciia practice clinical routine in use their limits which invasive are isoi n ytlcfnto,btn td a enal oso htti optimization this that show to able been outcomes. has clinical in study improvements no into but LV translates in function, improvements acute systolic in and result intervals diastolic optimized that demonstrated have methods tion adorpyi bevrdpnetadtm consuming time echo- However, and observer-dependent measurements. is cardiography hemodynamic or echocardiography e.g., using programmed eety a ere ta.soe htteVGcudas eue ootmz the optimize to used be also could VCG the that showed al. et Deursen van Recently, ni o oeo h uoai V n Vitra piiainagrtm a shown has algorithms optimization VV-interval and AV- automatic the of none now Until nte iavnaeo h ovninlotmzto rcdrsi htte r per- are they that is procedures optimization conventional the of disadvantage Another nodrt vi h iecnuigotmzto,svrlagrtm aebe de- been have algorithms several optimization, consuming time the avoid to order In .Cmaemaue Cswt Cscluae rma1-edEGi R candi- CRT in ECG 12-lead a from calculated VCGs with VCGs measured Compare 1. 30 dates. l loihsueacmiaino C n lcrga EM inl n are and signals (EGM) electrogram and ECG of combination a use algorithms All . 28 QuickOpt , 28,30–36 37 29 n patients and n ADAPTIVECRT and , rwako l urnl vial loihsis algorithms available currently all of drawback A / rdet hne ntedsaesaeo the of state disease the in changes to due or 27 akrudadgnrlamo h thesis the of aim general and Background 38 h ako adciia vdneand evidence clinical hard of lack The nti hss ehpteie hta that hypothesized we thesis, this In . 24 n eoyai measurements hemodynamic and , 25,26 TM utemr,aloptimiza- all Furthermore, . taeako Medtronic, of (trademark TM30 11

1 Chapter 1

2. Create a better understanding of how the heart adapts electrically to the presence of LBBB.

3. Improve patient selection for CRT.

4. Explore the feasibility of using a EGM-based VCG (derived from the pacing electrodes) to optimize CRT device settings continuously and individually.

1.7 Overview of the thesis

The strengths and weaknesses of ECG markers currently being used in guidelines for CRT patient selection, e.g. QRS duration and morphology, is reviewed in chapter 2. Furthermore, the current knowledge on the underlying electrical substrate and the mechanism of action of CRT will be discussed. In chapter 3 a comparison is made between the measured Frank-VCG39 and the synthesized Kors-VCG40 from a 12-lead ECG in patients with HF and conduction abnormalities who are candidate for CRT. Multiple vectorcardiographic variables related to either the depolarization or repolarization phase of a heartbeat will be evaluated. The way the heart adapts to the sudden presence of LBBB is investigated in chapter 4. Normally LBBB starts silently, making this investigation difficult. However, in patients receiving a new aortic valve through a transcatheter aortic valve implantation (TAVI) procedure, ∼33% develops LBBB as a result of this procedure. These patients are followed for ∼6 months, making it possible to investigate the electrical changes of the heart from the first moment of LBBB presence. The aim of reducing CRT non-response by improving patient selection using the VCG is addressed in chapter 5 and 6. In chapter 5, this assessment is based on echocardiographic outcome parameters, while in chapter 6 long-term clinical outcome parameters are used. A comparison is made between the uses of QRS- and T-wave derived variables as predictors for CRT response. The ability to optimize CRT device settings using the EGM-VCG is investigated in an animal model (chapter 7) and in humans (chapter 8). This thesis ends with a general discussion (chapter 9) in which the findings of the above- mentioned studies are linked and discussed in a broader scientific and clinical perspective.

12 iue1.2: Figure vriwo hsthesis this of Overview akrudadgnrlamo h thesis the of aim general and Background 13

1 References

1. MO Sweeney and FW Prinzen. Ventricular pump function and pacing: physiological and clinical integration. Circ Arrhythm Electrophysiol (2008) 1:127–139. 2. DG Strauss, RH Selvester, JAC Lima, H Arheden, JM Miller, G Gerstenblith, E Mar- bán, RG Weiss, GF Tomaselli, GS Wagner, and KC Wu. ECG quantification of myocardial scar in cardiomyopathy patients with or without conduction defects: correlation with cardiac magnetic resonance and arrhythmogenesis. Circ Arrhythm Electrophysiol (2008) 1:327–336. 3. JA Vassallo, DM Cassidy, FE Marchlinski, AE Buxton, HL Waxman, JU Doherty, and ME Josephson. Endocardial activation of left bundle branch block. Circulation (1984) 69:914–923. 4. DM Boyle and SS Fenton. Left bundle branch block. Ulster Med J (1966) 35:93–99. 5. NM Hawkins, D Wang, JJV McMurray, MA Pfeffer, K Swedberg, CB Granger, S Yusuf, SJ Pocock, J Ostergren, EL Michelson, FG Dunn, and CHARM Investigators and Committees. Prevalence and prognostic impact of bundle branch block in patients with heart failure: evidence from the CHARM programme. Eur J Heart Fail (2007) 9:510–517. 6. XAAM Verbeek, K Vernooy, M Peschar, RNM Cornelussen, and FW Prinzen. Intra- ventricular resynchronization for optimal left ventricular function during pacing in experimental left bundle branch block. J Am Coll Cardiol (2003) 42:558–567. 7. M Nahlawi, M Waligora, SM Spies, RO Bonow, AH Kadish, and JJ Goldberger. Left ventricular function during and after right ventricular pacing. J Am Coll Cardiol (2004) 44:1883–1888. 8. F Zannad, E Huvelle, K Dickstein, DJ van Veldhuisen, C Stellbrink, L Køber, S Cazeau, P Ritter, AP Maggioni, R Ferrari, and P Lechat. Left bundle branch block as a risk factor for progression to heart failure. Eur J Heart Fail (2007) 9:7–14. 9. JF Schneider, H Thomas Jr, BE Kreger, PM McNamara, and WB Kannel. Newly acquired left bundle-branch block: the Framingham study. Ann Intern Med (1979) 90:303–310. 10. GJ Fahy, SL Pinski, DP Miller, N McCabe, C Pye, MJ Walsh, and K Robinson. Natural history of isolated bundle branch block. Am J Cardiol (1996) 77:1185–1190. 11. JF Schneider, H Thomas Jr, P Sorlie, BE Kreger, PM McNamara, and WB Kannel. 2 eno,XA ebe,MPshr JMCin,TAt,RMCreusn and Cornelussen, RNM Arts, T Crijns, HJGM Peschar, M Verbeek, XAAM Vernooy, K 12. 3 JCte,DJyrj n SRsnam ada lcrclrmdln nhat and health in remodeling electrical Cardiac Rosenbaum. DS and Jeyaraj, D Cutler, MJ 13. 4 hikn ooi,BVji,IGsa,PZmtam n EJspsn Vec- Josephson. ME and Zimetbaum, P Gussak, I Vajdic, B Bojovic, B Shvilkin, A 14. 0 EEsen PDMro AElnoe,NMEts3d AFeda,L Gettes, LS Freedman, RA 3rd, Estes NAM Ellenbogen, KA DiMarco, JP Epstein, AE 20. 9 uln,R rm,TVra rsn,R trig LWilko BL Starling, RC Dresing, T Verga, T Grimm, RA Mullens, W 19. 1 rgoe uici,GBrnEqiis odca,GBrai ABreithardt, OA Boriani, G Bordachar, P Baron-Esquivias, G Auricchio, A Brignole, M 21. 5 ek,ARbls udh,M oe,adLBrflt ih etiua pacing- ventricular Right Bergfeldt. L and Rosen, MR Lundahl, G Rubulis, A Wecke, L 15. 8 uoenHatRyh soito,Erpa oit fCrilg,HatRhythm Heart Cardiology, of Society European Association, Rhythm Heart European Kocovic, DZ Loh, 18. E Leon, AR Delurgio, DB Smith, AL Fisher, WG Abraham, WT 17. Carson, P Marco, De T Kass, DA Krueger, S Boehmer, J Saxon, LA Bristow, MR 16. oprtv etrso el curdlf n ih udebac lc ntegen- the in block study. branch Framingham bundle the right population: and eral left acquired newly of features Comparative WPizn etbnl rnhbokidcsvnrclrrmdligadfunctional and remodelling ventricular hypoperfusion. induces septal block branch bundle Left Prinzen. FW disease. n otnosvnrclrpacing. activation ventricular ventricular continuous normal and during memory cardiac of determinants torcardiographic nue lcrpyilgclrmdln ntehmnhatadisrltosi ocar- to relationship its memory. and diac heart human the in remodeling electrophysiological induced MGlio,GGeoao,S aml,D ae,e l 02ACCF 2012 al. et Hayes, DL Hammill, SC Gregoratos, G Gillinov, AM heart a of part as program. clinic management optimization disease resynchronization failure cardiac a from Insights Tang. oue paeicroae noteACCF the into incorporated update focused lln,J eao egd,P lit,e l 03ECGieie ncardiac on Guidelines ESC 2013 al. et Elliott, PM Delgado, V Deharo, JC Cleland, J eto ada eycrnzto hrp nhatfiue mln n olwu re- follow-up and implant failure: management. heart and commendations in therapy resynchronization cardiac on ment ae hrp fcricryh bomlte:arpr fteAeia olg of College American the of report a abnormalities: Foundation Cardiology rhythm cardiac of therapy based n h er htmSociety. Rhythm Heart the and r soito,J abr,LSxn ta.21 EHRA 2012 al. et Saxon, L Fail- Daubert, Heart JC Echocardiography, Echocardiography, Association, of of ure Association Society European American Association, America, Heart of American Society Failure Heart heart Society, chronic in resynchronization Cardiac al. et Hayes, failure. DL Clavell, AL Packer, M failure. heart chronic defibrillator advanced implantable of in an without Comparison Invest- or (COMPANION) and with Failure therapy Feldman, Heart Cardiac-resynchronization AM in igators. Defibrillation DeVries, and DW Pacing, White, Therapy, BG Medical DeMets, D DiCarlo, L nlJMed J Engl N rnsPamclSci Pharmacol Trends er Rhythm Heart u er J Heart Eur 20)346:1845–1853. (2002) / mrcnHatAscainTs oc nPatc Guidelines Practice on Force Task Association Heart American 20)4:1477–1486. (2007) 21)32:174–180. (2011) Circulation nlJMed J Engl N er Rhythm Heart 20)26:91–98. (2005) er Rhythm Heart mCl Cardiol Coll Am J mJCardiol J Am 21)127:e283–e352. (2013) / AHA 20)350:2140–2150. (2004) 21)9:1524–1576. (2012) 20)6:943–948. (2009) / R 08gieie o device- for guidelines 2008 HRS 18)47:931–940. (1981) / 20)53:765–773. (2009) R xetcnessstate- consensus expert HRS ff n WHW and , / References AHA / HRS 15

1 Chapter 1

pacing and cardiac resynchronization therapy: the Task Force on cardiac pacing and resynchronization therapy of the European Society of Cardiology (ESC). Developed in collaboration with the European Heart Rhythm Association (EHRA). Eur Heart J (2013) 34:2281–2329. 22. MR Gold, C Thébault, C Linde, WT Abraham, B Gerritse, S Ghio, M St John Sutton, and JC Daubert. Effect of QRS duration and morphology on cardiac resynchronization therapy outcomes in mild heart failure: results from the Resynchronization Reverses Remodeling in Systolic Left Ventricular Dysfunction (REVERSE) study. Circulation (2012) 126:822–829. 23. W Zareba, H Klein, I Cygankiewicz, WJ Hall, S McNitt, M Brown, D Cannom, JP Daubert, M Eldar, MR Gold, et al. Effectiveness of Cardiac Resynchronization Therapy by QRS Morphology in the Multicenter Automatic Defibrillator Implantation Trial-Cardiac Resynchronization Therapy (MADIT-CRT). Circulation (2011) 123:1061–1072. 24. CE Raphael, A Kyriacou, S Jones, P Pabari, G Cole, R Baruah, AD Hughes, and DP Francis. Multinational evaluation of the interpretability of the iterative method of optimisation of AV delay for CRT. Int J Cardiol (2013) 168:407–413. 25. A Auricchio, C Stellbrink, M Block, S Sack, J Vogt, P Bakker, H Klein, A Kramer, J Ding, R Salo, B Tockman, T Pochet, and J Spinelli. Effect of pacing chamber and atrioventricular delay on acute systolic function of paced patients with congestive heart failure. The Pacing Therapies for Congestive Heart Failure Study Group. The Guidant Congestive Heart Failure Research Group. Circulation (1999) 99:2993–3001. 26. N Derval, P Steendijk, LJ Gula, A Deplagne, J Laborderie, F Sacher, S Knecht, M Wright, I Nault, S Ploux, et al. Optimizing hemodynamics in heart failure patients by systematic screening of left ventricular pacing sites: the lateral left ventricular wall and the coronary sinus are rarely the best sites. J Am Coll Cardiol (2010) 55:566–575. 27. WW Brabham and MR Gold. The role of AV and VV optimization for CRT. Journal of Arrhythmia (2013) 29:153–161. 28. MR Gold, I Niazi, M Giudici, RB Leman, JL Sturdivant, MH Kim, Y Yu, J Ding, and AD Waggoner. A prospective comparison of AV delay programming methods for hemodynamic optimization during cardiac resynchronization therapy. J Cardiovasc Electrophysiol (2007) 18:490–496. 29. JH Baker 2nd, J McKenzie 3rd, S Beau, GS Greer, J Porterfield, M Fedor, S Green- berg, EG Daoud, R Corbisiero, JR Bailey, and L Porterfield. Acute evaluation of programmer-guided AV/PV and VV delay optimization comparing an IEGM method and echocardiogram for cardiac resynchronization therapy in heart failure patients and dual-chamber ICD implants. J Cardiovasc Electrophysiol (2007) 18:185–191. 30. H Krum, B Lemke, D Birnie, KLF Lee, K Aonuma, RC Starling, M Gasparini, J

16 5 akr rpslfranwciia n on oeaut h e the evaluate to point end clinical new a for Proposal Packer. M 35. 8 J a ere,LWce MvnEedne,MSåleg H ase,F Janssen, MHG Ståhlberg, M Everdingen, van WM Wecke, L Deursen, van CJM 38. 7 J a ere,MSrk MRdmkr,AvnHni,MKie,LWecke, L Kuiper, M Hunnik, van A Rademakers, LM Strik, M Deursen, van CJM 37. 6 OMri,BLme ine rm L e,KAnm,MGsaii RC Gasparini, M Aonuma, K Lee, KLF Krum, H Birnie, D Lemke, B Martin, DO 36. 1 itr PMDlo,LPdlti uai agl,ABriBuet,adJ and Borri-Brunetto, A Naegele, H Lunati, M Padeletti, L Delnoy, PPHM Ritter, P 31. 4 ag u u,HZn,GYn,SWn,ZWn,QJn,adYHn Long- Han. Y and Jing, Q Wang, Z Wang, S Yang, G Zang, H Yun, T Yu, H Wang, D 34. Waggoner, AD Mittal, S Lozano, Fernndez I Meyer, TE Gold, MR Ellenbogen, KA 32. 3 CPrin,C a,MMci apli ogon,A eii il,G Lilli, A Perini, AP Bongiorno, G Cappelli, F Mochi, M Rao, CM Porciani, MC 33. niiulzdcricrsnhoiainteay ainl n eino h adaptive the trial. for of therapy algorithm resynchronization design novel cardiac and A rationale Martin. therapy: D resynchronization and cardiac Kalmes, individualized A Sambelashvili, A Rogers, T Gorcsan, p nsnsryh ainsuigapa noada ceeainsno s standard vs. sensor acceleration endocardial peak a methods. using ther- patients resynchronization rhythm cardiac sinus of in optimization apy of study pilot randomized A Silvestre. gah o piiaino tmlto nevl ncricrsnhoiaintherapy. Res resynchronization Transl cardiac Cardiovasc in J Vectorcardi- intervals stimulation Prinzen. of FW optimization and Vernooy, for ography K Crijns, HJGM Bergfeldt, L Braunschweig, failure. heart chronic of treatment the in devices emciia e clinical term three- 31:56–63. real-time therapy. A resynchronization cardiac Padeletti. optimizing L for method electrogram-based and intracardiac an Barold, of SS validation echocardiographic Silvestri, dimensional P Hashtroudi, L Ricciardi, ther- resynchronization cardiac apy. in therapy programming delay resynchronization atrioventricular cardiac algorithmic and in echocardiography-guided, used empirical, comparing methods trial randomized Rap- delay a J AV (SMART-AV) trial: Bedi, optimization: other AV M determined to Weiner, SmartDelay comparison S the from Whitehill, a results J Primary Eyk, Stein. Van KM JE and Spinale, kin, FG Singh, JP Lemke, B hoiainteayi ainswt er failure. heart resyn- with cardiac patients in for therapy echocardiography chronization versus electrography Intracardiac optimization: piiaino ada eycrnzto hrp ncnn etbnl rnhblock easy branch for bundle tool left a canine as in hearts. Vectorcardiography therapy resynchronization Prinzen. cardiac FW of optimization and Vernooy, K Crijns, HJGM hrp:rslso h dpieCTtrial. CRT adaptive the resynchronization of cardiac synchron- results of for therapy: optimization algorithm ambulatory novel and a pacing and left-ventricular of Houmsse, ized Investigation M 3rd, Investigators. Gorcsan Study J CRT Sambelashvili, Adaptive A Rogers, T Milasinovic, G Starling, Circulation icAryh Electrophysiol Arrhythm Circ Europace ﬀ cso rgamrgie tivnrclraditretiua delay interventricular and atrioventricular programmer-guided of ects 21)122:2660–2668. (2010) 21)14:1324–1333. (2012) 21)8:128–137. (2015) 21)5:544–552. (2012) mHatJ Heart Am er Rhythm Heart adFail Card J n e Res Med Int J aigCi Electrophysiol Clin Pacing 21)163:747–752.e1. (2012) 21)9:1807–1814. (2012) 20)7:176–182. (2001) 21)41:115–122. (2013) ﬃ ayo rg and drugs of cacy References (2008) 17

1 Chapter 1

39. E Frank. An accurate, clinically practical system for spatial vectorcardiography. Cir- culation (1956) 13:737–749. 40. JA Kors, G van Herpen, AC Sittig, and JH van Bemmel. Reconstruction of the Frank vectorcardiogram from standard electrocardiographic leads: diagnostic comparison of diﬀerent methods. Eur Heart J (1990) 11:1083–1092.

Chapter 2 Why QRS duration should be replaced by better measures of electrical activation to improve patient selection for cardiac resynchronization therapy

The content of this chapter is based on:

Elien B. Engels,1,∗ Masih Maﬁ-Rad,2,∗ Antonius M.W. van Stipdonk,2 Kevin Vernooy,2 and Frits W. Prinzen1 (2016). Why QRS duration should be replaced by better measures of electrical activation to improve patient selection for cardiac resynchronization therapy. J Cardiovasc Transl Res 9(4):257-265.

∗ Authors contributed equally. 1 Department of Physiology, CARIM School for Cardiovascular Diseases, Maastricht University, Maastricht, the Netherlands. 2 Department of Cardiology, Maastricht University Medical Center, Maastricht, the Netherlands. Chapter 2

Abstract

Cardiac resynchronization therapy (CRT) is a well-known treatment modality for patients with a reduced left ventricular ejection fraction accompanied by a ventricular conduction delay. However, a large proportion of patients does not beneﬁt from this therapy. Better patient selection may importantly reduce the number of non-responders. Here we review the strengths and weaknesses of the electrocardiogram (ECG) markers currently being used in guidelines for patient selection, e.g. QRS duration and morphology. We shed light on the current knowledge on the underlying electrical substrate and the mechanism of action of CRT. Finally, we discuss potentially better ECG-based biomarkers for CRT candidate selection, of which the vectorcardiogram may have high potential.

22 VF( LVEF UTCstudy. MUSTIC h CARE-HF the C fteeptet fe hwdapoogdP-nevladawdndQScomplex QRS widened a and PR-interval disturbances. prolonged conduction ventricular a to showed due often patients these of ECG n YAcas h IAL td ofimdteerslsi ainswt QRS a with patients in score, results life these of confirmed quality study MIRACLE uptake, The oxygen duration peak class. distance, NYHA walking and 6-min the improved pacing optlzto,adbte uvvl iia eut eesonb h COMPANION the by shown were results Similar survival. better and hospitalization, YAcasIII ept pia eia ramn,wt eue VF( LVEF reduced a with treatment, medical optimal despite III-IV class NYHA III-IV. class h lnclapiaino R ea n19 hntefis ae fatrio-biventricular of cases first the described. when were HF 1994 congestive severe in with patients began in implantations CRT pacemaker of application clinical The CRT of selection the in ECG 12-lead the of role The 2.2 elec- underlying the candidates. ECG-based CRT better on potentially of knowledge discuss selection and current for CRT biomarkers of the action of of mechanism light the the and substrate in trical markers ECG these of nesses in ndqaedvc etns uotmlmdcltetet rhtma,advariable resynchroniza- and to arrhythmias, amenable treatment, is that medical position. substrate suboptimal lead in settings, variation device are inadequate benefit tion, of range huge this for n naifigslcino ains nteoehn,asgicn oto 3-0)of (30-50%) portion significant guidelines a current to hand, whereas according one implanted the are On that patients patients. of selection unsatisfying and rmteEG R uainadmorphology. and duration QRS ECG: the from oecodntdade and coordinated more a ult flf,adrdcshatfiue(F ypos optlztosadmortality. and hospitalizations symptoms, (HF) failure heart reduces and life, of quality C Introduction 2.1 h rtrnoie rsoe ra netgtn h lncle clinical the investigating trial crossover randomized first The hs aoal n ossetrslsldt h eomnaino R nptet in patients in CRT of recommendation the to led results consistent and favorable These eetees hr ssila nopeeudrtnigo h ehns ftetherapy the of mechanism the of understanding incomplete an still is there Nevertheless, h otipratslcinciei ncretCTipatto udlnsaederived are guidelines implantation CRT current in criteria selection important most The candidates < ≥ ∼ ardiac odcindly seilydet etbnl rnhbok(BB.CTcreates CRT (LBBB). block branch bundle left to due especially ventricular delay, a with conduction combination in (LVEF) fraction ejection ventricular left decreased 5)adabodQScmlx( complex QRS broad a and 35%) 0 fptet hwcmlt omlzto fLVEF. of normalization complete show patients of 20% 3 ms. 130 6 1 10 eycrnzto hrp CT sa e an is (CRT) therapy resynchronization ras hc nlddptet ihQSduration QRS with patients included which trials, 2,11 hstili ainswt hoi eeeH NH I) reduced III), (NYHA HF severe chronic with patients in trial This hssuyas hwdacerrdcini Vvlms eue HF reduced volumes, LV in reduction clear a showed also study This ffi in otato ftehat mrvsL ytlcfnto and function systolic LV improves heart, the of contraction cient te esrst mrv ain eeto o CRT for selection patient improve to measures Other > 3,7 5 s,soe htbvnrclr(BiV) biventricular that showed ms), 150 eew eiwtesrntsadweak- and strengths the review we Here ff cieteayfrptet iha with patients for therapy ective 3,4 eetltl rmti therapy this from little benefit ffi 5 ≥ ayo R a the was CRT of cacy osbeexplanations Possible 2 sadNYHA and ms 120 8,9 h surface The < 5) in 35%), 12 1,2 and 23

2 Chapter 2 sinus rhythm, and a wide QRS complex (≥ 120 ms).13 Subsequent trials investigated the effect of CRT in less symptomatic patients (the REVERSE14, MADIT-CRT15 and RAFT trials16). Again, LV function improved and both all-cause mortality and non-fatal HF events improved. However, subgroup analyses of these three trials demonstrated that these effects were predominantly confined to patients with a QRS duration ≥ 150 ms (Fig. 2.1).17 This evidence resulted in the addition of a class I indication to CRT for patients presenting with NYHA class II, a reduced LVEF, and a QRS duration > 150 ms, in the 2010 guidelines.18

Figure 2.1: Eﬀect of CRT on composite clinical events in patients with moderately prolonged (QRS duration of 120 - 150 ms) and severely prolonged QRS duration (> 150 ms) (reprinted from Bryant et al. 17 ).

Even though most studies show an increased response rate after CRT in patients with a severely prolonged QRS duration, these studies used the fairly crude division of the cohorts in patients with a QRS duration < and > 150 ms. However, the best cut-off value for QRS duration is unclear. More recently, attention has shifted from QRS duration to QRS morphology. Small single- center studies19,20 and sub-analyses of the MADIT-CRT21, REVERSE22, and RAFT16 study showed that patients with a LBBB morphology benefit most from CRT. In contrast, patients with right bundle branch block (RBBB) or intra-ventricular conduction delays (IVCD) had no benefit or even a worse outcome from CRT (Fig. 2.2). These observations led to the

24 mrv h rdcino R epneadciia ucm,a es nsalsnl center single small in least at outcome, clinical significantly and studies. to response appears CRT slurring of or prediction notching the of improve presence the with morphology LBBB of ment edEGvre ewe uoenadAeia udlnsadbtenlreclinical large between and guidelines American and European trials between varies ECG lead iue2.2: Figure 2012 in guidelines the of adaptation R duration QRS hl R opooyi o n ftepiayidctr o R,arcn meta- recent a CRT, for indicators primary the of one now is morphology QRS While neetn n motn oee,i httedfiiino opeeLB rmte12- the from LBBB complete of definition the that is however, important and Interesting 21,22 20,24 rstudies or R opooyi h R ihdfirlao CTD r fteMDTCT(adjusted MADIT-CRT al. the et of Tian arm to (CRT-D) from according defibrillator (bottom) with alone CRT death the of in and morphology (top) QRS death or event HF of probability Cumulative > 5 sol fannLB opooyi present. is morphology non-LBBB a if only ms 150 23 htivsiae BBa rdco fCTe CRT of predictor a as LBBB investigated that 20 ). / 03 nldn BBa h rmr C rtro and criterion ECG primary the as LBBB including 2013, te esrst mrv ain eeto o CRT for selection patient improve to measures Other 3,4 ff ciees h refine- The ectiveness. 25

2 Chapter 2 analysis, combining data from CARE-HF, MIRACLE, MIRACLE ICD, REVERSE, and RAFT showed that QRS duration is a more powerful predictor of CRT outcomes (mortality and morbidity) than QRS morphology.25 This conclusion is in contrast to several reports derived from some of the individual trials and to a meta-analysis of the MADIT-CRT, RAFT, and REVERSE study (Fig. 2.1).26 One possible explanation for this discrepancy is the use of ‘liberal’ LBBB criteria. In that case, it is likely that QRS duration provides additional information. Indeed, when using ‘liberal’ LBBB criteria the non-LBBB patients tended to have a lower QRS duration than the LBBB patients21, but this difference could not be observed when stricter LBBB criteria were used.20 Furthermore, in the studies where strict LBBB criteria as defined by Strauss et al. 23 were used, QRS duration was not a predictor of response while LBBB was.20,27 In conclusion, currently it is not clear whether QRS duration or morphology should be preferred as primary marker for selection of CRT patients. QRS duration may not be specific, but LBBB criteria may be too complex and/or dependent. In order to come to a possible solution, it may be worthwhile to go back to the basic physiology of dyssynchronous HF and the mechanisms of CRT.

2.3 Electrophysiological evaluation of the electrical substrate for CRT

Delayed electrical activation of the LV is considered the underlying substrate of LV dysfunction in patients with systolic dysfunction and a conduction delay, mainly due to LBBB.28 CRT aims to correct the underlying electrical substrate by paced pre-excitation of late de- polarized and contracting LV regions, thereby restoring synchronous ventricular electrical activation and contraction.28 Experimental studies have confirmed that in hearts with delayed LV activation due to LBBB, LV-only or BiV pacing creates a more synchronous contraction pattern, which is accompanied by marked hemodynamic improvement.28,29 The clinical importance of LV activation delay has become evident in studies showing that a greater delay in time from onset of the QRS complex to the local intrinsic activation at the LV stimulation site (Q-LV) is associated with a greater likelihood of benefit from CRT. Singh et al. 30 measured Q-LV intra-procedurally as a percentage of the baseline QRS interval in 71 patients undergoing CRT device implantation. A longer Q-LV was related to superior acute LV hemodynamic improvement, whereas a reduced Q-LV (< 50% of QRS duration) was related to a worse clinical outcome.30 A secondary analysis of the prospective multi-center SMART-AV trial showed that patients with a Q-LV > 95 ms show significantly improved odds of reverse remodeling and quality of life response.31 Conversely, experimental studies and computer simulations have shown that pacing induced pre-excitation in a heart without a significant electrical delay (narrow QRS complex) widens the QRS complex and consequently worsens

26 Vpm function. pump LV eut ftercn coR trial. EchoCRT recent the of results fCTi ainswt arwQScmlx( complex QRS narrow a with patients in CRT of su eetbeciia eetadee hwdasgicn nraei otlt oprdto compared mortality in increase significant group. any a derive control showed the not even did and group CRT benefit the clinical because detectable stopped prematurely was trial The dyssynchrony. h aea alo h LV. towards the septum that of interventricular and wall the ventricle through lateral propagates right the slowly the then inside front electrical wave occurs performed, electrical activation was the electrical branch earliest bundle left that the showed of mapping ablation proximal where activation. LV hearts delayed canine with associated In is that disturbance conduction hallmark the considered ocnetoa C rtrahv hw httesqec fL noada ciainin activation endocardial LV of sequence heterogeneous. the is according that patients LBBB shown these and have HF criteria with ECG patients conventional in to mapping map- endocardial (EnSite) performed non-contact that three-dimensional Studies or or ping. NOGA) technique (CARTO, point-by-point contact conventional reconstruction using electro-anatomical mapping invasive by is nohr,L noada ciainocre sarsl frgtt-ettasetlsra of spread transseptal right-to-left of result branch, a bundle mid- as activation. left the occurred the activation in through endocardial system LV conduction others, slow conduction in by the activation of suggests vicinity which the region, septal in occurred breakthrough patients, some n ehncldsycrn hti uncue oso Vpm ucinadventricular and function pump LV of loss causes turn remodeling. in that dyssynchrony mechanical and rcewssont as Vedcrilbekhog ndi in breakthrough endocardial LV cause to shown was tricle n ih rsm.A icse ale,nmru ismlrte nEGciei o the for criteria ECG in as dissimilarities di straightforward numerous between as LBBB earlier, not of discussed is diagnosis As ECG the presume. on might LBBB one true remains delay identifying activation However, LV of ECG. extent surface the evaluate the and detect to investigation clinical key The niyi hrceie yhmgnospoaaino lcrclatvto hogotthe throughout activation electrical of propagation homogeneous by characterized is entity ersweepoia baino h etbnl rnhhsbe performed. been has branch bundle left canine the in of mapping ablation non-contact proximal endocardial where during hearts observed been around wall has lateral turns the that that activate pattern activation to activation order of in pattern wall U-shaped inferior a and apex by LV the characterized approximately is in observed patients, entity, first of The thirds patients. two these of LV the in propagation front wave rnspa odcintime. conduction transseptal ffi l h frmninddt upr h ointa neetia usrt,cnitn fa of consisting substrate, electrical an that notion the support data aforementioned the All h otacrt a oeaut h ada lcrclatvto eunei patients in sequence activation electrical cardiac the evaluate to way accurate most The noada o-otc apn a loietfidtodi two identified also has mapping non-contact Endocardial in muto Vatvto ea,nest epeetfrCTt ee be to CRT for present be to needs delay, activation LV of amount cient 40 37 hrceitcfidn ntu BBptet losest ealn ( long a be to seems also patients LBBB true in finding characteristic A nteehat,CTlreyrvre ucinladsrcua abnormalities. structural and functional reverses largely CRT hearts, these In 35 32–34 h lnclsgicneo hs nig a eoeeieti the in evident become has findings these of significance clinical The 36 42 nuto fLB nhatycnn erslast electrical to leads hearts canine healthy in LBBB of Induction ff 38–41 rn entoscmlct nfr diagnosis. uniform a complicate definitions erent 35 h ciainwv rn rgntn rmtergtven- right the from originating front wave activation The hswsarnoie ra hteautdtee the evaluated that trial randomized a was This te esrst mrv ain eeto o CRT for selection patient improve to measures Other < 3 s n vdneo mechanical of evidence and ms) 130 39,41,43 ff ff rn etlregions. septal erent rn atrso electrical of patterns erent hc ssmlrt the to similar is which , ffi in.LB is LBBB cient. 44 h second The > 39,40 0ms) 40 ff ect 27 In 28

2 Chapter 2 left ventricle.41,43 The varying conduction patterns observed in these mapping studies could be explained by variations in left bundle branch anatomy45 and the location of the block, but also by the fact that cellular uncoupling as a consequence of LV hypertrophy or fibrosis can give rise to a wide QRS complex with morphological features that meet conventional ECG criteria for LBBB.46,47 In contrast to LBBB, RBBB is typically associated with delayed RV activation, but not delayed LV activation. However, in some RBBB patients, the QRS morphology differs significantly from the characteristic RBBB pattern. These patients show a specific electrocardiographic pattern previously defined as RBBB masking LBBB48,49, which is characterized by precordial lead findings consistent with RBBB and limb lead findings consistent with LBBB. Extensive measurements of both RV and LV endocardial electrical activation in heart failure patients with RBBB using CARTO 3D contact mapping showed that patients with RBBB masking LBBB, have LV activation delay similar to that found in LBBB.50 Although the aforementioned mapping techniques provide accurate characterization of cardiac electrical activation, the application of these techniques in clinical practice is time- consuming, cumbersome, and not without risk. Measuring the Q-LV as described above provides a relatively simple manner of assessing the extent of LV activation delay. However, this technique provides limited information on LV electrical activation because usually measurements are only performed at the anatomically targeted region. A technique that provides a middle ground between complete mapping and single Q-LV measurement is intra-procedural coronary venous electro-anatomic mapping. In a recent study, we assessed the LV electrical activation in a cohort of 51 CRT candidates using this technique.51 A guidewire that allows for unipolar sensing and pacing was inserted into the coronary sinus and connected to an EnSite NavX system. The wire was then manipulated to various coronary sinus branches creating an anatomic map along with determining the electrical activation time associated with each anatomic region. Significant LV activation delay (> 75% of QRS duration) was found in 38 of 51 patients. QRS duration was shown to perform poorly in identifying delayed LV activation (area under the curve = 0.49). Twenty-nine of the 51 patients had LBBB according to specific ECG criteria which included broad, notched or slurred R waves in leads I, aVL,

V5 and V6, an occasional RS pattern in leads V5 and V6 attributed to displaced transition of the QRS complex, and absent q waves in lead I, V5 and V6 (in the absence of a large anterior- apical infarction). As described earlier, this refined LBBB definition, which includes the presence of QRS notching and slurring, has previously been shown to significantly improve the predictive value of LBBB QRS morphology for CRT response.52 Of the remaining 22 patients, 7 met ECG criteria for RBBB and 15 met neither criteria for LBBB nor RBBB and were classified as IVCD. QRS duration did not differ between different QRS morphologies. However, LV activation time was significantly larger in LBBB patients as compared to RBBB and IVCD patients. Significant LV activation delay was found in all patients diagnosed with

28 iue2.3: Figure is slurring, and notching QRS su lacks includes but which activation, LV from delayed definition, benefit for LBBB to specific highly substrate refined electrical the appropriate (2) include the likely and without most CRT, patients will mapping alone of delayed this duration of number marker QRS of substantial reliable on findings a a based not The selection is patient itself 2.3). even Thus, by Fig. and duration activation. QRS LV in IVCD prolonged shown with a (examples patients (1) RBBB 15 that: indicate with of study 8 patients in 7 also of but 1 criteria, in ECG specific to according LBBB usata ubro ainsta aedlydL ciainaentietfida uh and erroneously. such, withheld as be identified may not CRT are patients, activation LV these delayed in have that patients of number substantial etiua en ALV vein, ventricular AT eosrtn eae ciaino h Vatrltrlwl ( wall anterolateral LV the of activation delayed demonstrating RV h Vifrltrlwl ( wall inferolateral LV the ooayvnu lcr-ntmcmpo BBptetdmntaigdlydactiva- ( delayed wall demonstrating anterolateral LV patient the LBBB of a tion of map electro-anatomic venous Coronary = = lcrclatvto ie AP time, activation electrical ih etil Aatdfo a a tal. et Rad Mafi from (Adapted ventricle right = neoaea en ILV vein, anterolateral B ,adaRB ain ihaptnillf neirhemiblock anterior left potential a with patient RBBB a and ), te esrst mrv ain eeto o CRT for selection patient improve to measures Other A = ,a VDptetdmntaigdlydatvto of activation delayed demonstrating patient IVCD an ), neopseir L antero-posterior, ffi = in estvt.A osqec,a consequence, a As sensitivity. cient neoaea en CS vein, inferolateral 51 ). / RAO = C left .AIV ). / ih neiroblique, anterior right = = ooaysinus, coronary neirinter- anterior 29

2 Chapter 2

Instead, the above described technique of coronary venous electro-anatomic mapping can be used at the time of CRT implantation for a more precise characterization of the electrical substrate at only minor prolongation of procedure time (∼20 min).51,53 However, ideally the decision whether or not to implant a CRT device is made in advance. In this respect, electrocardiographic imaging (ECGi) provides an entirely non-invasive alternative.54 ECGi provides high-resolution non-invasive electrical mapping of the epicardial electrical activation. The technique acquires electrical data from more than 200 body surface electrodes using a multi- electrode vest. Epicardial anatomy and body-surface electrode positions are registered simultaneously by a thoracic computed tomography scan. The body-surface electrical data and the anatomical data are then processed with algorithms to construct epicardial depolarization and repolarization patterns, using a single heartbeat.54 In this way, detailed information on LV electrical activation can be readily obtained prior to CRT implantation, which may be used to guide the decision on whether or not to implant a CRT device. However, the requirement for a multi-electrode vest in combination with a computed tomography scan may preclude widespread application of this technique in clinical practice.

2.4 Better electrocardiographic identiﬁcation of the electrical substrate: new ECG parameters

The demand for easy and widely applicable non-invasive techniques that can be used to accurately characterize the electrical substrate in CRT candidates has renewed the interest in finding additional/alternative electrocardiographic markers of dyssynchrony. Sweeney et al. 19 carefully analyzed standard 12-lead ECGs of 202 CRT candidates with LBBB according to specific ECG criteria that included QRS notching/slurring and identified new measurements that predict volumetric CRT response.19 The time difference between the first notch after 40 ms of QRS onset and the end of the QRS on the baseline ECG was indicated as the LV activation time (LVATmax, Fig. 2.4). A longer LVATmax was shown to be predictive of CRT response (OR [CI] = 1.30 [1.11 – 1.52] for each 10 ms increase up to 125 ms). In addition, the Selvester QRS score for LBBB was used to quantify LV scar extent. A higher Selvester score was negatively associated with reverse remodeling (OR [CI] = 0.49 [0.27 – 0.88] for each 1-point increase from 0 to 4; 0.92 [0.83 – 1.01] for each 1-point increase > 4).19 Recently, the value of the vectorcardiogram (VCG) for characterizing the electrical substrate and predicting CRT response has been explored. VCG is a technique that records the magnitude and direction of the electrical forces that are generated by the heart over time, resulting in a resultant electrical force depicted by a vector for each time-point. Connecting the arrow heads of all vectors, a vector loop is constructed. The VCG thus contains 3D information of the electrical forces within the heart, which might provide more valuable information than the 1D-ECG. It was hypothesized that large electrical dyssynchrony, amenable to CRT,

30 ieo hs ocsmyb elrpeetdb h QRS the by represented well be the may that forces and these depolarization ventricular of during size forces electrical unopposed large to lead would (QRS iue2.4: Figure VDadta QRS that and IVCD QRS that observation the by n slata oda h otrfie BBdefinition. LBBB LBBB defined refined conventionally most the then as and good QRS as duration Moreover, least QRS as response. and than CRT better volumetric response long-term CRT of predicted odds high with associated was ntetrepicpedrcin.vnDusne al. et Deursen van directions. principle three the in h ointa QRS that notion The area nteVGi 1cneuieCTcniae n hwdta ag QRS large a that showed and candidates CRT consecutive 81 in VCG the on ) QRS stetm di time the as xml falf etiua ciaintm LA)maueet LVAT (LVAT) measurement. time activation ventricular left a of Example area ff area slwri shmcta nnnicei patients. non-ischemic in than ischemic in lower is rnebtentefis oc fe 0m fQSostadteedo the of end the and onset QRS of ms 40 after notch first the between erence ersnsteetn fuopsdeetia ocsi supported is forces electrical unopposed of extent the represents area slre nptet ihLB scmae optet with patients to compared as LBBB with patients in larger is te esrst mrv ain eeto o CRT for selection patient improve to measures Other 24 sesdteae fteQScomplex QRS the of area the assessed 24 area h rao h R complex QRS the of area the , 24 ute support Further max smeasured is area area 31

2 Chapter 2 comes from observations in the above mentioned study on coronary venous mapping. In this study, VCGs were constructed from pre-procedural standard 12-lead ECGs for all patients using the Kors algorithm. A large QRSarea (> 69 µVs) on the VCG was shown to be highly predictive of delayed LV lateral wall activation as determined by coronary venous mapping 51 (Fig. 2.5). On the other hand, QRSarea has been shown to be smaller in patients with heart failure of ischemic etiology, which may be explained by the presence of non-conductive 24 fibrotic tissue. Taken together, these observations suggest that QRSarea is not only useful to determine the extent of electrical dyssynchrony, but that it may also reflect the presence of determinants known to reduce the chance of CRT benefit, such as an ischemic etiology of heart failure. However, more research is required to better understand all determinants of

QRSarea.

Figure 2.5: QRSarea plotted as a function of maximal LV lateral wall activation time (maxLVLW-AT) expressed as % of QRS duration (QRSd) for all patients (each dot represents a patient, n = 51) with LBBB diagnosed according to the definition provided by the REVERSE trial (left) and the American Heart Association (AHA) definition (right). This figure demon-

strates the excellent diagnostic performance of QRSarea > 69 µVs for delayed LV lateral wall activation (defined as a maxLVLW-AT exceeding 75% of QRS duration), independent of the QRS morphology on the surface ECG, and illustrates the difference in QRS morphology classification caused by disparity in LBBB definitions (Adapted from Mafi Rad et al. 51 ).

Interestingly, two studies showed that VCG-derived measures of repolarization predict 55 CRT response even better than QRSarea. Engels et al. assessed the Tarea from VCGs of

244 CRT recipients (VCG examples shown in Fig. 2.6C and D). The VCG-derived Tarea 55 was shown to predict echocardiographic CRT response better than QRSarea. In a larger cohort consisting of 335 CRT recipients in which the primary endpoint was the composite of heart failure hospitalization, heart transplantation, left ventricular assist device implantation or death during a 3-year follow-up period, the predictive power of Tarea for CRT response

32 hneo survival. of chance a on ob rmrl vdn ntegopo ainswt BB(i.26 n F). and 2.6E (Fig. LBBB with patients of group T the in large evident A primarily be to found was QRS lopa oe nti td agrT larger a study this In role. a play also ocsdrn h eoaiainpae h T The phase. repolarization the during forces eeue.Frhroe h tde eae otepeito fCTrsos sn the using response CRT of prediction study. the prospective to related studies the QRS Furthermore, used. other were which investigate to needed is T research the in Further reflected JT-interval. exactly are longer factors a by much so not nojciemne n unie scniuu aibe,a poe oLB hc sa di is of which use LBBB the to to subject opposed is as that variables, measurement continuous dichotomous as quantified and manner objective an epeain fQSnotching QRS of terpretations h odsadr rn-C n aercnl lobe aiae o s nptet with patients in use for validated been failure. also heart recently dyssynchronous have and Frank-VCG standard gold inverse the the using transformation. ECGs regression 12-lead Kors’ standard or from Dower VCGs ECG construct available commercially to Most algorithms ECG. have 12-lead machines standard the from derived be easily can they ie ihteecletpeitv oe fQRS of power predictive excellent the with bined ciaindly h rmr lcrclsbtaefrCT n ntebte rdcinof prediction better the on and CRT, QRS for by QRS substrate response that electrical CRT mapping primary electro-anatomic the from delay, obtained activation evidence the on Based Conclusions 2.5 candidates therapy. appropriate this identify to response to improving practice potentially clinical thereby in CRT, for applied easily be can parameters these ntegieie saslcincieinfrCTipatto.Tepsil vnbetter even possibly The further requires complex QRS implantation. the than CRT investigation. rather T-wave for the using criterion by response selection CRT of a prediction as guidelines the in iiaino l hs tde eadn h QRS the regarding studies these all of limitation A h ra rcia eeto QRS of benefit practical great The area area 55 eealrtopcie hrfr,teerslsne ob aiae nalarger a in validated be to need results these Therefore, retrospective. all were u te atr uha hne nK in changes as such factors other but , area nLB ainswsascae ihls Fhsiaiain n higher a and hospitalizations HF less with associated was patients LBBB in 56 h ieo h T the of size The area scmae oQSdrto,w rps oicueQRS include to propose we duration, QRS to compared as 59 / lrig nte rcia etr fQRS of feature practical Another slurring. h o-naieadsml aueo C nlsscom- analysis VCG of nature simple and non-invasive The area area area . te esrst mrv ain eeto o CRT for selection patient improve to measures Other area n T and sarflcino h xeto npoe electrical unopposed of extent the of reflection a is 57,58 a rmrl asdb agrapiueand amplitude larger a by caused primarily was area area hs Cspoieago eebac of resemblance good a provide VCGs These area + sprilydtrie ytesz fthe of size the by determined partially is sta hs aaeesaemaue in measured are parameters these that is n T and n Ca and area ff sta eaieysalsml sizes sample small relatively that is area rn entosadsbetv in- subjective and definitions erent 2 + o R epneidctsthat indicates response CRT for o hne xrsinmight expression channel ion area n T and area eet LV reflects area sthat is area 33 56

2 Chapter 2

Figure 2.6: Typical example of VCGs constructed from standard 12-lead ECGs for a patient with a

large (A and C) and a patient with a small (B and D)Tarea, despite being both classified as having LBBB. Panels E and F show Kaplan-Meier estimates of the probability free from the composite endpoint HTLD (HF hospitalization, heart transplantation, LVAD implantation, death) after 3 years of CRT. Large QRS or T area are values ≥ median value and small QRS or T area are values < median value (adapted from Engels et al. 55 and Végh et al. 56 ). 34 References .C rc,A pti,DDra,J iac,S ubr A se r,T Fer- TB 3rd, Estes NAM Dunbar, SB Dimarco, JP Darbar, D Epstein, AE Tracy, CM 4. Breithardt, OA Boriani, G Bordachar, P Baron-Esquivias, G Auricchio, A Brignole, M 3. .Erpa er htmAscain uoenSceyo adooy er Rhythm Heart Cardiology, of Society European Association, Rhythm Heart European 5. .JFCead CDuet rmn,NFemnl,DGa,LKpebre,L Kappenberger, L Gras, D Freemantle, N Erdmann, E Daubert, JC Cleland, JGF 1. .S roi lmtö-udvs,AMzat,NBo,MBrgee am PM Camm, J Borggrefe, M Blom, N Mazzanti, A Blomström-Lundqvist, C Priori, SG 7. .W baa,W ihr LSih BDlri,A en o,D Kocovic, DZ Loh, E Leon, AR Delurgio, DB Smith, AL Fisher, WG Abraham, WT 2. .WMles AGim eg,TDeig CSaln,B Wilko BL Starling, RC Dresing, T Verga, T Grimm, RA Mullens, W 6. soito akFreo rcieGuidelines. Practice on Force Task Association aiis eoto h mrcnCleeo adooyFoundation Cardiology of College Abnor- American Rhythm the Cardiac of of report Therapy a Device-Based malities: for Guidelines 2008 the of Update uo r CHmil EKrsk SLn,e l 02ACCF 2012 al. et Link, MS Karasik, PE Hammill, SC Jr, guson (EHRA). Association Rhythm 34:2281–2329. Developed Heart (2013) European (ESC). the Cardiology and with of pacing collaboration Society cardiac in European cardiac on the on Force Guidelines of Task ESC therapy the 2013 resynchronization therapy: al. resynchronization et Elliott, cardiac PM and Delgado, pacing V Deharo, JC Cleland, J eto ada eycrnzto hrp nhatfiue mln n olwu re- follow-up and implant failure: management. heart and commendations in therapy resynchronization cardiac on ment r soito,J abr,LSxn ta.21 EHRA 2012 al. et Saxon, L Fail- Daubert, Heart JC Echocardiography, Echocardiography, Association, of of ure Association Society European American Association, America, Heart of American Society Failure Heart Society, aaz,adCricRsnhoiainHatFiue(AEH)SuyInvestigat- e Study The (CARE-HF) ors. Failure Resynchronization-Heart Cardiac and Tavazzi, ag nihsfo ada eycrnzto piiaincii spr faheart a of part as program. clinic management optimization disease resynchronization failure cardiac a from Insights Tang. et:TeTs oc o h aaeeto ainswt etiua Arrhythmias Ventricular with Patients of Management the cardiac for sudden Force of man- Task prevention the The the for and death: Guidelines arrhythmias ESC ventricular 2015 with al. patients et of Hindricks, agement G Hatala, R Fitzsimons, D Elliott, nlJMed J Engl N akr LCael LHys ta.Cricrsnhoiaini hoi heart chronic in resynchronization Cardiac al. et Hayes, failure. DL Clavell, AL Packer, M nlJMed J Engl N ff c fcricrsnhoiaino obdt n otlt nhatfailure. heart in mortality and morbidity on resynchronization cardiac of ect 20)352:1539–1549. (2005) 20)346:1845–1853. (2002) er Rhythm Heart mCl Cardiol Coll Am J er Rhythm Heart 21)9:1524–1576. (2012) / 20)53:765–773. (2009) R xetcnessstate- consensus expert HRS 21)9:1737–1753. (2012) / AHA / mrcnHeart American / ff R Focused HRS n WHW and , u er J Heart Eur

2 Chapter 2

and the Prevention of Sudden Cardiac Death of the European Society of Cardiology (ESC)Endorsed by: Association for European Paediatric and Congenital Cardiology (AEPC). Eur Heart J (2015) 36:2793–2867. 8. S Cazeau, P Ritter, S Bakdach, A Lazarus, M Limousin, L Henao, O Mundler, JC Daubert, and J Mugica. Four chamber pacing in dilated cardiomyopathy. Pacing Clin Electrophysiol (1994) 17:1974–1979. 9. PF Bakker, HW Meijburg, JW de Vries, MM Mower, AC Thomas, ML Hull, EO Robles De Medina, and JJ Bredée. Biventricular pacing in end-stage heart failure improves functional capacity and left ventricular function. J Interv Card Electrophysiol (2000) 4:395–404. 10. C Linde, C Leclercq, S Rex, S Garrigue, T Lavergne, S Cazeau, W McKenna, M Fitzgerald, JC Deharo, C Alonso, S Walker, F Braunschweig, C Bailleul, and JC Daubert. Long-term benefits of biventricular pacing in congestive heart failure: results from the MUltisite STimulation in cardiomyopathy (MUSTIC) study. J Am Coll Cardiol (2002) 40:111–118. 11. MGSJ Sutton, T Plappert, KE Hilpisch, WT Abraham, DL Hayes, and E Chinchoy. Sustained reverse left ventricular structural remodeling with cardiac resynchronization at one year is a function of etiology: quantitative Doppler echocardiographic evidence from the Multicenter InSync Randomized Clinical Evaluation (MIRACLE). Circula- tion (2006) 113:266–272. 12. MR Bristow, LA Saxon, J Boehmer, S Krueger, DA Kass, T De Marco, P Carson, L DiCarlo, D DeMets, BG White, DW DeVries, AM Feldman, and Comparison of Medical Therapy, Pacing, and Defibrillation in Heart Failure (COMPANION) Invest- igators. Cardiac-resynchronization therapy with or without an implantable defibrillator in advanced chronic heart failure. N Engl J Med (2004) 350:2140–2150. 13. PE Vardas, A Auricchio, JJ Blanc, JC Daubert, H Drexler, H Ector, M Gasparini, C Linde, FB Morgado, A Oto, R Sutton, M Trusz-Gluza, European Society of Cardi- ology, and European Heart Rhythm Association. Guidelines for cardiac pacing and cardiac resynchronization therapy. The Task Force for Cardiac Pacing and Cardiac Resynchronization Therapy of the European Society of Cardiology. Developed in collaboration with the European Heart Rhythm Association. Europace (2007) 9:959–998. 14. C Linde, WT Abraham, MR Gold, M St. John Sutton, S Ghio, C Daubert, and RE- VERSE (REsynchronization reVErses Remodeling in Systolic left vEntricular dysfunction) Study Group. Randomized Trial of Cardiac Resynchronization in Mildly Symptomatic Heart Failure Patients and in Asymptomatic Patients With Left Ventricu- lar Dysfunction and Previous Heart Failure Symptoms. Journal of the American Col- lege of Cardiology (2008) 52:1834–1843.

36 0 in hn,XL,YGo h,LWn,DL,JWn,CYa,adJGuo. J and Yuan, C Wang, J Li, D Wang, L Zhu, T Gao, Y Li, X Zhang, P Tian, Y 20. 2 RGl,CTéal,CLne TArhm erte ho tJh Sutton, John St M Ghio, S Gerritse, B Abraham, WT Linde, C Thébault, C Gold, MR 22. 1 aea li,ICgniwc,W al cit rw,DCannom, D Brown, M McNitt, S Hall, WJ Cygankiewicz, I Klein, H Zareba, W 21. 5 JMs,W al SCno,HKen WBon PDuet A se r,E 3rd, Estes NAM Daubert, JP Brown, MW Klein, H Cannom, DS Hall, WJ Moss, AJ 15. 7 RByn,S itn PLi n VEnr soito ewe R duration QRS between Association Exner. DV and Lai, MP Wilton, SB SH Bryant, AR Connolly, S 17. Sheldon, R Arnold, MO Talajic, M Wells, GA Tang, ASL 16. 9 OSeny JvnBme,M cai,CWBorle CJW Schalij, MJ Bommel, van RJ Sweeney, MO 19. P McMurray, J Linde, C Daubert, JC Auricchio, A Vardas, PE Dickstein, K 18. fhatfiueevents. heart-failure of prevention the for therapy Cardiac-resynchronization Investigators. Trial MADIT-CRT otr rebr,S ign,M Pfe MA Higgins, SL Greenberg, H Foster, n ucm ihcricrsnhoiainteay ytmtcrve n meta- and review systematic a therapy: analysis. resynchronization cardiac with outcome and and Rouleau, JL Healey, 363:2385–2395. failure. JS Investigators.(2010) heart Yee, Trial mild-to-moderate for R Failure therapy Sapp, Heart Cardiac-resynchronization JL Ambulatory Birnie, for DH Resynchronization-Defibrillation Nichol, G Hohnloser, etvnrclrrvrevlmti eoeigdrn ada eycrnzto ther- resynchronization predict cardiac to during apy. electrocardiography remodeling volumetric surface reverse using ventricular activation left ventricular of Analysis Bax. recmlt etbnl rnhbokmrhlg togypeit odresponse good predicts therapy. strongly resynchronization morphology cardiac block to branch bundle left complete True eycrnzto hrp.Dvlpdwt h pca otiuino h Heart and the cardiac Association. of for Rhythm contribution Guidelines Heart 12:1526–1536. European special ESC the the 2007 and with the Association treatment Failure Developed and and failure diagnosis the therapy. heart for for resynchronization chronic Guidelines in Committee ESC and therapy 2008 ESC device acute the on and of of Guidelines update Veldhuisen, ESC an of van failure: Update heart DJ Focused 2010 Sutton, Guidelines. R Practice Priori, SG Ponikowski, n CDuet E Daubert. JC and 123:1061–1072. eoeigi ytlcLf etiua yfnto RVRE study. Reverses 126:822–829. (REVERSE) Resynchronization (2012) Dysfunction the Ventricular from Left results Systolic in failure: Remodeling heart mild in outcomes therapy ra-ada eycrnzto hrp (MADIT-CRT). Therapy Resynchronization Implantation Defibrillator Automatic Multicenter Trial-Cardiac the in Morphology QRS by Therapy PDuet la,M od ta.E al. et Gold, MR Eldar, M Daubert, JP Circulation Electrocardiol J 21)121:626–634. (2010) ff c fQSdrto n opooyo ada resynchronization cardiac on morphology and duration QRS of ect nlJMed J Engl N 21)46:147–155. (2013) Europace 20)361:1329–1338. (2009) ff ff r DSlmn ibr aea and Zareba, W Wilber, D Solomon, SD er, cieeso ada Resynchronization Cardiac of ectiveness 21)15:1499–1506. (2013) ff ,A elap n JJ and Hellkamp, AS s, Circulation Europace nlJMed J Engl N Circulation References (2010) (2011) 37

2 Chapter 2

23. DG Strauss, RH Selvester, and GS Wagner. Defining left bundle branch block in the era of cardiac resynchronization therapy. Am J Cardiol (2011) 107:927–934. 24. CJM van Deursen, K Vernooy, E Dudink, L Bergfeldt, HJGM Crijns, FW Prinzen, and L Wecke. Vectorcardiographic QRS area as a novel predictor of response to cardiac resynchronization therapy. J Electrocardiol (2015) 48:45–52. 25. JG Cleland, WT Abraham, C Linde, MR Gold, JB Young, J Claude Daubert, L Sherfe- see, GA Wells, and ASL Tang. An individual patient meta-analysis of five randomized trials assessing the effects of cardiac resynchronization therapy on morbidity and mortality in patients with symptomatic heart failure. Eur Heart J (2013) 34:3547–3556. 26. R Zusterzeel, KA Selzman, WE Sanders, DA Caños, KM O’Callaghan, JL Carpenter, IL Piña, and DG Strauss. Cardiac resynchronization therapy in women: US Food and Drug Administration meta-analysis of patient-level data. JAMA Intern Med (2014) 174:1340–1348. 27. G Mascioli, L Padeletti, B Sassone, M Zecchin, E Lucca, S Sacchi, G Boggian, AL Tondo, C Belvito, N Bakhtadze, A Borrelli, and G Sinagra. Electrocardiographic criteria of true left bundle branch block: a simple sign to predict a better clinical and instrumental response to CRT. Pacing Clin Electrophysiol (2012) 35:927–934. 28. K Vernooy, RNM Cornelussen, XAAM Verbeek, WYR Vanagt, A van Hunnik, M Kuiper, T Arts, HJGM Crijns, and FW Prinzen. Cardiac resynchronization therapy cures dyssynchronopathy in canine left bundle-branch block hearts. Eur Heart J (2007) 28:2148–2155. 29. L Liu, B Tockman, S Girouard, J Pastore, G Walcott, B KenKnight, and J Spinelli. Left ventricular resynchronization therapy in a canine model of left bundle branch block. Am J Physiol Heart Circ Physiol (2002) 282:H2238–H2244. 30. JP Singh, D Fan, EK Heist, CR Alabiad, C Taub, V Reddy, M Mansour, MH Picard, JN Ruskin, and T Mela. Left ventricular lead electrical delay predicts response to cardiac resynchronization therapy. Heart Rhythm (2006) 3:1285–1292. 31. MR Gold, U Birgersdotter-Green, JP Singh, KA Ellenbogen, Y Yu, TE Meyer, M Seth, and PJ Tchou. The relationship between ventricular electrical delay and left ventricular remodelling with cardiac resynchronization therapy. Eur Heart J (2011) 32:2516–2524. 32. MF van Oosterhout, FW Prinzen, T Arts, JJ Schreuder, WY Vanagt, JP Cleutjens, and RS Reneman. Asynchronous electrical activation induces asymmetrical hypertrophy of the left ventricular wall. Circulation (1998) 98:588–595. 33. BT Wyman, WC Hunter, FW Prinzen, OP Faris, and ER McVeigh. Effects of single- and biventricular pacing on temporal and spatial dynamics of ventricular contraction. Am J Physiol Heart Circ Physiol (2002) 282:H372–H379.

38 6 ahrv,AMtai,RKas,F rne,AArcho n os.The Potse. M and Auricchio, A Prinzen, FW Krause, R Mateasik, A Bacharova, L 46. 5 CDmui n EKlets itptooia xmnto fcneto left of concept of examination Histopathological Kulbertus. HE therapy: and resynchronization Cardiac Demoulin JC Prinzen. FW and 45. Vernooy, K Ploux, S Strik, M 44. 3 W ug Y hn W i,HKCa,WLCa,QZag n MY.Ef- Yu. CM and Zhang, Q Chan, WWL Chan, HCK Yip, GWK Chan, JYS Fung, JWH 43. useful dyssynchrony of assessment echocardiographic Is Auricchio. A and Prinzen FW Variable Sanderson. JE 42. and Kum, CC Chan, H Zhang, Y Yip, G Yu, CM Fung, JWH pat- Variable 41. Wellens. HJJ and Beatty, G Nabar, A Timmermans, C Rodriguez, LM 40. 9 uici,CFnoi eoi abccho ote elr ls,and Kloss, M Geller, C Goette, A Carbucicchio, C Regoli, F Fantoni, C Auricchio, A 39. 7 eno,XA ebe,MPshr JMCin,TAt,RMCreusn and Cornelussen, RNM Arts, T Crijns, HJGM Peschar, M Verbeek, XAAM Vernooy, K 37. 8 AVsal,D asd,F aclnk,A utn LWxa,J oet,and Doherty, JU Waxman, HL Buxton, AE Marchlinski, FE Cassidy, DM Vassallo, JA 38. 6 ti,L a idnop n eno.Aia oeso dyssynchrony. of models Animal Vernooy. K and Middendorp, van LB Strik, M 36. 5 uciza TArhm PSnh JBx SBrr rgd,KDcsen I Dickstein, K Brugada, J Borer, JS Bax, JJ Singh, JP Abraham, WT Ruschitzka, F 35. 4 ues eha,BKr,adTAt.Trewl emn Tie)mdldescrib- model (TriSeg) segment Three-wall Arts. T and Kirn, B Delhaas, T Lumens, J 34. hypertrophy. e hemiblock. substrate. electrical the on refocus epnet ada eycrnzto therapy. clinical resynchronization and cardiac echocardiographic to on response pattern activation endocardial ventricular left of fect available. is duration QRS if not Imaging therapy is Cardiovasc resynchronization Echocardiography cardiac therapy? before resynchronization useful cardiac for candidates select to branch bundle left and failure heart with patients block. in pattern activation ventricular left failure. heart and block branch bundle left with Electrophysiol Cardiovasc patients in activation septal of terns li.Caatrzto flf etiua ciaini ainswt er failure heart with patients in activation block. ventricular bundle-branch left left and of Characterization Klein. H 69:914–923. WPizn etbnl rnhbokidcsvnrclrrmdligadfunctional and remodelling ventricular hypoperfusion. induces septal block branch bundle Left Prinzen. FW adoacTas Res Transl Cardiovasc EJspsn noada ciaino etbnl rnhblock. branch bundle left of activation Endocardial Josephson. ME nlJMed complex. J QRS Study Engl narrow Echo-CRT a N and with Holzmeister, failure heart J in Sogaard, therapy P Cardiac-resynchronization Krum, Group. H Gras, D 3rd, Gorcsan J Ford, 37:2234–2255. n ehnc n eoyaiso etiua interaction. ventricular of hemodynamics and mechanics ing ﬀ c frdcditrellrculn neetoadorpi in flf ventricular left of signs electrocardiographic on coupling intercellular reduced of ect Heart rHatJ Heart Br Electrocardiol J 20)90:17–19. (2004) 21)369:1395–1405. (2013) 20)1:70–77. (2008) u er J Heart Eur 21)5:135–145. (2012) 17)34:807–814. (1972) 20)14:135–141. (2003) Circulation 21)44:571–576. (2011) icJ Circ 20)26:91–98. (2005) 21)75:1297–1304. (2011) 20)109:1133–1139. (2004) Heart 20)93:432–437. (2007) n imdEng Biomed Ann Circulation References (1984) (2009) Circ 39 J J

2 Chapter 2

47. L Bacharova, V Szathmary, and A Mateasik. Electrocardiographic patterns of left bundle-branch block caused by intraventricular conduction impairment in working myocardium: a model study. J Electrocardiol (2011) 44:768–778. 48. PN Unger, ME Lesser, VH Kugel, and M Lev. The concept of masquerading bundle-branch block; an electrocardiographic-pathologic correlation. Circulation (1958) 17:397–409. 49. JL Richman and L Wolff. Left bundle branch block masquerading as right bundle branch block. Am Heart J (1954) 47:383–393. 50. C Fantoni, M Kawabata, R Massaro, F Regoli, S Raffa, V Arora, JA Salerno-Uriarte, HU Klein, and A Auricchio. Right and left ventricular activation sequence in patients with heart failure and right bundle branch block: a detailed analysis using three- dimensional non-fluoroscopic electroanatomic mapping system. J Cardiovasc Electro- physiol (2005) 16:112–9, 112–9. 51. M Mafi Rad, GWM Wijntjens, EB Engels, Y Blaauw, JGLM Luermans, L Pison, HJ Crijns, FW Prinzen, and K Vernooy. Vectorcardiographic QRS area identifies delayed left ventricular lateral wall activation determined by electroanatomic mapping in candidates for cardiac resynchronization therapy. Heart Rhythm (2016) 13:217–225. 52. CJM van Deursen, Y Blaauw, MI Witjens, L Debie, L Wecke, HJGM Crijns, FW Prinzen, and K Vernooy. The value of the 12-lead ECG for evaluation and optimization of cardiac resynchronization therapy in daily clinical practice. J Electrocardiol (2014) 47:202–211. 53. AMW van Stipdonk, MM Rad, JGLM Luermans, HJ Crijns, FW Prinzen, and K Vernooy. Identifying delayed left ventricular lateral wall activation in patients with non-specific intraventricular conduction delay using coronary venous electroanatomical mapping. Neth Heart J (2016) 24:58–65. 54. C Ramanathan, RN Ghanem, P Jia, K Ryu, and Y Rudy. Noninvasive electrocardiographic imaging for cardiac electrophysiology and arrhythmia. Nat Med (2004) 10:422–428. 55. EB Engels, EM Végh, CJM Van Deursen, K Vernooy, JP Singh, and FW Prinzen. T- wave area predicts response to cardiac resynchronization therapy in patients with left bundle branch block. J Cardiovasc Electrophysiol (2015) 26:176–183. 56. EM Végh, EB Engels, CJM van Deursen, B Merkely, K Vernooy, JP Singh, and FW Prinzen. T-wave area as biomarker of clinical response to cardiac resynchronization therapy. Europace (2016) 18:1077–1085. 57. JA Kors, G van Herpen, AC Sittig, and JH van Bemmel. Reconstruction of the Frank vectorcardiogram from standard electrocardiographic leads: diagnostic comparison of different methods. Eur Heart J (1990) 11:1083–1092.

40 8 dnrntadOPhm etradormsnhszdfo 2la C:su- ECG: 12-lead a from synthesized Vectorcardiogram Pahlm. O and Edenbrandt L 58. 9 BEgl,SAser,CMvnDusn ek,LBrflt eno,adFW and Vernooy, K Bergfeldt, L Wecke, L Deursen, van CJM Alshehri, S Engels, EB 59. eirt fteivreDwrmatrix. Dower inverse the of periority nptet ihdsycrnu er failure. heart dyssynchronous with vectorcardiogram patients measured in the resembles vectorcardiogram synthesized The Prinzen. te esrst mrv ain eeto o CRT for selection patient improve to measures Other Electrocardiol J Electrocardiol J 18)21:361–367. (1988) 21)48:586–592. (2015) 41

Chapter 3 The synthesized vectorcardiogram resembles the measured vectorcardiogram in patients with dyssynchronous heart failure

The content of this chapter is based on:

Elien B. Engels,1 Salih Alshehri,1 Caroline J.M. van Deursen,1 Liliana Wecke,2 Lennart Bergfeldt,2 Kevin Vernooy,3 and Frits W. Prinzen1 (2015). The synthesized vectorcardiogram resembles the measured vectorcardiogram in patients with dyssynchronous heart failure. J Electrocardiol 48(4):586-592.

1 Department of Physiology, CARIM School for Cardiovascular Diseases, Maastricht University, Maastricht, the Netherlands. 2 Department of Cardiology, Karolinska Institutet, Karolinska University Hospital, Stockholm, Sweden. 3 Department of Cardiology, Maastricht University Medical Center, Maastricht, the Netherlands. Chapter 3

Abstract

Introduction: The use of vectorcardiography (VCG) has regained interest, however, original Frank-VCG equipment is rare. This study compares the measured VCGs with those synthesized from the 12-lead electrocardiogram (ECG) in patients with heart failure and conduction abnormalities, who are candidate for cardiac resynchronization therapy (CRT).

Methods and Results: In 92 CRT candidates, Frank-VCG and 12-lead ECG were recorded before CRT implantation. The ECG was converted to a VCG using the Kors method (Kors- VCG) and the two methods were compared using correlation and Bland-Altman analyses. Variables calculated from the Frank- and Kors-VCG showed correlation coeﬃcients between 0.77 and 0.90. There was a signiﬁcant but small underestimation by the Kors-VCG method, relative bias ranging from -1.9 ± 4.6% (QRS-T angle) to -9.4 ± 20.8% (Tarea).

Conclusions: The present study shows that it is justiﬁed to use Kors-VCG calculations for VCG analysis, which enables retrospective VCG analysis of previously recorded ECGs in studies related to CRT.

44 piiaino h R settings CRT the of optimization mrsieo ru ee,tebnfisi niiul aycnieal;tenon-response the 30-50%. considerably; still vary is individuals therapy in this benefits to rate the level, group a on impressive htteVGcudpa nipratrl nteslcino ainsfrCRT for patients of selection the in role important shown have an studies play Recent could improved. VCG be the should that selection patient products, expensive of use huhtems omnyue n ste8eetoesse codn oFrank. to according system 8-electrode the is one used commonly most the though VCG. the the of from revival synthesized a be in can resulting loops ECG vector 12-lead the technology computer today’s with and subsided diagnosis. eebe h rn-C npeiu studies previous in Frank-VCG the resembles hs nomto ewe hs ed.Ti ehiu a rtdsrbd11yasago years 101 described first was technique This Williams. leads. by these between information phase W Introduction 3.1 C n h osVGi ainswt er alr n Vcnuto ea ousn on ventricles. focussing the delay of conduction LV resynchronization and to failure relevant heart variables with vectorcardiographic patients in Kors-VCG the and VCG ayo hs ytm,adtecmlxt oitrrttedi the interpret to complexity the and equip- systems, system recording these VCG-lead VCG of standard special many a for of need the lack of the because ment, standard, clinical the became (ECG) hrp CT.CThsbe hw oipoecricpm ucin er alr symp- failure survival. heart and function, life, pump of cardiac quality improve resynchronization to toms, shown cardiac been with has treated CRT commonly (CRT). bundle are therapy left patients to due These mostly (LBBB). delay, conduction block (LV) ventricular branch left a and failure heart with tients ad o lcrcrigah CE utla irr,icuigbt ainsadhealthy and regression. linear patients multiple both by generated including was library, and individuals, multilead (CSE) Electrocardiography for dards ed n l i rcrillas yamti.Rcn tde eosrtdta h Kors- the that demonstrated Frank-VCG. the studies of Recent approximation best matrix. the limb a in (two by results leads VCG leads) ECG derived precordial independent 8 six multiplying all by and achieved synthesized leads is commonly This is ECG. VCG 12-lead the the systems, from recording Frank-VCG of availability limited to eea ytm ftreotoomllashv endsrbdfrrcrigteVCG, the recording for described been have leads orthonormal three of systems Several h arxpooe yKr tal. et Kors by proposed matrix The 5,6 1 in.TeVGcnit ftreotoomllasX ,adZ containing Z, and Y, X, leads orthonormal dimen- three three of in consists displayed and VCG recorded The are heart sions. the by generated forces trical ith h neeti h igotcvleo h C a,hwvr ee completely never however, has, VCG the of value diagnostic the in interest The h C ehiu a lotaadndadte1-edelectrocardiogram 12-lead the and abandoned almost was technique VCG The h s fvcocrigah VG,tesz n ieto fteelec- the of direction and size the (VCG), vectorcardiography of use the 14 11 10 hrfr,tecretsuyam ocmaeteFrank- the compare to aims study current the Therefore, . ordc h iko opiain n h unnecessary the and complications of risk the reduce To lhuhtee the Although 8 sbsdo erigstfo h omnStan- Common the from set learning a on based is 5,7–9 2–4 oprsnhsnvrbe aefrpa- for made been never has comparison a , h mrciaiyo akeetoein electrode back a of impracticality the , ff cso R nlreciia rasare trials clinical large in CRT of ects osVGrsmlsFrank-VCG resembles Kors-VCG 8 lhuhteKr-C nicely Kors-VCG the Although ff rn opmrhlge for morphologies loop erent 5,7–9 12,13 ri the in or 4 Due 45

3 Chapter 3

3.2 Methods

3.2.1 Patient population

The patient population used in this study has been described previously.13 It consisted of 138 consecutive patients with heart failure, who were scheduled for implantation of a CRT device at the Maastricht University Hospital between September 2010 and June 2012. Excluded were patients with an intrinsic QRS duration < 120 ms (n = 13) or previous RV pacing during either the VCG (n = 22) or ECG recording (n = 2). Another 9 patients were excluded due to technical disturbances, multiple ectopic beats or missing ECGs. This resulted in a ﬁnal population of 92 patients. The project was approved by the ethics committee of Maastricht University Hospital and was conducted in accordance with the Declaration of Helsinki. All participants gave written informed consent prior to investigation.

3.2.2 Study design

One day before CRT implantation, a VCG and 12-lead ECG were recorded. In some cases (n = 25) the 12-lead ECG was not recorded at the same date, but near this date (with a median of 12 days, ranging between 1 and 57 days). Both VCG and ECG were recorded at rest and in supine position. The VCGs were recorded as described earlier13, using 8-electrodes according to the Frank orthogonal lead system at a sampling frequency of 500 Hz for 5 minutes after which the complexes were averaged over a period of one minute (Coronet II System, Ortivus AB, Danderyd, Sweden). The ECGs were recorded at a frequency of 250 Hz and stored digitally as PDF ﬁles in the MUSE Cardiology Information system (GE Medical System). The digital data of the 12- lead ECGs were extracted from the PDF ﬁles using Inkscape version 2 (Boston, MA, USA) as described previously.12 A VCG was then synthesized by multiplying the voltages in the digital ECG leads by the Kors matrix.7

3.2.3 VCG analysis

Both the Frank-VCG and the Kors-VCG were analyzed offline using customized software described previously.12,15 This software calculates the different VCG variables described here. The QRS duration and QT interval were defined as the duration between the beginning of the QRS complex and the end of the QRS complex or T-wave, respectively. Subsequently, the maximal QRS vector and T vector were found by the maximal distance between the origin of the 3D loop and a point on the QRS- and T-vector loop, respectively. The size and direction of these maximal vectors were expressed by the vector amplitude, azimuth (angle in the transversal plane with backward vector directions being negative) and elevation (angle

46 n acltdvle sdfie sba n h 5 pe n oe iiso agreement of limits lower and upper 95% the bias and as bias defined were as (LoA) defined is values calculated and ietoso h aia etr.I ain h hp fteloswr lgtydi slightly were loops the and of amplitudes shape as the well 2 as patient loops In the resemblances of vectors. large shape maximal showed the similar 1 Patient of very directions with 3.3. methods, Fig. of two plots the Bland-Altman indicated between also the are in They squares 3.1. Fig. red in by presented are Kors-VCG and Frank-VCG the of Examples Kors-VCG versus Frank-VCG 3.3.2 fibrillation atrial CRT. had receiving patients quarter These (LVEF). for fraction a typical ejection and are ventricular left characteristics etiology reduced a failure Strauss had heart patients the were the ischemic to Furthermore, patients (AF). according had Most LBBB patients and the 3.1. III Table of or in half II shown class are NYHA patients had 92 male, the of characteristics baseline The characteristics Patient 3.3.1 Results 3.3 calculated. were measurements Frank-VCG the to relative agreement h di the rn-C.Mroe,tedrcino h aia R-etri h rna ln was plane frontal the in the QRS-vector in maximal larger considerably the being of amplitudes direction the the loops, Moreover, VCG the Frank-VCG. of amplitudes the in served ucmswt h osVGotoe,apie -etwspromdadtePasncor- Pearson the and performed was t-test coe paired relation a outcomes, Frank-VCG Kors-VCG recorded the the with compare To outcomes respectively. (percentages), counts and (SD) deviation hp hl h osVGQSlo a pn utemr,lredi large Furthermore, open. was loop figure-of-eight a QRS had Kors-VCG loop QRS the Frank-VCG while the where shape plane transversal the in especially ent, ntecaicua ieto ihdwwr etrdrcin being directions vector downward with direction craniocaudal the in n,Ciao lios.Cniuu n iceevralsaepeetda mean as presented are variables discrete (SPSS 21 and version Continuous software statistics Illinois). SPSS Chicago, IBM Inc, using performed was analysis statistical The Analysis Statistical T-integrals.3.2.4 and QRS- the of vectors two the from derived was angle QRS-T The ek tal. et Wecke clitgaina h rabtentecreadtebsln nteX ,adZdirection calcu- Z and respectively, and T-wave, Y, or X, complex the QRS QRS in as the baseline lated of the ending and and curve beginning the the between between area the as integration ical ff rne ewe h w C ehd.Tema di mean The methods. VCG two the between erences ffi 16 area in a acltd utemr,aBadAta nlsswsue oshow to used was analysis Bland-Altman a Furthermore, calculated. was cient utemr,teaeso h R-adTloswr optdb numer- by computed were T-loops and QRS- the of areas the Furthermore, . = q QRS area 2 ± , x 1.96 + QRS · D(ftedi the (of SD area 2 , y + QRS ff rne.I diin h isadlmt of limits and bias the addition, In erence). area 2 18 , z rT or osVGrsmlsFrank-VCG resembles Kors-VCG area ff rnebtentemeasured the between erence = q T < 17 area 2 90 rtra naddition, In criteria. , x ff ◦ + ,a ecie by described as ), rne eeob- were erences T area 2 , y ± + standard T area 2 ff , er- z 47 . 7

3 Chapter 3

Figure 3.1: Two examples of VCGs: a recorded Frank-VCG and a Kors-VCG. Patient 1 showed comparable results between the Kors-VCG and Frank-VCG while Patient 2 showed large differences between the calculated and measured VCGs.

48 aibe r hw scut pretg)o mean or (percentage) counts as shown are Variables 3.1: Table eelwrta h R mltd,laigt ie itiuino h onsaon the coe around correlation points lower the somewhat high of a distribution a to wider and in a identity to resulting of leading identity, line amplitude, values of QRS amplitude QRS the T line the than absolute the lower The for were values. around results Kors-VCG the well and Frank-VCG shows align the 3.2A between points Fig. correlation the 3.2). which (Fig. of other amplitude, each against plotted were VCG, uha h R zmt Fg .C n h R- nl Fg .D losoe good showed also 3.2D) (Fig. angle QRS-T the Frank-VCG. and and Kors-VCG 3.2C) the (Fig. between correlations azimuth QRS the as such C.Tesm a eosre ntergtsgta ln.A hw nFg .,ti patient this 3.3, Fig. in shown As plane. sagittal di Frank- right largest the the the in showed in y-axis observed the be above can slightly was same it The while VCG. Kors-VCG the in y-axis the below slightly antagonist. aldosterone left angiotensin LVEF: Ald-antagonist: ARB: block, blocker, enzyme, branch receptor angiotensin-converting bundle ACE: 1 left volume, type LBBB: diastolic II association, end ventricular heart York left LVEDV: New fraction, NYHA: ejection failure, ventricular heart HF: index, mass body BMI: lgtylre (-10.3 larger slightly iei h ln-lmnpo.FrteQRS bias the zero identity. For the of around plot. line points Bland-Altman the the of the along distribution distributed in random nicely line a are by plot substantiated scatter is observation the This in points the Again, (right). yteKr-C ehdcmae oteFakVGvalues. Frank-VCG the to compared method Kors-VCG the by l niiulvle fvrosvrals xrce rmteKr-C n h Frank- the and Kors-VCG the from extracted variables, various of values individual All i.33sostesatrposadBadAta lt o h QRS the for plots Bland-Altman and plots scatter the shows 3.3 Fig. aeiecaatrsiso h 2icue patients. included 92 the of characteristics Baseline ﬀ ± rnei C aibe ewe h w methods. two the between variables VCG in erence 17.1 µ s o h T the for Vs) g yas 67 n (years) Age characteristics Patient er ae(p)69 (36) 33 (bpm) rate Heart %) (n, Female VF()26 (79) 73 (%) LVEF %) (n, LBBB (33) 30 (25) 23 class NYHA (54) %) 50 (n, mellitus Diabetes %) (n, Fibrillation Atrial %) (n, etiology HF Ischemic M (kg BMI VD m)195 (ml) LVEDV R uain(s 168 (ms) duration QRS Titra m)458 (ms) interval QT ACE-inhibitor β l-naoit(,% 9(32) 29 (67) 62 %) (n, Ald-antagonist %) (n, diuretics Loop bokr(,% 3(90) 83 %) (n, -blocker V(,% (1) 1 (10) 9 (36) 33 (43) 40 (10) 9 %) (n, Unknown %) (n, IV %) (n, III %) (n, II %) (n, I ± / m tnaddvainwe appropriate. when deviation standard 2 27 ) / R n )8 (88) 81 %) (n, ARB area area h atridctsasalunderestimation small a indicates latter The . h isws-3.4 was bias the ﬃ = ± ± ± ± ± ± ± osVGrsmlsFrank-VCG resembles Kors-VCG 92 in Fg .B.Agevariables Angle 3.2B). (Fig. cient 9 5 12 7 58 17 39 ± 20.2 area µ s hl twas it while Vs, lf)adT and (left) area 49

3 Chapter 3

Figure 3.2: Scatter plots of diﬀerent VCG variables as derived from either the Kors-VCG (x-axis) or the Frank-VCG (y-axis). The dotted line represents the line of identity. Figure (A) displays the QRS amplitude, (B) T amplitude, (C) QRS azimuth, and (D) QRS-T angle.

Table 3.2 shows the various indices calculated from the Frank- and Kors-VCG analysis and results from the Bland-Altman analysis. The values of most indices differed significantly between the two VCG methods when using paired t-test analysis, but absolute differences were small. The largest difference was seen for the Tarea, where the Kors method showed an underestimation of the Tarea with a large variability, as evidenced by the 95% LoA (-44 µVs to 23 µVs) and relative LoA (-51% to 26%). This resulted in a relative bias of (-9.4 ± 20.8%). Most other VCG-variables showed relative biases below 10%, the smallest one being for the

50 R- nl (-1.9 angle QRS-T 3.3: Figure h rn-C n h osVGrne rm07 o09,adwr l infiatwt a with significant all were and 0.90, to 0.77 from P ranged Kors-VCG the and Frank-VCG the < 0.01. hw ttebto.Terdsursidct h ainsta r hw nFg 3.1. Fig. in shown are that patients the indicate squares red for The Frank-VCG bottom. and the Kors-VCG at shown the between correlation QRS the the show either top on figures two The ± .%.Fnly orlto coe correlation Finally, 4.6%). area lf)o h T the or (left) area rgt.TecrepnigBadAta grsare figures Bland-Altman corresponding The (right). ffi inso l di all of cients osVGrsmlsFrank-VCG resembles Kors-VCG ff rn aibe between variables erent 51

3 Chapter 3

Table 3.2: Mean values from Frank- and Kors-VCG analysis for diﬀerent variables are shown. In addition, the bias, relative bias, 95% limits of agreement, relative limits of agreement, and the Pearson correlation coeﬃcient (R) are given. The relative parameters are compared to the Frank-VCG.

VCG variable Frank- Kors- Bias Relative 95% LoA Relative R VCG VCG bias (%) LoA (%)

QRS amplitude (mV) 1.8 ± 0.5 1.7 ± 0.5† −0.1 ± 0.2 −3.4 ± 14.1 −0.5 to 0.3 −28 to 17 0.90 QRS elevation (◦) 88 ± 18 80 ± 16† −7.8 ± 8.6 −8.2 ± 9.2 −25 to 9 −28 to 10 0.88 QRS azimuth (◦) −63 ± 15 −66 ± 16† −3.5 ± 10.3 8.0 ± 24.3 −24 to 17 −27 to 38 0.77

QRSarea (µVs) 124 ± 46 121 ± 46 −3.4 ± 20.2 −2.0 ± 16.6 −43 to 36 −35 to 29 0.90 T amplitude (mV) 0.6 ± 0.2 0.5 ± 0.2⋆ −0.03 ± 0.13 −1.8 ± 25.0 −0.3 to 0.2 −50 to 33 0.82 T elevation (◦) 88 ± 14 93 ± 15† 5.1 ± 9.1 6.6 ± 15.6 −13 to 23 −15 to 26 0.81 T azimuth (◦) 112 ± 15 106 ± 18† −6.3 ± 10.6 −5.7 ± 9.6 −27 to 14 −24 to 13 0.81 † Tarea (µVs) 87 ± 35 77 ± 30 −10.3 ± 17.1 −9.4 ± 20.8 −44 to 23 −51 to 26 0.88 QRS-T angle (◦) 165 ± 11 161 ± 12† −3.1 ± 7.2 −1.9 ± 4.6 −17 to 11 −10 to 7 0.81 LoA: Limits of agreement. ⋆P-value < 0.05 compared to Frank-VCG using the paired t-test. †P-value < 0.01 compared to Frank-VCG using the paired t-test.

3.4 Discussion

This study shows a good resemblance between the recorded Frank-VCG and the Kors-VCG in patients with left ventricular conduction abnormalities. Also variables calculated from the Frank-VCG and the Kors-VCG method showed good correspondence.

3.4.1 Good resemblance between Kors-VCG and Frank-VCG

This report illustrates many similarities between the Kors-VCG and the recorded Frank-VCG in patients with ventricular conduction abnormalities who are candidate for CRT. These findings are in line with an earlier comparison between these two methods in healthy individuals and in patients with narrow QRS complex8, as well as studies comparing only the QRS-T angle computed using these two methods in different patient populations.5,7,9 In these articles, the recorded Frank-VCG is compared to both the Kors method and the inverse Dower method19, resulting in the superiority of the Kors method. Although not shown here, for the current patient population the Kors method also performed better than the inverse Dower. Comparing the Frank-VCG with the inverse Dower VCG resulted in correlation coefficients ranging from 0.52 to 0.84, while these ranged from 0.77 to 0.90 for the Kors method. Although statistically significant, absolute differences between the Frank-VCG and Kors- VCG derived variables were small and when calculating the bias using the Bland-Altman analysis, no relative bias was larger than 10%. The differences in values between the two methods were small enough that they would not influence diagnosis or therapy predictions. For instance, a QRS elevation and azimuth found to be 88◦ and -63◦, respectively, in the

52 V oadi a to h C s C lcrds ieo h rn lcrdsne ob oiinda h level the at positioned be to need electrodes space Frank intercostal the fifth of the Five of electrodes. VCG vs. ECG the oreo aiblt a aebe nrdcdb h atta o oeptet (n patients some for that first fact A the methods. by introduced both been by have obtained may variables variability the of in variability, source or bias by up made Kors-VCG, 2la C n C eercre ndi on recorded were VCG and ECG 12-lead iue3.4: Figure 80 and Frank-VCG lgtydwwr.As,i rvossuisw hwdta h QRS the that showed we studies previous in Also, downward. slightly C ntepeetsuyadapyn tt h usto ainswt nw response known a with patients of subset the to (n echocardiogram it their applying to and according status study present the in VCG rdcosfrCTrsos.Freape ainswt QRS a with patients example, For response. CRT for predictors niae hti 0 fteptet h ml di This small Frank-VCG. the the patients to the according of responders 90% in as that predicted indicates borderline were Kors-VCG, others the to these 2 according of while non-responders Two as 3.4). predicted (Fig. borderline predictions were Frank-VCG cases negative to false compared cases negative false extra 5 and etrod o R response. CRT for odds better o edt lnclydi clinically to lead not 4 hr r eea xlntosfrtesaldi small the for explanations several are There n V and ff rn hsooia tt ftepteta ela di as well as patient the of state physiological erent 6 Di . iia gr stetplf gr nFg .,bticuigol ainswt available with patients only including QRS but derived Frank-VCG 3.3, optimal Fig. the in are Indicated figure data. left echocardiographic top the as figure Similar niae ycrlsadtennrsodr ycrosses. by non-responders the and circles by indicated swl steba-orce osVGtrsod(94.6 threshold Kors-VCG bias-corrected the as well as itnus R epnesfo o-epnesfudb a ere tal. et Deursen van by found non-responders from responders CRT distinguish ff rne neetoepsto ih euti di in result might position electrode in erences ◦ n -66 and ff rn predictions. erent 4 h aepsto od o h lcrdsue omauelead measure to used electrodes the for holds position same the , ◦ 13 nteKr-C si ohcsspitn etpseiryand posteriorly left pointing cases both in is Kors-VCG the in dutn hscut-o this Adjusting ff = rn as hsmyhv e odi to led have may This days. erent 8,teewudb xr as oiiecases positive false extra 2 be would there 68), ff ff rne ewe h rn-C n the and Frank-VCG the between erences rne ewe h w ehd would methods two the between erences ff au o h isfudfrteKors- the for found bias the for value osVGrsmlsFrank-VCG resembles Kors-VCG ff rne npaeeto atof part of placement in erences µ ff area s.TeCTrsodr are responders CRT The Vs). rn mltdsa elas well as amplitudes erent agrta 98 than larger area 13 n h T the and area ff rne due erences 13 hehl to threshold µ = area shave Vs (98 5 the 25) 12 µ Vs) are 53

3 Chapter 3 directions of the loops.20,21 Secondly, the Frank-VCGs were averaged over 1 minute, while the 12-lead ECG only contains 10 seconds of a recording. This allows less averaging for the Kors-method, leading to a lower signal-to-noise ratio.

3.4.2 Potential Clinical Implications

The concordance between the Frank-VCG and Kors-VCG in patients with LV conduction delay supports the use of the Kors-VCG method. This method avoids the need for special equipment to record the VCG, while the extra information from the VCG is still obtained. Using this method, every 12-lead ECG recording that measured the eight independent leads simultaneously or that includes a running lead during a stable RR interval (multiplexed ECG) could be used to synthesize a VCG. It should be noted that there are some limitations to the use of a multiplexed ECG. First, the calculation of an averaged heartbeat is often hampered by the fact that only 1 or 2 complete heartbeats are present per lead, this might lead to a lower signal-to-noise ratio. Furthermore, because of the limited amount of heartbeats per lead, the computation of a VCG is impossible if too much premature ventricular contractions or other artifacts are present. Therefore, it would be preferable to save the digital recording of the 12-lead ECG or to save a PDF with a parallel display of the eight independent leads I, II, and

V1 -V6. Use of the synthesized VCG might result in a better diagnosis or treatment plan for the patient. Moreover, VCGs can be reconstructed from already existing ECGs, using the method to extract the digital signal out of PDF-ﬁles generated by the ECG recording systems.

3.5 Conclusions

In patients with heart failure and LV conduction delay who are candidate for CRT, the Frank- VCG and the Kors-VCG show a good resemblance. Furthermore, the Kors-VCG method avoids the need for special recording equipment, while the extra information from the VCG is still obtained. The Kors-VCG enables retrospective as well as prospective VCG analysis of routinely recorded 12-lead ECGs in studies related to CRT.

3.6 Funding

This research was performed within the framework of CTMM (Center for Translational Molecular Medicine; www.ctmm.nl), Project COHFAR (Congestive Heart Failure and Ar- rhythmia; Grant 01C-203), and supported by the Dutch Heart Foundation.

54 References 1 uoenHatRyh soito,Erpa oit fCrilg,HatRhythm Heart Cardiology, of Society European Association, Rhythm Heart European 11. 0 TArhm GFse,A mt,D eugo RLo,ELh ZKocovic, DZ Loh, E Leon, AR Delurgio, DB Smith, AL Fisher, WG Abraham, WT 10. .C crus MAga CMn CCneitr EvndrWl,M Schalij, MJ Wall, der van EE Cannegieter, SC Man, SC Algra, AM Schreurs, CA 9. .D otzadT clgl hndrvn h pta R- nl rmte12-lead the from angle QRS-T spatial the deriving When Schlegel. TT and Cortez DL 7. .J os a epn CSti,adJ a eml eosrcino h Frank the of Reconstruction Bemmel. van JH and Sittig, AC Herpen, van G Kors, JA 8. .LEebad,AHutn n WMcaln.Vcocrigassnhszdfrom synthesized Vectorcardiograms Macfarlane. PW and Houston, A Edenbrandt, L 6. .SMn MAga ASher,CWBorle CJW Schreurs, CA Algra, AM Man, S 5. vectorcardiography. of status present The Simonson. E and Schmitt leads. OH and Heartvector Herpen. 3. van and Brummelen, van A Burger, HC 2. .EFak nacrt,ciial rcia ytmfrsailvectorcardiography. spatial for system practical clinically accurate, An Frank. E 4. .H ilm.O h as ftepaedi phase the of cause the On Willams. HB 1. oit,HatFiueSceyo mrc,Aeia oit fEchocardiography, of Society American America, of Society Failure Heart Society, akr LCael LHys ta.Cricrsnhoiaini hoi heart chronic in resynchronization Cardiac al. et Hayes, failure. DL Clavell, AL Packer, M vectorcardiogram: electrocardiogram. Frank 12-lead the the 43:294–301. in (2010) from angle derived QRS-T estimates spatial of The accuracy Swenne. CA and Kors, JA lcrcriga,wihtasomi oeFak ersino nes Dower? inverse or regression Frank: Electrocardiol more is transform which electrocardiogram, 15:21–26. etradormfo tnadeetoadorpi ed:dansi oprsnof comparison di diagnostic leads: electrocardiographic standard from vectorcardiogram 2la Cs e ehdapidi 72hatychildren. healthy 1792 in applied method new a ECGs: 12-lead R- nl opeitaryhisi ainswt shmchatdsaeadsystolic vec- and disease the dysfunction. heart ventricular ischemic of with left patients Influence spatial in electrocardiogram-derived arrhythmias predict the Swenne. to of angle CA power QRS-T and the on Schalij, matrix MJ synthesis Cannegieter, torcardiogram SC Wall, der van culation Med Intern 61:317–323. (1961) electrocardiogram. the of peaks onymous ff rn methods. erent nlJMed J Engl N 15)13:737–749. (1956) 15)96:574–590. (1955) 21)43:302–309. (2010) u er J Heart Eur 20)346:1845–1853. (2002) Electrocardiol J 19)11:1083–1092. (1990) m .Physiol. J. Am. ff rnefeunl bevdbtenhom- between observed frequently erence 21)44:410–415. (2011) ff ,RCShrtn,LvnEvn EE Erven, van L Scherptong, RWC s, 11)35:292–300. (1914) eit Cardiol Pediatr Electrocardiol J mHatJ Heart Am M Arch AMA (1994) Cir- J

3 Chapter 3

American Heart Association, European Association of Echocardiography, Heart Fail- ure Association, JC Daubert, L Saxon, et al. 2012 EHRA/HRS expert consensus state- ment on cardiac resynchronization therapy in heart failure: implant and follow-up re- commendations and management. Heart Rhythm (2012) 9:1524–1576. 12. EB Engels, EM Végh, CJM Van Deursen, K Vernooy, JP Singh, and FW Prinzen. T- wave area predicts response to cardiac resynchronization therapy in patients with left bundle branch block. J Cardiovasc Electrophysiol (2015) 26:176–183. 13. CJM van Deursen, K Vernooy, E Dudink, L Bergfeldt, HJGM Crijns, FW Prinzen, and L Wecke. Vectorcardiographic QRS area as a novel predictor of response to cardiac resynchronization therapy. J Electrocardiol (2015) 48:45–52. 14. CJM van Deursen, M Strik, LM Rademakers, A van Hunnik, M Kuiper, L Wecke, HJGM Crijns, K Vernooy, and FW Prinzen. Vectorcardiography as a tool for easy optimization of cardiac resynchronization therapy in canine left bundle branch block hearts. Circ Arrhythm Electrophysiol (2012) 5:544–552. 15. L Wecke, A Rubulis, G Lundahl, MR Rosen, and L Bergfeldt. Right ventricular pacing- induced electrophysiological remodeling in the human heart and its relationship to cardiac memory. Heart Rhythm (2007) 4:1477–1486. 16. L Wecke, F Gadler, C Linde, G Lundahl, MR Rosen, and L Bergfeldt. Temporal characteristics of cardiac memory in humans: vectorcardiographic quantification in a model of cardiac pacing. Heart Rhythm (2005) 2:28–34. 17. DG Strauss, RH Selvester, and GS Wagner. Defining left bundle branch block in the era of cardiac resynchronization therapy. Am J Cardiol (2011) 107:927–934. 18. GL Botto, CD Dicandia, M Mantica, C La Rosa, A D’Onofrio, MG Bongiorni, G Molon, R Verlato, GQ Villani, A Scaccia, G Raciti, and E Occhetta. Clinical characteristics, mortality, cardiac hospitalization, and ventricular arrhythmias in patients undergoing CRT-D implantation: results of the ACTION-HF study. J Cardiovasc Elec- trophysiol (2013) 24:173–181. 19. L Edenbrandt and O Pahlm. Vectorcardiogram synthesized from a 12-lead ECG: superiority of the inverse Dower matrix. J Electrocardiol (1988) 21:361–367. 20. M Kania, H Rix, M Fereniec, H Zavala-Fernandez, D Janusek, T Mroczka, G Stix, and R Maniewski. The effect of precordial lead displacement on ECG morphology. Med Biol Eng Comput (2014) 52:109–119. 21. RR Bond, DD Finlay, CD Nugent, C Breen, D Guldenring, and MJ Daly. The effects of electrode misplacement on clinicians’ interpretation of the standard 12-lead electrocardiogram. Eur J Intern Med (2012) 23:610–615.

Chapter 4 Electrical remodeling in patients with iatrogenic left bundle branch block

The content of this chapter is based on:

Elien B. Engels,1 Thomas T. Poels,2 Patrick Houthuizen,3 Peter P.T. de Jaegere,4 Jos G. Maessen,2 Kevin Vernooy,5 and Frits W. Prinzen1 (2016). Electrical remodeling in patients with iatrogenic left bundle branch block. Revision under review for Europace.

1 Department of Physiology, CARIM School for Cardiovascular Diseases, Maastricht University, Maastricht, the Netherlands. 2 Department of Cardiothoracic Surgery, Maastricht University Medical Center, Maastricht, the Netherlands. 3 Department of Cardiothoracic Surgery, Catharina Hospital, Eindhoven, the Netherlands. 4 Department of Cardiology, Erasmus Medical Center, Rotterdam, the Netherlands. 5 Department of Cardiology, Maastricht University Medical Center, Maastricht, the Netherlands. Chapter 4

Abstract

Introduction: Left bundle branch block (LBBB) is induced in approximately one third of all transcatheter aortic valve implantation (TAVI) procedures. We investigated electrophysiological remodeling in patients with TAVI-induced LBBB.

Methods and Results: This retrospective study comprises 107 patients with initially narrow QRS complex of whom 40 did not and 67 did develop persistent LBBB after TAVI. 12-lead electrocardiograms (ECGs) taken before TAVI, within 24 hours (‘acute’), and 1-12 months after TAVI (‘chronic’) were used to reconstruct vectorcardiograms. From these vectorcardiograms, QRSarea and Tarea were calculated as comprehensive indices of depolarization and repolarization abnormalities, respectively. TAVI-induced LBBB resulted in signiﬁcant acute depolarization and repolarization changes while further repolarization changes were observed with longer lasting LBBB. The amount of long-term repolarization changes (remodeling) was highly variable between patients. The change in Tarea between acute and chronic LBBB ranged from +57% to -77%. After dividing the LBBB cohort into tertiles based on the change in Tarea, only baseline QRSarea was larger in the tertile with no signiﬁcant change in Tarea. During longer lasting LBBB the spatial ventricular gradient (SVG) changed orientation towards the direction of the QRS-vector, indicating that later-activated regions developed shorter action potential duration.

Conclusions: This study in patients with TAVI-induced LBBB shows that repolarization changes develop within months after onset of LBBB, and that these changes are highly variable between individual patients.

60 tv ada disease. cardiac ative nldn p n onrglto fvrosincanl n vnls fsarcoplasmic of loss even and channels ion various of regulation reticulum. failure, down heart rapid and dyssynchronous and in up- LBBB remodeling including of cellular and combination molecular a run. extensive long causes by the pacing, induced on remodeling failure, cardiac heart and dyssynchronous function pump Experimental cardiac of reduction rapid a induces igi nw,sm nomto bu h aua itr fdsycrn a enderived been has dyssynchrony of studies. history pacemaker natural the from about information some known, is cing of duration the that known. not so usually silently, is develops patient usually a in LBBB LBBB because partly LBBB, of presence A Introduction 4.1 etrbtenJnay20 n pi 04(i.41.Fo l ains electrocardio- patients, all referred TAVI (further From after hours 24 4.1). within (Fig. baseline, at 2014 available were April measurements graphic and Medical University 2006 Maastricht January or between Rotterdam Center Catharina Center the Medical in TAVI underwent Erasmus who Eindhoven, patients Hospital 998 of consisted population patient index The population Study 4.2.1 Methods 4.2 LBBB. lasting longer during (‘acute TAVI variables after same of the immediately with found images variables LBBB’) one-dimensional repolarization multiple compared We the ECG. than 12-lead image information a one valuable in it more heart that the provide is within might VCG forces and the electrical vectorcar- the of and of advantage information The (ECG) three-dimensional LBBB. TAVI-induced electrocardiogram contains with 12-lead patients the in analyzed (VCG) diogram we purpose this To LBBB. h ontqaiyfroe hs ugr.I uhpoeue prxmtl n hr fall of third one approximately procedures patients such in In LBBB. surgery. develops stenosis chest patients aortic open for symptomatic qualify severe, not for is do TAVI treatment who invasive procedure. (TAVI) less implantation new, valve relatively aortic a transcatheter a during LBBB develop pacing of weeks several within occurring reduction main memory). (cardiac the amplitude, T-wave the of tion eie BB lorgtvnrclrpcn rae ysnhoy ic h ne fpa- of onset the Since dyssynchrony. creates pacing ventricular right also LBBB, Besides h i ftepeetsuywst netgt lcrclrmdln nptet with patients in remodeling electrical investigate to was study present the of aim The e nih ntentrlhsoyo BBmyb civdb netgtn ainsthat patients investigating by achieved be may LBBB of history natural the in insight New rnhbok(LBBB). block branch pproximately a obdt n otlt n tmytu eamre o rgesv,degener- progressive, for marker a be thus may it and mortality and morbidity lar 4,5 oee,i ssilntwl nw nwihwytehmnhataat othe to adapts heart human the way which in known well not still is it However, 2 utemr,aia xeiet aesonta nuto fLBBB of induction that shown have experiments animal Furthermore, 5 fptet ihhatfiue(F r igoe ihlf bundle left with diagnosed are (HF) failure heart with patients of 25% 6,7 8 nptet,eetia eoeigwscaatrzdb reduc- a by characterized was remodeling electrical patients, In 1 BBi soitdwt nicesdrs fcardiovascu- of risk increased an with associated is LBBB lcrclrmdln nTV-nue BBpatients LBBB TAVI-induced in remodeling Electrical 61 3

4 Chapter 4 to as ‘acute’), and one to twelve months after TAVI (‘chronic’). Patients were included in the current study if the baseline QRS duration was smaller than 120 ms and the ECGs recorded at the diﬀerent time points were stored as PDF-ﬁles in order to enable VCG conversion (see below). Exclusion criteria were pre- or postoperative ventricular pacing, the presence of postoperative QRS widening other than LBBB, and/or transient LBBB that was only present postoperative and not at 12 months follow-up. A total of 293 patients were eligible for analysis (Fig. 4.1).

Figure 4.1: Study population selection. LBBB indicated left bundle branch block.

The study population consisted of 67 patients who developed LBBB directly after TAVI (acute LBBB) and had persistent LBBB for at least 12 months after the procedure (chronic LBBB). Out of 226 patients who did not develop LBBB upon TAVI, a group of 40 age and

62 or fe AI n )11 otsatrTV.We h remnts hwdsignificant showed test Friedman the TAVI. When di after months 1-12 3) TAVI, and after hours eetse sn h rsa-alsadWloo aksmts ihBnern correction Bonferroni P-value with two-sided test A rank-sum consecutively. Wilcoxon and Kruskal-Wallis the using tested were e hc iepit eedi were points time which see n iceevralsa ons(ecnae.Tecmaio ewe iceevariables discrete between comparison The the using (percentage). performed was counts as variables discrete and eecsi C aibe ttedi the at variables VCG in ferences C nlssa ecie ndti before. detail in described as analysis VCG R n mltd hl h R- nl ste3 nl ewe h aia R-and QRS- maximal the between angle 3D the QRS is The angle T-vector. QRS-T the while amplitude T and QRS o a edsrbdb t mltd,aiuh(retto ntetases ln) n eleva- QRS and The plane), plane). transverse frontal the the in in vec- (orientation (orientation maximal azimuth tion The amplitude, loop. its VCG by 3-dimensional described the be on max- QRS can point tor the the any as of to defined origin end the were the T-vector from and and distance imal complex QRS- maximal QRS The the respectively. of T-wave, beginning or complex the between duration the as defined aeieoe iefrteprso neeti edX ,adZadaecluae sn the using calculated are and Z and Y, X, lead in interest of QRS formulas parts the for time over baseline rga rte nMTA 21b(ahok,Ntc,M)adwr yteie to synthesized were method. and Kors MA) the Natick, using (MathWorks, VCGs computer R2010b custom-made MATLAB a in using analyzed written semi-automatically program were ECGs all of files PDF The VCG and ECG Helsinki. 4.2.2 of Declaration the of group). principles QRS the (narrow with LBBB accordance of in electrocardio- absence conducted was the the project in in This procedure TAVI changes the the to investigate due to properties order graphic in selected were patients sex-matched n,Ciao L.Cniuu aibe r rsne smean as presented are (SPSS variables 22 version Continuous software IL). statistics Chicago, SPSS IBM Inc, using performed was analysis statistical The analysis spatial Statistical the 4.2.3 by represented is T-integral and QRS- the of (SVG). sum gradient ventricular vectorial The T-integral. or h etr opsdo h rai h di the in area the of composed vectors The ff rne,teWloo indrn etwt ofroiajsmn a efre to performed was adjustment Bonferroni a with test signed-rank Wilcoxon the erences, area = area q QRS n T and χ 2 11 area 2 ts.TeFida etwsue ots hte hr eedif- were there whether test to used was test Friedman The -test. area , ff x 9 rn.Psil di Possible erent. r h tredmninl ra ewe h uv n the and curve the between areas ‘three-dimensional’ the are + utpevralswr xrce rmbt h C and ECG the both from extracted were variables Multiple ff QRS rn iepit:1 aeie(eoeTV) )wti 24 within TAVI), 2) (before baseline 1) points: time erent < lcrclrmdln nTV-nue BBpatients LBBB TAVI-induced in remodeling Electrical .5wscniee ttsial significant. statistically considered was 0.05 area 2 , ff y 10 rn ietosX ,adZaecle h QRS- the called are Z and Y, X, directions erent + / QRS rey R uainadQcitra were interval QTc and duration QRS Briefly, -mltd ai ersnstertobetween ratio the represents ratio T-amplitude ff area 2 rne ewe di between erences , z rT or area = ± q tnaddvain(SD), deviation standard T area 2 ff rn ain groups patient erent , x + T area 2 , y + T area 2 63 , z .

4 Chapter 4

4.3 Results

4.3.1 Patient characteristics

Baseline characteristics of patients who developed LBBB (n = 67) and the group of 40 age and sex-matched patients who did not are summarized in Table 4.1. Almost all patients were septua- and octogenarians with an even gender distribution. Most patients were in NYHA class III and approximately 20% had a prior myocardial infarction. The QRS duration at baseline as extracted from the VCG was around 110 ms for both the LBBB group and the narrow QRS group.

Table 4.1: Baseline characteristics of the included LBBB and narrow QRS patients.

Study population Entire LBBB Narrow QRS n = 107 n = 67 n = 40

Demographics Age (years) 79 ± 7 79 ± 6 80 ± 8 Male gender (n, %) 49 (46) 32 (48) 17 (43) Clinical Previous myocardial infarction (n, %) 23 (22) 14 (21) 9 (23) Diabetes Mellitus (n, %) 24 (22) 17 (25) 7 (18) CVA (n, %) 19 (18) 13 (19) 6 (15) CABG (n, %) 29 (27) 17 (25) 12 (30) NYHA class I/II/III/IV (%) 3/17/68/12 3/15/70/12 3/20/65/13 Electrocardiography QRS duration (ms) 110 ± 15 110 ± 15 110 ± 13 QRS axis (◦) 19 ± 30 20 ± 29 17 ± 32 Variables are shown as counts (percentage) or mean ± standard deviation when appropriate. CVA = cerebrovascular accident; CABG = coronary artery bypass graft; NYHA = New York Heart Association.

4.3.2 ECG and VCG changes in narrow QRS patients

VCG variables of the narrow QRS group did not significantly change between baseline and directly after TAVI (Table 4.2). The depolarization variables showed some minor changes between briefly and 1-12 months after TAVI. QRS-vector amplitude and QRS-area decreased significantly (-12% and -16%, respectively) over time while QRS duration did not change significantly. Furthermore, the elevation of the T-vector decreased significantly (Table 4.2).

4.3.3 ECG and VCG changes in TAVI-induced LBBB patients

An example of ECG and VCG changes in one patient with TAVI-induced LBBB is shown in Fig. 4.2. From the 12-lead ECG the widening of the QRS complex within 24 hours after TAVI can be observed. Also, the orientation of the maximal QRS-vector changed from pointing to the left before TAVI to pointing posteriorly directly after TAVI, which is expressed by a more

64 nl.Hwvr h zmt fteSGdcesdsgicnl ongtv aus becoming values, QRS-T negative and to elevation) significantly decreased and SVG (azimuth the T-vector of azimuth the the However, of angle. direction associated in not change were significant changes any repolarization with These LBBB. lasting longer during significantly Tbe43.I otat l eoaiainvrals uha h T nevl -etram- T-vector interval, QTc the as such QRS T variables, or and repolarization plitude, direction, all QRS-vector contrast, In amplitude, QRS-vector 4.3). (Table duration, QRS as such variables direction or size either in significantly stage. change acute not the did in SVG the However, LBBB. in T-wave bu 40 about cn nrae nQSdrto n T nevl(al .)a ela oeta doubling than more as well as 4.3) (Table interval T QTc of and duration QRS in increases ficant nl nrae rm100 from increased angle ope n R opwr bevdadas h ieto fteTvco i o change. T not the did and T-vector loop T-wave the the of of direction QRS size the the the also in contrast, and changes In observed minor were LBBB size T-wave loop chronic increased QRS and and an acute complex show Between VCG 4.2). and (Fig. ECG LBBB the acute both during addition, In azimuth. QRS negative ⋆ † aibe r hw smean as shown are Variables 4.2: Table value P value P ewe ct n hoi BB osgicn hne eeosre ndepolarization in observed were changes significant no LBBB, chronic and acute Between o h niechr fLB ains eeomn fLB uigTV asdsigni- caused TAVI during LBBB of development patients, LBBB of cohort entire the For area < < T , .5cmae to compared 0.05 .5cmae obsln,cluae sn h icxnsge aktest. rank signed Wilcoxon the using calculated baseline, to compared 0.05 ◦ area hne neetoadorpi aibe vrtm ntenro R ru.Dt r as are Data group. QRS narrow VCG. the the from in calculated time over variables electrocardiographic in Changes hl tddntcag h ieto fteTwv.A osqec,teQRS-T the consequence, a As T-wave. the of direction the change not did it while V zmt ( azimuth SVG V mltd ( amplitude SVG QRS V lvto ( elevation SVG R- nl ( angle QRS-T QRS R-etrapiue(V 1 (mV) amplitude QRS-vector T lvto ( elevation T R zmt ( azimuth QRS zmt ( azimuth T QRS T nevl(s 322 (ms) interval JTc -etrapiue(V 0 (mV) amplitude T-vector T nevl(s 442 (ms) interval QTc R lvto ( elevation QRS R uain(s 110 (ms) duration QRS er ae(p)72 Baseline (bpm) rate Heart Variable n h -etrapiue BBsgicnl erae h R zmt by azimuth QRS the decreased significantly LBBB amplitude. T-vector the and area area / / area ( erae infiaty codnl,teQRS the Accordingly, significantly. decreased , rarto()1 (-) ratio area T -mltd ai - 6 (-) ratio T-amplitude µ s 62 Vs) ( µ < ± s 59 Vs) ◦ 4husatrTV,cluae sn h icxnsge aktest. rank signed Wilcoxon the using calculated TAVI, after hours 24 ◦ 74 ) tnaddvain SVG deviation. standard 81 ) ◦ ◦ ◦ 104 ) ◦ ) ◦ 4 ) µ 71 ) 70 ) ◦ s 75 Vs) o155 to ◦ Tbe43,cnitn ihtecaatrsi discordant characteristic the with consistent 4.3), (Table − 23 lcrclrmdln nTV-nue BBpatients LBBB TAVI-induced in remodeling Electrical . . . . 3 2 9 4 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± = 354 63 964 29 0 2111 52 01 60 2 352 53 0 4341 34 685 26 747 77 25 657 26 0 662 16 9461 39 3107 13 076 10 pta etiua gradient. ventricular spatial . . . . 1 7 1 9 5 9 0 3 < fe AIatrTV etP-value test TAVI after TAVI after − 24 4hus1-1 otsFriedman months 12 - 1 hours 24 . . . . area 4 9 3 8 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 562 25 0 184 41 212 62 2 862 28 0 844 28 7319 57 169 31 0 62 103 29 547 25 0 869 18 7437 57 3107 13 474 14 erae vrtm Fg 4.2). (Fig. time over decreased , . . . . 1 9 5 5 1 6 0 2 − 25 . . . . 4 4 5 3 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 90.31 29 1 20.06 42 70.27 37 3 50.43 15 20.39 22 24 30.34 43 23 0 31 0 30.11 23 36 70.08 17 20.18 32 10.30 11 / . . . . mltd ai increased ratio amplitude T 0.27 0 0.04 7 5 0.80 2 † † † ⋆ ⋆ † † < < < 0.01 0.01 0.01 0.01 0.01 area 65

4 Chapter 4

Figure 4.2: Typical example of one patient undergoing TAVI and developing LBBB. From top to bottom the 12-lead ECG (left) and 3D VCG (right) at baseline, within 24 hours after TAVI and 12 months after TAVI are shown. The black arrows in the 3D-VCG image indicate the maximal QRS- and T-vector. 66 etls uigaueLB,teol di only the LBBB, acute During tertiles. bu,n (blue, h nyosre di observed only The C aibe rvralsmnindi al . hwdasgicn di significant a showed 4.1 Table in mentioned variables or variables VCG ie1ad3 a infiatysalrQRS smaller significantly a was 3, and 1 tile osgicn hnei T T co- in in LBBB change changes the relative significant 4.3, the remodeling, no T-wave on (Fig. in based measured variability tertiles, in was the divided explore VCG was further hort LBBB to chronic order the was In which changes panel). these at right of time-point size the The of 4.3). dependent (Fig. not patients LBBB individual between variable highly atn BB seilyepesdb la euto nteT the in reduction clear a by longer expressed with variables especially repolarization in LBBB, decrease lasting significant a showed cohort LBBB the While changes repolarization Individual 4.3.4 azimuth. QRS the towards more directed ⋆ † aibe r hw smean as shown are Variables 4.3: Table value P value P < < .5cmae oaueLB,cluae sn h icxnsge aktest. rank signed Wilcoxon the using calculated LBBB, acute to compared 0.05 .5cmae obsln,cluae sn h icxnsge aktest. rank signed Wilcoxon the using calculated baseline, to compared 0.05 = n n wlemnh fe AI n h atclm ersnstePvleo h Fried- the VCG. the of from P-value calculated the as represents are column Data last between test. the results man and TAVI, the after represents the months column TAVI, twelve third and before the TAVI, one after performed hours measurements 24 within electrocardiographic columns column first the second The of group. results LBBB the the in shows time over variables electrocardiographic in Changes 2 hwda vrgdrdcino 2 n 8 euto nT in reduction 58% and 32% of reduction averaged an showed 22) V lvto ( elevation SVG V zmt ( azimuth SVG V mltd ( amplitude SVG QRS R- nl ( angle QRS-T R-etrapiue(V 1 (mV) amplitude QRS-vector R zmt ( azimuth QRS er ae(p)71 Baseline (bpm) rate Heart Variable QRS T nevl(s 448 (ms) interval QTc R uain(s 109 (ms) duration QRS T nevl(s 329 (ms) interval JTc R lvto ( elevation QRS T QRS lvto ( elevation T -etrapiue(V 0 (mV) amplitude T-vector zmt ( azimuth T area / / area ( rarto()1 (-) ratio area T -mltd ai - 5 (-) ratio T-amplitude µ s 56 Vs) ( µ s 54 Vs) ± ◦ ff ◦ 81 ) tnaddvain SVG deviation. standard 80 ) rnebtentetregop tbsln,seiclybtenter- between specifically baseline, at groups three the between erence ◦ ◦ ◦ 108 ) ◦ ) ◦ 6 ) µ 71 ) 68 ) s 62 Vs) area rd n (red, − = 26 . . . . lcrclrmdln nTV-nue BBpatients LBBB TAVI-induced in remodeling Electrical 2 3 7 4 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 2.I otat etl bak n (black, 2 tertile contrast, In 22). = 262 32 38 63 064 40 0 3156 43 8341 38 0 28 2 177 11 880 18 8520 48 5159 15 0124 40 1114 21 685 26 0 399 63 pta etiua gradient. ventricular spatial . . . . 1 5 1 6 2 3 0 2 ff rne bevdbtentrie1ad3were 3 and 1 tertile between observed erences area < fe AIatrTV etP-value test TAVI after TAVI after − 70 4hus1-1 otsFriedman months 12 - 1 hours 24 . . . . 0 3 7 8 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ntrie3(i.44.Nn fteother the of None 4.4). (Fig. 3 tertile in 569 25 65 559 35 0 14 0 15 0 14 16 56 22 0307 40 43 37 985 19 0 2100 22 . . . . 3 1 5 5 2 ⋆ ⋆ ⋆ ⋆ ⋆ ⋆ ⋆ ⋆ ⋆ ⋆ ⋆ − − 156 488 159 113 1 3 0 29 67 78 80 84 area . . . . 5 1 7 6 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± aual,Trie1showed 1 Tertile Naturally, . 38 60.04 26 30.15 23 0 14 18 1 17 39 16 51 34 20 0 35 70.19 17 0 90.40 19 . . . . 5 0.53 4 1 2 ⋆ ⋆ ⋆ ⋆ ⋆ ⋆ ⋆ ⋆ ⋆ ⋆ † ⋆ ⋆ † † † † † † area hs hne were changes these , < < < < < < < < < < < 0.02 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 = ff area 3 n etl 3 tertile and 23) rnebetween erence respectively. , 67

4 Chapter 4

Tarea and T/QRS area ratio (Fig. 4.4). From acute to chronic LBBB, the T-vector amplitude,

Tarea, and T/QRS area ratio remained unchanged in tertile 1, while these variables decreased significantly in the other two groups. It should be noted that at baseline the T-wave was concordant to the QRS complex, but that the T-wave became discordant once LBBB developed. Despite decreasing T-wave size, the T-wave remained discordant (see also Fig. 4.2), even in subgroups 2 and 3. The JTc interval was not significantly different between tertiles at any time-point and shortened for all subgroups during chronic LBBB compared to acute LBBB. Finally, the SVG azimuth showed a large (∼80◦) change in tertile 3 but no change in tertile 1.

Figure 4.3: Absolute values of Tarea (left) and relative changes over time (right panel), in which the patients were divided into tertiles. In red the patients in tertile 1 are indicated, black represents the patients in tertile 2, and blue the patients in tertile 3.

4.4 Discussion

The present study demonstrates that within one month after onset of TAVI-induced LBBB, electrical remodeling develops, that is mainly expressed as reduction in repolarization variables (like JTc interval, T-wave amplitude, and Tarea). The redirection of the SVG indicates an adaptation in repolarization sequence opposite to the sequence of activation. This remodeling is highly variable between patients, despite similar baseline characteristics.

4.4.1 Electrical remodeling in LBBB patients

A reduction of T-wave amplitudes in longer lasting LBBB has been reported earlier by Shvilkin et al. 12 . These investigators retrospectively compared ECGs from patients with LBBB lasting <24 hours versus >24 hours. The same investigators also reported a similar

68 iue4.4: Figure r ieydet hne ntehtrgniyo h cinptnilmrhlg ic the since morphology potential action the of T-wave heterogeneity in the changes in observed changes T-wave the to and due LBBB QRS likely chronic in are changes and huge acute despite between LBBB However, acute morphology. and baseline between angles or tude rgniyo cinptnilmrhlg repcieo ciainsqec atr that factors sequence factors’). activation het- (‘secondary T-wave of i.e. the irrespective T-wave, observed to the morphology contribute the determining potential by factors’ action ‘primary supported of the is assesses erogeneity changes SVG The depolarization SVG. to in due changes not and changes polarization a n ellrcags nldn hne nincanl n acu adigproteins, handling calcium and models. channels these ion in in reported been changes have including changes, cellular and lar hoyo n as LB,R aig per ob omnbhvo.Ta LBBB That LBBB behavior. of model common canine a the in be studies and to from known appears is changes pacing) electrophysiological RV induce can (LBBB, dyssyn- lasting cause severe longer less any during (presumably) variables of and repolarization chrony disease, in valve reduction a aortic Therefore, no failure. had heart younger, were they that in ulation euto nTwv mltd vrtm uiglne atn ih etiua aigin pacing ventricular right lasting longer conduction. AV 1:1 during a time with patients over amplitude T-wave in reduction h data h bevdrdcin nTwv raadapiueaesll u ore- to due solely are amplitude and area T-wave in reductions observed the that idea The / rtcyaigidcdhatfiue osnn fpm ucinadetniemolecu- extensive and function pump of Worsening failure. heart induced tachypacing or eeto feetoadorpi nig fteLB ru fe h iiinaccording division T the after their group to LBBB the of ﬁndings electrocardiographic of selection A qaetemdl ru,adtedaodtelretdces nT in decrease largest the diamond the and group, middle the square etl n 3, and 1 tertile sn h icxnrn u test. sum rank Wilcoxon the using area erae h iceidctstegopwt h oetT lowest the with group the indicates circle The decrease. ◦ P < .5btentrie1ad2 and 2, and 1 tertile between 0.05 7 3,5 motnl,teeptet di patients these Importantly, lcrclrmdln nTV-nue BBpatients LBBB TAVI-induced in remodeling Electrical 11 ned hr sn hnei V ampli- SVG in change no is there Indeed, • P < .5btentrie2ad3all 3 and 2 tertile between 0.05 ﬀ rdfo h rsn pop- present the from ered area . ⋆ P area < .5between 0.05 erae the decrease, 69

4 Chapter 4 azimuth of the SVG changed. This indicates that the observed electrical remodeling may be attributed to changes in the heterogeneity and/or shape of ventricular action potentials. The most pronounced change in SVG concerns a shift in SVG azimuth towards negative values, implying a rotation in direction towards the direction of the QRS vector. Because repolarization is an electrophysiological process that is opposite to depolarization, the redirection of the T-wave towards the QRS complex (less discordant) indicates that the sequence of repolarization changes in a direction that is opposite to the sequence of depolarization. This is consistent with the observation in rabbit hearts that the sequence of repolarization adapts slowly after onset of ventricular pacing.13 An intriguing observation in this study is that although all patients developed a similar, iatrogenic, proximal LBBB, the change in Tarea ranged between +57% to -77% compared to acute LBBB. In addition, the subgroup of patients showing the smallest amount of Tarea reduction (tertile 1) does not show any change in the azimuth of the SVG, while changes are observed in the other two patient groups. This indicates that electrical remodeling is occurring in the tertile 2 and 3 group, but not in the tertile 1 group. The only diﬀerence between the two extreme groups (tertile 1 and tertile 3) that could be observed at baseline was that the

QRSarea was signiﬁcantly larger for patients showing almost no signs of electrical remodeling. Larger QRS amplitudes observed on the ECG may be associated with a higher degree of left ventricular hypertrophy (LVH)14 or a larger end systolic volume15. Both conditions could indicate a worse left ventricular function. This would suggest that more diseased hearts are less capable of electrical remodeling, an idea that requires further investigation.

4.4.2 Possible explanation

A possible mechanism for the observed electrical remodeling might be mechano-electrical (MEC) coupling. This hypothesis was previously tested in a computer model.16 In this model study it was assumed that the myocardium strives to maintain local myocardial mechanical work constant, despite the abnormal electrical activation. Because in LBBB the early septal activation lowers contractility in the septum, the MEC simulation imposes an increase in L-type calcium channel conductance during longer lasting LBBB in an attempt to increase contractility. This remodeling of the calcium channel also leads to an increase in action potential duration in the early-activated septum and therefore a smaller dispersion of ventricular repolarization. Applying the simulation predictions to the present clinical observations, more pronounced ‘remodeling’ would imply a stronger MEC. The latter was also observed in isolated rabbit hearts.17 In this study the investigators applied local stretch to the left ventricle of ∼15%, while the heart was continuously paced from the paced right atrium at a constant cycle length. It was observed that 5 minutes of applied local stretch led to changes in T-wave morphology and longer stretch led to more electrical remodeling.

70 ouainwr iia ota fteetr td ouainb otuzne al. et Houthuizen by population study entire study the our of of that characteristics to baseline dyssynchrony similar However, of were (30%). type population LBBB the A developed dyssynchrony, follow-up later at pre-existing or analysis. or to transient for entirely related or either eligible were ECGs, was reasons missing Other population to due patient (31%). TAVI excluded were original patients the of miss- of portion and large 26% collection Only data in imperfections data. with ing study observational retrospective a was This Limitations 4.4.5 generating, mechanism. hypothesis exact the is find above duration to described short necessary as is a model research after The more (observed unknown. memory also cardiac is dyssynchrony) of of function and mechanism the thermore, ntenro R ains e oasgicn euto nQRS in reduction significant a to led patients, unloading. QRS occurring narrow mechanical as the the abnormalities, in of changes conduction consequence the without a of unloading as some mechanical occurred Accordingly, comparable have However, valve. also aortic could development the LBBB of in time: time stenosis same over the the of (almost) removal and at LBBB occur of events two patients LBBB TAVI-induced LBBB In versus unloading to due remodeling Electrical 4.4.3 sacmaidb ihrrs fvnrclrtcyrhtmaslk osd ePointes, de Torsade di like the tachyarrhythmia’s between ventricular interval interval of JTc JTc longer risk the a higher but that studies a some elec- by in of accompanied suggested mechanism is is the It unknown. and still significance amount are functional the remodeling quantify trical the However, to reported, variables are remodeling. important intervals be electrical QTc also of or might JTc area the or only group. amplitude T-wave remodeling, non-remodeling whereas electrical the at in looking patients sig- studies to a most compared show In interval not JTc did in group, remodeling decrease electrical larger the nificantly decrease, largest the showing tertile The euto nteT the in reduction T the in tion nti td ag ag fT of range large a study this In relevance Potential 4.4.4 LBBB. to due of changes some the fact, by In overruled procedure. TAVI are the alone TAVI of to not due and changes LBBB the of result a predominantly were group etrudrtnigo h eainbtensrcua,eetia n ehnclchanges mechanical and electrical structural, between relation achieve the to of and results study understanding our better smaller confirm even a to required an is echocardiographic to measures systematic outcome led clinical including have and study measurements would prospective study multicenter this larger, A to data population. these adding consequence, patients. a these As in inconsistently collected were measurements echocardiographic fortunately, area h BBptet hwdopst hne,caatrzdb significant a by characterized changes, opposite showed patients LBBB The . area hrfr,tecagsosre fe h AIpoeuei h LBBB the in TAVI procedure the after observed changes the Therefore, . area hnebtenaueadcrncLB a observed. was LBBB chronic and acute between change ff rn ruswr iia uigaltm-ons Fur- time-points. all during similar were groups erent lcrclrmdln nTV-nue BBpatients LBBB TAVI-induced in remodeling Electrical area u osgicn reduc- significant no but 18 Un- . 71

4 Chapter 4 in the heart. Such a study will also allow performing a multivariate analysis to adjust for covariates such as age, gender, comorbidities, and echo parameters. VCGs were reconstructed from the 12-lead ECG using the Kors method rather than direct recording of the Frank-VCG19. This was dictated by the available 12-lead ECG data. How- ever, the Kors method shows a good resemblance to the true 3-dimensional Frank-VCG.9,20 For the chronic measurements, ECG recordings were used that were acquired between one month and one year after TAVI (6 ± 5 months). This wide range was accepted because repolarization changes associated with primary electrical remodeling only occur within the ﬁrst week after initiation of an altered electrical activation.6 Furthermore, no signiﬁcant changes were observed between ECGs measured during one to twelve months after the onset of TAVI- induced LBBB (Fig. 4.3).

4.5 Conclusions

This study in patients with TAVI-induced LBBB shows that on a population level LBBB induces electrical remodeling. However, there is a large diﬀerence in change in Tarea on a patient level. The amount of electrical remodeling might be associated with the functional status of a patient.

4.6 Funding

This work was supported within the framework of Center for Translational Molecular Medi- cine (www.ctmm.nl), Project COHFAR (Congestive Heart Failure and Arrhythmia) [Grant number 01C-203], and supported by the Dutch Heart Foundation.

72 References .LWce uui,GLnal RRsn n eged.Rgtvnrclrpacing- ventricular Right Bergfeldt. L and Rosen, MR Lundahl, G Rubulis, A Wecke, L 6. .ASvli,BBjvc adc usk iebu,adM oeho.Vec- Josephson. ME and Zimetbaum, P Gussak, I Vajdic, B Bojovic, B Shvilkin, A 7. .PHuhie,LF a ase TPes eJeee M a e on BM Boon, der van RMA Jaegere, de P Poels, TT Garsse, Van LAFM Houthuizen, P 8. .J os a epn CSti,adJ a eml eosrcino h Frank the of Reconstruction Bemmel. van JH and Sittig, AC Herpen, van G Kors, JA 9. .KVroy AMVrek eca,HG rjs rs N onlse,and Cornelussen, RNM Arts, S T Crijns, HJGM Køber, Peschar, M L Verbeek, XAAM Vernooy, Stellbrink, K C Veldhuisen, 3. van DJ Dickstein, K Huvelle, E Zannad, F 2. Campana, C Marini, M Marchionni, N Lucci, D Gorini, M Opasich, C Baldasseroni, S 1. .G oael n abn lcrpyilgclrmdln nhprrpyand hypertrophy in remodeling Electrophysiological Marbán. E and Tomaselli GF 4. .TAb,G ekt,A at,TLu aa hkr LDmao PAb- TP Dimaano, VL Chakir, K Daya, S Liu, T Barth, AS Hesketh, GG Aiba, T 5. nue lcrpyilgclrmdln ntehmnhatadisrltosi ocar- to relationship its memory. and diac heart human the in remodeling electrophysiological induced n otnosvnrclrpacing. activation ventricular ventricular continuous normal and during memory cardiac of determinants torcardiographic wnes MTnBr,FvndrKe,M cai,JBa r ta.Lf bundle- Left al. et Jr, Baan death. J of risk Schalij, increases implantation MJ Circulation valve Kley, aortic transcatheter der by induced van block F branch Berg, Ten JM Swinkels, etradormfo tnadeetoadorpi ed:dansi oprsnof comparison di diagnostic leads: electrocardiographic standard from vectorcardiogram WPizn etbnl rnhbokidcsvnrclrrmdligadfunctional and remodelling ventricular hypoperfusion. induces septal block branch bundle Left block Prinzen. branch FW bundle Left failure. Lechat. heart to P progression and for factor Ferrari, risk R a as Maggioni, AP Ritter, P Cazeau, failure. heart congestive congestive on with network 143:398–405. outpatients Italian (2002) the 5517 from in report rate a with mortality on failure: associated heart total Network is and Italian block sudden and bundle-branch 1-year Maggioni, Left increased AP Tavazzi, Investigators. L Failure Masotti, Heart G Congestive Deorsola, A Perini, G er failure. heart aa,BORuk,F kr AKs,adG oael.Electrophysiological Tomaselli. resynchronization by GF restoration its and therapy. and Kass, failure heart DA dyssynchronous of Akar, consequences FG O’Rourke, B raham, ﬀ rn methods. erent Circulation 21)126:720–728. (2012) adoacRes Cardiovasc er Rhythm Heart u er J Heart Eur 20)119:1220–1230. (2009) u er J Heart Eur 20)4:1477–1486. (2007) 19)42:270–283. (1999) 19)11:1083–1092. (1990) 20)26:91–98. (2005) er Rhythm Heart u er Fail Heart J Eur 20)6:943–948. (2009) 20)9:7–14. (2007) mHatJ Heart Am

4 Chapter 4

10. EB Engels, EM Végh, CJM Van Deursen, K Vernooy, JP Singh, and FW Prinzen. T- wave area predicts response to cardiac resynchronization therapy in patients with left bundle branch block. J Cardiovasc Electrophysiol (2015) 26:176–183. 11. HHM Draisma, MJ Schalij, EE van der Wall, and CA Swenne. Elucidation of the spatial ventricular gradient and its link with dispersion of repolarization. Heart Rhythm (2006) 3:1092–1099. 12. A Shvilkin, B Bojovic, B Vajdic, I Gussak, KK Ho, P Zimetbaum, and ME Josephson. Vectorcardiographic and electrocardiographic criteria to distinguish new and old left bundle branch block. Heart Rhythm (2010) 7:1085–1092. 13. A Costard-Jäckle, B Goetsch, M Antz, and MR Franz. Slow and long-lasting modulation of myocardial repolarization produced by ectopic activation in isolated rabbit hearts. Evidence for cardiac "memory". Circulation (1989) 80:1412–1420. 14. PY Courand, P Lantelme, and P Gosse. Electrocardiographic detection of left ventricular hypertrophy: Time to forget the Sokolow-Lyon index? Arch Cardiovasc Dis (2015) 108:277–280. 15. EG Caiani, A Auricchio, M Potse, R Krause, A Pellegrini, RM Lang, and P Vaïda. Evaluation of the relation between changes in R-wave amplitude and LV mass and dimensions in a model of ‘Reversed Hypertrophy’. In: Computing in Cardiology 2013. 2013, pp. 867–870. 16. NHL Kuijpers, E Hermeling, J Lumens, HMM ten Eikelder, T Delhaas, and FW Prin- zen. Mechano-electrical coupling as framework for understanding functional remodeling during LBBB and CRT. Am J Physiol Heart Circ Physiol (2014) 306:H1644– H1659. 17. EA Sosunov, EP Anyukhovsky, and MR Rosen. Altered ventricular stretch contributes to initiation of cardiac memory. Heart Rhythm (2008) 5:106–113. 18. P Houthuizen, RMA van der Boon, M Urena, N Van Mieghem, GBR Brueren, TT Poels, LAFM Van Garsse, J Rodés-Cabau, FW Prinzen, and P de Jaegere. Occurrence, fate and consequences of ventricular conduction abnormalities after transcatheter aortic valve implantation. EuroIntervention (2014) 9:1142–1150. 19. E Frank. An accurate, clinically practical system for spatial vectorcardiography. Cir- culation (1956) 13:737–749. 20. EB Engels, S Alshehri, CJM van Deursen, L Wecke, L Bergfeldt, K Vernooy, and FW Prinzen. The synthesized vectorcardiogram resembles the measured vectorcardiogram in patients with dyssynchronous heart failure. J Electrocardiol (2015) 48:586–592.

Chapter 5 T-wave area predicts response to cardiac resynchronization therapy in patients with left bundle branch block

The content of this chapter is based on:

Elien B. Engels,1,∗ Eszter M. Végh,2,3,∗ Caroline J.M. van Deursen,1 Kevin Vernooy,4 Jagmeet P. Singh,2 and Frits W. Prinzen1 (2015). T-wave area predicts response to cardiac resynchronization therapy in patients with left bundle branch block. J Cardiovasc Electrophysiol 26(2):176-183.

∗ Authors contributed equally. 1 Department of Physiology, CARIM School for Cardiovascular Diseases, Maastricht University, Maastricht, the Netherlands. 2 Cardiac Arrhythmia Service, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA. 3 Semmelweis University Heart and Vascular Center, Budapest, Hungary. 4 Department of Cardiology, Maastricht University Medical Center, Maastricht, the Netherlands. Chapter 5

Abstract

Introduction: Chronic heart failure patients with a left ventricular (LV) conduction delay, mostly due to left bundle branch block (LBBB), generally derive beneﬁt from cardiac resynchronization therapy (CRT). However, 30-50% of patients do not show a clear response to CRT. We investigated whether T-wave analysis of the ECG can improve patient selection.

Methods and Results: The study population comprised 244 CRT recipients with baseline 12-lead electrocardiogram recordings. Echocardiographic response after 6-months CRT was deﬁned as a ≥5% increase in LV ejection fraction (LVEF). Vectorcardiograms (VCGs) were constructed from the measured 12-lead ECGs using an adapted Kors algorithm on digitized ECGs. Logistic regression models indicated repolarization variables as good predictors of

CRT response. The VCG-derived Tarea predicted CRT response (odds ratio [OR] per 10 µVs increase 1.172 [P < 0.001]), even better than QRSarea (OR = 1.116 [P = 0.001]). Tarea had especially predictive value in the LBBB patient group (OR = 2.77 in LBBB vs. 1.09 in non-LBBB). This predictive value persisted after adjustment of multiple covariates, such as gender, ischemia, age, hypertension, coronary artery bypass graft, and the usage of diuretics and beta-blockers. In LBBB patients the increase in LVEF was 6.1 ± 9.7% and 11.3 ± 9.1% in patients with Tarea below and above the median value, respectively (P < 0.01).

Conclusions: In patients with LBBB morphology of the QRS complex, a larger baseline Tarea is an important independent predictor of LVEF increase following CRT.

78 h er n ih rvd oevlal nomto hnteoedmninlEG This ECG. within one-dimensional forces the di than electrical the information compares the valuable study of more information provide might three-dimensional and advantage contains heart The it the (VCG). that vectorcardiograms is analysis into VCG converted comprehensive the were more of For signals ECG patients. T-wave, CRT the of of cohort large a from electrocardiograms in- regulate relaxation. that and those contraction including mediate channels, that ion concentrations, many calcium of tracellular These activity potential. the action by myocardial determined the are of phases it phases repolarization since and information, plateau additional the elec- of the provide reflective of may to is T-wave phase the central repolarization that vital clearly hypothesized equally We is the of myocardium. sequence importance activation the this ignores it While dyssynchrony, trical wall. ventricular the throughout C rdco fQSdrto.W loivsiae h oeta rdcievle fthese of values predictive di potential within the parameters investigated also repolarization We duration. QRS of predictor ECG prvdb h G ntttoa eiwBadadEhc omte n a conducted was was project and Committee The Ethics Clinic. and Multidisciplinary Board Review MGH Institutional the MGH at the a 6-month by had approved and Patients 3-, (MGH). 1-, Hospital at General visit followup Massachusetts the at implantation CRT underwent ewe 04ad21,59cneuieptet h e salse niain NwYork III (New class indications established Association met Heart who patients consecutive 569 2010, and 2004 Between population Study 5.2.1 Methods 5.2 ti hrfr motn oipoeptetslcin ordc h neesr s fex- of use itself. unnecessary procedure implantation the the reduce to to related complications selection, prevent patient and remodeling. devices improve reverse pensive cardiac to of important evidence therefore no show is 50% It to up and CRT after response clinical sasrn rdco fCTrsos,wievle ewe 2 n 5 srqiefurther response. require CRT ms of 150 predictor and good 120 between values while response, evidence. CRT of predictor strong a is hrp CT.CThsbe hw oipoecricpm ucin Fsmtm,qual- survival. symptoms, and HF hospitalization function, life, pump of cardiac improve ity to shown been has CRT (CRT). therapy C Introduction 5.1 nodrt xlr h rdcievleo h -ae ertopcieyivsiae the investigated retrospectively we T-wave, the of value predictive the explore to order In depolarization, is: that activation, electrical of sequence the reflects complex QRS The R sidctdfrptet ihH n ieQScmlx R duration QRS A complex. QRS wide a and HF with patients for indicated is CRT otydet etbnl rnhbok(BB,addcesdlf etiua ejec- ventricular left decreased (LVEF and fraction (LBBB), tion block branch bundle left to due mostly hronic 3,4 oercnl,LB opooyo h R ope a ensont ea be to shown been has complex QRS the of morphology LBBB recently, More er alr H)ptet ihalf etiua L)cnuto delay, conduction (LV) ventricular left a with patients (HF) failure heart ff rn eoaiainadrplrzto aaeest h established the to parameters repolarization and depolarization erent / Vsmtm,LVEF symptoms, IV < 5)gnrlydrv eetfo ada resynchronization cardiac from benefit derive generally 35%) 5–9 ff rn R morphologies. QRS erent 1 oee,2-0 fidvda ainsd o show not do patients individual of 20-30% However, -aepeit R epnei BBpatients LBBB in response CRT predicts T-wave < 5 n R duration QRS and 35% > 2 milliseconds), 120 ≥ 5 ms 150 79 2

5 Chapter 5 in accordance with the Declaration of Helsinki. Patients with a preimplantation ECG available in digital archives, either in sinus rhythm or atrial ﬁbrillation, were enrolled in this retrospective analysis. ECGs with < 3 normal beats (e.g., due to premature ventricular complexes [PVCs]), remarkable interfering noise, or ventricular pacing were excluded from the analysis.

5.2.2 Echocardiography

Transthoracic echocardiography was performed at baseline and at 6 months clinical follow-up. The images were taken by Philips iE33 (Eindhoven, The Netherlands), SONOS 5500/7500 (Andover, MA, USA) and General ElectricVivid 7 (GE, Milwaukee, WI,USA) ultrasound machines. CRT response was deﬁned as absolute increase in LVEF of ≥ 5% after 6 months of CRT, determined using the biplane method of discs (modiﬁed Simpson’s method).

5.2.3 ECG analysis

Before CRT implantation, supine 12-lead ECGs were recorded digitally by a MAC 5500 ECG Machine (GE Healthcare, Waukesha, WI, USA) at a paper speed of 25 mm/s and a frequency of 250 Hz. All ECGs were stored digitally as PDF files in the MUSE Cardiology Information system (GE Medical System). ECGs recorded up to 1 month before CRT implantation were included into the analysis. The vector graphics of these PDF files were obtained by converting the files to an .svg file using Inkscape version 2 (Boston, MA, USA). The calibration pulses, as indicated in Fig. 5.1, were used to scale the signals to real time and amplitude. The digital ECG signals were semiautomatically analyzed using a custom-made computer program written in MATLAB R2010b (MathWorks, Natick, MA, USA). After band-pass filtering between 0.5 – 40 Hz and baseline wander removal, the beginning and end of the QRS complex could be detected using the curve length transformation (cLT).10 The beginning of the QRS complex corresponded to the last found minimum value and the end of the QRS complex was the first point with maximal cLT value. Additionally, the end of the T-wave was identified as the intersection between baseline and the tangent of the slope of the T-wave. The QRS axis was calculated using lead I and II, where angles between -30◦ and -90◦ were defined as left axis deviation (LAD). Furthermore, patients were classified as LBBB or non-LBBB according to the MADIT-CRT criteria.5

5.2.4 VCG analysis

From the digital 12-lead ECG signals, a VCG was synthesized using the Kors method.11 The matrix proposed by Kors et al. 12 is based on a learning set from the Common Standards for

80 ewe h aia R-adTvco a enda h QRS the as defined was T-vector and QRS- maximal the between cie o h C analysis. ECG the for scribed iue5.1: Figure patients. and subjects The healthy regressions. both linear included multiple by populations generated was and library multilead Electrocardiography h nl ntefotlpaedfie rm0 from defined plane frontal the in angle the 3D rabtentecreadtebsln rmQSbgnigt n nlasX ,and Y, X, leads in end to beginning QRS from QRS baseline Z: the and curve the between area ‘3-D’ ac rmteoii oapito h op,adisapiueadlcto nsaewere space in angle location the (0 and is Azimuth plane amplitude elevation. transverse its and the and azimuth as loop), in expressed the were on angles vector point dis- The a (maximal method calculated. to defined Bazett’s origin were using T-vector the changes and from QRS rate tance T-wave, maximum heart the the for or Subsequently, complex corrected interval). QRS was the (QTc interval of end QT the The and complex respectively. QRS the of beginning the between ed 1 ,I,II :aR V,aF :V 3: aVF; aVL, aVR, 3 of 2: groups III; 4 II, were I, there case (1: our leads in However, simultaneously. measured leads independent aebgnigadedo nRRitra nec ed fe eetn etfrec lead, each for beat 1 selecting the After select lead. to each synthesized. needed in was interval was VCG R-R the lead an running of This end and 5.1. beginning Fig. same in shown as measured, simultaneously o ahptet h R opadTlo eeietfiduigtesm ehda de- as method same the using identified were loop T and loop QRS the patient, each For h C aeu h potnt oas nlz h rao h op.TeQRS The loops. the of area the analyze also to opportunity the us gave VCG The area yia xml ftelyu famaue 2la C,soigclbainple and pulses calibration showing ECG, 12-lead V measured lead a of layout the of example Typical = q QRS 1 srniglead. running as area 2 , x ◦ + left, QRS 10 h R uainadQ nevlwr enda h time the as defined were interval QT and duration QRS The + area 2 90 , ◦ y + rn,-90 front, QRS -aepeit R epnei BBpatients LBBB in response CRT predicts T-wave 1 area 2 ◦ V , dwwrs o180 to (downwards) , ◦ z h T The . 2 ak 180 back, V , 3 :V 4: ; 11 area h osmto sue l 8 all assumes method Kors The ◦ a acltdsmlry using similarly, calculated was 4 ih) h lvto nl is angle elevation The right). V , / angle. T 5 V , ◦ uwrs.Teangle The (upwards). 6 n unn lead running 1 and ) area sthe is 81

5 Chapter 5 the part from the end of the QRS complex to the end of the T-wave. The sum QRST area was the sum of the absolute values of the QRSarea and Tarea (Fig. 5.2). The ratio between QRS and T areas was deﬁned as the QRS/T area ratio.

Figure 5.2: Calculation of the 3-D QRS and Tarea using the integral between the signal and the baseline from beginning to end of the QRS complex or from the end of the QRS complex to end of

the T-wave, respectively. From the QRS and Tarea, the sum QRST area was calculated.

5.2.5 Statistical analysis

The statistical analysis was performed using IBM SPSS statistics software version 21 (SPSS Inc., Chicago, IL, USA). Continuous and discrete variables are presented as mean ± standard deviation (SD) and counts (percentages), respectively. Linear correlations were evaluated by Pearson’s correlation or Kendall’s tau coefficient when appropriate. Comparison between different patient groups was performed using either 1-way ANOVA (continuous variables) or the χ2-test (discrete variables). Follow-up paired comparisons were made using the Tukey test. The classification performance of electrical parameters in identifying CRT response was evaluated by odds ratios (OR) calculated using logistic regression models. The likelihood ratio test was used for comparison of the goodness-of-fit of hierarchical models. A 2-sided P-value < 0.05 was considered statistically significant.

5.3 Results

5.3.1 Patient characteristics

During the study period, a total of 569 patients underwent CRT implantation. Eighty patients were excluded because of missing ECGs, 91 because of missing baseline or follow-up echocardiograms, 116 due to pre-existent pacing, and 38 because of frequent PVCs or unac-

82 one otelf n h akwieteTvco a onigt h etadtefront, the and left the to pointing was vector T the while QRS-vector back the the general, the and In of longer. left areas lasted the T-wave the the to but because pointed similar, amplitude, T-vector were loop the T than and larger QRS times 3 was amplitude QRS-vector h ainshdapoogdQSdrto.TeaeaeQSai a ntenra axis normal the in was axis QRS average The duration. (-30 QRS range prolonged a had patients The analysis VCG and ECG 5.3.2 aibe r hw scut pretg)o mean or (percentage) counts as shown are Variables 5.1: Table ischemic had 5.1). patients (Table symptoms the IV of or half III class about NYHA and male, years had were 67 most of and patients cardiomyopathy, age 24%.Most mean a of with LVEF population baseline CRT typical a a entire the represent to patients similar 244 was The group patient cohort. presents included 5.1 the Table that analysis. indicating for characteristics, remained baseline patients the 244 of total a Consequently, noise. ceptable rcin NYHA fraction; steEGadVGaayi eut fol h nlddcohort included the only of results analysis VCG and ECG the as aeieptetcaatrsiso h nldd(n included the of characteristics patient Baseline ◦ = o-90 to e okHatAscain HF Association; Heart York New ◦ u h Dwslre niaigalrevrainbtenptet.The patients. between variation large a indicating large, was SD the but ) C measurements VCG C measurements ECG aibeIcue nieCohort Entire Included Characteristics Baseline Variable R xs( axis QRS shmcH tooy(,% 3 (55 136 %) (n, etiology HF Ischemic R uain(ilscns 175 (milliseconds) duration QRS BB(,% 5 (63 154 %) (n, LBBB er ae(p)75 (bpm) rate Heart hoi F(,% 3(21 53 %) (n, AF Chronic QRS -etreeain( elevation T-vector Titra mlieod)472 (milliseconds) interval QT u RTae ( area QRST Sum -etraiuh( azimuth T-vector T T nevl(ilscns 520 (milliseconds) interval QTc -etrapiue(V 0 (mV) amplitude T-vector R-etreeain( elevation QRS-vector R-etraiuh( azimuth QRS-vector QRS aeieLE % 24 (%) LVEF Baseline YAclass NYHA g yas 66 (years) Age eae(,% 1(20 51 %) (n, Female area nnw n )2 (9 24 %) (n, Unknown I n )10(77 190 %) (n, III V(,% 2(9 22 %) (n, IV I(,% (3 8 %) (n, II / area ( nl ( angle T µ s 84 Vs) ( µ s 90 Vs) ◦ ) ◦ 156 ) µ ± ◦ = s 174 Vs) ◦ 78 ) tnaddvainwe prpit.LVEF appropriate. when deviation standard 87 ) er alr;LBBB failure; heart ◦ ◦ ) 90 ) -aepeit R epnei BBpatients LBBB in response CRT predicts T-wave − − 15 65 n . . . 5 3 9 = ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 4 n 244 = . . . 61 15 83 26 63 33 29 0 45 88 24 7 53 47 77 12 )18(20 118 8) )4 (8 49 0) )2 (4 24 3) ...... )37(59 337 7) )28(48 258 1) )18(29 168 7) )38(66 378 9) )16(20 116 9) 2 24 0 etbnl rnhbok AF block; branch bundle left = . 68 8 4)adetr oot(n cohort entire and 244) . . 2 6 = ± ± 596 . . 7 12 6) 2) ...... 7) 3) 5) 5) 4) 4) 2 . 4 = etvnrclrejection ventricular left = tilﬁbrillation. atrial = 9)a well as 596) 83

5 Chapter 5 indicating a large angle between the 2 vectors (Table 5.1). Typical examples of 12-lead ECGs and corresponding 3-D vector loops for 2 patients, diagnosed with LBBB with a large or small Tarea, are shown in Fig. 5.3A, Fig. 5.3B, respectively.

Figure 5.3: Typical examples of 12-lead ECGs and corresponding 3-D vector loops for a patient with a

large (A) and a small (B)Tarea, despite being both classiﬁed as having LBBB.

5.3.3 Echocardiographic response predicted by ECG and VCG variables

Logistic regression models indicated VCG-derived repolarization variables as good predictors of CRT response (Table 5.2). An increase in T-vector amplitude or Tarea was associated with a greater probability of ≥ 5% or ≥ 10% points increase in LVEF. Prediction of LVEF increase by the T-vector amplitude and Tarea was even stronger than by the QRS-vector amplitude or QRSarea (Table 5.2). Adding QRSarea to a model including only the Tarea did not

84 al,T larly, (88.6 agrT larger (R LBBB and BBadicei tooy T etiology. ischemic and LBBB rbblt feh epne ainswtotLDo h R-etrddntso better a show not did QRS-vector the (OR of response LAD CRT without Patients response. echo of probability fQRS of .5 (P 0.753 bewsfrhreautdi t eaint R uain QRS duration, QRS to relation its in evaluated further was able al 5.2: Table (QRS vector T and QRS Tbe53adFg .) h Ro h QRS the of OR The 5.4). Fig. and 5.3 (Table orlto coe correlation rcin OR fraction; o-BBptet lohdfil ml T small fairly had also patients Non-LBBB responder. CRT a becoming of nodrt etrudrtn h odpeito fCTrsos yteT the by response CRT of prediction good the understand better to order In T 5.3.4 a rdciepwrcmaal ota fteT the of that to comparable power predictive a had rsne r einvle n 5 ofiec nevl C) h nraeclm niae h tpsz ftevralsue to used LVEF variables Wald-test. the the of using size calculated step were the values indicates P column Corresponding Increase (OR). The ratios (CI). odds intervals the confidence calculate 95% and values median are Presented (P significantly goodness-of-fit the improve otptet eersodr n T and responders were patients most Fg 5.4B). (Fig. sepce,nnLB ainswr oecmol o-epnesta responders. than non-responders commonly more were patients non-LBBB expected, As ± area area area 49v.105.5 vs. 44.9 = = area ierRgeso oe o rdcinof Prediction for Model Regression Linear mltd m)126(1.115–1.413) 1.256 (mV) amplitude T aibeO 9%C)frPVleO 9%C)frPVleIncrease ms 0 10 P-Value 0.011 for 0.679 CI) (95% OR (1.015–1.124) 1.068 (0.907–1.066) 0.983 P-Value 0.009 0.890 for CI) (95% OR (1.017–1.129) 1.071 (0.919–1.077) 0.994 (mV) amplitude QRS (ms) duration QRS Variable QRS QRS T QRS a mle nptet ihicei tooycmae onnicei etiology non-ischemic to compared etiology ischemic with patients in smaller was dsratio. odds .3,idctn htalreT large a that indicating 0.13), area ln,tegons-ffi mrvdsgicnl (P significantly improved goodness-of-fit the alone, hnptet ihu BB(97.8 LBBB without patients than + area / ( nl ( angle T µ ra( area T ffi s .7 (1.083–1.255) 1.172 Vs) = sa diinlpeitro coadorpi response echocardiographic of predictor additional an as ( µ in f06 niae htmr atr hnQRS than factors more that indicated 0.63 of cient s .1 101115 .0 .9 100119 .0 10 0.003 (1.030–1.149) 1.094 0.001 (1.051–1.195) 1.116 Vs) .6.Nvrhls,ptet igoe ihLB a significantly a had LBBB with diagnosed patients Nevertheless, 0.36). = µ ◦ s .7 (1.090–1.266) 1.175 Vs) .9 090115 .6 .0 090121 .6 10 0.063 (0.990–1.231) 1.105 0.067 (0.990–1.195) 1.094 ) .5[.919] P [0.69–1.91]; 1.15 ± / nl)wsascae ihannsgicnl (P non-significantly a with associated was angle) T 73 P 47.3; ∆ LVEF area = ≥ area a oryrltdt ohteQSdrto (R duration QRS the both to related poorly was .1.T 0.01). 5% n QRS and area = area -aepeit R epnei BBpatients LBBB in response CRT predicts T-wave = < < < 0.59). eaiet h QRS the to relative area .0 .3 (1.101–1.377) 1.231 0.001 .0 .2 (1.062–1.207) 1.127 0.001 .0 .3 (1.057–1.210) 1.131 0.001 / .0,wieadn T adding while 0.40), rartofrpeitn R epnewas response CRT predicting for ratio area T n QRS and ± orltdbte ihQRS with better correlated area ≥ %or 5% 54v.60.6 vs. 45.4 eelre hni h o-BBgroup non-LBBB the in than larger were area ∆ LVEF ≥ area ln.Alre nl ewe the between angle larger A alone. 0 boueIces nLVEF in Increase Absolute 10% ≥ Fg .A.I h BBgroup LBBB the In 5.4A). (Fig. 10% < area ± area .1.TesmQS area QRST sum The 0.01). ed oices h odds the increase to tends area 36 P 33.6; area < < < swl speec of presence as well as , .0 0 0.001 .0 10 0.001 .0 20 0.001 = etvnrclrejection ventricular left ln eemn T determine alone oamdlconsisting model a to area < . . = mV 1 mV 1 .0) Simi- 0.001). lhuhthe although , area ◦ µ µ µ .6 higher 0.06) Vs Vs Vs hsvari- this , = 0.34) area 85

5 Chapter 5

Figure 5.4: Scatter plot of the relation between QRSarea and Tarea within non-LBBB (A) and LBBB patients (B). The responders are indicated by blue triangles, non-responders by red circles.

Table 5.3: Baseline and outcome patient characteristics after dividing patients according to their QRS

morphology and Tarea

LBBB Non-LBBB

Tarea Tarea Tarea Tarea < Median ≥ Median < Median ≥ Median (n = 51) (n = 103) (n = 61) (n = 29)

Age (years) 66.1 ± 10.6 68.1 ± 12.0 66.9 ± 13.5 64.3 ± 16.8 Female (n, %) 8 (15.7) 32 (31.1)⋆ 8 (13.1) 3 (10.3) Ischemic HF etiology (n, %) 33 (64.7) 50 (48.5) 37 (60.7) 16 (55.2) Hypertension (n, %) 36 (70.6) 83 (80.6) 45 (73.8) 17 (58.6) CABG (n, %) 22 (43.1) 34 (33.0) 24 (39.3) 12 (41.4) QRS duration (ms) 173 ± 30 183 ± 31 162 ± 32 174 ± 39 QTc interval (ms) 524 ± 92 533 ± 75 495 ± 61 518 ± 78 † † QRSarea (µVs) 78.0 ± 32.7 122.1 ± 44.0 50.1 ± 22.5 83.4 ± 36.3 † † Tarea (µVs) 53.0 ± 14.1 120.0 ± 38.6 41.7 ± 13.7 100.5 ± 27.5 NYHA class baseline (-) 3.0 ± 0.4 3.1 ± 0.4 3.0 ± 0.3 3.2 ± 0.4 NYHA class follow-up (-) 2.3 ± 0.8 2.0 ± 0.8 2.4 ± 0.9 2.3 ± 0.4 ∆NYHA class (-) −0.8 ± 0.8 −1.0 ± 0.8 −0.7 ± 0.8 −0.9 ± 0.7 LVEF baseline (%) 25.1 ± 7.1 24.5 ± 6.6 24.3 ± 7.5 22.4 ± 7.2⋆ LVEF follow-up (%) 31.2 ± 11.1 35.8 ± 10.9⋆ 28.6 ± 10.8 28.9 ± 13.4 ∆LVEF (%) 6.1 ± 9.7 11.3 ± 9.1⋆ 4.3 ± 10.0 6.5 ± 14.3 ∆LVEF ≥ 5% (n, %) 27 (52.9) 78 (75.7)† 24 (39.3) 12 (41.4) ∆LVEF ≥ 10% (n, %) 19 (37.3) 58 (56.3)⋆ 14 (23.0) 10 (34.5) Continuous variables are presented as mean ± standard deviation (SD), discrete variables as counts (percentages). HF = heart failure; CABG = coronary artery bypass graft; NYHA = New York Heart Association; LVEF = left ventricular ejection fraction. ⋆ P value < 0.05 compared to the Tarea < median group with the same LBBB conditions. † P value < 0.01 compared to the Tarea < median group with the same LBBB conditions.

86 oe nbt h BB(OR LBBB the both in power o-BBptetgop (OR groups patient non-LBBB iue5.5: Figure hssuydmntae htteTwv otisifrainta mrvsteprediction T VCG-derived the months. improves 6 that at information response CRT contains echocardiographic T-wave for the that demonstrates study This Discussion 5.4 ≥ BBadnnLB ain ruswr udvddi ugop ihabsln T baseline a with subgroups in subdivided were groups patient non-LBBB and LBBB eeahge ecnaeo eae (P females of percentage higher a were shmcH tooy(P etiology HF ischemic 11 .9;P 5.39]; – [1.16 ugopas,tedces nNH ls 10v.07–09 a lgtylre n the and larger slightly 6.5%; – was 4.3 0.9) vs. (11.3% – subgroups 3 0.7 other P the vs. in than (1.0 larger significantly class was LVEF NYHA in increase in decrease the also, subgroup fehcrigahcrsodr 7 s 9–5% i.55.Tbe53adFg . also 5.5 Fig. and 5.3 Table 5.5). Fig. 53%; – T 39 that indicate vs. (75 responders echocardiographic of fmlil oaits uha edr shma g,hpreso,crnr reybypass artery and coronary diuretics hypertension, of adjustment usage age, after the ischemia, and similar gender, graft, remained as ORs such of covariates, values multiple covariates, of to adjustments before ORs the < h ein ewe hs ugop,teol ain hrceitcsoigdi showing characteristic patient only the subgroups, 4 these Between median. the h s fQRS of use The nodrt netgt hte T whether investigate to order In .1.Ti agrices nLE lotasae noasgicnl ihrpercentage higher significantly a into translated also LVEF in increase larger This 0.01). ihn1gopi niae ntebr.TeOsbtente2QSmrhlge and morphologies QRS 2 the between ORs The T bars. large the and small in between indicated is group 1 within h ecnaeo coadorpi epnesi BBadnnLB R morpho- T QRS with non-LBBB cohorts and into LBBB subdivided in logy, responders echocardiographic of percentage The area = a opeitv au ntennLB ugop hl i.55shows 5.5 Fig. While subgroup. non-LBBB the in value predictive no had 0.02). area rsmQS raisedo T of instead area QRST sum or = .6 nteLB-ihTwv ugop(al .) nthis In 5.3). (Table subgroup T-wave LBBB-high the in 0.06) = = .9[.. n OR and [N.S.] 1.39 .6[.. n OR and [N.S.] 1.76 areas area β bokr ajse Rwti BBptet a 2.50 was patients LBBB within OR (adjusted -blockers r indicated. are a rdcievlecmlmnayt BB the LBBB, to complementary value predictive a had = -aepeit R epnei BBpatients LBBB in response CRT predicts T-wave area .4 n oe ecnaeo ainswith patients of percentage lower a and 0.04) < or † ≥ niae P indicates area = h einvle h ubro patients of number The value. median the = .0[..,respectively). [N.S.], 1.50 i o rvd ihrdistinctive higher a provide not did .0[P 2.40 < 0.01. = area .2,rsetvl)and respectively) 0.02], rdcsCTresponse CRT predicts ff area erences < 87 or

5 Chapter 5 even better than any QRS-complex related parameter. This better outcome prediction especially applies to patients diagnosed with LBBB.

5.4.1 Tarea as an additional predictor of CRT response This report clearly illustrates the added value of the VCG derived T-wave in predicting CRT response in LBBB patients. The data indicate that the chance of echocardiographic CRT response in patients with LBBB and a large Tarea is 75%, as opposed to 53% in LBBB patients with small Tarea and 40% in non-LBBB patients. Considering the fact that large studies show an average echocardiographic response rate of ∼50%2, the combination of LBBB and T-wave criteria strongly improves patient selection for CRT. The sum QRST area showed similar prognostic value within LBBB patients (unadjusted OR = 2.40 vs. 2.77 for Tarea alone) but this parameter was probably mainly determined by the Tarea itself. The lack of additional predictive value of Tarea in the non-LBBB group might be due to the fact that the LBBB sequence of electrical activation is the dominant electrical substrate for CRT.5

There may be several reasons why Tarea improves predicted CRT response. First of all, there is, to some extent, a positive relation between Tarea and QRS duration. It is well known that QRS duration is an independent predictor of CRT response, especially when differentiat- ing between QRS duration values below and over 150 milliseconds.3,4 Moreover, the T-wave amplitude is known to be smaller in patients with large BMI13, possibly simply related to the larger distance of the surface to the heart. Large BMI is known, for not well-understood reasons, to reduce CRT response.14 Notably, in this study the ORs were not adjusted for BMI. In contrast to Brenyo et al. 15 we did not find a significant effect of the QRS axis deviation on the benefit of CRT in this study. A possible explanation for these discrepant findings might be that the patients in this study had severe HF (mostly NYHA class III–IV) while the MADIT-CRT study consisted of patients with NYHA class I–II. Another explanation may be that the MADIT-CRT study consisted of a larger number of patients.

Even though different heart conditions might also reduce QRSarea, in this study the QRS/T area ratio showed a tendency to predict CRT response, pointing toward specific repolarization related factors. One possible explanation might be hypertrophy. In patients with narrow QRS complex hypertrophy is known to lead to smaller T waves: T-wave flattening, also referred to as ‘strain-pattern’ of the T-wave.16 It is incompletely understood whether similar conditions also reduce the T-wave in patients with wide QRS complex. Finally, it is known that HF in general and dyssynchronous HF in particular leads to extensive changes in expression of many ion channels, such as K+ and Ca2+ channels.17 In short, the predictive power of the T-wave reflective of electrical recovery suggests that this component of the action potential may give additional information of the responsiveness of the myocardial substrate. Regardless of the exact mechanism, this study shows that several categories of patients, which are known to respond well to CRT, are overrepresented in the higher Tarea category,

88 niaeta oe epn etrt CRT. to better respond women that indicate aeyptet h r eae and females are who patients namely tooyo Fadmr BB h esnfrti e-eae eeca e beneficial sex-related this for reason the LBBB, more and HF of etiology diinly h IAL study MIRACLE the Additionally, h C,a nlzdi hssuy a ecntutdfo eua 2la C.Sub- ECG. 12-lead regular T a and from morphology QRS constructed be can T study, the this sequently, in analyzed as VCG, The implications clinical Potential 5.4.2 hc ol lordc h T the reduce also would which nptet ihLB opooyo h R ope,alre aeieT baseline larger a complex, QRS the of morphology LBBB with patients In Conclusions 5.5 Frank-VCG. estimation the an to resemblances is Kors best the ECG the that the shows demonstrated from have method VCG studies using previous the However, Moreover, calculate Frank-VCG. gold-standard to results. the method our of Kors confirm the to of version required implan- adapted CRT is methods an after study the and prospective in before multicenter available detailed A echocardiograms as had tation. data patients missing of 73% with Only study section. single-center retrospective, a was This Limitation 5.4.3 approaches. these to information add or replace can analysis T-wave our whether investigate betv esr o etrslcino ainswt BBfrCT te tde showed studies Other CRT. for echocardiography LBBB of with value patients the of selection better for measure objective iselaigt mle muto iseta a ersnhoie yCT ned as Indeed, CRT. by resynchronized be can that study modeling tissue computer of a by amount shown During smaller a to cardiomyopathy. occur leading can non-ischemic tissue coupling with cellular patients or degeneration than ischemia, response volumetric smaller a uoaial acltd tcnesl eapidi h lnc ofrhrehneorability our enhance patients. further LBBB to within clinic, CRT be the to in can response predict applied variable to be this easily Since can CRT. it following calculated, increase automatically LVEF of predictor independent portant eto fCTcniae.Teavnaeo hsT this of advantage The candidates. CRT of lection otnosvral htpoie diinlojcieeiec sapeitro reverse of predictor a as T evidence of addition objective the additional Therefore, remodeling. provides that variable continuous a area a eatmtclycluae.Cneunl,cmie nlssof analysis combined Consequently, calculated. automatically be can area rvdsa ayadwdl plcbeapoc oipoese- improve to approach applicable widely and easy an provides 23,24 area . n R nptetselection. patient in MRI and 20 22 / hwdta shmccrimoah ainsshow patients cardiomyopathy ischemic that showed rhv o-shmceilg fH.Svrlstudies Several HF. of etiology non-ischemic have or nopiglast euto nTwv amplitude, T-wave in reduction a to leads uncoupling , area -aepeit R epnei BBpatients LBBB in response CRT predicts T-wave nptetslcinfrCTi nesl available easily an is CRT for selection patient in 18,19 lhuhte edt aels ischemic less have to tend they Although area 21 12,27,28 sta ti oha3Dmaueand measure 3-D a both is it that is eutn nnnodcigfibrotic nonconducting in resulting 25,26 twl ewrhhl to worthwhile be will It ff c sunclear. is ect area sa im- an is 89

5 References

1. WT Abraham, WG Fisher, AL Smith, DB Delurgio, AR Leon, E Loh, DZ Kocovic, M Packer, AL Clavell, DL Hayes, et al. Cardiac resynchronization in chronic heart failure. N Engl J Med (2002) 346:1845–1853. 2. European Heart Rhythm Association, European Society of Cardiology, Heart Rhythm Society, Heart Failure Society of America, American Society of Echocardiography, American Heart Association, European Association of Echocardiography, Heart Fail- ure Association, JC Daubert, L Saxon, et al. 2012 EHRA/HRS expert consensus state- ment on cardiac resynchronization therapy in heart failure: implant and follow-up re- commendations and management. Heart Rhythm (2012) 9:1524–1576. 3. S Ploux, Z Whinnett, J Lumens, A Denis, A Zemmoura, M De Guillebon, K Ramoul, P Ritter, P Jaïs, J Clementy, M Haïssaguerre, and P Bordachar. Acute hemodynamic response to biventricular pacing in heart failure patients with narrow, moderately, and severely prolonged QRS duration. Heart Rhythm (2012) 9:1247–1250. 4. A Auricchio, C Stellbrink, C Butter, S Sack, J Vogt, AR Misier, D Böcker, M Block, JH Kirkels, A Kramer, E Huvelle, Pacing Therapies in Congestive Heart Failure II, Study Group, and Guidant Heart Failure Research Group. Clinical efficacy of cardiac resynchronization therapy using left ventricular pacing in heart failure patients stratified by severity of ventricular conduction delay. J Am Coll Cardiol (2003) 42:2109–2116. 5. W Zareba, H Klein, I Cygankiewicz, WJ Hall, S McNitt, M Brown, D Cannom, JP Daubert, M Eldar, MR Gold, et al. Effectiveness of Cardiac Resynchronization Therapy by QRS Morphology in the Multicenter Automatic Defibrillator Implantation Trial-Cardiac Resynchronization Therapy (MADIT-CRT). Circulation (2011) 123:1061–1072. 6. Y Seo, H Ito, S Nakatani, M Takami, S Naito, T Shiga, K Ando, Y Wakayama, K Aonuma, and J-CRT investigators. The role of echocardiography in predicting responders to cardiac resynchronization therapy. Circ J (2011) 75:1156–1163. 7. MO Sweeney, RJ van Bommel, MJ Schalij, CJW Borleffs, AS Hellkamp, and JJ Bax. Analysis of ventricular activation using surface electrocardiography to predict left ventricular reverse volumetric remodeling during cardiac resynchronization therapy. Circulation (2010) 121:626–634. 8 rhd JMs,EFse,LPdlti aseht odneg Greenberg, H Goldenberg, I Barsheshet, A Padeletti, L Foster, E Moss, AJ Arshad, A 18. 3 MNsr JRbl OJns n DSa.Tee The Shah. AD and Jones, SO Rubal, BJ Nasir, JM 13. 4 CHu DSlmn oron cit odneg li,A os E Moss, AJ Klein, H Goldenberg, I McNitt, S Bourgoun, M Solomon, SD Hsu, JC 14. 1 a,A lr,C crus J Borle CJW Schreurs, CA Algra, AM Man, S 11. 7 ia GHseh SBrh i,SDy,KCai,V ian,T Ab- TP Dimaano, VL Chakir, K Daya, S Liu, T Barth, AS Hesketh, GG Aiba, T 17. 2 AKr,GvnHre,A itg n HvnBme.Rcntuto fteFrank the of Reconstruction Bemmel. van JH and Sittig, AC Herpen, van G Kors, JA 12. 6 ey aoo,R ’gsio JBlne,adW anl rgotcim- Prognostic Kannel. WB and Belanger, AJ D’Agostino, RB Salomon, M Levy, D 16. 5 rno a,ABrhse,DCno,AQeaa cit THag AJ Huang, DT McNitt, S Quesada, A Cannom, D Barsheshet, A Rao, M Brenyo, A 15. 0 og BMoy n in.Arbs pnsuc loih odtc ne and onset detect to algorithm open-source robust A Jiang. D and Moody, GB Zong, W 10. .KVroy N onlse,XA ebe,WRVng,AvnHni,M Hunnik, van A Vanagt, WYR Verbeek, XAAM Cornelussen, RNM Vernooy, K 8. .YTa,PZag i a,TZu ag i ag un n Guo. J and Yuan, C Wang, J Li, D Wang, L Zhu, T Gao, Y Li, X Zhang, P Tian, Y 9. upr rs JMCin,adF rne.Cricrsnhoiaintherapy resynchronization Cardiac hearts. block Prinzen. bundle-branch left FW canine and in dyssynchronopathy Crijns, cures HJGM Arts, T Kuiper, recmlt etbnl rnhbokmrhlg togypeit odresponse good predicts therapy. strongly resynchronization morphology cardiac block to branch bundle left complete True 28:2148–2155. omte.Cricrsnhoiainteayi oee more is Executive MADIT-CRT therapy and resynchronization Steinberg, JS Cardiac Solomon, S Committee. Zareba, W McNitt, S Hall, WJ AI-R mlietratmtcdfirlao mlnaintilwt ada re- cardiac with study. trial therapy) the synchronization implantation outcome: defibrillator clinical automatic in (multicenter improvement MADIT-CRT car- associated to and super-response therapy of resynchronization Predictors diac Committee. Executive MADIT-CRT and Foster, adults. young in electrocardiograms surface uaino R opee.In: complexes. QRS of duration aa,BORuk,F kr AKs,adG oael.Electrophysiological Tomaselli. resynchronization by GF restoration its and therapy. and Kass, failure heart DA dyssynchronous of Akar, consequences FG O’Rourke, B raham, etradormfo tnadeetoadorpi ed:dansi oprsnof comparison di diagnostic leads: electrocardiographic standard from vectorcardiogram lctoso aeieeetoadorpi etrsadtersra hne nsubjects in hypertrophy. changes ventricular serial left their with and features electrocardiographic baseline of plications ainswt idysmtmtchatfiueerle nMADIT-CRT. in in enrolled Electrophysiol therapy failure resynchronization heart cardiac symptomatic of mildly benefit with the and patients axis QRS Zareba. W and Moss, R- nl opeitaryhisi ainswt shmchatdsaeadsystolic vec- and disease the dysfunction. heart ventricular ischemic of with left patients Influence spatial in electrocardiogram-derived arrhythmias predict the Swenne. to of angle CA power QRS-T and the on Schalij, matrix MJ synthesis Cannegieter, torcardiogram SC Wall, der van ff rn methods. erent Circulation 21)24:442–448. (2013) u er J Heart Eur 20)119:1220–1230. (2009) Electrocardiol J 19)11:1083–1092. (1990) mCl Cardiol Coll Am J optr nCrilg,2003 Cardiology, in Computers Circulation Europace Electrocardiol J 21)44:410–415. (2011) 19)90:1786–1793. (1994) ff ,RCShrtn,LvnEvn EE Erven, van L Scherptong, RWC s, 21)15:1499–1506. (2013) 21)59:2366–2373. (2012) ff cso oyms ne on index mass body of ects ff ciei oe hnin than women in ective 21)45:646–651. (2012) 03 p 737–740. pp. 2003, . u er J Heart Eur Cardiovasc J References (2007) 91

5 Chapter 5

men: the MADIT-CRT (Multicenter Automatic Defibrillator Implantation Trial with Cardiac Resynchronization Therapy) trial. J Am Coll Cardiol (2011) 57:813–820. 19. S Zabarovskaja, F Gadler, F Braunschweig, M Ståhlberg, J Hörnsten, C Linde, and LH Lund. Women have better long-term prognosis than men after cardiac resynchronization therapy. Europace (2012) 14:1148–1155. 20. MGSJ Sutton, T Plappert, KE Hilpisch, WT Abraham, DL Hayes, and E Chinchoy. Sustained reverse left ventricular structural remodeling with cardiac resynchronization at one year is a function of etiology: quantitative Doppler echocardiographic evidence from the Multicenter InSync Randomized Clinical Evaluation (MIRACLE). Circula- tion (2006) 113:266–272. 21. LR Dekker, H Rademaker, JT Vermeulen, T Opthof, R Coronel, JA Spaan, and MJ Janse. Cellular uncoupling during ischemia in hypertrophied and failing rabbit ventricular myocardium: effects of preconditioning. Circulation (1998) 97:1724–1730. 22. M Potse, D Krause, L Bacharova, R Krause, FW Prinzen, and A Auricchio. Similarities and differences between electrocardiogram signs of left bundle-branch block and left- ventricular uncoupling. Europace (2012) 14 Suppl 5:v33–v39. 23. GE Leenders, MJ Cramer, MD Bogaard, M Meine, PA Doevendans, and BW De Boeck. Echocardiographic prediction of outcome after cardiac resynchronization therapy: conventional methods and recent developments. Heart Fail Rev (2011) 16:235– 250. 24. J Gorcsan 3rd. Role of echocardiography to determine candidacy for cardiac resynchronization therapy. Curr Opin Cardiol (2008) 23:16–22. 25. JA White, R Yee, X Yuan, A Krahn, A Skanes, M Parker, G Klein, and M Drangova. Delayed enhancement magnetic resonance imaging predicts response to cardiac resynchronization therapy in patients with intraventricular dyssynchrony. J Am Coll Cardiol (2006) 48:1953–1960. 26. H Cochet, A Denis, S Ploux, J Lumens, S Amraoui, N Derval, F Sacher, P Reant, S Lafitte, P Jais, F Laurent, P Ritter, M Montaudon, and P Bordachar. Pre- and intra- procedural predictors of reverse remodeling after cardiac resynchronization therapy: an MRI study. J Cardiovasc Electrophysiol (2013) 24:682–691. 27. DL Cortez and TT Schlegel. When deriving the spatial QRS-T angle from the 12-lead electrocardiogram, which transform is more Frank: regression or inverse Dower? J Electrocardiol (2010) 43:302–309. 28. CA Schreurs, AM Algra, SC Man, SC Cannegieter, EE van der Wall, MJ Schalij, JA Kors, and CA Swenne. The spatial QRS-T angle in the Frank vectorcardiogram: accuracy of estimates derived from the 12-lead electrocardiogram. J Electrocardiol (2010) 43:294–301.

Chapter 6 T-wave area as biomarker of clinical response to cardiac resynchronization therapy

The content of this chapter is based on:

Elien B. Engels,1,∗ Eszter M. Végh,2,3,∗ Caroline J.M. van Deursen,1 Béla Merkely,3 Kevin Vernooy,4 Jagmeet P. Singh,2 and Frits W. Prinzen1 (2016). T-wave area as biomarker of clinical response to cardiac resynchronization therapy. Europace 18(7):1077-1085.

Abstract

Introduction: There is increasing evidence that left bundle branch block (LBBB) morphology on the ECG is a positive predictor for response to cardiac resynchronization therapy (CRT).

We previously demonstrated that the vectorcardiography(VCG)-derived Tarea predicts echocardiographic CRT response in LBBB patients. In the present study we investigate whether the Tarea also predicts long-term clinical outcome to CRT.

Methods and Results: This is a retrospective study consisting of 335 CRT recipients. Primary endpoint was the composite of heart failure (HF) hospitalization, heart transplantation, left ventricular assist device implantation or death during a 3-year follow-up period. HF hospitalization or death alone were secondary endpoints. The patient subgroup with a large Tarea and LBBB 36% reached the primary endpoint, which was considerably less

(P < 0.01) than for patients with LBBB and a small Tarea or non-LBBB patients with a small or large Tarea (48%, 57%, and 51%, respectively). Similar diﬀerences were observed for the secondary endpoints HF hospitalization (31% vs. 51%, 51%, and 38%, respectively, P < 0.01) and death (19% vs. 42%, 34%, and 42%, respectively, P < 0.01). In multivariate analysis a large Tarea and LBBB were the only independent predictor of the combined endpoint beside high creatinine levels and use of diuretics.

Conclusions: Tarea may be useful as an additional biomarker to stratify CRT candidates and improve selection of those most likely to beneﬁt from CRT. A large Tarea may derive its predictive value from reﬂecting good intrinsic myocardial properties and a substrate for CRT.

96 C Introduction 6.1 eueti ubri oipoeteptetselection. patient the improve to is number this reduce a endmntae htCTipoe ohtemriiyadmortality. and morbidity the both improves CRT that demonstrated been has hr eanacnieal ubro o-epnest R 3 50%) - (30 CRT to non-responders of number considerable a remain there ycrilato oeta,aenttknit con.I rvosstudy previous a In the account. of into taken phase not repolarization are the potential, reflect action that myocardial sequence variables T-wave depolarization However, the cycle. of heart representation a one are during variables Both duration. QRS the and plex oCTi BBptet.Ptet ihbt BBmrhlg n ag T large a and morphology LBBB a both with Patients patients. LBBB in CRT to T the that 5 hneo coadorpi epne(boueices nLVEF in increase (absolute response echocardiographic of chance 75% olwu) sopsdt 3 nptet ihLB n ml T small a and LBBB with patients in 53% to opposed as follow-up), T pcieyivsiae odtrieteptnilpeitv au fteT the of value predictive potential the determine to investigated spectively clinical known years. with 3 patients least CRT at with study of previous follow-up the from group patient the extended we ihnnLB morphology. non-LBBB with ihdi with ahstsGnrlHsia MH.Alptet a e okHatAscainclass Association Mas- Heart York the New at CRT had underwent patients who All III patients (MGH). 569 Hospital of General consisted sachusetts population study current The Population Study 6.2.1 Methods 6.2 parameters. electrocardiographic relevant ihpeiu aig(n pacing previous with os nteEG(n ECG the on noise oa f35ptet nlddi h nlss hssuypplto seult h co- the to equal is population study This analysis. the in previously included described patients hort 335 of total a adorpyiaigwr vial.Tepoetwsapoe yteMHInstitutional MGH the by approved was echo- project and The specialist available. (HF) failure were the heart imaging in control, cardiography visits device follow-up where had clinic, Patients multidisciplinary follow-up. MGH echocardiographic not but clinical long-term area / h urn guidelines current The h lcrcrigas(Cs n etradorm VG)o l ainswr retro- were patients all of (VCGs) vectorcardiograms and (ECGs) electrocardiograms The Vsmtm,LVEF symptoms, IV nLB ainsas rdcsteln-emciia ucm oCT nodrt oso, do to order In CRT. to outcome clinical long-term the predicts also patients LBBB in ff rn R opoois hsptnilpeitv au a locmae oother to compared also was value predictive potential This morphologies. QRS erent ardiac etiua odcindly seilydet etbnl rnhbok(BB.It (LBBB). block branch bundle left to due left especially a delay, with combined conduction (LVEF) ventricular fraction ejection ventricular left decreased a with tients area sdtrie yvcocrigah rdcsteehcrigahcresponse echocardiographic the predicts vectorcardiography by determined as eycrnzto hrp CT a mre sa e an as emerged has (CRT) therapy resynchronization = = 8,o isn aeieEG (n ECGs baseline missing or 38), 3 < 4 mhsz h motneo BBmrhlg fteQScom- QRS the of morphology LBBB of importance the emphasize u twsetne ih9 R ainso hmw i have did we whom of patients CRT 91 with extended was it but , 1) rqetpeauevnrclrcnrcin runacceptable or contractions ventricular premature frequent 116), 5 n R duration QRS and 35% 4 h i ftecretsuywst netgt hte the whether investigate to was study current the of aim The -aepeit ogtr lnclCTresponse CRT clinical long-term predicts T-wave > 2 s ntepeetsuypatients study present the In ms. 120 = 0 eeecue,laigto leading excluded, were 80) area ff ≥ cieteayi pain therapy ective n 0 npatients in 40% and 4 area %atr6months 6 after 5% edemonstrated we , 2 1 ncombination in n a to way One . Nevertheless, area a a had 97

6 Chapter 6

Review Board and Ethics Committee and was conducted in accordance with the Declaration of Helsinki.

6.2.2 Study endpoints

The patients were followed for an average of 2.4 years with the combined endpoint of HF hospitalization, heart transplantation, left ventricular assist device (LVAD) implantation, and all-cause death (HTLD) as primary endpoint. HF hospitalization was deﬁned as necessity of in-patient admission due to acute cardiac decompensation, symptoms of shortness of breath, signs of congestion on the chest radiograph, peripheral oedema, or relieve of shortness of breath after intravenous medical therapy. Secondary endpoints included HF hospitalization or death alone. Clinical outcome data were gained with the review of hospital medical records and social security death index.

6.2.3 ECG and VCG analysis

All 12-lead ECGs measured up to 6 months prior to CRT implantation were recorded in supine position at a frequency of 250 Hz. All ECGs were measured by MAC 5500 ECG Machine (GE Healthcare, Waukesha, WI, USA) and stored in the MUSE Cardiology Infor- mation system (GE Medical System). These digital PDF files contain vector graphics which were used to extract the original digital ECG, as previously described.4 The ECGs could be analyzed and VCGs could be constructed by the Kors method5 and analyzed using the methods described earlier in detail.4 Briefly, the digital ECGs were semi-automatically analyzed using a custom-made computer program written in MATLAB R2010b (MathWorks, Natick, MA). After band-pass filtering between 0.5 and 40 Hz, the QRS complex was identified using the curve length transformation6 and the end of the T-wave as the intersection between the maximum T-wave slope after the final T-wave peak and the isoelectric line in the TP segment. The QRS axis was calculated using lead I and II. Patients were classified as LBBB or non-

LBBB using the MADIT-CRT criteria (QRS duration ≥ 130 ms; QS or rS in V1; notched or 7 slurring R waves in leads I, aVL, V5, or V6; absent q waves in leads V5 and V6). The digital ECGs were subsequently converted to VCGs using the matrix multiplica- tion provided by Kors et al. 5 After deﬁning the begin and end of the QRS complex and the end of the T-wave, QRS duration, QT interval, and QTc intervals could be measured. The maximum distance between the origin (0,0,0) and a point on the three-dimensional QRS- or T-vector loop were represented by the variables QRS- and T-vector amplitude, respectively. The maximal vectors could also be described by their orientation using the azimuth (angle in the transversal plane with backward vector direction being negative) and elevation (angle in craniocaudal direction with upward vector directions being >90◦).8 The angle between the two maximal vectors was deﬁned by the QRS/T angle. The area of the

98 ration icnl etrta ainswtotLB (P LBBB without patients than better nificantly h uv n h aeiei h ,YadZdrcinadcluae sn h formulas the using calculated and direction Z and Y X, between the areas in ’three-dimensional’ QRS the baseline as the VCG and the curve from the conducted were T-loop and QRS- TDfrptet ihaQSduration QRS a with patients for HTLD otnosaddsrt aibe r rsne smean as presented are variables discrete and Continuous Analysis Statistical 6.2.4 agrta htfrQSae u iia ota o T for that to similar but QRS-area for that than larger h ue et h iceevralswr oprdwt the with compared using were made variables discrete were The comparisons paired test. follow-up Tukey and the variables one-way the continuous with for corre- compared test were Pearson’s ANOVA subgroups by between evaluated variables characteristic were Clinical correlations lation. Linear respectively. (percentages), counts uvsfrteHL npit hr sn infiatdi significant no is Kaplan-Meier using There 6.1A Fig. endpoint. in HTLD shown the is for duration curves QRS and LBBB VCG of the power predictive from The derived areas of ability predictive had The patients the 6.3.2 of two-thirds LBBB. had addition, patients In of amount same rhythm. CRT. the sinus and for etiology indications in HF established and ischemic the male low with were a consistent had patients patients duration, Most The QRS 6.1. Table prolonged in a listed and are LVEF patients 335 the of characteristics baseline The Characteristics Patient 6.3.1 Results 6.3 statistics SPSS IBM Illinois). using Chicago, performed Inc, was (SPSS 21 analysis version software statistical The significant. statistically etduigteWl-et h o ersinmdl eefitdfrcvrae (P covariates for fitted were models regression and Cox analyses The pre- regression Wald-test. hazard in the proportional parameters using to Cox tested the electrical used by was of evaluated was test performance response log-rank predictive CRT The dicting The endpoints. values. VCG-derived secondary between probability or associations determine primary the the evaluate reaching to and used variables was function survival of estimator fsbeswspromdi ainswt aeieT baseline a with analysis patients purposes, in descriptive performed For was used. subsets was of approach selection stepwise backward a where h rso-a etfrhtrgniytsig w-ie P-value two-sided A testing. heterogeneity for test Breslow-Day the area < = 5 s(P ms 150 q QRS area 2 = , x + .1.Hwvr ainswt BBmrhlg efre sig- performed morphology LBBB a with patients However, 0.41). QRS area 2 , y + QRS area 2 ≥ -aepeit ogtr lnclCTresponse CRT clinical long-term predicts T-wave , 5 scmae optet ihaQSdu- QRS a with patients to compared ms 150 z rT or < area .1.Ti itnto yLB was LBBB by distinction This 0.01). = area area q ff n u RTae Fg 6.1B). (Fig. area QRST sum and rnei h hneo reaching of chance the in erence T < ± area 2 and tnaddvain(D and (SD) deviation standard , x χ + 2 ≥ ts.TeKaplan-Meier The -test. T h einwt s of use with median the < area 2 .5wsconsidered was 0.05 , y + T area 2 , z . 9 < 0.05), 99

6 Chapter 6

Table 6.1: Baseline ECG and VCG analysis results of the entire cohort (n = 335) as well as the entire cohort divided by primary endpoint (HTLD) or survival status at the end of a 3-year follow- up

Variable Entire cohort -HTLD +HTLD Survivors Non-survivors (n = 335) (n = 203) (n = 132) (n = 262) (n = 73)

Baseline characteristics Age (years) 67 ± 13 66 ± 13 69 ± 12 66 ± 13 71 ± 11§ Female (n, %) 70 (21) 50 (25) 20 (15)⋆ 59 (23) 11 (15) Baseline LVEF (%) 24 ± 7 24 ± 7 23 ± 7 24 ± 7 23 ± 7 NYHA class II (n, %) 8 (3) 6 (3) 2 (2) 8 (3) 0 (0) III (n, %) 236 (70) 148 (73) 88 (67) 190 (73) 46 (63) IV (n, %) 31 (9) 16 (8) 15 (11) 22 (8) 9 (12) Unknown (n, %) 60 (18) 33 (7) 27 (20) 42 (16) 18 (25) Ischemic HF etiology(n, %) 192 (57) 105 (52) 87 (66)⋆ 143 (55) 49 (67) LBBB (n, %) 199 (59) 132 (65) 67 (51)† 164 (63) 35 (48)‡ Chronic AF (n, %) 85 (25) 47 (23) 38( 29) 66 (25) 19 (26) ECG measurements Heart rate (bpm) 75 ± 15 75 ± 15 76 ± 14 75 ± 15 75 ± 14 QRS axis (◦) −20 ± 58 −17 ± 57 −24 ± 59 −22 ± 56 −11 ± 64 VCG measurements QRS duration (ms) 172 ± 33 172 ± 32 172 ± 35 173 ± 34 170 ± 31 QT interval (ms) 467 ± 86 467 ± 83 466 ± 90 466 ± 86 469 ± 85 QTc interval (ms) 516 ± 77 515 ± 74 517 ± 82 515 ± 78 518 ± 75 QRS amplitude (mV) 1.5 ± 0.5 1.5 ± 0.5 1.4 ± 0.5 1.5 ± 0.5 1.4 ± 0.5 ‡ QRSarea (µVs) 88 ± 46 92 ± 47 82 ± 45 91 ± 47 78 ± 40 QRS azimuth (◦) −62 ± 57 −69 ± 47 −49 ± 70† −66 ± 52 −44 ± 73‡ QRS elevation (◦) 90 ± 24 89 ± 24 92 ± 24 90 ± 23 90 ± 24 T amplitude (mV) 0.5 ± 0.2 0.5 ± 0.3 0.4 ± 0.2† 0.5 ± 0.3 0.4 ± 0.2§ † § Tarea (µVs) 82 ± 47 88 ± 47 74 ± 45 87 ± 49 66 ± 35 T-vector azimuth (◦) 77 ± 65 77 ± 62 76 ± 71 77 ± 64 76 ± 72 T-vector elevation (◦) 85 ± 26 87 ± 26 83 ± 26 85 ± 26 85 ± 27 QRS/T angle (◦) 155 ± 30 156 ± 30 154 ± 27⋆ 155 ± 31 156 ± 21 QRS+T area (µVs) 170 ± 85 179 ± 84 156 ± 82⋆ 177 ± 87 144 ± 68§ Variables are shown as counts (percentage) or mean ± standard deviation when appropriate. AF: atrial ﬁbrillation, HF: heart failure, LBBB: left bundle branch block, LVEF: left ventricular ejection fraction, HTLD: composite endpoint of HF hospitalization, heart transplantation, left ventricular assist device, or death. ⋆P < 0.05; †P < 0.01 compared to the patient-group whom did not reach HTLD. ‡P < 0.05; §P < 0.01 compared to the survivor-group.

Patients with QRS-, T-, and sum QRST areas ≥ than their median values all reached the primary endpoint significantly less than patients with areas < the median values. Moreover, patients reaching the primary endpoint had a significantly smaller Tarea compared to patients not reaching the primary endpoint, while QRSarea was not significantly different between the two groups (Table 6.1). Since LBBB morphology of the QRS complex is a well-known predictor for CRT response and because the percentage of patients with LBBB morphology was significantly higher in the patient group not reaching the primary endpoint (Table 6.1), the patient population was divided according to QRS morphology and subsequently subdivided into cohorts with < or

100 rsn n7 u f17ptet 4% ihasalT small a with (42%) patients 167 of out 70 in present 6 ains(7)wt ag T large a with (77%) T patients with was 168 etiology group HF sub- patient ischemic other with the patients the in of to found percentage compared highest the etiology Indeed, HF 6.2). (Table ischemic groups with patients less significantly contained iue6.1: Figure atyiflec h hneo ecigteHL npita erflo-p(P follow-up year 3 at endpoint HTLD P the reaching of chance the influence cantly T ≥ h BBptet ihasalT small a with patients LBBB the (P o ml T small a for area = einQRS median h ootwt BBadalreT large a and LBBB with cohort The < .5 epciey i.61) ncnrs,tesbru fptet ihLB n large a and LBBB with patients of subgroup the contrast, In 6.1C). Fig. respectively; 0.35, .1(i.61) T 6.1D). 0.01)(Fig. a infiatylwricdneo ecigteHL npit(6)cmae to compared (36%) endpoint HTLD the reaching of incidence lower significantly a had h eut ftenwVGvralsT variables VCG new the of results the h ainswr hnsbiie nofu rusacrigt hi R morphology QRS their to according groups four QRS into and subdivided then were patients The h eut fcretgieievralsLB n R uainaesoni ( in shown are duration QRS and LBBB variables years. guideline 3 after current HTLD of endpoint results composite The of free probability the of estimates Kaplan–Meier QRS area area area s 1 o agrT larger a for 51% vs. rT or rT or area area ( area C area rT or ) r values are ihnnnLB n BBsbrus QRS subgroups, LBBB and non-LBBB Within . i o infiatya significantly not did area area ( area areas D < .LreQRS Large ). 4% P (48%, einvalue. median a BBQSmrhlg hl BBwsonly was LBBB while morphology QRS LBBB a had area nisfis urie(i.62.I diin 2 u of out 129 addition, In 6.2). (Fig. quartile first its in area P , -aepeit ogtr lnclCTresponse CRT clinical long-term predicts T-wave rne ocnanmr eae (P females more contain to trended = = area 0.65). .1 swl stennLB ain groups patient non-LBBB the as well as 0.01) area QRS , ff c TDi h o-BBptet (57% patients non-LBBB the in HTLD ect rT or area area area n u RTae r hw n( in shown are area QRST sum and , r values are (P < .1 i.62.TeQRS The 6.2). Fig. 0.01, ≥ einvleadsmall and value median area i o signifi- not did = = .9 and 0.09) .1and 0.81 A and ) 101 B ).

6 Chapter 6

duration increases when the Tarea increases, but the association between these two variables is not as high as the association between Tarea and the number of LBBB patients.

Table 6.2: Baseline patient characteristics after dividing patients according to their QRS morphology

and Tarea. The median cut-oﬀ value for the Tarea was based on the entire cohort

LBBB Non-LBBB T-area T-area T-area T-area < Median ≥ Median < Median ≥ Median (n = 70) (n = 129) (n = 97) (n = 39)

Age (years) 67.3 ± 10.4 67.6 ± 12.8 67.1 ± 13.3 65.7 ± 15.9 Female (n, %) 13 (18.6) 38 (29.5) 14 (14.4) 5 (12.8) Ischemic HF etiology (n, %) 45 (64.3) 63 (48.8)⋆ 61 (62.9) 23 (59.0) Hypertension (n, %) 53 (75.7) 104 (80.6) 75 (77.3) 25 (64.1) CABG (n, %) 33 (47.1) 46 (35.7) 44 (45.4) 18 (46.2) Heart rate (bpm) 74 ± 14 75 ± 15 77 ± 14 74 ± 15 QRS duration (ms) 170 ± 29 184 ± 30† 158 ± 32 173 ± 39 QTc interval (ms) 514 ± 87 532 ± 74 492 ± 64 526 ± 87 † † QRSarea (µVs) 77 ± 34 121 ± 44 50 ± 22 88 ± 36 † † Tarea (µVs) 52 ± 15 121 ± 40 42 ± 15 108 ± 42 HF: Heart failure, CABG: Coronary artery bypass surgery. ⋆P < 0.05 compared with the T-area < median group with the same LBBB conditions. †P < 0.01 compared with the T-area < median group with the same LBBB conditions.

Figure 6.2: Histogram showing the associations between LBBB (Turquoise, left y-axis) or QRS dura-

tion (red, right y-axis) with Tarea. The percentage of ischaemic patients was indicated by

shading in the LBBB bars. Tarea was divided into quartiles, indicated by Q1, Q2, Q3, and Q4 on the x-axis.

In a univariate analysis of the total population ischemic HF etiology, LBBB, QRSarea, T amplitude, and Tarea were some of the signiﬁcant predictors for reaching HTLD (Table 6.3).

Since T amplitude and Tarea are mutually dependent (R = 0.87), only Tarea was included in the multivariate model. Multivariate analysis indicated that besides baseline creatinine levels and diuretics, only LBBB and Tarea remained in the model. The multivariate model indicated

102 o-ersinhzr oe eeldteT the revealed model hazard Cox-regression T 6.3.4 ( echo-responders Similarly, o-BB(2) o ainswt BB otlt a infiatyadconsiderably and significantly with T was patients large in mortality a with highest LBBB, patients was with in risk lower patients mortality For the subgroups, (42%). all non-LBBB Analyzing 6.4). (Fig. LBBB nLE a ny3.4 only was LVEF in e hnteeh-o-epnesdd(4 s 1,P 51%, vs. (24% did echo-non-responders the than ten .706,P 0.17-0.67, infiatcvrae hzr ai [HR] ratio (hazard covariates significant ecignn ftecmie npitHL a 10.4 was HTLD endpoint combined patients in the LVEF of in increase none echo- mean reaching on The based examined. echocardiography also and was remodeling outcome reverse clinical cardiographic adverse any reaching of relation The outcome clinical vs. Echocardiographic 6.3.3 hneo et oprdt BBptet ihasalT small a with patients LBBB to compared death of chance otlt a osattruhu di throughout constant was mortality T ihasalT small a with ain h hneo Fhsiaiainwsams wc shg o ainswt small a with patients for high medi- as and twice almost gender was as hospitalization T such HF covariates of significant chance for The adjusting cation. after even mortality, and tion htptet ihalreT large a with patients that 5 I .5-09,P 0.96, - 0.35 CI: 95% eeaaye,ie BBptet ihalreT large a with patients LBBB i.e. analyzed, were Tbe61.Smlry h hneo uvvlwsapoiaeytiea ag nthe in large as twice approximately T was large survival a of with chance group the patient Similarly, 6.1). (Table fH optlztosatr3yaso R a ihs ntennLB ru n LBBB T and large group a non-LBBB with the patients T in highest small was a CRT with of subgroups years 3 after hospitalizations HF of area area ihrgr omraiy uvvr a ihrpeaec fLB n ag T large and LBBB of prevalence higher a had survivors mortality, to regard With ≥ (HR or hospitalization area < = h einvlews05,artota eandsgicn fe dutetfor adjustment after significant remained that ratio a 0.54, was value median the .3 5 ofiec nevl[I:03-.3 P 0.35-0.83, [CI]: interval confidence 95% 0.53, < area .1.Atog ecitv ustaayi hwdta h e the that showed analysis subset descriptive Although 0.01). sadtoa rdco fmraiyadhatfailure heart and mortality of predictor additional as diinly ihnLB ainstehzr ai o TDbtena between HTLD for ratio hazard the patients LBBB within Additionally, . area ± = .%i ainswoddrahtecmie npit(P endpoint combined the reach did who patients in 8.9% area Fg 6.3A). (Fig. .3.Ti di This 0.03). area ∆ hra Fhsiaiainwssgicnl oe nLBBB in lower significantly was hospitalization HF whereas , LVEF area area a 7 oe iko ecigHL oprdt those to compared HTLD reaching of risk lower 37% a had 1% hni hs ihasalT small a with those in than (19%) oprdt ainswt ml T small a with patients to compared ff ≥ rn ugop,i a o rsn o ainswithout patients for present not was it subgroups, erent %atr6mnh olwu)rahdHL esof- less HTLD reached follow-up) months 6 after 5% ff = rnewsee agrwe nyLB patients LBBB only when larger even was erence .5 P 0.45, -aepeit ogtr lnclCTresponse CRT clinical long-term predicts T-wave area loa togpeitro Fhospitaliza- HF of predictor strong a as also < .1 al 6.3). Table 0.01; area < a naeaea average on had 0.01). ± 06 hl hsimprovement this while 10.6% < area .1.Ide,teamount the Indeed, 0.01). (HR area 4% Fg 6.3B). (Fig. (42%) = area ∼ ff .4 5 CI: 95% 0.34, ie lower times 3 c fT of ect (HR < = area 0.01). 0.57, 103 area on

6 Chapter 6 C:agoesncnetn-nye F tilfirlain R:agoesnI ye1rcpo lce,CB:crnr reybps rf,LB:lf udebac lc,LE:left LVEF: block, branch bundle left LBBB: parameters(P graft, significant bypass statistically artery indicate coronary figures intervention. CABG: Bold blocker, coronary receptor percutaneous 1 PCI: type fraction, P-values. II ejection and angiotensin ventricular CIs ARB: 95% fibrillation, corresponding atrial their AF: with angiotensin-converting-enzyme, HRs ACE: unadjusted and adjusted the are Shown 6.3: Table ucm parameter outcome nvraeadmliait nlsso di of analysis multivariate and Univariate aeiecetnn .511 1.39 - 1.13 1.25 creatinine Baseline QRS R .206 .207 .606 .90.83 0.09 1.89 0.52 - 0.60 1.00 - 0.93 1.38 1.06 - 0.53 0.97 0.85 0.35 2.11 - 0.70 1.27 0.71 1.80 - 0.67 1.10 0.04 0.71 5.48 0.26 - 1.02 T 1.42 0.44 - 0.60 1.01 2.37 - 0.96 1.23 0.92 - 0.62 0.03 0.99 0.87 4.47 0.08 - 1.11 1.93 2.22 - 0.96 0.01 1.35 duration QRS 4.13 - (%) 1.20 LVEF 2.23 ARB 0.74 ACEi 1.35 - 0.66 β 0.94 Digoxin antagonist Aldosterone Diuretics Titra m)10 .9-10 .510 .9-10 0.34 1.01 - 0.99 1.00 0.95 1.00 - 0.99 1.00 QRS (ms) interval QT C .609 .601 .408 .90.16 2.39 - 0.86 1.44 0.16 0.41 2.27 - 0.87 1.54 - 1.41 0.35 0.02 0.26 0.75 0.41 0.87 - 0.17 2.35 2.05 0.34 - - 0.02 0.80 0.75 0.39 1.37 2.04 1.24 - 1.06 0.78 - 0.01 1.01 1.26 0.09 1.03 0.83 0.01 - 0.20 3.69 - 0.02 3.46 0.91 0.41 - 1.23 1.83 1.05 0.11 - 2.06 1.01 0.83 - 1.96 1.03 0.42 - 0.94 0.59 1.36 population LBBB 0.15 0.13 1.81 - 0.91 1.14 - 1.28 0.38 0.45 0.66 0.64 population LBBB 1.65 1.60 0.05 - - 0.79 0.75 LBBB 1.16 1.99 1.10 - 1.01 0.07 PCI 1.42 0.02 surgery Valve 1.04 0.02 - population 0.40 Total 2.85 - 0.10 CABG 2.18 1.12 0.64 - 1.06 AF Paroxysmal 1.79 1.03 - 1.52 0.99 AF Chronic 1.01 etiology HF Ischaemic population Total Diabetes Hypertension Female (years) Age Variable R mltd m)07 .3-10 .907 .7-12 0.29 1.25 - 0.47 0.77 0.68 - 0.14 0.09 0.31 1.05 - 0.53 0.75 (mV) amplitude T (mV) amplitude QRS bokr06 .0-10 .707 .2-15 0.38 1.54 - 0.32 0.71 0.07 1.04 - 0.40 0.65 -blocker area area / ≥ nl ( angle T ein05 .1-0.82 - 0.41 0.58 median ≥ ein06 .8-09 .307 .7-13 0.35 1.31 - 0.47 0.79 0.03 0.95 - 0.48 0.68 median ≥ ◦ .009 .003 .909 .00.16 1.00 - 0.99 0.99 0.34 1.00 - 0.99 1.00 ) 5 s08 .1-12 .110 .2-18 0.79 1.89 - 0.62 1.08 0.41 1.22 - 0.61 0.86 ms 150 R9%C R9%C R9%C R9%C P CI 95% HR P CI 95% HR P CI 95% HR P CI 95% HR ± 0.05). nvraeMliait nvraeMultivariate Univariate Multivariate Univariate ff rn aaeesi h niechr lf)adteLB ain ru rgt aigHL sthe as HTLD taking (right) group patient LBBB the and (left) cohort entire the in parameters erent < < < < .106 .2-09 .305 .4-08 .104 .6-0.77 - 0.26 0.45 0.01 0.88 - 0.34 0.54 0.03 0.96 - 0.42 0.63 0.01 1.56 - 1.21 1.38 0.01 .106 .4-09 0.04 0.98 - 0.44 0.65 0.01 .103 .1-10 0.05 1.00 - 0.11 0.33 0.01 < .112 .8-1.40 - 1.08 1.23 0.01 < .112 .5-14 0.01 1.41 - 1.05 1.22 0.01 < 0.01

104 a oeovosi h T the in obvious more was nayotoe Fhsiaiain(1 s 1,3% n 1)addah(9 s 34%, vs. (19% death and alone. 51%) 42%) and and 38%, 42%, 51%, vs. (31% hospitalization HF outcomes ondary sopsdt 7,5% n 8 epciey.Smlrdi Similar respectively). 48% and 51%, 57%, to compared opposed CRT T as large receiving or when small (HTLD) with endpoints patients composite non-LBBB with the of one reaching of chance otepeito fCTrsos ntpo BBmrhlg.WieLB ainswt a with patients LBBB While morphology. LBBB of QRS top large on response CRT of contribute prediction to the impor- appears to analysis an T-wave interestingly, is However, LBBB response. that CRT trials of predictor randomized tant large from findings corroborates study present The T 6.4.1 selection improve and CRT. candidates from CRT benefit stratify to likely to most biomarker those additional of an as useful be may rdcieo agrices nLE nLB patients LBBB in LVEF in increase larger a of predictive iue6.3: Figure olwu.I obnto ihtedt rmapeiu td,soigta ag T large a that showing study, previous a from 3-year data during the with death) LVAD, combination transplantation, In hospitalization, follow-up. T (HF baseline outcome large clinical a good LBBB, a with of patients CRT in that shows study This Discussion 6.4 CRT area area ainsit hs ihLB rnnLB R opooy n udvdn hs with those subdividing and morphology, QRS T non-LBBB or LBBB with those into patients h ecnaeo ainsrahn ihrH optlzto ( hospitalization HF either reaching patients of percentage The bars. area eddt epn etrt R hntoewt ml QRS small with those than CRT to better respond to tended nadtoa rdco fln-emciia epneto response clinical long-term of predictor additional an , < or ≥ h einvle h ubro ainswti n ru sidctdi the in indicated is group one within patients of number The value. median the area ainswt ag T large a with Patients . -aepeit ogtr lnclCTresponse CRT clinical long-term predicts T-wave area n BBptet ihasalT small a with patients LBBB and area 4 n BBmrhlg a lower a had morphology LBBB and h rsn td mle htT that implies study present the , ff rne eese o h sec- the for seen were erences A rdah( death or ) area sasrn predictor strong a is areas hsdi this , B fe dividing after ) area ff area erence (36% was 105 area

6 Chapter 6

Figure 6.4: Subset analysis of all-cause mortality. Odds ratio and 95% CI are plotted for the secondary endpoint of all-cause mortality at 3-year follow-up, comparing patients with a large and

small Tarea. CABG: coronary artery bypass graft, HF: heart failure.

Because the present study was based on retrospective analysis and did not contain a control

(non-paced) group, the better outcome in the LBBB-large-Tarea patients may be explained by either a better baseline condition of these patients or a larger beneﬁt of CRT. The latter idea is supported by the observation in our previous study that LBBB patients with large Tarea showed a larger increase in LV ejection fraction.4 It seems plausible that this larger increase in cardiac function further translates into a better clinical outcome, as observed in the present study. Using the Tarea in the LBBB patient could help predict whether CRT may be useful, especially in a subset of LBBB patients who have additional co-morbidities in whom the risk-beneﬁt ratio may not be as clear as in other LBBB patients.

A large Tarea (as well as QRSarea) may also imply a better baseline condition, because it is known that electrical uncoupling10 or RV dilatation leads to lower ECG amplitudes. RV dilatation could lead to loss of myocardial tissue due to the replacement by ﬁbrofatty tissue or could lead to a rotation of the heart aﬀecting both QRS and T-wave amplitudes. However, in a previous study we showed that a large QRSarea is commonly accompanied by a large Tarea,

106 eeto ftebgnigadedo h -ae hrfr,assmn fteT the of assessment Therefore, T-wave. the of end and beginning the of detection addtsfrCTdvc mlnaini ay o-naieadcnb mlmne without implemented investments. be can significant and non-invasive easy, is implantation device CRT for candidates yteEGeupet tcnb aiycluae rmeey1-edEGta ssaved is that ECG 12-lead every from calculated easily pdf-format. be in can or it digitally equipment, ECG the by T the prahstesrc rtradvlpdb tas tal. et Strauss by developed criteria strict the approaches nti td h AI-R rtrawt aenthn rsurn nlasI aVL, I, leads in slurring or CRT. notching after wave outcome R clinical with long-term criteria of MADIT-CRT V prediction the the study T for this the morphology, biomarker In LBBB valuable a besides a that, be time first may the for demonstrates study present The Implications Clinical Potential potential. 6.4.2 action the in phases later determining processes depolarization of ventricular that of of also sequence but understanding the needed, on better information is for only that not CRT, indicate of also action of data mode these the finding, the this Beside CRT. of to use outcome practical better possible known with subgroups various revealing variable, tinuous nopeeyunderstood. incompletely ceitc r nw ohv oiieiflec ntersos oCRT to response the on influence positive a have to known are acteristics u htteei locnieal variability. considerable also is there that but ie yFlmne al. et Feldman by given eetdi h ain ru ihalreT large a with group patient the in LBBB. of resented case in is rarer even this and T-waves, complex, concordant QRS wide by with accompanied patients commonly in rare are complexes QRS narrow while i o hne hs oee,cno ea xlnto h ainswt eaieylarge relatively a with patients why explanation an QRS be the T cannot while however, diameter This, cavity change. increasing with not increased did amplitude wave T:R of ratio the ple optet ihawd R ope.A motn di important An complex. QRS wide is a phenomenon with this patients to complexes applies QRS narrow flattening. T-wave and as known hypertrophy also LV the severe in with role patients a For play dyssynchrony. potential, and action failure myocardial T a and QRS- of between phases variability repolarization and plateau the CRT. area 5 hrfr,teT the Therefore, ial,fmlsadptet ihu itr ficei adoyptywr overrep- were cardiomyopathy ischemic of history a without patients and females Finally, rV or , area 12–14 epn etrt R ic ti elkonta o iae etilsrsodls to less respond ventricles dilated too that known well is it since CRT to better respond C ed ob yteie rmte1-edEG fteVGi o constructed not is VCG the If ECG. 12-lead the from synthesized be to needs VCG a , 6 hssget htohrfcos uha oi hne rprispeetduring present properties channel ionic as such factors, other that suggests This a sda BBciei.Icuigti ocigo lrig h sdcriteria used the slurring, or notching this Including criteria. LBBB as used was area per ob nojcieydtrie imre,epesda con- a as expressed biomarker, determined objectively an be to appears 11 5 15 h hwdta nauemaueet nptet h T the patients in measurements acute in that showed who fe hscneso h nlssol eursasemi-automatic a requires only analysis the conversion this After diinly h ml T small the Additionally, area 16 ti,hwvr o nw hte -aeflteigalso flattening T-wave whether known not however, is, It hs o hne rprismycag u oheart to due change may properties channel ion These . -aepeit ogtr lnclCTresponse CRT clinical long-term predicts T-wave area 4 osbeepaainfrti aiblt was variability this for explanation possible A n BBmrhlg.Bt ain char- patient Both morphology. LBBB and areas 20 oecoey nodrt determine to order In closely. more ih erltdt hypertrophy. to related be might ff rnei hsrseti that is respect this in erence 17–19 o reasons for , area area 107 and area area in

6 Chapter 6

6.4.3 Limitations

The present study was a retrospective single center study in which 41% of the original cohort of consecutive patients had to be excluded because of missing or too low quality measurements. To conﬁrm our results, a prospective multicenter study is required. Such multicenter study should also include systematic coverage of the cause of death, which was not included in the present study. In the multivariate analysis the type of device, CRT-D vs. CRT-P, was not included because the data was not available in the database. However, it is expected that the vast majority were CRT-D devices and thus would not inﬂuence the result. Furthermore, factors related to LV lead position, such as Q-LV, were also not included because they were unknown and because focus of the current study was on pre-procedure predictors of response.

6.5 Conclusions

A vectorcardiographically derived Tarea assessed before the start of CRT was able to strengthen the prediction of long-term clinical outcome in a CRT patient cohort, especially in patients with LBBB morphology on the ECG. The combined analysis of QRS morphology and T-wave provides an easy and widely applicable approach to improve selection of CRT candidates.

6.6 Funding

This work was supported within the framework of Center for Translational Molecular Medi- cine (www.ctmm.nl), Project COHFAR (Congestive Heart Failure and Arrhythmia) (Grant number 01C-203), and supported by the Dutch Heart Foundation.

108 References .LWce alr id,GLnal RRsn n eged.Temporal Bergfeldt. L and Rosen, MR Lundahl, G Linde, C Gadler, F Wecke, L 8. .WZn,G od,adDJag outoe-oreagrtmt eetostand onset detect to algorithm open-source robust A Jiang. D and Moody, GB Zong, W 6. .WZrb,HKen yakeiz JHl,SMNt,MBon Cannom, D Brown, M McNitt, S Hall, WJ Cygankiewicz, I Klein, H Zareba, W 7. .JFCead CDuet rmn,NFemnl,DGa,LKpebre,L Kappenberger, L Gras, D Freemantle, N Erdmann, E Daubert, JC Cleland, JGF 1. .Erpa er htmAscain uoenSceyo adooy er Rhythm Heart Cardiology, of Society European Association, Rhythm Heart European 2. .J os a epn CSti,adJ a eml eosrcino h Frank the of Reconstruction Bemmel. van JH and Sittig, AC Herpen, van G Kors, JA T- Prinzen. 5. FW and Singh, JP Vernooy, K Deursen, Van CJM Végh, EM Engels, EB 4. Breithardt, OA Boriani, G Bordachar, P Baron-Esquivias, G Auricchio, A Brignole, M 3. hrceitc fcricmmr nhmn:vcocrigahcqatfiaini a in quantification vectorcardiographic humans: in memory cardiac of characteristics 123:1061–1072. ra-ada eycrnzto hrp (MADIT-CRT). Therapy Resynchronization Implantation Defibrillator Automatic Multicenter Trial-Cardiac the in Morphology QRS by Therapy uaino R opee.In: complexes. QRS of duration PDuet la,M od ta.E al. et Gold, MR Eldar, M Daubert, JP eto ada eycrnzto hrp nhatfiue mln n olwu re- follow-up and implant failure: management. heart and commendations in therapy resynchronization cardiac on ment aaz,adCricRsnhoiainHatFiue(AEH)SuyInvestigat- e Study The (CARE-HF) ors. Failure Resynchronization-Heart Cardiac and Tavazzi, nlJMed J Engl N r soito,J abr,LSxn ta.21 EHRA 2012 al. et Saxon, L Fail- Daubert, Heart JC Echocardiography, Echocardiography, Association, of of ure Association Society European American Association, America, Heart of American Society Failure Heart Society, etradormfo tnadeetoadorpi ed:dansi oprsnof comparison di diagnostic leads: electrocardiographic standard from vectorcardiogram left with patients in therapy block. resynchronization branch cardiac bundle to response predicts area wave (EHRA). Association Rhythm 34:2281–2329. Developed Heart (2013) European (ESC). the Cardiology and with of pacing collaboration Society cardiac in European cardiac on the on Force Guidelines of Task ESC therapy the 2013 resynchronization therapy: al. resynchronization et Elliott, cardiac PM and Delgado, pacing V Deharo, JC Cleland, J ff rn methods. erent ff c fcricrsnhoiaino obdt n otlt nhatfailure. heart in mortality and morbidity on resynchronization cardiac of ect 20)352:1539–1549. (2005) u er J Heart Eur adoacElectrophysiol Cardiovasc J 19)11:1083–1092. (1990) optr nCrilg,2003 Cardiology, in Computers er Rhythm Heart ff cieeso ada Resynchronization Cardiac of ectiveness 21)9:1524–1576. (2012) 21)26:176–183. (2015) / R xetcnessstate- consensus expert HRS 03 p 737–740. pp. 2003, . Circulation u er J Heart Eur (2011)

6 Chapter 6

model of cardiac pacing. Heart Rhythm (2005) 2:28–34. 9. DL Cortez and TT Schlegel. When deriving the spatial QRS-T angle from the 12-lead electrocardiogram, which transform is more Frank: regression or inverse Dower? J Electrocardiol (2010) 43:302–309. 10. M Potse, D Krause, L Bacharova, R Krause, FW Prinzen, and A Auricchio. Similarities and differences between electrocardiogram signs of left bundle-branch block and left- ventricular uncoupling. Europace (2012) 14 Suppl 5:v33–v39. 11. T Feldman, RW Childers, KM Borow, RM Lang, and A Neumann. Change in ventricular cavity size: differential effects on QRS and T wave amplitude. Circulation (1985) 72:495–501. 12. M Gasparini and P Galimberti. Device therapy in heart failure: has CRT changed "the sickest benefit the most" to "the healthiest benefit the most?". J Am Coll Cardiol (2013) 61:945–947. 13. J Rickard, DM Brennan, DO Martin, E Hsich, WHW Tang, BD Lindsay, RC Starling, BL Wilkoff, and RA Grimm. The impact of left ventricular size on response to cardiac resynchronization therapy. Am Heart J (2011) 162:646–653. 14. E Díaz-Infante, L Mont, J Leal, I García-Bolao, I Fernández-Lozano, A Hernández- Madrid, N Pérez-Castellano, M Sitges, R Pavón-Jiménez, J Barba, MA Cavero, JL Moya, L Pérez-Isla, J Brugada, and SCARS Investigators. Predictors of lack of response to resynchronization therapy. Am J Cardiol (2005) 95:1436–1440. 15. T Aiba, GG Hesketh, AS Barth, T Liu, S Daya, K Chakir, VL Dimaano, TP Ab- raham, B O’Rourke, FG Akar, DA Kass, and GF Tomaselli. Electrophysiological consequences of dyssynchronous heart failure and its restoration by resynchronization therapy. Circulation (2009) 119:1220–1230. 16. D Levy, M Salomon, RB D’Agostino, AJ Belanger, and WB Kannel. Prognostic implications of baseline electrocardiographic features and their serial changes in subjects with left ventricular hypertrophy. Circulation (1994) 90:1786–1793. 17. A Arshad, AJ Moss, E Foster, L Padeletti, A Barsheshet, I Goldenberg, H Greenberg, WJ Hall, S McNitt, W Zareba, S Solomon, JS Steinberg, and MADIT-CRT Executive Committee. Cardiac resynchronization therapy is more effective in women than in men: the MADIT-CRT (Multicenter Automatic Defibrillator Implantation Trial with Cardiac Resynchronization Therapy) trial. J Am Coll Cardiol (2011) 57:813–820. 18. S Zabarovskaja, F Gadler, F Braunschweig, M Ståhlberg, J Hörnsten, C Linde, and LH Lund. Women have better long-term prognosis than men after cardiac resynchronization therapy. Europace (2012) 14:1148–1155. 19. MGSJ Sutton, T Plappert, KE Hilpisch, WT Abraham, DL Hayes, and E Chinchoy. Sustained reverse left ventricular structural remodeling with cardiac resynchronization at one year is a function of etiology: quantitative Doppler echocardiographic evidence

110 0 GSrus HSletr n SWge.Dﬁiglf udebac lc nthe in block branch bundle left Deﬁning Wagner. GS and Selvester, RH Strauss, DG 20. r fcricrsnhoiaintherapy. resynchronization cardiac of era (MIRACLE). Evaluation Clinical tion Randomized InSync Multicenter the from 20)113:266–272. (2006) -aepeit ogtr lnclCTresponse CRT clinical long-term predicts T-wave mJCardiol J Am 21)107:927–934. (2011) Circula- 111

Chapter 7 Prediction of optimal cardiac resynchronization by device-based vectorcardiography in dyssynchronous canine hearts

Elien B. Engels,1 Marc Strik,1 Lars B. van Middendorp,1 Arne van Hunnik,1 Marion Kuiper,1 Kevin Vernooy,2 and Frits W. Prinzen1 (2016). Prediction of optimal cardiac resynchronization by vectorcardiography in dyssynchronous canine hearts. [In preparation]

Abstract

Introduction: The response to CRT is heterogeneous between patients. Proper optimization of atrioventricular (AV) and interventricular (VV) intervals may improve CRT response. In the present study, we explored the possibility of extracting a VCG from the electrograms (EGMs) obtained from unused pacing electrodes positioned on the epicardium. Furthermore, the possibility of using this EGM-based VCG (EGM-VCG) to optimize acute hemodynamic CRT response is investigated.

Methods and Results: CRT was performed in 8 canines with chronic left bundle branch block (LBBB). 100 randomized AV-VV settings were tested. Settings providing an increase in

LV dP/dtmax ≥ 90% of the highest achieved value were deﬁned as optimal. The prediction capabilities of the area of the QRS complex (QRSarea), derived from body surface VCG (ECG-QRSarea) and EGM-VCG (EGM-QRSarea) were compared. The ECG-QRSarea was moderately correlated with change in LV dP/dtmax (R = 0.59 ± 0.23), while the EGM-

QRSarea correlated well with LV dP/dtmax (R = 0.73 ± 0.19). This resulted in a signiﬁcantly better performance of the EGM-QRSarea to predict the optimal CRT settings (AUC = 0.90

[0.87 – 0.93] vs. 0.81 [0.70 – 0.91] for ECG-QRSarea,P = 0.01)

Conclusions: In canine hearts with chronic LBBB, the EGM-QRSarea predicts hemodynamic response and identiﬁes optimal settings accurately. These data support the potency of electrogram based vectorcardiography as a non-invasive and easy tool to individually and continuously optimize the AV- and VV-intervals in CRT.

114 nams otnosoptimization continuous almost an loihsipeetdi R eie.Teeagrtm ihruepa noada accel- endocardial peak use signals either eration algorithms These devices. CRT in implemented limitations. algorithms own their having them of all variables, eiesettings. device ae nteidvda u atal na vrg eainfudi ag ain popula- patient large a in found ADAPTIVECRT only relation algorithms, average these an Of on partially tion. but individual the on based mortality. I Introduction 7.1 insa non-responders. as tients ysnhoyi aiehat ihLB.TeQRS The LBBB. with hearts Deursen canine van LV. in and dyssynchrony (RV) ventricle right the al. from et originating fronts wave electrical of sion pacing. during found values on not thus h uoenDrciefrtePoeto fVrert nml sdfrEprmna and Experimental for Used Animals Vertebrate of Protection the and Experimentation for Animal Directive on European Law Dutch the the to according performed was handling Animal Methods 7.2 EGM-VCG the using chronic device with CRT hearts canine the LBBB. in optimize performed goal can were ultimate experiments than The The system individually. response. and a CRT continuously develop hemodynamic to the is to study VCGs this these of behavior relate its to compare and to VCG EGM-VCG), the VCG, to (EGM-based electrodes pacing unused sleep, the from (e.g. activity patient myocardial and process) (e.g. disease time exercise). altering with and or activity, change normal therapy may the settings rest to during optimal occur due the measurements remodeling while such position, However, supine hospital. in the to and come they when patients of patients. in VV-optimization and AV- for Vpcn n BBcrepne ootmlhmdnmcperformance. hemodynamic optimal to corresponded LBBB and pacing LV td,i a hw htbt h QRS the both that shown was it study, urnl,ms piiaintcnqe ihruevnrclrfiln rssoi function systolic or filling ventricular use either techniques optimization most Currently, ecniee httemi ehns fCTi lcrclrsnhoiain ycolli- by resynchronization, electrical is CRT of mechanism main the that considered We ntepeetsuy eepoe h osblt fetatn C rmEM measured EGMs from VCG a extracting of possibility the explored we study, present the In ada eycrnzto hrp CT a ensont mrv obdt and morbidity improve to shown (LBBB), been block has branch bundle (CRT) left therapy to due resynchronization mostly activation, cardiac LV delayed a and (EF) tion n 11 ainswt er alr obndwt eue etvnrclr(V jcinfrac- ejection (LV) ventricular left reduced a with combined failure heart with patients led hwdta etradorpy(C)rflcseetia interventricular electrical reflects (VCG) vectorcardiography that showed already 1 oee,tersos oCTi eyhtrgnos evn 05%o h pa- the of 30-50% leaving heterogeneous, very is CRT to response the However, 7 3 rla lcrgas(EGMs) electrograms lead or 2 etradorpyfrCTotmzto nLB aiemodel canine LBBB in optimization CRT for Vectorcardiography nipratfco o o-epnemyb uotmlCRT suboptimal be may non-response for factor important An 10 u hsotmzto sbsdo aeievle and values baseline on based is optimization this but , ampl 12 n QRS and hs ehd a eue o eua check-up regular for used be can methods These 8–10 oee,nn fteemtosaesolely are methods these of none However, . TM 4–6 area taeako etoi,Ic)provides Inc.) Medtronic, of (trademark lentv prahsaepoie by provided are approaches Alternative ampl r eibeadrpouil variables reproducible and reliable are afwybtenta enduring seen that between half-way 11 nasubsequent a In 115

7 Chapter 7

Other Scientiﬁc Purposes. The protocol was approved by the Animal Experimental Commit- tee of Maastricht University.

7.2.1 Experimental model

The experiments were performed on 8 mongrel canines of either sex and unknown age. The extensive protocol has already been described earlier.13 In short, in all canines radiofrequency ablation was used to create LBBB. CRT studies were performed 16 weeks after creation of LBBB, to allow for ventricular remodeling to occur. During these CRT studies, RV and LV pressure catheters were positioned and the right atrial pacing lead was inserted transvenously. In addition, an octapolar electrode catheter was placed against the right side of the septum, from which the most distal electrode was used for RV apical pacing.13 After opening the chest, two multi-electrode arrays holding 102 contact electrodes were placed around the heart to measure epicardial electrical potentials and a basal posterolateral electrode was selected for LV-pacing (see bottom-left panel in Fig. 7.1).

Figure 7.1: Flow-chart of the ECG-VCG derived QRSarea (top) and EGM-VCG derived QRSarea determination (bottom).

116 hs aeiemaueet eeue ocluaerltv hne nL d LV rate. in sinoatrial changes intrinsic above relative BPM calculate 10 to rate used heart were a measurements at measure- baseline pacing baseline These atrial a to by equal intertwined was were which settings ment pace of steps Four programmed. randomly rxmtn )aantaF(prxmtn )rsle na2dmninlVCG 2-dimensional MA) (ap- Natick, a (MathWorks, I R2010b in MATLAB lead custom resulted QRS plotting Using derived and ECG-VCG Y) the panel). electrodes software, (approximating right lead top aVF limb 7.1, (Fig. against the from X) recorded proximating were ECGs Surface analysis Data 7.2.3 minimum a for cycles. recorded respiratory were 2 data of electrophysiological and hemodynamic Both surements. QRS AL eas agdfo 0m o20m nseso 0m,rsligi atro 10x10 of raster a in resulting ms, 20 of steps di in 100 ms LV thus 230 or to and delays) (A-RV ms RV 50 the from either ranged and atrium delays) (A-LV between delay the protocol, pacing the During protocol Pacing 7.2.2 ueetvraiiy h etnsrsligi ecnaeices fL d LV of increase mea- percentage for a in account resulting to settings applied The was variability. fitting surement Quadratic plots. surface three-dimensional in interval ie(Vd (LV rise ECG-QRS the account. for into described 7.1, as (Fig. formula QRS VCG same 2-dimensional derived the a EGM-VCG using in The results other panel). each 7.1). right by Fig. against panel direction bottom signal (bottom ‘Y’ ‘Y’ EGM LV the and apical sub- and ‘X’ the EGM, by the from LV EGM Plotting calculated RV basal basal is the the from EGM-VCG of EGM signal the RV the of subtracting apical direction the of ‘X’ electrode signal pacing The LV the the tracting bands. below just electrode one contact and the elec- above octapolar just on the electrode one from and electrodes (RV) two catheter from trode measured are EGMs required The EGM-VCG. h uv n baseline. and curve the 0 ftemxmm(ufc ek eeietfida pia (CRT optimal as identified were peak) (surface maximum the of 90% otnosdt r rsne smean as presented are data Continuous analysis Statistical 7.2.4 lt fpretg hnei Vd LV in change percentage of plots h lcrpyilgclvralsECG-QRS variables electrophysiological the sn h Gs etradormwsetatd hc eefe ilb eerdt as to referred be will hereafter which extracted, was vectorcardiogram a EGMs, the Using h eoyai epnewsasse ymauigtemxmlrt fL pressure LV of rate maximal the measuring by assessed was response hemodynamic The area = P / q d t QRS max ff rn aestig.Teocrec fal10di 100 all of occurrence The settings. pace erent .Frec etn,vralswr lte nrlto oteLV the to relation in plotted were variables setting, each For ). area 2 14 , I + QRS etradorpyfrCTotmzto nLB aiemodel canine LBBB in optimization CRT for Vectorcardiography area 2 , aVF P / hr QRS where area d ± t max tnaddvain orltosbtensurface between Correlations deviation. standard (ECG-QRS oprdt aeieadtesraeposof plots surface the and baseline to compared area n EGM-QRS and area area , I area area n QRS and (EGM-QRS a optduigteformula the using computed was ) u o aigtelasXadY and X leads the taking now but area area ff , aVF area rn obntoswas combinations erent opt eecluae using calculated were ol ecalculated be could ) ). r h rabetween area the are P / d t P max / d t fa least at of max / VAV- RV mea- 117

7 Chapter 7 the Pearson correlation coefficient. The classification performance of electrical variables in identifying the CRTopt settings was evaluated by receiver operating characteristic (ROC) curve analysis. The significance of the difference in classification performance between variables was evaluated by comparing the area under the ROC curves with the method proposed by DeLong et al. 15 . A P-value < 0.05 was considered statistically significant.

7.3 Results

7.3.1 Hemodynamic eﬀect of diﬀerent stimulation intervals during CRT

Two examples of the hemodynamic changes at all 100 tested A-RV/A-LV combinations are shown in Fig. 7.2. LV- and RV-only pacing at short AV-intervals (left and right corner in

Fig. 7.2, respectively) caused only small changes in LV dP/dtmax and this was also the case during biventricular (BiV) pacing at long AV-intervals (top corner) Several combinations of A-RV and A-LV intervals led to good hemodynamic improvement. At short AV-intervals, high LV dP/dtmax values were found during simultaneous BiV pacing, while at longer AV- intervals the optimum was found during LV pre-excitation, most likely related to partial fusion of LV pacing with the wave front originating from the right bundle branch (RBB).13 The location of the leftward turn of the ridge was dependent on intrinsic conduction properties, e.g. a longer PR-interval results in fusion with the intrinsic conduction at longer AV-intervals (Fig. 7.2A vs. 7.2B). Indeed, when for all dogs the eﬀective VV-interval (the actual time delay between onset of activation of the RV apex and LV lateral wall16) was calculated, all points transferred to one parabola with the optimum around an eﬀective VV-interval of 0 (Fig. 7.3).

In the 8 canines, the number of A-LV/A-RV combinations which resulted in CRTopt varied between 1 and 6 (median of 4 combinations).

7.3.2 ECG-VCG and EGM-VCG prediction performance

Fig. 7.4A and Fig. 7.4B show the surface plots for the QRSarea as derived from the ECG-VCG and the EGM-VCG and the corresponding LV dP/dtmax plots during all 100 tested A-RV/A- LV combinations of two canines with diﬀerent intrinsic PR-intervals. To improve the visual comparison between the QRSarea and LV dP/dtmax plot, the z-axis of the QRSarea plot was inverted. During most of the 100 A-RV and A-LV settings there was a good match between values of ECG-QRSarea and EGM-QRSarea, with low values occurring during settings ranging from BiV-pacing at short AV-intervals to pronounced LV pre-excitation at longer AV intervals and higher values during LV-only pacing and LBBB. The number of CRTopt settings during BiV- pacing before the leftward shift was higher for the canine with a longer PR-interval, in which the leftward shift occurred later (Fig. 7.4B). In both experiments the ECG-QRSarea during

118 n BBwsntfudfrteEGM-QRS the for found not was LBBB and iue73 A 7.3: Figure 7.2: Figure Vpcn a oe hndrn BB hsdi This LBBB. during than lower was pacing RV n -Vitrasi aiewt Ritra f11m ( ms 141 of PR-interval a with canine a in intervals A-LV and aiu aeo Vpesr ie(Vd (LV rise pressure LV of rate maximum figurations. n -Vvle ls oec te sBVpcn ihacranVV-interval. certain a with pacing BiV is A-LV to other and equal each pacing from is to only A-RV left RV close long represents points values A-LV a A-RV long at and and a A-LV and and short A-RV pre-excited short a was a of RV while combination the pacing, LV-only A line pre-excited. this was LV from line right this points the at while 0, xmlso h eaiecag nL d LV in change relative the of Examples f21m ( ms 201 of obsln o l animals. all for baseline to yia xml fterltosi ewe e between relationship the of example Typical : B B eainhpo e of Relationship : .A h ignlln ntehrzna ln (y plane horizontal the in line diagonal the At ). etradorpyfrCTotmzto nLB aiemodel canine LBBB in optimization CRT for Vectorcardiography ff cieV-nevladcag nL d LV in change and VV-interval ective area P P sacneune h orlto between correlation the consequence, a As . ff / / d rnei QRS in erence d t t max max oprdt aeiefralpcn con- pacing all for baseline to compared ) oprdt aeiefrtetse A-RV tested the for baseline to compared ff cieV-nevladipoeetin improvement and VV-interval ective area A n n ihaPR-interval a with one and ) ewe Vol pacing only RV between = )teV-nevlwas VV-interval the x) P / d t max compared 119

7 Chapter 7

the ECG and EGM derived QRSarea was moderate (R = 0.44 ± 0.17 for all experiments combined).

The small QRSarea found during RV pre-excitation for the ECG-VCG did not match with an increase in LV dP/dtmax, although small QRSareas during other pace settings did correlate with a LV dP/dtmax increase. As a consequence, the correlation between these two variables was moderate (R = 0.59 ± 0.23 for all experiments), despite the fact that the location of minimal QRSarea matched quite well with that of the optimal hemodynamic beneﬁt. For the EGM-QRSarea this low value during RV-pacing was absent and there was also a match between minimal QRSarea and optimal hemodynamic beneﬁt leading to a higher correlation with the change in LV dP/dtmax compared to baseline (R = 0.73 ± 0.19 for all experiments).

7.3.3 ROC curves

Fig. 7.5 presents ROC curves that show the classifying abilities of ECG-QRSarea and

EGM-QRSarea to identify CRTopt. The area under the curve for the EGM-QRSarea was high (0.90 [0.87 – 0.93]), while the ECG-QRSarea and QRS duration were signiﬁcantly less accurate in identifying the settings resulting in best hemodynamic improvement (AUC = 0.81 [0.70 – 0.91] and 0.74 [0.64 – 0.84], respectively; P ≤ 0.01 comparing both

ECG-QRSarea and QRS duration to EGM-QRSarea). Indeed, the variation in QRS duration between diﬀerent settings can be quite small, making it diﬃcult to select the best settings using QRS duration (Fig. 7.6).

120 iue7.4: Figure h aiewt Ritra f11m ( ms 141 of PR-interval a with canine QRS the the for plots surface of Examples w ufc lt.Tepoiiyo h iia QRS minimal the of proximity The plots. surface two oprsnbtenteQRS the between comparison etridctrfrrsos rdcinadcreae etrwt epneobservation response with d LV better in correlated change % and as prediction (deﬁned response for indicator better a QRS of methods two the between agreement The eie QRS derived area . etradorpyfrCTotmzto nLB aiemodel canine LBBB in optimization CRT for Vectorcardiography P area / d t max n Vd LV and oprdt aeie oprdwt h ECG-VCG the with compared baseline) to compared area eie rmteEGVGadEMVGfor EGM-VCG and ECG-VCG the from derived A P raP-nevlo 0 s( ms 201 of PR-interval a or ) / d t max area ufc lt,tezai a inverted. was z-axis the plots, surface area aclto sidctdbtenthe between indicated is calculation eie rmteEMVGwas EGM-VCG the from derived B .Frvisual For ). 121

7 Chapter 7

Figure 7.5: Receiver operating characteristics (ROC) curves for QRS duration, ECG-VCG QRSarea,

and EGM-VCG QRSarea for classifying CRTopt (settings ≥ 90% of maximal LV dP/dtmax).

ROC area under the curve signiﬁes performance of each variable in identifying the CRTopt settings.

Figure 7.6: Relation between QRS duration and LV dP/dtmax for the canine with a PR-interval of 141 ms. For visual comparison, the z-axis of the QRS duration was inverted.

122 o otn piiaino V n Vitrasi R.TeEGM-QRS The CRT. tool in VV-intervals reproducible AV- and and of accurate, optimization in- non-invasive, routine the easy, for from an obtained as electrograms electrodes the pacing from unused derived tracardiac vectorcardiogram the of use the support h rsn td efre ncnn erswt hoi BBdmntae htteei a is QRS derived there EGM-VCG that and demonstrates ECG-VCG LBBB between chronic relation with moderate hearts canine in performed study present The Discussion 7.4 itn h pia R eiestig eutn ntebs ct eoyai response, hemodynamic acute best QRS the derived in EGM-VCG resulting the pre- settings with of device capable CRT are methods optimal both However, the pacing. dicting only RV during methods two the between hw htas h QRS canines the in also pacing that LV-only shown during that demonstrated already QRS group the our study, previous a In vectorcardiographic using CRT of optimization Individual 7.4.1 individually. and continuously settings device CRT optimize to used be R eiestig.TelwrcreainbtenECG-QRS between correlation lower ECG-QRS The the settings. canines device in CRT pacing BiV during also eepandb h s fol iblas hnol okn ntefotlpae sis as QRS derived plane, EGM-VCG frontal supported the seems the idea dogs in This in looking lost. that be only finding might When the information some by leads. study, canine limb current only the in of done use the by explained be inmgtb htQSdrto nyrflcstettlvnrclratvto ie hl the while time, activation ventricular total the reflects only QRS duration explana- An QRS ROC-curves. that the be under might area tion higher the by evidenced as response, hemodynamic iainmto,bt ECG-QRS both method, mization plots. response hemodynamic ueet n yapyn udai tig eody n ol usinwehrchanges whether question could d one LV Secondly, mea- in observed baseline fitting. repeated quadratic performing applying by by minimum and a surements to kept was variability setting, perimental a ligas ly oe na ale td efre ySrke al. et Strik by performed study earlier an In role. a plays also filling lar aiblt nhmdnmc n esrmn rosa errors measurement and hemodynamics in variability rteEGM-QRS the or n R tlne Vitras hssget htvnrclrfiln osntpa role a play not e does hemodynamic filling acute AV-intervals, ventricular optimal short that the at suggests achieving CRT This in baseline, AV-intervals. longer between at change CRT not and did volume end-diastolic LV the eea atr a xli h o-efc orlto ewe ihrteECG-QRS the either between correlation not-perfect the explain may factors Several vntog oepyiin s h etn ihtesals R uaina nopti- an as duration QRS smallest the with setting the use physicians some though Even area measurements etrrflcsteatvto waves. activation the reflects better ampl sapeitrfrotmzto fsiuainintervals stimulation of optimization for predictor a is P area / d t ufc lt ihteL d LV the with plots surface max area r nieyrltdt lcrcldsycrn,o hte ventricu- whether or dyssynchrony, electrical to related entirely are etradorpyfrCTotmzto nLB aiemodel canine LBBB in optimization CRT for Vectorcardiography ol eue o hspurpose this for used be could area area en h infiatybte n.Teeobservations These one. better significantly the being n EGM-QRS and 11,17 ff P c n,teeoe lcrcldyssynchrony electrical therefore, and, ect / area d t max area ff a eue o h piiainof optimization the for used be can c hscreain ntepeetex- present the In correlation. this ect area ufc lt biul,biological Obviously, plot. surface eebte rdcoso maximal of predictors better were 12 orltdbte ihteacute the with better correlated h urn td hw that shows study current The . area 11 area hl nptet twas it patients in while n Vd LV and ihtelreto largest the with 13 area twssonthat shown was it ol potentially could P / d t max may ff 123 area set

7 Chapter 7

seems to be the leading factor for changes in LV dP/dtmax. Finally, as discussed above, the use of only the limb leads for the VCG determination on the body surface may have led to the lower predictive values of the ECG-QRSarea.

7.4.2 Potential clinical implications

Extrapolation of experimental data to the clinical situation should always be done with care.

However, it has already been shown that the VCG-derived QRSampl and QRSarea during LV- only pacing at a range of AV-delays and during BiV pacing at diﬀerent VV-delays are good 12 predictors of CRT response , a ﬁnding similar to that in our study regarding the QRSarea. Therefore, the results of this animal study might well be applicable in the clinical situation.

The EGM-QRSarea provides a non-invasive and easy method to optimize the AV- and VV-intervals continuously and individually. In patients, one could use the electrodes on a quadripolar LV lead that are not being used for pacing as the two LV electrodes and the signal of the two RV electrodes might be extracted from the RV ring and the coil of the ICDs. This approach may be superior to currently available algorithms for automated optimization of CRT device settings using electrograms. All these algorithms use an estimated relation between measured variables and the optimal AV- and VV-interval, estimated using electro- 8–10 grams obtained in the unpaced (CRT-oﬀ) situation. Instead, the concept of EGM-QRSarea as presented here, continuously determines the eﬀect of ventricular pacing on ventricular resynchronization.

7.4.3 Limitations

The current study only took the acute response to CRT into account using hemodynamic measurements. The acute CRT response may not correspond to CRT response on the long- term.18 Moreover, while the canines had long-term LBBB, the response in patients with a higher degree of heart failure and other comorbidities might be diﬀerent.

7.5 Conclusions

In canine hearts with chronic LBBB, the QRSarea derived from a VCG extracted from myocardial EGM’s from two RV and two LV electrodes predicts hemodynamic response and identifies optimal settings accurately. The worse performance of the QRS duration shows that variables reflecting patterns of activation perform better than those reflecting absolute electrical dyssynchrony. These data support the potency of device based vectorcardiography as a non- invasive and easy tool to individually and continuously optimize the AV- and VV-intervals in CRT.

124 ubr0C23,adspotdb h uc er Foundation. Heart Dutch [Grant the Arrhythmia) by supported and and Failure 01C-203], Medi- Heart number Molecular (Congestive Translational COHFAR for Project Center (www.ctmm.nl), of framework cine the within supported was work This Funding 7.6 etradorpyfrCTotmzto nLB aiemodel canine LBBB in optimization CRT for Vectorcardiography 125

7 References

1. JGF Cleland, JC Daubert, E Erdmann, N Freemantle, D Gras, L Kappenberger, L Tavazzi, and Cardiac Resynchronization-Heart Failure (CARE-HF) Study Investigat- ors. The effect of cardiac resynchronization on morbidity and mortality in heart failure. N Engl J Med (2005) 352:1539–1549. 2. European Heart Rhythm Association, European Society of Cardiology, Heart Rhythm Society, Heart Failure Society of America, American Society of Echocardiography, American Heart Association, European Association of Echocardiography, Heart Fail- ure Association, JC Daubert, L Saxon, et al. 2012 EHRA/HRS expert consensus state- ment on cardiac resynchronization therapy in heart failure: implant and follow-up re- commendations and management. Heart Rhythm (2012) 9:1524–1576. 3. W Mullens, RA Grimm, T Verga, T Dresing, RC Starling, BL Wilkoff, and WHW Tang. Insights from a cardiac resynchronization optimization clinic as part of a heart failure disease management program. J Am Coll Cardiol (2009) 53:765–773. 4. CE Raphael, A Kyriacou, S Jones, P Pabari, G Cole, R Baruah, AD Hughes, and DP Francis. Multinational evaluation of the interpretability of the iterative method of optimisation of AV delay for CRT. Int J Cardiol (2013) 168:407–413. 5. A Auricchio, C Stellbrink, M Block, S Sack, J Vogt, P Bakker, H Klein, A Kramer, J Ding, R Salo, B Tockman, T Pochet, and J Spinelli. Effect of pacing chamber and atrioventricular delay on acute systolic function of paced patients with congestive heart failure. The Pacing Therapies for Congestive Heart Failure Study Group. The Guidant Congestive Heart Failure Research Group. Circulation (1999) 99:2993–3001. 6. N Derval, P Steendijk, LJ Gula, A Deplagne, J Laborderie, F Sacher, S Knecht, M Wright, I Nault, S Ploux, et al. Optimizing hemodynamics in heart failure patients by systematic screening of left ventricular pacing sites: the lateral left ventricular wall and the coronary sinus are rarely the best sites. J Am Coll Cardiol (2010) 55:566–575. 7. P Ritter, PPHM Delnoy, L Padeletti, M Lunati, H Naegele, A Borri-Brunetto, and J Silvestre. A randomized pilot study of optimization of cardiac resynchronization therapy in sinus rhythm patients using a peak endocardial acceleration sensor vs. standard methods. Europace (2012) 14:1324–1333. 8. MR Gold, I Niazi, M Giudici, RB Leman, JL Sturdivant, MH Kim, Y Yu, J Ding, and AD Waggoner. A prospective comparison of AV delay programming methods for 6 eno,X ebe,R onlse,BDjmn JCin,TAt,adFW and Arts, T Crijns, HJ Dijkman, B Cornelussen, RN Verbeek, XA Vernooy, K two under 16. areas the Comparing Clarke-Pearson. DL and DeLong, DM DeLong, ER 12-lead 15. the from angle QRS-T spatial the deriving When Schlegel. TT and Cortez DL 14. 8 DBgad otuzn ABak,P ovnas WPizn en,and Meine, M Prinzen, FW Doevendans, PA Bracke, FA Houthuizen, heart P Bogaard, with MD patients of imaging 18. Electrocardiographic Rudy. Y and Jia, P Varma, N 17. 0 rm ek,DBri,KFLe oua CSaln,MGsaii J Gasparini, M Starling, RC Aonuma, K Lee, KLF Birnie, D Lemke, B Krum, H 10. 1 J a ere,MSrk MRdmkr,AvnHni,MKie,LWecke, L Kuiper, M Hunnik, van A Rademakers, LM Strik, M Deursen, van CJM 11. 3 ti,L a idnop otuzn lu,AvnHni,MKie,A Kuiper, M Hunnik, van A Ploux, S Houthuizen, P Middendorp, van LB Strik, M 13. F Janssen, MHG Ståhlberg, M Everdingen, van WM Wecke, L Deursen, van CJM 12. .J ae n,JMKni r,SBa,G re,JPrefil,MFdr Green- S Fedor, M Porterfield, J Greer, GS Beau, S 3rd, McKenzie J 2nd, Baker JH 9. eoyai piiaindrn ada eycrnzto therapy. resynchronization Electrophysiol cardiac during optimization hemodynamic eg GDod obseo RBie,adLPrefil.Aueeauto of evaluation Acute Porterfield. L and Bailey, JR AV Corbisiero, programmer-guided R Daoud, EG berg, n coadormfrcricrsnhoiainteayi er alr ainsand patients failure implants. heart ICD in dual-chamber therapy resynchronization cardiac for echocardiogram and niiulzdcricrsnhoiainteay ainl n eino h adaptive the trial. for of therapy algorithm resynchronization design novel cardiac and A rationale Martin. therapy: D resynchronization and cardiac Kalmes, individualized A Sambelashvili, A Rogers, T Gorcsan, piiaino ada eycrnzto hrp ncnn etbnl rnhblock easy branch for bundle tool left a canine as in hearts. Vectorcardiography therapy resynchronization Prinzen. cardiac FW of optimization and Vernooy, K Crijns, HJGM Vitra ncricrsnhoiaintherapy. resynchronization cardiac in interval VV rne.Cluaino e of Calculation Prinzen. approach. nonparametric a curves: Biometrics characteristic operating receiver correlated more or Dower? inverse or regression Frank: Electrocardiol more is transform which electrocardiogram, hearts. the canine of dyssynchronous Electrophysiol determinant in Arrhythm as therapy wavefronts resynchronization electrical cardiac of to Interplay response Prinzen. FW and Auricchio, therapy. Res resynchronization Transl cardiac Cardiovasc in J Vectorcardi- intervals stimulation Prinzen. of FW optimization and Vernooy, for ography K Crijns, HJGM Bergfeldt, L Braunschweig, ndP in MvnGle.Bsln etvnrclrdP ventricular left Baseline Gelder. van BM therapy. resynchronization cardiac Electrocardiol to J response and block branch bundle left with failure / ta rdcsciia ucm nptet ihcricrsnhoiainther- resynchronization cardiac with patients in outcome clinical predicts dtmax icAryh Electrophysiol Arrhythm Circ 18)44:837–845. (1988) 21)43:302–309. (2010) 20)18:490–496. (2007) 20)40:S174–S178. (2007) / VadV ea piiaincmaiga EMmethod IEGM an comparing optimization delay VV and PV 21)8:128–137. (2015) 21)6:924–931. (2013) ff cieV nevlfclttsotmzto fA ea and delay AV of optimization facilitates interval VV ective adoacElectrophysiol Cardiovasc J 21)5:544–552. (2012) mHatJ Heart Am / ta ahrta h ct improvement acute the than rather dtmax er Rhythm Heart 21)163:747–752.e1. (2012) 20)18:185–191. (2007) 20)4:75–82. (2007) Cardiovasc J References Circ 127 J

7 Chapter 7

apy. Eur J Heart Fail (2011) 13:1126–1132.

128

Chapter 8 Tailoring device settings in cardiac resynchronization therapy by vectorcardiography using the implanted pacing electrodes.

Elien B. Engels,1 Masih Maﬁ-Rad,2 Ben Hermans,1 Alfonso A. Hernandez,3 Antonius M.W. van Stipdonk,2 Michiel Rienstra,4 Coert Scheerder,3 Alexander H. Maass,4 Kevin Vernooy,2 and Frits W. Prinzen1 (2016). Tailoring device settings in cardiac resynchronization therapy by vectorcardiography using the implanted pacing electrodes. [In preparation]

1 Department of Physiology, CARIM School for Cardiovascular Diseases, Maastricht University, Maastricht, the Netherlands. 2 Department of Cardiology, Maastricht University Medical Center, Maastricht, the Netherlands. 3 Medtronic Bakken Research Center, Maastricht, the Netherlands. 4 Department of Cardiology, University Medical Center Groningen, Groningen, the Netherlands. Chapter 8

Abstract

Introduction: Left ventricular (LV) pacing fused with intrinsic ventricular activation aug- ments CRT beneﬁt as compared to conventional biventricular pacing. In the present study we explored the possibility to use an electrogram (EGM)-based vectorcardiogram (EGM-VCG) for patient-speciﬁc LV fusion pacing.

Methods and Results: During CRT device implantation of 28 patients, hemodynamic measurements and 12-lead ECG recordings were performed during various AV-delays. In addition, unipolar EGMs were recorded from the implanted ventricular pacing leads. The VCG was reconstructed from the 12-lead ECG and the EGM-VCG from ventricular EGM signals. QRS area (QRSarea) and amplitude (QRSampl) were extracted from both methods. Optimal hemodynamic response was deﬁned as the largest increase in LV systolic pressure

(LVPsyst) or in the maximal rate of LV pressure rise (LV dP/dtmax). Over all patients and pacing modes there was a good agreement between the surface VCG and EGM-VCG derived

QRSarea (R = 0.74) and QRSampl (R = 0.80). VCG and EGM-VCG derived QRSarea predicted the AV-delay resulting in highest LV systolic pressure with reasonable accuracy. However, prediction of the AV-delay resulting in highest LV dP/dtmax was poor, because in one third of patients highest LV dP/dtmax occurred at short AV-delays. Determination of the onset of right ventricular (RV) activation to employ LV fusion pacing could be extracted individually using the QRSampl derived from either the VCG or EGM-VCG.

Conclusions: QRSarea derived from the VCG or EGM-VCG, can be used to predict the AV-delay resulting in highest LV systolic pressure. The onset of contribution of the intrinsic RV activation to obtain the best fusion with LV pacing can be determined using the VCG or EGM-VCG.

132 ouainlvl R skont mrv ohmriiyadmortality. and morbidity both improve to known is CRT level, population C Introduction 8.1 niiullvlteei osdrbevraiiyi h response. the in variability considerable is there level individual omddrn in-o during formed conduction intrinsic of fusion of amount the on waves. activation cardiac impact pacing-generated and has with AV-delay volume stroke addition, to In contribute that output. characteristics filling LV the determines delay) acceleration esrs(d measures satiue osbpia tivnrclr(V timing. (AV) atrioventricular suboptimal to attributed is di n h ciainwv rgntn rmtepcn iela otelrethemodynamic largest the to lead site wave pacing activation the intrinsic from improvement. the originating wave between activation fusion the optimal and because activation (RV) ventricular ADAPTIVECRT rmtevcocriga eet h ereo etiua eycrnzto uigvarious during resynchronization extracted ventricular vector AV-delays. of QRS degree the the that reflects indicating vectorcardiogram group, the ap- our from this from studies of previous basis from The data optimization. is device proach patient-specific for pacing ventricular during leads ocspeeti h er n ih hspoieavlal ecito fteaon fre- of amount (QRS the area of QRS description minimal valuable The a pacing. provide LV thus during might synchronization and heart the in present forces rvdsa lotcniuu uoai piiainadealsbt Vpcn and pacing LV both enables and optimization automatic pacing. BiV continuous almost an provides ino h Vdly uha di as AV-delay, such the of tion naeaedt rmagopo patients. of group a from based based stimulation data are optimal average algorithms estimating and these on activation of intrinsic Most during measured devices. parameters have in on algorithms implemented purpose, been this have To that desirable. developed more been be may fashion, automated an in preferably ogtr e long-term r ieadrsuc osmn n ujc olremaueetvraiiy Therefore, (‘out-of-the-box’). values variability. nominal the measurement at large settings to device CRT subject leave and clinicians many consuming resource and time are V n Votmzto oprdt oia etnshv endescribed been have settings nominal to compared VV-optimization and AV- ff nCT h iedlybtensiuaino h ih timadtevnrce (AV- ventricles the and atrium right the of stimulation between time-delay the CRT, In hl igeA-ea piiaini rbbyvlal,rglroptimization, regular valuable, probably is optimization AV-delay single a While ntepeetsuyw xlrdtepsiiiyo sn aadrvdfo h implanted the from derived data using of possibility the explored we study present the In rne uigitisccnuto o uigpcn.Teeetorm(EGM)-based electrogram The pacing. during nor conduction intrinsic during erences ncmiainwt eue etvnrclr(V jcinfato E) tthe At (EF). fraction ejection (LV) ventricular left reduced a (LBBB), with block combination branch bundle in left to due mainly failure, heart induced dyssynchrony ardiac 19,20 ff 12 17 P csi ag lncltil aentbe observed. been not have trials clinical large in ects 4,18 / ihecpino h attcnqe hs esrmnscnol eper- be only can measurements these technique, last the of exception With . ntecs fL aig h -Vdlyi e oteosto nrni right intrinsic of onset the to set is A-LV-delay the pacing, LV of case the In d etradorpy(C)i -iesoa ersnaino h electrical the of representation 3-dimensional a is (VCG) Vectorcardiography t eycrnzto hrp CT sa salse hrp o ainswith patients for therapy established an is (CRT) therapy resynchronization toework) stroke , TM ffi taeako etoi,Ic)agrtmi h nymto that method only the is algorithm Inc.) Medtronic, of (trademark evst.Wieauehmdnmcipoeet fu to up of improvements hemodynamic acute While visits. ce ff 7–9 rn coadorpi measures echocardiographic erent ne photoplethysmography finger , 4,5 alrn R eiestig ydvc ae VCG based device by settings device CRT Tailoring utpetcnqe aebe sdfroptimiza- for used been have techniques Multiple 15–17 on o hyd o osdrindividual consider not do they so, Doing 3 2 ato orCTresponse CRT poor a of Part area 6 10,11 1,6 oevr otmethods most Moreover, n h R amplitude QRS the and ) naiehemodynamic invasive , n ekendocardial peak and 1 oee,a the at However, 13 favorable , ∼ 14 5 for 15% 133

8 Chapter 8

(QRSampl) closest to a value halfway between LV pacing and LBBB were shown to predict the AV-delay settings resulting in best hemodynamic improvement in patients. In addition, in a previous animal study we showed that the principle of optimization based on the body surface VCG could be extended to a VCG derived from the EGMsø obtained from the intracardiac pacing electrodes (EGM-VCG).21 The objectives of the current study were to investigate 1) whether EGM-VCG derived

QRSarea or QRSampl can be used to determine the AV-delay that provides the best hemodynamic eﬀect and 2) how the patient-speciﬁc onset of intrinsic activation of the RV can be extracted from the EGM-VCG to enable LV fusion pacing.

8.2 Methods

8.2.1 Study population

The current study population consisted of 28 consecutive patients referred for CRT implantation with a class I indication according to the European society of cardiology (ESC) guidelines (New York Heart Association class II, III or ambulatory IV despite adequate medical treatment, in sinus rhythm, LVEF ≤ 35% and QRS duration > 120 ms with LBBB morphology).22 All patients were prospectively enrolled in this dual center study. Patients presenting with ≥ 4 premature ventricular complexes (PVCs) on the 12-lead ECG and with moderate to severe aortic valve stenosis were excluded. In addition, all participants had to be between 18 and 80 years old. This study was performed according to the principles of the Declaration of Helsinki and approved by the ethics committee of Maastricht University Medical Center+ (MUMC+) and University Medical Center Groningen (UMCG). All participants gave written informed consent prior to investigation.

8.2.2 Procedure

Standard digital 12-lead ECGs were recorded throughout the entire procedure. All participants underwent routine CRT-deﬁbrillator implantation; all with a quadripolar LV lead (QuartetTM Model 1458Q, St. Jude Medical, St. Paul, USA). The RV lead was placed in the apex, the right atrial lead in the right auricle, and the LV lead in the most optimal vein. After implantation of all leads, the pressure wire was introduced via the femoral artery into the LV cavity and the unpaced electrodes were connected to an electrophysiological system to enable EGM registration during the pacing protocol (described below). Once the pacing protocol was completed, the leads were connected to the CRT device and the procedure was completed.

134 aeR edfo h nplrEMo h itleetoeo h Vla Fg 8.1). (Fig. lead LV the of the electrode on of distal signal the electrode EGM of proximal unipolar EGM the the unipolar subtracting the of by from EGM calculated lead RV was unipolar same B the lead uni- from bipolar the and subtracting ring lead by RV LV calculated the the was from EGM-VCG signal the EGM of A polar lead Bipolar other. each against 2la C n G eodnswr aea apigfeunyo tlat10 Hz constructed 1000 were least (VCGs) at vectorcardiograms matrix. ECGs Kors of 12-lead the frequency the using sampling From a seconds. at 10 least made at were for recordings EGM and ECG 12-lead Vectorcardiography 8.2.5 the at pacing AAI setting, pace baseline. ventricular as each used after (pseudo- was and ECG rate Before heart intrinsic same ms. the 30 resembled of almost steps paced-ECG in the fusion), where AV-delay an and 30 to AV-delay (between ms) short very 50 a from varied AV-delays were rate). heart intrinsic above V ol aiga di at pacing LV-only protocol Pacing 8.2.4 AV-delays. various the during obtained data aeo Vpesr ie(Vd (LV rise pressure LV of rate o tlat1 eod ihu n rmtr etiua otatos nodrt identify to order In d LV contractions. in ventricular increase largest premature measured the was any with AV-delay pressure without LV the the seconds which 10 least after least at stabilize, at pressure transition, the each for let After to pacing). used (AAI were seconds pacing 10 baseline atrial by to alternated equal St. were was measurements which Certus, pacing measurements (PressureWire, Ventricular guidewire USA). Paul, transluminal St. tipped Medical, sensor Jude pressure performed were inch measurements 0.014 pressure LV a The with seconds. 10-20 over averaged and beat heart nMTA 21b(ahok,Ntc,MA). Natick, (MathWorks, R2010b MATLAB in ewe h etrada xspredclrt h rnvra ln) nodrt express to order In plane). transversal the (angle QRS to elevation the of and perpendicular direction plane) axis the angle transversal an the in and VCG, (angle vector the azimuth the For the between by angle. described and was amplitude vector as the expressed of were space in vector QRS imum oad h oiieBai.Frhroe h rao h R opwscluae rmthe from calculated was directed loop was QRS it the when of EGM-VCG area the the of Furthermore, case B-axis. the positive in the towards or azimuth), (negative back the towards et.Fo h Vpesr inl,tessoi Vpesr (LVP pressure LV systolic the signals, pressure measure- LV pressure LV the invasive From using assessed ments. was CRT to response hemodynamic acute The measurements Pressure 8.2.3 h CsadEMVG eeaaye o analyzed were EGM-VCGs and VCGs The B, and A leads, bipolar two plotting by constructed was EGM-VCG dimensional two A ff rn Vdly a efre uigara vrrv aig(0BPM (10 pacing overdrive atrial during performed AV-delays was erent 24 ampl h QRS the , P / d t max eedtrie.Teevralswr eemndper determined were variables These determined. were ) ampl alrn R eiestig ydvc ae VCG based device by settings device CRT Tailoring a endngtv hntevco a directed was vector the when negative defined was 23 P ffl / d n sn utmzdsfwr programmed software customized using ine 25 t max h antd n ieto ftemax- the of direction and magnitude The rLVP or syst aaoawsfitdt the to fitted was parabola a , syst n h maximum the and ) 135

8 Chapter 8

Figure 8.1: Schematic representation of the EGM-VCG construction. The LV-pacing electrode is des- ignated with a star. LVd and LVp are the most distal and proximal LV electrodes, respectively. area under the curve from beginning to end of the QRS complex in the three orthonormal 2 2 2 axes X, Y, and Z, using the equation QRSarea = qQRSarea,x + QRSarea,y + QRSarea,z. Be- cause in EGM-VCG the third dimension was missing, the QRS-vector angle was expressed in the plane formed by the unipolar EGMs and an area was calculated using the equation QRS = QRS2 + QRS2 . area q area,A area,B

8.2.6 Determination of onset of intrinsic ventricular activation

In order to obtain fusion of LV-only pacing with intrinsic RV activation, it is important to determine the exact onset of intrinsic RV activation. This onset in each individual patient was determined visually during RV-only pacing at diﬀerent AV-delays. The AV-delay at which the shape of the QRS complex of the 12-lead ECG changed, indicating contribution of intrinsic activation of the RV, was deﬁned as the delay between atrial and RV activation (A-RVvis; 4 Fig. 8.2A). The onset of intrinsic ventricular activation was assessed as A-QRSonset: the interval between atrial pace spike and the onset of the QRS complex (Fig. 8.2B). Finally, the ADAPTIVECRTTM algorithm uses a formula that estimates the A-RV-delay as the onset of contribution of intrinsic RV activation (A-RVaCRT): the delay between atrial sensing or pacing and RV sensing (A-RVsense) is pre-empted by 40 ms or 70% of this amount, whichever is smaller (Fig. 8.2C).26

136 sn h rsa-alsadWloo aksmts ihBnern orcinconsecu- correction Bonferroni with test Di rank-sum tively. Wilcoxon and Kruskal-Wallis the using rt aibe r rsne scut pretgs.Lna orltoswr evaluated were correlations Linear di Possible (percentages). correlation. counts Pearson’s as by presented are variables crete otnosvralsaepeetda envalues mean as presented are variables Continuous analysis Statistical 8.2.7 iue8.2: Figure ff rn -Vmtoswr ttsial etduigacmiaino h Friedman the of combination a using tested statistically were methods A-RV erent G inlfrRV for signal EGM h pxadRO,rsetvl,admauigissga.I ae h e o indicates dot RVOT red the A panel setting. In optimal signal. the its measuring and respectively, RVOT, and apex the Panel he di Three A = ff lutae h A-RV the illustrated Voto tract. outflow RV rn ehd odtrieteosto otiuino etiua contraction. ventricular of contribution of onset the determine to methods erent apex n RV and vis ff rne ewe di between erences method, RVOT alrn R eiestig ydvc ae VCG based device by settings device CRT Tailoring eeotie ytmoaiypaigteR edin lead RV the placing temporarily by obtained were B A-QRS ) ± tnaddvain(D hra dis- whereas (SD) deviation standard onset ff rn ain ruswr tested were groups patient erent and , C h A-RV the ) aCRT ehd The method. 137

8 Chapter 8 test and the Wilcoxon signed rank test with a Bonferroni correction. Agreement between the various predictions of AV-delays was evaluated with a Bland-Altman analysis, containing the mean differences (bias) and the limits of agreement (defined as ± 1.96·SD). A two-sided P- value < 0.05 was considered statistically significant. The statistical analysis was performed using IBM SPSS statistics software version 21 (SPSS Inc., Chicago, IL).

8.3 Results

8.3.1 Patient characteristics

Of the 28 included patients, 25 patients completed all measurements. Failure to acquire all measurements in three patients occurred due to an early stop because of back pain as a result of the prolonged procedure time in one, the inability to cross the aortic bioprosthesis with the pressure wire in one, and technical problems with the LV pressure measurement device in one patient. The baseline characteristics of the 25 patients are presented in Table 8.1. The patient population was a typical CRT population with mostly males, half of the patients with ischemic cardiomyopathy, reduced LVEF, and prolonged QRS duration. During the procedure, the LV lead was aimed at a postero-lateral wall and 30% of the patients were acute non-responders

(maximal change in LV dP/dtmax ≤ 10%).

Table 8.1: Patient characteristics (n = 25)

Baseline characteristics Age (years) 68 ± 9 Male gender (n, %) 15 (60) Ischemic heart disease (n, %) 14 (44) NYHA functional class (n, %) II 21 (84) III 4 (16) LVEF (%) 26 ± 6 QRS duration (ms) 160 ± 15 Treatment (n, %) Diuretics 15 (60) ACE-I/ARB 23 (92) β-blockers 21 (84) Spironolactone 17 (68) Nitrates 4 (16) Digoxin 0 (0) Amiodarone 1 (4) During procedure LV lead location (n, %) Lateral 15 (60) Posterolateral 9 (36) Posterior 1 (4) Acute-responder (n, %) 18 (72) NYHA class: New York heart association class; LVEF = left ventricular ejection fraction; ACE-I = angiotensin-converting enzyme inhibitor; ARB = angiotensin receptor blocker.

138 hp fteEMVGloswr iia oteVGlos huhmr reua and irregular more though loops, VCG the to similar 8.3). were (Fig. narrower loops EGM-VCG the of shape ai epnet Vpcn:1 o-epnesi hmteices nL d LV in increase the whom in non-responders 1) pacing: LV to response namic (n h ihs Vd LV highest the iue8.3: Figure di at pacing LV during EGM-VCG and VCG, ECG, in changes the displays 8.3 Fig. during measurements pressure and VCG between Relation 8.3.2 (n h nl ftemxmlQSvco xrce rmteEMVGcagdby changed EGM-VCG Similarly, the from LBBB. extracted during vector back QRS the maximal vector, towards the QRS of and the angle pacing the LV with branch during corresponded bundle complex front right QRS the the towards the from pointing of starting orientation in front wave changes activation The (RBB). intrinsic paced the the between and fusion front increasing to wave due LV likely were changes These polarity. QRS negative ttesm Vdlya h ihs LVP highest the AV-delay as same the at V dly noeptet hl nraigA-eas edV lead AV-delays, increasing While patient. one in AV-delays xmlsteA-ea orsodn otelws C-adEMVGdrvdQRS EGM-VCG-derived and VCG- lowest was the to corresponding AV-delay the examples ihnteetr oottresbruso ainswr dnie,bsdo hi hemody- their on based identified, were patients of subgroups three cohort entire the Within = = ∼ ) n )mthrsodr nwo h ihs Vd LV highest the whom in match-responders 3) and 8), ) )msac-epnesi hmteL d LV the whom in mismatch-responders 2) 7), 0m ae hnteA-ea ihtehgetLVP highest the with AV-delay the than later ms 30 V dlyoptimization AV-delay G-C lwrrw uigL aigwt di with pacing LV during row) (lower EGM-VCG bevdcagsi C edV lead ECG in changes Observed P / d t max curda osdrbysotrA-ea hntehgetLVP highest the AV-delay than shorter considerably a at occurred syst alrn R eiestig ydvc ae VCG based device by settings device CRT Tailoring 1 (n uprrw,VGaiuhpae(iderw,and row), (middle plane azimuth VCG row), (upper = 0 eape hw nFg .) nalthree all In 8.4). Fig. in shown (examples 10) P / d ff t max rn Vdly noepatient. one in AV-delays erent syst P increased / nteohrhn,L pacing LV hand, other the On . d 1 t hne rmapstv oa to positive a from changed max was > > 0,adi whom in and 10%, 0 n occurred and 10% P / d ∼ t max 180 ≤ ff ◦ The . erent 10% 139 area syst

8 Chapter 8

at the AV-delay at which QRSampl was the halfway value of the QRSampl derived from the VCG or EGM-VCG between LV-only pacing and intrinsic conduction was associated with a large increase in LVPsyst. In all three examples the QRSampl halfway value derived from the EGM-VCG occurred at a 30 ms longer AV-delay than the QRSampl halfway value derived from the VCG, but the diﬀerences in LVPsyst between these two AV-delay steps were small.

Due to the mismatch between LVPsyst and LV dP/dtmax in the ‘mismatch-responders’, the

AV-delay with the lowest QRSarea did not predict the AV-delay accompanied by the highest

LV dP/dtmax (Fig. 8.4). Overall, throughout the LV pacing protocol there was a good agreement between the surface VCG and EGM-VCG derived QRSarea (R = 0.74) and QRSampl (R = 0.80). The VCG or EGM-VCG derived QRS areas were able to predict the AV-delay resulting in highest LV systolic pressure with reasonable accuracy (Fig. 8.5A and C), but not the AV-delay resulting in highest LV dP/dtmax, especially in the mismatch responders (Fig. 8.5B and D). In the mismatch-responders group, the predicted AV-delays, according to the QRSarea, EGM-

QRSarea, QRSampl, or EGM-QRSampl, resulted in an underestimation in LV dP/dtmax compared to the maximal increase (50 ± 22%, 67 ± 20%, 35 ± 21%, and 36 ± 20%, respectively).

For the match-responders group, this underestimation in LV dP/dtmax was less (32 ± 29%, 51 ± 38%, 22 ± 31%, and 23 ± 30%, respectively). In an attempt to explain the different hemodynamic responses to LV pacing, the baseline characteristics of the three different groups were compared (Table 8.2). It was observed that the match-responder group had a lower baseline LVEF and baseline LV dP/dtmax than the non-responder group. Furthermore, there was a trend towards a lower baseline QRSarea for the non-responder patients (P = 0.06 compared to mismatch-responder group; P = 0.10 compared to match-responder group). The only observed difference between the mismatch and match responder groups was the lack of patients with ischemic cardiomyopathy (ICM) in the mismatch-responder group, while in the match-responder group 70% of the patients had ICM.

140 iue8.4: Figure eut r lutae yagendaod odA-easadtecrepnigvertical corresponding the A-RV patients and AV-delays the Bold indicate lines diamond. dashed green a by illustrated are results n mltd xrce rmteVGo G-C r hw.AIpcn ( AAI-pacing shown. are area EGM-VCG QRS or in VCG changes the Corresponding from extracted pacing. amplitude LV and during AV-delays various at baseline to xmlso he ains n o-epnes n imthrsodr n n match- d LV one relative and the mismatch-responder, are Shown one responder. non-responders, one patients: three of Examples alrn R eiestig ydvc ae VCG based device by settings device CRT Tailoring P vis / d . t max n Vssoi rsueicesscompared increases pressure systolic LV and ∼ LBBB) 141

8 Chapter 8

Figure 8.5: Agreement between VCG or EGM-VCG derived QRSarea and hemodynamic predicted

AV-delays deﬁned as maximal increase in LV systolic pressure (LVPsyst; A and C) or

LV dP/dtmax (B and D). Patients with a maximal LV dP/dtmax increase of ≤ 10% are indicated by red dots (CRT non-responders). Shown is the line of identity (Y = X) and the range within 30 ms of this line.

142 LVEF 8.2: Table nitni-ovrigezm niio;ARB inhibitor; enzyme angiotensin-converting † ⋆ § ntetorsodr hw nFg .,temmn fosto nrni Vactivation, RV intrinsic of onset of moment the A-RV 8.4, by Fig. determined in shown as responders two the In e ntehgetLVP highest the in ted ‡ .. oprn di Comparing 8.3.3 EGM-QRS twihQRS which at P P P P < = < < .5cmae onnrsodr sn the using non-responders to compared 0.05 .6cmae onnrsodr sn h icxnrn-u etwt ofroicorrection. Bonferroni with test rank-sum Wilcoxon the using non-responders to compared 0.06 .5cmae omthrsodr sn the using match-responders to compared 0.05 .5cmae onnrsodr sn h icxnrn-u etwt ofroicorrection. Bonferroni with test rank-sum Wilcoxon the using non-responders to compared 0.05 = etvnrclreeto rcin C:icei adoypty YAcas e okhatascaincas ACE-I class; association heart York New class: NYHA cardiomyopathy; ischemic ICM: fraction; ejection ventricular left activation ain hrceitc ftetresbrus(o-epnes imthrsodr;match- responders) mismatch-responders; (non-responders; subgroups three the of characteristics Patient ampl irtc n )4(7 5)7(70) 7 (50) 4 (0) 0 (57) 4 (57) 4 ACE-I %) (n, Diuretics class NYHA %) (n, ICM ae(,% 8)3(8 (60) 6 (38) 3 31 (86) 6 (%) LVEF %) (n, Male g y 64 (y) Age ioi n )0()0()0(0) 0 (0) 0 (10) 1 (40) 4 (0) 0 (0) 0 (25) 2 LVP (75) 6 (0) 0 (14) 1 (14) 1 regurgitation (100) 7 Mitral %) (n, Amiodarone %) (n, Digoxin %) (n, Nitrates %) (n, Spironolactone β Ritra m)200 (ms) interval PR Vd LV aelocation Pace R uain(s 152 (ms) duration QRS QRS EGM-QRS bokr n )6(6 8)8(80) 8 (88) 7 (86) 6 %) (n, -blockers ampl I n )1(4 1)2(20) 2 (80) 8 (13) 1 (88) 7 (14) 1 (86) 6 %) (n, III %) (n, II otro n )1(4 0 (0) 0 (80) 8 (20) 2 (0) 0 (63) 5 (38) 3 (14) 1 (29) 2 (57) 4 %) (n, Posterior %) (n, Posterolateral %) (n, Lateral rd 0 0 (10) 1 (10) 1 (0) 0 (10) 1 (70) 7 (0) (25) 0 2 (0) 0 (25) 2 (50) 4 (43) 3 (0) 0 (28) 2 (0) 0 (28) 2 Unknown 3 Grade 2 Grade 1 Grade No uigL-nypcn tavr hr Vdly hsidctsta h EGM- the that indicates This AV-delay. short very a at pacing LV-only during syst area P / / R n )7(0)8(0)8(80) 8 (100) 8 (100) 7 %) (n, ARB d mH)122 (mmHg) a tl oiieada hc h EGM-QRS the which at and positive still was ( t µ max s 68 Vs) area (mmHg syst ( µ vis s 949 Vs) ﬀ motnl,A-RV Importantly, . rn esrso ne fitiscventricular intrinsic of onset of measures erent / ace elwt h Vdlydrn Vpcn htresul- that pacing LV during AV-delay the with well matched )1088 s) o-epnesMsac-epnesMatch-responders Mismatch-responders Non-responders = χ nitni eetrbokr LVP blocker; receptor angiotensin 2 χ ts ihBnern correction. Bonferroni with -test (n 2 ts ihBnern correction. Bonferroni with -test = ± ± ± ± ± ± ± ± )(n 7) 28 3 070 10 130 9 6 1045 160 8179 38 4158 14 3107 23 7 1318 777 alrn R eiestig ydvc ae VCG based device by settings device CRT Tailoring vis orsoddwt h ogs AV-delay longest the with corresponded = ± ± ± ± ± ± ± ± )(n 8) 22 5 69 9 2168 12 0119 30 39 1 835 219 3 1771 335 3189 23 †‡ § syst = etvnrclrssoi pressure. systolic ventricular left ampl 111 a tl qa othe to equal still was (70) 7 = ± ± ± ± ± ± ± ± 10) 7 8 16 139 712 27 15 43 ⋆ † ⋆ 143 =

8 Chapter 8

VCG and VCG-derived QRSampl could also be used to find the delay between atrial activation and onset of RV activation (A-RVEGM−VCG and A-RVVCG, Fig. 8.4). The different methods to find the values of the time to onset of contribution of intrinsic ventricular activation (A-RV) were compared to the A-RVvis, the AV-delay at which the shape of the QRS complex of the 12-lead ECG changed during DDD-RV pacing. There was no significant difference between A-RVEGM−VCG, A-RVVCG and A-RVvis (Table 8.3). A-RVaCRT was significantly longer than A-RVvis. Furthermore, A-QRSonset was significantly longer than the other four indices of intrinsic RV activation. These values held for the RV apex as site of implantation of the RV lead. In order to investigate the sensitivity of the four methods for the location of the RV lead, in five patients the RV lead was temporarily placed at the RV outflow tract (RVOT). A-RVvis, A-RVVCG, and A-QRSonset were not affected by this location change of the RV lead. However, A-RVaCRT was generally shorter at the RVOT position, the difference with the RV apex position ranging from -27 ms to +9 ms (example in Fig. 8.2B). Moreover, during LV pacing some patients revealed a time delay between pacing stimulus and onset of QRS complex, often referred to as LV pacing latency (example is shown in Fig. 8.6). In this example the LV pacing latency was 44 ms, leading to a relative prolongation of the A-RVaCRT compared to A-RVvis, while A-RVEGM−VCG and A-RVVCG adjust for this LV pacing latency by shortening the A-RV compared to A-RVaCRT.

Table 8.3: Optimal AV-delay using diﬀerent methods

AV-delay method Mean ± SD

A-RVvis 170 ± 33 A-RVEGM−VCG 180 ± 43 A-RVVCG 181 ± 38 ⋆ A-RVaCRT 189 ± 29 ⋆†‡• A-QRSonset 228 ± 36

A-RVvis = A-RV as visually observed; A-RVEGM−VCG = A-RV according to the EGM-VCG; A-RVVCG = A-RV according to the TM VCG; A-RVaCRT = A-RV according to ADAPTIVECRT algorithm; A-QRSonset = time between atrial pace spike and QRS onset; most deﬁnitions are illustrated in Fig. 8.2. ⋆ P < 0.05 compared to A-RVvis using the Wilcoxon signed rank test. † P < 0.05 compared to A-RVEGM−VCG using the Wilcoxon signed rank test. ‡ P < 0.05 compared to A-RVVCG using the Wilcoxon signed rank test. • P < 0.05 compared to A-RVaCRT using the Wilcoxon signed rank test.

Performances of the various algorithms were investigated by comparing the measured

LV dP/dtmax or LVPsyst during LV pacing with an AV-delay equal to the calculated A-RV delays. The longer AV-delay found using A-QRSonset resulted in a signiﬁcantly smaller increase in LV dP/dtmax than using the other four methods (A-RVvis, A-RVEGM−VCG, A-RVVCG, and A-RVaCRT; Fig. 8.7A). Furthermore, A-RVvis, A-RVEGM−VCG, and A-RVVCG resulted in a comparable increase in LV dP/dtmax, while A-RVaCRT led to a lower increase in LV dP/dtmax compared to A-RVvis (P < 0.01), A-RVEGM−VCG (P = 0.24), and A-RVVCG (P = 0.19). These diﬀerences between A-RVvis, A-RVEGM−VCG or A-RVVCG and A-RVaCRT were also present

144 V dlyrsligi ihs Vssoi rsuewt esnbeacrc,btntthat not but accuracy, reasonable with d LV pressure highest systolic with LV highest in resulting AV-delay h urn td hw htteEMVGrsmlstenra C n ht during that, and VCG normal the resembles QRS EGM-VCG the the pacing, LV that shows study current The Discussion 8.4 A-RV tteidvda ee Fg .B.Aslt hne nLVP LVP in lower towards changes trend Absolute a 8.7B). still (Fig. level individual the at iue8.6: Figure sn ihrteA-RV the either Using nrni Vatvto (A-RV activation RV intrinsic h eosrto htteEMVGrsmlsteVG seilywt ead othe to regards with especially VCG, QRS the of resembles behavior EGM-VCG the that demonstration The EGM-VCG and VCG between Relation 8.4.1 latency. LV and placement lead RV by induced since variability especially overcomes response, it hemodynamic in improvement possible a to leading individualized, h VadR ed,friga2 ln.Ti ln aisprptetsneeeyptetis patient every since patient on per varies electrodes plane three This from plane. derived 2D di been a only forming has leads, RV EGM-VCG and the LV the while vector 3D a has VCG ﬀ rn n h Vla a lcda aiu oain.Tego efrac fteEGM- the of performance good The locations. various at placed was lead LV the and erent VCG Fg .CadD). and 8.7C (Fig. iigo h aeatfc.Tetm nevlbtenpcmkrsiuu n h onset was the the complex and indicates stimulus QRS line pacemaker paced green between earliest interval The the time of The pacing. LV-only pace-artefact. during the latency of with timing patient a of Example A-RV aCRT area P / f3 s A-RV ms. 39 of area d n QRS and t EGM max n QRS and lo ohmtosse sflt eemn h xc ne of onset exact the determine to useful seem methods both Also, . − VCG VCG ampl syst rA-RV or n a hrfr sitt niiulz Vfso pacing. fusion LV individualize to assist therefore may and ) ampl sn A-RV using uiga Vpcn rtcl snttiil fe l the all After trivial. not is protocol, pacing LV an during EGM eie rmteVGo G-C a rdc the predict can EGM-VCG or VCG the from derived − VCG alrn R eiestig ydvc ae VCG based device by settings device CRT Tailoring VCG n A-RV and h ADAPTIVECRT the aCRT ∼ 4m.Teltnycue rlnainof prolongation a caused latency The ms. 44 oprdt A-RV to compared VCG eenta not were syst eesalr u hr was there but smaller, were ﬀ TM ected. vis loih a efully be can algorithm A-RV , EGM − VCG and , 145

8 Chapter 8

Figure 8.7: Hemodynamic response (relative change compared to baseline in LV dP/dtmax [A and B]

and LVPsyst [C and D]) at settings with an AV-delay equal to A-RVvis (black), A-RVEGM−VCG

(blue), A-RVVCG (green), A-RVaCRT (yellow), and A-QRSonset (red). Overall diﬀerences in change in hemodynamic measures are shown in the left panels, individual changes are ⋆ † shown in the right panels. P < 0.05 compared to A-RVvis, P < 0.05 compared to A- ‡ • RVaCRT, P < 0.05 compared to A-RVVCG, P < 0.05 compared to A-RVEGM−VCG using the Wilcoxon signed rank test with a Bonferroni correction.

LVPsyst = LV systolic pressure; A-RVvis = A-RV as visually observed; A-RVEGM−VCG = A-

RV according to the EGM-VCG; A-RVVCG = A-RV according to the VCG; A-RVaCRT = A- TM RV according to ADAPTIVECRT algorithm; A-QRSonset = time between atrial pace spike and QRS onset.

VCG in this regard may be explained by the fact that the plane of the EGM-VCG is exactly the plane containing most information. This can be explained by the fact that in LBBB the ventricular activation occurs from RV to LV and the LV lead is positioned in a vein on the LV lateral wall.

146 h rsn td orbrtspeiu nig nptet htdrn Vpcn h min- the pacing LV during that patients QRS in imal findings previous corroborates study present The measurements pressure and VCG between Relation 8.4.2 itteA-ea eutn nhgetL ytlcpesr ihraoal accuracy. reasonable with pressure systolic LV highest in resulting AV-delay the dict oae yams 180 almost by rotates rsn td xed h atrosrainb hwn htas h G-C derived EGM-VCG the also that showing by observation latter QRS the extends study present o dutfrtepsil rsneo aec,rsligi noeetmto ftera time real the of overestimation activation. an RV in intrinsic resulting of latency, onset of of presence possible the for adjust not eod fe hnigtepc etn,tu rbbylaigt mle bevdchanges. observed smaller to leading probably thus setting, pace the changing after seconds eedo er aecags n Vla oainde o hne igeA-RV single a su may change, device not the of does implantation location the lead after RV briefly or and during changes, measurement rate heart on depend ue1 et eoeadatrec rniinwr vrgdt aclt h hnei re to order in change baroreflex. the the calculate pres- to of blood averaged influence were finger avoid transition the each after for and before Furthermore, beats 10 based. sure present pressure the purely whereas was information hemodynamic response flow of and hemodynamic measure pressure a both as includes pressure combination This blood finger response. and integral) time velocity al. tract et outflow Deursen van that were study previous oprdt hntela a lcdi h Vae.Fnly h A-RV the Finally, apex. RV the in placed was lead the when to compared ed eprr lcmn fteR edi h VT e oadi a to led RVOT, the in lead RV the of placement Temporary lead. prtrdpnet stemmn fRsnehst eidctdmnal nteEMsig- EGM the in A-RV manually indicated Instead, be to nal. has RVsense of moment the as operator-dependent, .. Di 8.4.3 AV-delays. shortest the hssuysosta ohA-RV both that shows study This iemauet sesteielA-ea uigL-nypcn ocet fso’pacing. ‘fusion’ create to pacing ADAPTIVECRT LV-only the during AV-delay to ideal Compared the assess to measure tive te tde hs aaeesapa omthadtehgetL d LV highest the and match to appear parameters these studies other lt utemr,ti td hw htA-RV that shows study this Furthermore, plot. the at A-RV relation addition, average In such response. using hemodynamic shows, suboptimal study in present result may the level As A-RV. individual gen- find averaged, to an uses relation ADAPTIVECRT it AV-delay, The eral but AV-delay. of adaptation the automated of continuous optimization allows individualized algorithm fully a to leads VCG suigta naptettedi the patient a in that Assuming h imthbtenL d LV between mismatch The area activation n QRS and area ff rne ewe di between erences cussmweebtenL-nypcn n BB htteQSvector QRS the that LBBB, and pacing LV-only between somewhere occurs ampl EGM ◦ a eue oti ups.Di purpose. this to used be can − ewe Vol aigadLB,adta h QRS the that and LBBB, and pacing LV-only between VCG n A-RV and P / 27 EGM d t TM max norsuyhwvr esrmnswr efre 10 performed were measurements however, study our In ff ff − rn esrso ne fitiscRV intrinsic of onset of measures erent loih,otmzto efre yEMVGor EGM-VCG by performed optimization algorithm, VCG rnebtenARsneadA-RV and A-RVsense between erence VCG n LVP and alrn R eiestig ydvc ae VCG based device by settings device CRT Tailoring n A-RV and 19 r ietyotie rmaQRS a from obtained directly are sdacmiaino coadorpy(LV echocardiography of combination a used aCRT syst n8 in sdpneto h oaino h RV the of location the on dependent is VCG ff / rne ewe h rsn n the and present the between erences 5ptet srmral,bcuein because remarkable, is patients 25 a eue sarbs n objec- and robust a as used be can P / ff d rnei A-RV in erence t aCRT max ffi srrl enat seen rarely is et dutthe adjust to ce loih does algorithm ampl VCG area − AV-delay osnot does a pre- can aCRT 19 aCRT The VCG 147 TM is

8 Chapter 8

ADAPTIVECRTTM algorithm at the individual level. Another possibility would be that the algorithm to ﬁnd A-RVEGM−VCG is embedded in the CRT device, resulting in the possibility to adjust for the delay between A-RVsense and A-RVEGM−VCG more regularly. Obviously, future studies are required to investigate whether the diﬀerence between A-RV and A-RVsense indeed is independent of heart rate.

8.4.4 Limitations

The current study was a relatively small study, designed to proof the principle of electrical optimization of CRT using VCG and EGM-VCG. The study was performed in two centers (MUMC+ and UMCG) and the patients were consecutive patients. Furthermore, as indicated by the baseline patient characteristics (Table 8.1), the patient population is representative for the CRT population. A larger multicenter trial should be performed to conﬁrm our results. A limitation of the current study is that it only investigated acute hemodynamic CRT response. Acute changes in LV dP/dtmax or LVPsyst may not correspond to the chronic long- term changes.28 The current study only investigated the response in patients with a LBBB, while the response and the A-RV determination could be diﬀerent in right bundle branch block and intraventricular conduction delay patients. However, the current indication for CRT regarding these conduction disturbances is class I for LBBB and class II for non-LBBB patients.22

8.5 Conclusions

The present study shows that during LV pacing the QRSarea or QRSampl derived from the VCG or EGM-VCG, can be used to predict the AV-delay resulting in highest LV systolic pressure. Furthermore, the onset of contribution of the intrinsic RV activation can be determined using the VCG or EGM-VCG, which can be used to objectively and easily tailor the ADAPTIVECRTTM algorithm, possibly leading to a further increase in hemodynamic response by CRT.

8.6 Funding

148 References .JFCead CDuet rmn,NFemnl,DGa,LKpebre,L Kappenberger, L Gras, D Freemantle, N Erdmann, E Daubert, JC Cleland, JGF 1. .Erpa er htmAscain uoenSceyo adooy er Rhythm Heart Cardiology, of Society European Association, Rhythm Heart European 2. .WMles AGim eg,TDeig CSaln,B Wilko BL Starling, RC Dresing, T Verga, T Grimm, RA Mullens, W 3. .KVroy AVrek NCreusn ika,H rjs rs n FW and Arts, T Crijns, HJ Dijkman, B Cornelussen, RN Verbeek, XA Vernooy, K 4. .K lebgn RGl,T ee,IFrne oao itl DWaggoner, AD Mittal, S Lozano, Fernndez I Meyer, TE Gold, MR Ellenbogen, KA A 6. Kuiper, M Hunnik, van A Ploux, S Houthuizen, P Middendorp, van LB Strik, M 5. .AArcho tlbik lc,SSc,JVg,PBke,HKen Kramer, A Klein, H Bakker, P Vogt, J Sack, S Block, M Stellbrink, C Auricchio, A 7. eto ada eycrnzto hrp nhatfiue mln n olwu re- follow-up and implant failure: management. heart and commendations in therapy resynchronization cardiac on ment aaz,adCricRsnhoiainHatFiue(AEH)SuyInvestigat- e Study The (CARE-HF) ors. Failure Resynchronization-Heart Cardiac and Tavazzi, nlJMed J Engl N r soito,J abr,LSxn ta.21 EHRA 2012 al. et Saxon, L Fail- Daubert, Heart JC Echocardiography, Echocardiography, Association, of of ure Association Society European American Association, America, Heart of American Society Failure Heart Society, Vitra ncricrsnhoiaintherapy. resynchronization cardiac in interval VV ag nihsfo ada eycrnzto piiaincii spr faheart a of part as program. clinic management optimization disease resynchronization failure cardiac a from Insights Tang. rne.Cluaino e of Calculation Prinzen. alr.TePcn hrpe o ogsieHatFiueSuyGop h Guidant The Group. Study Failure Heart Congestive for heart Therapies congestive Pacing with The patients paced failure. of function systolic acute on delay atrioventricular n loihi tivnrclrdlypormigi ada eycrnzto ther- resynchronization cardiac apy. in therapy programming delay resynchronization atrioventricular cardiac algorithmic and in echocardiography-guided, used empirical, comparing methods trial randomized Rap- delay a J AV (SMART-AV) trial: Bedi, optimization: other AV M determined to Weiner, SmartDelay comparison S the from Whitehill, a results J Primary Eyk, Stein. Van KM JE and Spinale, kin, FG Singh, JP Lemke, B hearts. the canine of dyssynchronous Electrophysiol determinant in Arrhythm as therapy wavefronts resynchronization electrical cardiac of to Interplay response Prinzen. FW and Auricchio, ig ao oka,TPce,adJSiel.E Spinelli. J and Pochet, T Tockman, B Salo, R Ding, J Circulation ff c fcricrsnhoiaino obdt n otlt nhatfailure. heart in mortality and morbidity on resynchronization cardiac of ect 20)352:1539–1549. (2005) 21)122:2660–2668. (2010) 21)6:924–931. (2013) ff cieV nevlfclttsotmzto fA ea and delay AV of optimization facilitates interval VV ective er Rhythm Heart mCl Cardiol Coll Am J 21)9:1524–1576. (2012) er Rhythm Heart / 20)53:765–773. (2009) ff R xetcnessstate- consensus expert HRS c fpcn hme and chamber pacing of ect 20)4:75–82. (2007) ff n WHW and , Circ

8 Chapter 8

Congestive Heart Failure Research Group. Circulation (1999) 99:2993–3001. 8. A Auricchio, C Stellbrink, S Sack, M Block, J Vogt, P Bakker, P Mortensen, H Klein, PATH-CHF Study Group, et al. The Pacing Therapies for Congestive Heart Failure (PATH-CHF) study: rationale, design, and endpoints of a prospective randomized multicenter study. The American journal of cardiology (1999) 83:130–135. 9. DA Kass, CH Chen, C Curry, M Talbot, R Berger, B Fetics, and E Nevo. Improved left ventricular mechanics from acute VDD pacing in patients with dilated cardiomyopathy and ventricular conduction delay. Circulation (1999) 99:1567–1573. 10. ZI Whinnett, JE Davies, K Willson, AW Chow, RA Foale, DW Davies, AD Hughes, DP Francis, and J Mayet. Determination of optimal atrioventricular delay for cardiac resynchronization therapy using acute non-invasive blood pressure. Europace (2006) 8:358–366. 11. C Butter, C Stellbrink, A Belalcazar, D Villalta, M Schlegl, A Sinha, F Cuesta, and C Reister. Cardiac resynchronization therapy optimization by finger plethysmography. Heart Rhythm (2004) 1:568–575. 12. P Ritter, PPHM Delnoy, L Padeletti, M Lunati, H Naegele, A Borri-Brunetto, and J Silvestre. A randomized pilot study of optimization of cardiac resynchronization therapy in sinus rhythm patients using a peak endocardial acceleration sensor vs. standard methods. Europace (2012) 14:1324–1333. 13. ZI Whinnett, DP Francis, A Denis, K Willson, P Pascale, I van Geldorp, M De Guil- lebon, S Ploux, K Ellenbogen, M Haïssaguerre, et al. Comparison of different invasive hemodynamic methods for AV delay optimization in patients with cardiac resynchronization therapy: implications for clinical trial design and clinical practice. International journal of cardiology (2013) 168:2228–2237. 14. D Gras, MS Gupta, E Boulogne, L Guzzo, and WT Abraham. Optimization of AV and VV Delays in the Real-World CRT Patient Population: An International Survey on Current Clinical Practice. Pacing and Clinical Electrophysiology (2009) 32:S236– S239. 15. MR Gold, I Niazi, M Giudici, RB Leman, JL Sturdivant, MH Kim, Y Yu, J Ding, and AD Waggoner. A prospective comparison of AV delay programming methods for hemodynamic optimization during cardiac resynchronization therapy. J Cardiovasc Electrophysiol (2007) 18:490–496. 16. JH Baker 2nd, J McKenzie 3rd, S Beau, GS Greer, J Porterfield, M Fedor, S Green- berg, EG Daoud, R Corbisiero, JR Bailey, and L Porterfield. Acute evaluation of programmer-guided AV/PV and VV delay optimization comparing an IEGM method and echocardiogram for cardiac resynchronization therapy in heart failure patients and dual-chamber ICD implants. J Cardiovasc Electrophysiol (2007) 18:185–191.

150 6 OMri,BLme ine rm L e,KAnm,MGsaii RC Gasparini, M Aonuma, K Lee, KLF Krum, H Birnie, D Lemke, B Martin, DO 26. 5 BEgl,E éh J a ere,KVroy PSnh n WPizn T- Prinzen. FW and Singh, JP Vernooy, K Deursen, Van CJM Végh, EM Engels, EB 25. 4 AKr,GvnHre,A itg n HvnBme.Rcntuto fteFrank the of Reconstruction Bemmel. van JH and Sittig, AC Herpen, van G Kors, JA 24. 3 hnet ais ilo,CMnsy hw ol,D ais Hughes, A Davies, DW Foale, R Chow, A Manisty, C Willson, K Davies, J Whinnett, Z 23. 7 rm ek,DBri,KFLe oua CSaln,MGsaii J Gasparini, M Starling, RC Aonuma, K Lee, KLF Birnie, D Lemke, B Krum, H 17. 8 MvnGle,F rce ejr n HPjs h eoyai e hemodynamic The Pijls. NH and Meijer, A Bracke, FA Gelder, van BM 18. 0 J a ere,MSrk MRdmkr,AvnHni,MKie,LWecke, L Kuiper, M Hunnik, van A Rademakers, LM Strik, M Deursen, van CJM 20. F Janssen, MHG Ståhlberg, M Everdingen, van WM Wecke, L Deursen, van CJM 19. 2 rgoe uici,GBrnEqiis odca,GBrai ABreithardt, OA Boriani, G Bordachar, P Baron-Esquivias, G Auricchio, A Brignole, M and Vernooy, K 22. Kuiper, M Hunnik, van A Middendorp, van LB Strik, M Engels, EB 21. niiulzdcricrsnhoiainteay ainl n eino h adaptive the trial. for of therapy algorithm resynchronization design novel cardiac and A rationale Martin. therapy: D resynchronization and cardiac Kalmes, individualized A Sambelashvili, A Rogers, T Gorcsan, dpieCTSuyIvsiaos netgto fanvlagrtmfrsynchron- for algorithm novel a and of Houmsse, Investigation M 3rd, Investigators. Gorcsan Study J CRT Sambelashvili, Adaptive A Rogers, T Milasinovic, G Starling, aeae rdcsrsos ocricrsnhoiainteayi ainswt left with patients in therapy block. resynchronization branch cardiac bundle to response predicts area wave ora fteAeia olg fCardiology of College pacing. American biventricular the to of compared Journal as pacing ventricular left during conduction intrinsic piiaino ada eycrnzto hrp ncnn etbnl rnhblock easy branch for bundle tool left a canine as in hearts. Vectorcardiography therapy resynchronization Prinzen. cardiac FW of optimization and Vernooy, K Crijns, HJGM therapy. Res resynchronization Transl cardiac Cardiovasc in J Vectorcardi- intervals stimulation Prinzen. of FW optimization and Vernooy, for ography K Crijns, HJGM Bergfeldt, L Braunschweig, etradormfo tnadeetoadorpi ed:dansi oprsnof comparison di diagnostic leads: electrocardiographic standard from vectorcardiogram interventricu- and atrioventricular delay. of ana- lar importance relative pattern: and consistent magnitude a shape, show of lysis therapy resynchronisation cardiac in delay ventricular ae,adDFacs amdnmce Haemodynamic Francis. D and Mayet, J (EHRA). Association Rhythm 34:2281–2329. Developed Heart (2013) European (ESC). the Cardiology and with of pacing collaboration Society cardiac in European cardiac on the on Force Guidelines of Task ESC therapy the 2013 resynchronization therapy: al. resynchronization et Elliott, cardiac PM and Delgado, pacing V Deharo, JC Cleland, J in vectorcardiography by Unpublished. resynchronization hearts. cardiac canine optimal dyssynchronous of Prediction Prinzen. F ff rn methods. erent icAryh Electrophysiol Arrhythm Circ Heart 20)92:1628–1634. (2006) u er J Heart Eur adoacElectrophysiol Cardiovasc J 21)8:128–137. (2015) 19)11:1083–1092. (1990) 21)5:544–552. (2012) mHatJ Heart Am ff cso hne naroetiua n inter- and atrioventricular in changes of ects 20)46:2305–2310. (2005) 21)26:176–183. (2015) 21)163:747–752.e1. (2012) u er J Heart Eur References ff c of ect 151

8 Chapter 8

ized left-ventricular pacing and ambulatory optimization of cardiac resynchronization therapy: results of the adaptive CRT trial. Heart Rhythm (2012) 9:1807–1814. 27. A Kyriacou, PA Pabari, ZI Whinnett, S Arri, K Willson, R Baruah, B Stegemann, J Mayet, P Kanagaratnam, AD Hughes, et al. Fully automatable, reproducible, noninvasive simple plethysmographic optimization: proof of concept and potential for im- plantability. Pacing and Clinical Electrophysiology (2012) 35:948–960. 28. MD Bogaard, P Houthuizen, FA Bracke, PA Doevendans, FW Prinzen, M Meine, and BM van Gelder. Baseline left ventricular dP/dtmax rather than the acute improvement in dP/dtmax predicts clinical outcome in patients with cardiac resynchronization therapy. Eur J Heart Fail (2011) 13:1126–1132.

152

Chapter 9 General discussion Chapter 9

9.1 Introduction

he ultimate goal of this thesis is to improve the response to cardiac resynchronization therapy (CRT) using vectorcardiographic (VCG) analysis. T In chapter 3 we showed that the Kors-VCG can be calculated from the 12-lead electrocardiogram (ECG) and closely resembles the measured Frank-VCG in patients eligible for CRT. This enables retrospective as well as prospective VCG analysis of routinely recorded 12-lead ECGs, which was used in all other chapters in the present thesis. Patients with heart failure and left bundle branch block (LBBB) benefit from CRT. To better understand the natural history of LBBB we took advantage of the fact that LBBB develops in part of the patients undergoing transcatheter aortic valve implantation (TAVI). In chapter 4 we showed that in the majority of patients who developed LBBB during a TAVI procedure, electrical remodeling occurred in the first month of LBBB, as evidenced by a reduction in repolarization variables. However, the extent of electrical remodeling was highly variable between patients. In chapters 5 and 6 we investigated the potential role of the VCG in patient selection for CRT. CRT response was assessed echocardiographically (left ventricular [LV] ejection fraction [EF]) as well as clinically (e.g., mortality and HF hospitalization). We showed that in patients presenting with LBBB morphology, a larger baseline Tarea is an important predictor of both LVEF increase and of improved long-term clinical outcomes following CRT. This effect was larger than the prediction ability of the QRSarea. Once a CRT device is implanted, the benefit of CRT might be increased by optimizing the CRT device settings continuously and individually using the VCG or an electrogram (EGM) based VCG (EGM-VCG). In canines with LBBB (chapter 7) as well as in patients undergoing a CRT implantation (chapter 8), the EGM-VCG proved to be able to predict the AV settings providing the largest increase in LV pump function, especially during LV fusion pacing. In chapter 8 it was also shown that both the VCG and EGM-VCG can be used to find the onset of intrinsic right ventricular (RV) activation in the individual patient, which is useful to optimize LV fusion pacing.

9.2 Use and advantages of the VCG

As was pointed out in chapter 3 of this thesis, it is justiﬁed to use the Kors-VCG as calculated from the 12-lead ECG for VCG analysis in patients eligible for CRT, so with wide QRS complexes. This good correlation in this patient group extends previously published articles on the good performance of the Kors-VCG in other patient cohorts.1–4 To be able to calculate a VCG from the 12-lead ECG it is preferable to have a digital ECG. However, in many hospitals the ECG is only stored semi-digitally in a pdf-ﬁle. In chapter 5

156 C hntecretaayi fte1-edEG twudee epsil oadthe add at to information VCG. present possible the already be showing the without even to ECG, would an area, of It the top ECG. by the depicted 12-lead as the the size of analyzing loop of analysis vectorcardiographic way current objective more the a than provides This VCG programs. computer using analyzed images. be 1D 12 in than image 3D orientation one vector in and and detected shape, easily easily loop to more size, more easier are loop calculated it in changes making be Overall, directions, can changes. all T-wave size 3D in T-wave observe information in provides com- per VCG change T-wave done the global the be Furthermore, the only of accurately. can VCG change size T-wave the directional in Using the changes and studying lead. However, intervals, complex. QTc QRS or the to JTc pared the variables are conventional the used ECG, are 12-lead a that using remodeling electrical this quantify to aiming C ol esandaddgtzduigtepormEGcn(MSLC e York, New LLC, (AMPS ECGScan program the using USA) NY, digitized and scanned be could ECG artefact, pace added be a can as it such systems point, present. ECG reference be most common should in one Even first, least necessary. at at is present Alternatively, to lead always retrospectively. running not heartbeats common is all one lead align least running at to this VCG, same though Therefore, a the calculate not to often column(s). able are subsequent be case column eventually the first in the in so However, in graphics, registered registered leads. vector heartbeats as 4x3 the the heartbeats leads, or in 4x3 6x2 embedded or of be leads format will 6x2 a pdf-file of in the stored on are information ECGs visible most only that USA). MA, is Natick, aspect (MathWorks, MATLAB critical using One the reconstructed graphics, be vector editor, can extracted graphics these ECG vector Using 12-lead USA). a digital MA, using (Boston, pdf-files. extracted Inkscape these program be from the can as information which such ECG graphics 12-lead vector digital include the pdf-files extract The to method a described we rclrmdln seiecdb hne nrplrzto ( repolarization in changes by evidenced as ECG. remodeling scalar trical a of of interpretation projections 1D improved 12 an the for to allows compared as image activity phe- single electric 3D a the the in of activation visualization the spatial electrical from the calculated of information, be nomenon same can the 3-dimensional VCG contains a the thus vectors, and though all even ECG of 12-lead Therefore, depicted heads constructed. force arrow be electrical the can Connecting resultant loop a vector time-point. in each resulting for heart, vector the a by by generated are that forces trical h C aibe eae olo ie opsae rvco retto a automatically can orientation vector or shape, loop size, loop to related variables VCG The ncs nypprEGrcrig r vial n omnrnigla speet the present, is lead running common a and available are recordings ECG paper only case In h C a eo pca motnewe hne htocroe ie uha elec- as such time, over occur that changes when importance special of be can VCG The elec- the of direction and magnitude the records that technique forgotten almost an is VCG 5 aigi osbet locluaeaVGfo ae 2la ECGs. 12-lead paper from VCG a calculate also to possible it making , hpe 4 chapter ,aeo neet When interest. of are ), eea discussion General 157

9 Chapter 9

9.3 Electrical remodeling in patients with LBBB

Until now the exact natural history of LBBB in patients was poorly understood because the onset of LBBB is silent and, consequently, it is usually not known when LBBB started. The TAVI-induced LBBB patients established the first human ‘model’ of LBBB in that respect. In chapter 4 we showed that in these patients electrical remodeling, as indicated by changes in repolarization variables, is present, but also that the extent of these changes differs greatly between patients. Possibly, this variability may be caused by the fact that these old (∼80 years) patients may have multiple other co-morbidities. However, we did not discover significant baseline differences between subgroups of TAVI-LBBB patients presenting with minor or major electrical remodeling. Therefore it is not known whether this variability is due to the aging and disease processes of the patients or due to other factors such as genetic variation. The phenomenon of electrical remodeling can be divided into primary and secondary remodeling. Primary remodeling, better known as cardiac memory, describes persistent T-wave changes after the resumption of normal activation following a period of altered ventricular activation.6,7 Secondary remodeling develops as a consequence of a structural alteration such as heart failure and is believed to play a role in the resynchronization of ventricular repolarization in the new activation sequence.7,8 Cardiac memory was first introduced by Rosenbaum et al. 6 in 1982 and has previously been observed in patients after a period of RV pacing9, intermittent LBBB10, pre-excitation with Wolff-Parkinson-White11, and in patients receiving CRT12. Shvilkin et al. 13 demonstrated that the repolarization changes, associated with cardiac memory, can be observed not only when normal ventricular activation is restored but even when abnormal activation (e.g., ventricular pacing) continues. These repolarization changes occur between one day and one week.12,14 Shvilkin et al. 15 also showed that patients with longer lasting LBBB (> 24 hours) have a smaller T-wave amplitude compared to patients with new LBBB (≤ 24 hours), which is consistent with our finding in the TAVI-induced LBBB population (chapter 4). Importantly, the patient population described by Shvilkin et al. 15 differed from our population in that they were younger, had no aortic valve disease, and (presumably) less severe heart failure. Furthermore, in our study the time course of LBBB was more precise than in the study presented by Shvilkin et al. 15 . There- fore, a reduction in repolarization variables (like JTc interval, T-wave amplitude, and Tarea) during longer lasting dyssynchrony of any cause (LBBB, RV pacing) appears to be a common behavior. To better understand the changes in the T-wave following the presence of LBBB, we first need to go back to the T-wave in the normal situation. The exact reason for the concordant T-wave in normal individuals is not clear. It appears that the sequence depends on segmental and transmural differences in action potential duration. In the normally activated ventricles, the sequences of depolarization and repolarization across the ventricular wall are opposite,

158 hrceitc a ebs oprdwt h hoi aao h AIidcdLB pa- LBBB TAVI-induced the of di data several chronic are the There with tients. compared best be can characteristics . Di 9.4 investigated. is mortality term, long xli h euto nteTwv aibe uha h T nevl -aeapiue and amplitude, T-wave interval, JTc the as such T variables T-wave the in reduction the explain h R ope ugssta losm ftoeptet aeapoia LBBB. proximal a have patients those shorten of can some pacing is also His-bundle block patients that the the LBBB suggests of on of complex subset location valve QRS a exact the least artificial the at the in candidates that CRT of fact In the pressure but branch. unknown, the bundle left to the due of LBBB, part most TAVI proximal proximal because ‘typical’ candidates, CRT a with causes patients TAVI-LBBB likely compare to interesting is It h usinrmiswehrteocrec feetia eoeiga remodeling electrical of occurrence the whether remains question The perspectives Future severely action so in is adaptation patients by totally these reversed in be duration. phase cannot potential depolarization repolarization of the sequence that the fact discordant that are the delayed T-waves to the due population, LBBB be general might the which in However, clear. not is expected oevr otlkl jcinfato a osdrbylre n Vvlmssalrin smaller volumes LV CRT-population. and the larger in patients considerably patients. the was TAVI than fraction III ejection ischemia class likely less NYHA most in had Moreover, often men, more more were included patients the older, but was TAVI-population the 9.1) (Table teristics ieadi hc h eainbtencagsi h T the in over prospective changes followed large between are relation A patients the which LBBB TAVI-induced demonstrated in deleterious. which and been or in time never performed beneficial has be is should it memory trial memory cardiac clinical cardiac of for presence Also the whether rates. mortality eventually and ein eeo ograto oeta uainwieteltratvtdrgosdevelop regions activated later the duration while potential duration action potential shorter action longer develop regions vr tde aesonta loi aeo uhsgetldi segmental how- such activation, of abnormal case in sustained also During that T-wave. shown have discordant studies a ever, of onset the explain may .. h aletatvtdrgosaerplrzdlts,laigt ocrac ewe the between concordance a T-wave. to and leading complex latest, QRS repolarized are regions activated earliest the i.e., rnmrldi transmural area eas h R addtsms ieyhdLB o ogrpro ftm,terECG their time, of period longer a for LBBB had likely most candidates CRT the Because hte loi hs ae opeenraiaint ocratTwv a be can T-wave concordant to normalization complete a cases these in also Whether . R candidates CRT ff rne ewe AIidcdLB ainsand patients LBBB TAVI-induced between erences ff 21–26 rne.We BBi nue,lresgetldi segmental large induced, is LBBB When erences. epn hs di these Keeping 16,17 ff rne ewe hs ains okn ttebsln charac- baseline the at Looking patients. these between erences ne hs odtossgetldi segmental conditions these Under 8 edn oamr ycrnu eoaiain hsmay This repolarization. synchronous more a to leading , ff rne nmn,i sitrsigt bev htQRS that observe to interesting is it mind, in erences area n ees eoeigo,o the on or, remodeling reverse and ff rne,teeriractivated earlier the erences, ff ff rne r mle than smaller are erences rne eeo which develop erences ff csteL function LV the ects eea discussion General 18–20 159

9 Chapter 9

duration was shorter but QRSarea and QRSampl were larger for the TAVI-population. Simi- larly, QTc and JTc interval were shorter while Tarea and direction of the maximal T-vector were similar between the two patient groups (Table 9.1). The shorter QRS duration and larger QRSarea in the TAVI-induced LBBB population could be explained by the fact that the TAVI patients are subject to concentric LV hypertrophy (LVH)27 while CRT candidates often show signs of dilated cardiomyopathy. The smaller LV cavity in the TAVI patients may have led to a shorter QRS duration. This idea is supported by the notion that, in patients with a narrow QRS complex, LVH is known to result in a larger R-wave due to a higher 28,29 LV mass. The larger QRSarea in the TAVI-induced LBBB patients may relate further to the lower number of patients with ischemic HF etiology. van Deursen et al. 30 previously showed that the QRSarea is lower in LBBB patients with an ischemic HF etiology compared to the ones without ischemic HF etiology. Moreover, the TAVI-induced population all had LBBB, while 43% of the CRT candidates had non-LBBB. Non-LBBB has been associated with a lower QRSarea than LBBB patients, resulting in an overall lower average QRSarea in the CRT candidates.30–32

Table 9.1: Patient characteristics in chronic TAVI-induced LBBB patients and CRT candidates. Ex- tracted from Table 4.1, Table 4.3, and Table 6.1.

Variable Chronic TAVI-induced LBBB CRT candidates n = 67 n = 335

Baseline characteristics Age (years) 79 ± 6 67 ± 13 Male gender (n, %) 32 (48) 265 (79) Ischemic HF etiology (n, %) 14 (21) 192 (57) NYHA class I (n, %) 2 (3) 0 (0) II (n, %) 10 (15) 8 (3) III (n, %) 47 (70) 236 (70) IV (n, %) 8 (12) 31 (9) Unknown (n, %) 0 (0) 60 (18) LBBB (n, %) 67 (100) 199 (59) VCG measurements QRS duration (ms) 159 ± 20 172 ± 33 QTc interval (ms) 488 ± 51 516 ± 77 JTc interval (ms) 307 ± 39 325 ± 62 QRS amplitude (mV) 1.7 ± 0.4 1.5 ± 0.5

QRSarea (µVs) 113 ± 35 88 ± 46 T amplitude (mV) 0.6 ± 0.2 0.5 ± 0.2

Tarea (µVs) 84 ± 34 82 ± 47 Variables are shown as counts (percentage) or mean ± standard deviation when appropriate. HF = heart failure; NYHA = New York Heart Association.

160 lcrclrmdln ih lob xetdt hwrvrermdln fe R.However, CRT. in after shown remodeling reverse as of show signs to showing expected be Patients also outcome. might clinical remodeling of and electrical amount function the LV in whether changes unknown to is relates it However, remodeling remodeling. of amount the regarding patients fhatfiuewt Vdsucindvlp npeec fpritn BB R a be may CRT LBBB, closely. persistent of monitored presence LVEF. be in of should irrespective develops considered patients dysfunction LV these with mortality, failure and heart If function LV on LBBB induced ees eoeigams l a T a had all almost T remodeling lower reverse a with patients than response CRT lnclsuis h eeto R osntdpn nLVEF. and on experimental depend not various does on CRT based and of LVEF benefit Nevertheless, the reduced implantation. studies, a CRT clinical have for not qualify do not patients current do LBBB TAVI-induced to thus the according but accompanied LVEF, failure Furthermore, reduced heart a dyssynchronous with hold. by patients not in indicated does is CRT CRT guidelines, against CRT argument age the patients o ugr u nwo AIi aoe yteHatTeam. Heart the by favored suitable be is TAVI still whom may in who stenosis but IIa a aortic surgery already symptomatic is for TAVI there with of also patients use will the high-risk procedures for TAVI for guidelines that ESC often indication expected the and In be old may patients. such younger it in in term applied another be long after the device on one However, of patients. implantation frail add to questionable be may it rmpeiu tde ti nw htTV-nue BBa TAVI-induced LBBB that known is it studies previous From TAVI? after CRT 9.5 htotgnrasdrv iia eet rmCTt one patients. younger to CRT from benefits similar derive octogenarians that BBalremlietrpopciesuyi eurdta olcsdt nelectrocardio- on develop data and collects procedure that TAVI required a is undergo study who prospective and multicenter patients structural, large of electrical, a between heart LBBB relation the the in of changes understanding better mechanical a achieve to order In perspectives Future may LBBB of onset after responders. CRT remodeling best electrical the least be the underwent who patients LBBB that tnadoe hs ugr.Svrlpirsuiseautn h e for the patients evaluating risk studies fact high prior the considered Several to are due surgery. whom is patients This chest in open patients. applied standard CRT only most is than TAVI older currently is TAVI-LBBBthat the cohort TAVI in CRT current of The application form the typical patients. withhold most may factors the several have patients However, in LBBB. patients CRT of LBBB TAVI-induced for substrate and electrical cardiomyopathy ideal dilated the with considered is LBBB as CRT, receive should soitdwt nrae mortality. increased with associated In hpe 4 chapter hpe 5 chapter esoe htteewsahg aiblt nteTV-nue LBBB TAVI-induced the in variability high a was there that showed we and 6 BBptet ihaT a with patients LBBB 33 area hrfr h usinrsswehrteepatients these whether rises question the Therefore oe hnti cut-o this than lower area h AIptet hwn ihdge of degree high a showing patients TAVI The . area ≥ 72 38–40 ff csL ucinadhsbeen has and function LV ects ff 37 µ shv agrcac of chance larger a have Vs au.Ti ol suggest would This value. ie h e the Given lal,i hs younger these in Clearly, ff cso R reported CRT of ects eea discussion General 34–36 ff Nevertheless, cso TAVI- of ects 161

9 Chapter 9 graphy, echocardiography, and clinical outcome. On the basis of this data patients may be selected who might beneﬁt from CRT after developing TAVI-LBBB.

9.6 The role of repolarization in patient selection for CRT

In chapter 5 and 6 we showed that Tarea predicts CRT response even better than QRSarea. The predictive power for both echocardiographic and long term clinical CRT response was found to be primarily evident in the group of patients with LBBB, who already are known to benefit most from CRT. As a logical consequence also sum QRST is a good predictor. This latter observation is supported by a recent study in which it was shown that the sum absolute QRST integral (SAI QRST) can also be used for prediction of CRT response.41 The method used to calculate SAI QRST was however different from our calculation of the sum QRST area; the SAI QRST was equal to the absolute area under the QRST curve, while for our sum QRST area the area under the QRST curve below baseline was subtracted from the positive area under the curve. Moreover, in the article of Tereshchenko et al. 41 the inverse Dower method was used to calculate the VCG which might induce a larger error compared to a measured Frank-VCG. Despite these differences with our studies, this study supports the possible importance of the Tarea in the patient selection for CRT. The use of VCG for CRT response prediction was initiated by van Deursen et al. 30 who showed that a large QRSarea is associated with high odds of long-term volumetric CRT response. Moreover, QRSarea predicted CRT response better than QRS duration and then conventionally defined LBBB and at least as good as the most refined LBBB definition.30 It was hypothesized that large electrical dyssynchrony, amenable to CRT, would lead to large unopposed electrical forces during ventricular depolarization and that the size of these forces may be well represented by the QRSarea. Similarly, the size of the Tarea is a reflection of the extent of unopposed electrical forces during the repolarization phase. The Tarea is partially 32 + 2+ determined by the size of the QRSarea , but other factors such as changes in K and Ca ion channel expression might also play a role. Important to note is also that a larger Tarea was primarily caused by larger amplitude and not so much by a longer JTc-interval, the most commonly applied measure of repolarization. Further research is needed to investigate which other factors are exactly reflected in the Tarea.

Future perspectives

The exact mechanism why the T-wave is such a good predictor of CRT response is still unknown. Therefore, the diﬀerent roles of ion channels such as the L-type Ca2+ channel, transient outward potassium current or the rapidly activated delayed rectifying potassium current should be investigated. This can for instance be performed using a dedicated computer

162 rdsuigtoeetoe nteR n n lcrd utaoeadoejs eo the below just one and above just electrode one and elec- ( RV non-paced electrode the the pacing LV from in extracted electrodes be two could using VCG trodes device-based the that showed we canines h a h G-C a acltddi calculated was EGM-VCG the way The optimization CRT VCG-mediated of Robustness 9.7.1 pacing. fusion LV in programming (A-RV) activation for RV helpful intrinsic extremely of is that onset time of a time individually, the patient identify each to used be can EGM-VCG the h G-C a bet rdc h Vdlyrsligi ihs Vssoi rsuewith pressure systolic LV ( highest accuracy in reasonable AV-delay resulting the predict to able was patients EGM-VCG in the However, AV-delay (’mismatch-responders’). short unphysiologically an at curred R etn stecnn ehd(i.91.Teeoe sln stercre Gsaein are EGMs recorded the optimal as work. long the to as predicting seems Therefore, EGM-VCG in 9.1). the good RV-LV (Fig. direction, as method almost canine the performed as method setting patient CRT methods the lead. both B study, performing and canine when A However, the the loss. both in electrode information for RV some electrode in reference proximal RV resulted the as have ring might Since RV This the use mm). to from 47 chose electrodes of we proximal missing (distance most was available and distal were most lead the quadripolar electrodes, LV the the For ring. RV the being scmlctdb h bevto hto n hr fteptet h ihs Vd LV highest the patients the of third one studies on these that between observation comparison the devices, by CRT complicated of AV-delay is of optimization for VCG of use eitrsigt e o h -aecagsatrCTipatto;d h ainswt a with patients the do implantation; CRT after changes T T-wave large the how see to interesting be omneo h G-C ndg ihifrto.Terslswr iia compared QRS similar derived cardiomyopathy. were EGM-VCG ischemic results of Therefore, presence The the 9.2). in (Fig. works infarction. ischemia per- without with the canines dogs tested LBBB also in to we EGM-VCG cardiomyopathy, ischemic the have of patients CRT formance the of half Since used. T utemr,w nysoe h motneo h T the of importance the showed only we Furthermore, hl h eut rmdg ihLB ( LBBB with dogs from results the While intervals stimulation optimizing Tailor-CRT: 9.7 trial. clinical prospective larger a in performed oe,a a rvosydn o h -yeCa L-type the for done previously was as model, eivsiae hte h motneo h T the of importance the whether investigated be area ntecnn td ecie in described study canine the In hsaayi scretybigpromdi u ru nasalptetpopulation. patient small a in group our in performed being currently is analysis This ? area hwalre euto nT in reduction larger a show hpe 7 chapter hpe 8 chapter .Hwvr nptet,ol n Veetoewsavailable, was electrode RV one only patients, in However, ). .Moreover, ). hpe 7 chapter area hpe 7 chapter fe R oprdt h ainswt small a with patients the to compared CRT after ﬀ hpe 8 chapter rdbtentecnnsadteptet.In patients. the and canines the between ered BBdg ihu n oobdte were comorbidities any without dogs LBBB , 2 + area hne yKipr tal. et Kuijpers by channel n R ains( patients CRT and ) vrteQRS the over area rvdseiec htteVGand VCG the that evidence provides nartopciesuy tshould it study, retrospective a in area hpe 8 chapter ssilpeetwhen present still is eea discussion General 42 twudalso would It . upr the support ) P / d area t max also 163 oc-

9 Chapter 9

Figure 9.1: ROC-curves for QRSarea derived from the EGM-VCG in LBBB dogs, using the dog (2 RV and 2 LV EGM signals; red) or patient (1 RV and 2 LV EGM signals; grey) method for

classifying CRTopt (settings ≥ 90% of maximal LV dP/dtmax).

Figure 9.2: ROC-curves for QRSarea derived from the EGM-VCG for classifying CRTopt (set-

tings ≥ 90% of maximal LV dP/dtmax) in LBBB canines with (blue line) and without (red line) ischemia. MI: myocardial infarction.

164 h eut bevdin observed results VCG The surface body the and VCG Device-based 9.7.2 al. uigtee3mnh,btti i o nuneteaiiyo h R etrt piiethe optimize to vector QRS the of ability the influence not AV-delay. did this but place months, taken 3 already these have during might remodeling myocardial Some implantation. CRT after months pia Vdly nadto,otmzto norsuywspromddrn R imal. CRT et during Deursen van the performed by of was performed estimation study study accurate the our in more in while a optimization plantation in addition, resulting In ratio study AV-delay. signal-to-noise our measures optimal lower outcome in a of whereas combination in A pressures result based. might blood pressure finger purely was and echocardiography response of integral) hemodynamic combination the time a velocity used they tract response, outflow of (LV measure acute the As patients. in aigeetoe.Ti hrcmn a esle yadn oeeetoe nteleads, pacing. the of on sites electrodes the more from adding distance the by to larger close solved at are be especially EGM-VCG may the shortcoming derive to This used electrodes. electrodes three pacing all that QRS fact derived the EGM-VCG by the explained be in observed are changes no still changes no almost protocol, pacing BiV QRS the for During observed (BiV) were 9.3). biventricular (Fig. simultaneous investigated during also settings was device pacing CRT the optimizing in EGM-VCG C n G-C aibe rvddago siaino h pia Vdly con- VV-delay, al. optimal et Deursen the van QRS of by optimization estimation VV-delay the during good both observations Therefore, the a firming provided measured. be variables could EGM-VCG electrode depolarization and ventricles, pacing VCG the the of from activation directly between than delay other the waves to Due ms. A-RV-60 to set was Vpeectto y4 si tp f2 s uigaldi all During ms. until 20 ms of 60 steps by in pre-excitation ms interventricular LV 40 the between by varied changing pre-excitation VV-delay RV by The pacing 9.4). BiV Fig. (VV-delay; sequential delay during investigated also was VCG V dly ogrta h nrni ciaino h V h nrni ciaino h right At QRS the the of in pattern. activation changes activation intrinsic to leading the unchanged RV, contributing is the an branch of in bundle activation resulting originating intrinsic only the electrodes, than is pacing longer activation LV AV-delays electrical and RV the AV-delays, the short from relatively these At A-RV. the msac responders’. ‘mismatch eieL-nypcn,a ersne in represented as pacing, LV-only Beside nalptet nlddfrtesuydsrbdin described study the for included patients all In 43 area h hwdta h R etri eibeto o piiigsiuainintervals stimulation optimizing for tool reliable a is vector QRS the that showed who n QRS and ampl eeal oetmt h ea ihhgetL d LV highest with delay the estimate to able were area hpe 8 chapter n QRS and ampl r nln ihterslsrpre yvnDusnet Deursen van by reported results the with line in are ne odtoswe h Vdlywssotrthan shorter AV-delay was the when conditions under hpe 8 chapter hpe 8 chapter h efrac fteVGand VCG the of performance the , area ff area rn Vdly,teAV-delay the VV-delays, erent 43 h osberl fteEGM- the of role possible the eie rmteVGwhile VCG the from derived h atrosraincan observation latter The . twspromda es 3 least at performed was it P eea discussion General / d 43 t max Interestingly, vni the in even 165

9 Chapter 9

Figure 9.3: Examples of the same three patients as in Fig. 8.4: one non-responders, one mismatch-

responder, and one match-responder. Relative LV dP/dtmax and LV systolic pressure increase at various AV-delays during BiV pacing. Corresponding changes in QRS area and amplitude extracted from the VCG or EGM-VCG are shown. AAI-pacing (∼LBBB) results are illustrated by a green diamond. Bold AV-delays and the corresponding vertical dashed lines indicate the individual onset of RV activation. 166 iue9.4: Figure orsodn hne nQSae n mltd xrce rmteVGo EGM-VCG or VCG ( the AAI-pacing from shown. extracted are amplitude and area QRS in pre-excitation. changes RV Corresponding indicates VV-delay positive a pre-excitation, LV indicates VV-delay tive imthrsodr n n ac-epne.TeA-ea a e oAR-0ms. A-RV-60 one to non-responders, set one was AV-delay 9.3: d Fig. The LV and Relative 8.4 match-responder. Fig. one in and as patients mismatch-responder, three same the of Examples P / d t max n Vssoi rsueices tvrosV-eas nega- A VV-delays. various at increase pressure systolic LV and ∼ BB eut r lutae yagendiamond. green a by illustrated are results LBBB) eea discussion General 167

9 Chapter 9

9.7.3 Mismatch between LV dP/dtmax and LV systolic pressures measures

Even though the QRS vector was able to ﬁnd the optimal AV-delay when LV systolic pressure is the outcome measure, this was not the case when using LV dP/dtmax as outcome measure. This discrepancy was mainly due to the fact that in the patient population presented in chapter 8, there was a ‘mismatch’ subgroup in which the optimal AV-delay according to the maximal increase in LV dP/dtmax and the AV-delay matching the maximal increase in LV systolic pressure did not match (n = 8). In fact, in these patients the AV-delay corresponding to a maximal increase in LV dP/dtmax was very short. The only observed diﬀerence between these ‘mismatch’ patients and the ‘match’ patients is that none of the ‘mismatch’ patients had ischemic cardiomyopathy, whereas in the ‘match’ patients, 7 out of 10 had ischemic cardiomyopathy.

In the literature, most studies show a bell-shaped curve for LV dP/dtmax as function of 44–46 AV-delay. However, in some patients in these studies, LV dP/dtmax only decreases to a limited extent at very short AV-delays compared to their maximal increase in LV dP/dtmax. Whinnett et al. 47 observed an unphysiologically short optimal AV-delay of 40 ms for some patients. However, they did not provide an explanation for this short AV-delay and it is unknown whether these patients with a very short optimal AV-delay also showed a mismatch between LV dP/dtmax and LV systolic pressure. When comparing the relative increases in LV systolic pressure by CRT in our patient study with those from others, our values were smaller. This may be explained by the fact that others measured the relative change in LV systolic pressure measured at the transition between a reference and tested AV-delay43,48,49, whereas we measured this relative change in LV systolic pressure at a certain delay after the transition. This delay allows some time for stabilization before recording but also allows the start of the influence of the baroreflex, leading to smaller differences in LV systolic pressure.

9.7.4 Future application of the EGM-VCG for optimization of the CRT device settings

To some extent the idea of using electrical parameters for optimization has already been used by several device manufacturers to automatically optimize CRT device settings. The SmartDelay algorithm calculates the optimal AV-interval using an equation that was formed based on the results of several clinical studies and uses as input the intrinsic sensed or paced AV-interval, QRS duration, and a series of constants.50 The QuickOpt method is an empirical method also resulting from clinical observations, in which the AV-interval is calculated as the width of atrial intrinsic depolarization added to an oﬀset factor.51 This oﬀset is equal to 30 ms when the electrogram duration is > 100 ms and 60 ms when it is smaller than 100 ms, which

168 h xc -V loi hsmto,tedi find the again method, to AV-delay this automatic programmed in this the Also A-RV. During around exact A-RV varied week. the of a be determinations once will AV-delay multiple instance the Therefore, for period performed, measurement, a heart). be over the could change of EGM-VCG may optimal remodeling the LV individual to using and due an RV with between (e.g. but conduction time that performed, of possible be is still present It now already 2) can process AV-delay. algorithm optimization CRT continuous adaptive The the device. in CRT the into programmed be sazga eainhp hrfr,sm ainswt ogEMdrto aetesame di around the very duration have have EGM duration with can patients EGM ms Furthermore, long 100 duration. with short patients a with some those Therefore, as AV-interval relationship. zigzag a is aigbsdo eidcabltr vlaino nrni lcrclcnuto intervals CRT conduction adjusts electrical that method intrinsic ADAPTIVECRT A of CRT. the evaluation relatively for is ambulatory optimize rate periodic to response on able higher based being a pacing in as result well potentially as parasympathetic could optimization and fast sympathetic continuous the The of activation to conduction system. due when nervous change exercise, may physical heart during the arise appa- of may and properties condition method similar loop A closed a adapting. from may for AV-interval benefit ratus optimal would the thus and Furthermore, during visits therapy. otherwise physician because and between or or change treatment failure event failure heart heart of decompensation of progression failure course the heart the disease in occurs patient’s acute that The an remodeling to deleterious activity. due of patient example and for time evolves, benefit with clinical state change long-term full may into settings translate not optimal may because this while once, only delays the optimize h envle ae rmagopo ainsadbsdo esrmnswe h CRT the when measurements on based and patients of group o a was from taken values mean the ainprmtrdrn aigfrec ain niiuly sn h QRS the Using individually. patient each for pacing during parameter zation rpeep -Ves ih4 s h xc di A-RVsense exact of the 70% ms, taking of 40 ECG Instead with regular A-RVsense calculated. the pre-empt be using or can after, vector shortly QRS or maximal implant the of which time from at A-RV, of determination single A 1) ADAPTIVECRT the individualize factors. these on dependent While not is determined. EGM-VCG be can activation ADAPTIVECRT RV the intrinsic in of used contribution equation of the onset exact the plot, oainadltnydrn Vpcn,teQRS the pacing, LV during latency and location ARsne n speepe ya es 0m r7%o h oa interval. total the of 70% or ms 40 sensing RV least and at AV- sensing by the or pre-empted pacing studies, is atrial earlier and between in (A-RVsense) interval found intrinsic the relation as a calculated on is based interval AV-interval optimal of definition a uses hl h bv ecie loihsuepeitoso pia uinetaoae from extrapolated fusion optimal of predictions use algorithms described above the While h niiuladeatAR on sn h C rEMVGcudb sdto used be could EGM-VCG or VCG the using found A-RV exact and individual The ff u ehduigteEMVGo C rcstepeitdotmlresynchroni- optimal predicted the traces VCG or EGM-VCG the using method our , TM ff rn Vitras ohteSatea n ucOtmethod QuickOpt and SmartDelay the Both AV-intervals. erent taeako etoi,Ic)agrtm hsagrtmalso algorithm This algorithm. Inc.) Medtronic, of (trademark TM loih vnmr.Oecudiaietooptions: two imagine could One more. even algorithm TM ff rnebtenARsneadAR ilbe will A-RV and A-RVsense between erence loih eed mn teso Vlead RV on others among depends algorithm ff rnebtenARsneadAR can A-RV and A-RVsense between erence ampl ihrdrvdfo h C rthe or VCG the from derived either eea discussion General 52 ampl − AV-delay 169

9 Chapter 9 programmed into the device.

9.8 Conclusions

Throughout this thesis, we have tried to ﬁnd ways to reduce the number of non-responders to cardiac resynchronization therapy using vectorcardiography. We showed that repolarization might play an important role in how the heart adapts to LBBB as well as in the patient selection for CRT. Furthermore, for those patients suitable for CRT, a device-based VCG can be extracted from the leads which could eventually be used for continuous and individual optimization of the stimulation intervals.

170 References 0 REgl hh ADPdsa SFak,adR rue -aeabnormalities T-wave Krause. RL and Frankl, WS DePodesta, LA Shah, R Engel, TR 10. .KCatre,AHri,GDve,adALahm lcrcrigahccagssub- changes Electrocardiographic Leatham. A and Davies, G Harris, A Chatterjee, K 9. .ACsadJcl,BGesh nz n RFaz lwadln-atn mod- long-lasting and Slow Franz. MR and Antz, M Goetsch, B Costard-Jäckle, A 8. .M ulr eaa,adD oebu.Criceetia eoeigi elhand health in remodeling electrical Cardiac Rosenbaum. DS and Jeyaraj, D Cutler, MJ 7. Electrotonic Davidenko. JM and Lázzari, JO Elizari, MV Blanco, HH Rosenbaum, MB of conversion for 6. method a ECGScan: Moss. AJ and Zareba, W Erdem, T Badilini, F Schalij, MJ 5. Wall, der van EE Cannegieter, SC Man, SC Algra, AM Schreurs, CA 4. .SMn MAga ASher,CWBorle CJW Schreurs, CA Algra, AM Man, S 1. .D otzadT clgl hndrvn h pta R- nl rmte12-lead the from angle QRS-T spatial the deriving When Schlegel. TT and Cortez DL 2. .J os a epn CSti,adJ a eml eosrcino h Frank the of Reconstruction Bemmel. van JH and Sittig, AC Herpen, van G Kors, JA 3. 206. block. bundle-branch left intermittent of depolarization. ventricular artificial to sequent lto fmoada eoaiainpoue yetpcatvto nioae rabbit isolated "memory". in cardiac activation for ectopic Evidence by hearts. produced repolarization myocardial of ulation disease. memory. cardiac 50:213–222. and (1982) wave T the of modulation files. diol electrocardiographic digital to printouts electrocardiographic paper vectorcardiogram: electrocardiogram. Frank 12-lead the the 43:294–301. in (2010) from angle derived QRS-T estimates spatial of The accuracy Swenne. CA and Kors, JA lcrcriga,wihtasomi oeFak ersino nes Dower? inverse or regression Frank: Electrocardiol more is transform which electrocardiogram, etradormfo tnadeetoadorpi ed:dansi oprsnof comparison di diagnostic leads: electrocardiographic standard from vectorcardiogram R- nl opeitaryhisi ainswt shmchatdsaeadsystolic vec- and disease the dysfunction. heart ventricular ischemic of with left patients Influence spatial in electrocardiogram-derived arrhythmias predict the Swenne. to of angle CA power QRS-T and the on Schalij, matrix MJ synthesis Cannegieter, torcardiogram SC Wall, der van ff rn methods. erent 20)38:310–318. (2005) rnsPamclSci Pharmacol Trends 21)43:302–309. (2010) u er J Heart Eur Electrocardiol J 21)32:174–180. (2011) 19)11:1083–1092. (1990) Circulation naso nenlmedicine internal of Annals 21)44:410–415. (2011) ff rts er journal heart British ,RCShrtn,LvnEvn EE Erven, van L Scherptong, RWC s, h mrcnjunlo cardiology of journal American The 18)80:1412–1420. (1989) 16)31:770. (1969) Electrocardiol J 17)89:204– (1978) Electrocar- J J

9 Chapter 9

11. P Nicolai, J Medvedowsky, M Delaage, C Barnay, E Blache, and A Pisapia. Wolff- Parkinson-White syndrome: T wave abnormalities during normal pathway conduction. Journal of electrocardiology (1981) 14:295–300. 12. L Wecke, CJ van Deursen, L Bergfeldt, and FW Prinzen. Repolarization changes in patients with heart failure receiving cardiac resynchronization therapy-signs of cardiac memory. Journal of electrocardiology (2011) 44:590–598. 13. A Shvilkin, B Bojovic, B Vajdic, I Gussak, P Zimetbaum, and ME Josephson. Vec- torcardiographic determinants of cardiac memory during normal ventricular activation and continuous ventricular pacing. Heart Rhythm (2009) 6:943–948. 14. L Wecke, A Rubulis, G Lundahl, MR Rosen, and L Bergfeldt. Right ventricular pacing- induced electrophysiological remodeling in the human heart and its relationship to cardiac memory. Heart Rhythm (2007) 4:1477–1486. 15. A Shvilkin, B Bojovic, B Vajdic, I Gussak, KK Ho, P Zimetbaum, and ME Josephson. Vectorcardiographic and electrocardiographic criteria to distinguish new and old left bundle branch block. Heart Rhythm (2010) 7:1085–1092. 16. D Jeyaraj, LD Wilson, J Zhong, C Flask, JE Saffitz, I Deschênes, X Yu, and DS Rosen- baum. Mechanoelectrical feedback as novel mechanism of cardiac electrical remodeling. Circulation (2007) 115:3145–3155. 17. E Hermeling, T Delhaas, FW Prinzen, and NH Kuijpers. Mechano-electrical feedback explains T-wave morphology and optimizes cardiac pump function: Insight from a multi-scale model. Progress in biophysics and molecular biology (2012) 110:359– 371. 18. OS Narula. Longitudinal dissociation in the His bundle. Bundle branch block due to asynchronous conduction within the His bundle in man. Circulation (1977) 56:996– 1006. 19. DL Lustgarten, S Calame, EM Crespo, J Calame, R Lobel, and PS Spector. Electrical resynchronization induced by direct His-bundle pacing. Heart Rhythm (2010) 7:15–21. 20. R Barba-Pichardo, AM Sánchez, JM Fernández-Gómez, P Moriña-Vázquez, J Venegas-Gamero, and M Herrera-Carranza. Ventricular resynchronization therapy by direct His-bundle pacing using an internal cardioverter defibrillator. Europace (2013) 15:83–88. 21. P Houthuizen, RMA van der Boon, M Urena, N Van Mieghem, GBR Brueren, TT Poels, LAFM Van Garsse, J Rodés-Cabau, FW Prinzen, and P de Jaegere. Occurrence, fate and consequences of ventricular conduction abnormalities after transcatheter aortic valve implantation. EuroIntervention (2014) 9:1142–1150. 22. TM Nazif, MR Williams, RT Hahn, S Kapadia, V Babaliaros, J Rodés-Cabau, WY Szeto, H Jilaihawi, WF Fearon, D Dvir, et al. Clinical implications of new-onset left bundle branch block after transcatheter aortic valve replacement: analysis of the PART-

172 0 J a ere,KVroy uik eged,HG rjs WPizn and Prinzen, FW Crijns, HJGM Bergfeldt, L Dudink, E Vernooy, K Deursen, van CJM 30. 1 a a,GMWjtes BEgl,YBau,JL uras io,HJ Pison, L Luermans, JGLM Blaauw, Y Engels, EB Wijntjens, GWM Rad, Mafi M 31. 2 BEgl,E éh J a ere,KVroy PSnh n WPizn T- Prinzen. FW and Singh, JP Vernooy, K Deursen, Van CJM Végh, EM Engels, EB 32. 9 ooo n PLo.Tevnrclrcmlxi etvnrclrhypertrophy ventricular left in left complex of ventricular diagnosis The ECG Lyon. the TP for and system Sokolow point-score M A 29. Estes. EH and Romhilt DW 28. 4 zks MvnDln MVnMehm LGterzCio JNi,FKur C Kauer, F Nuis, RJ Gutierrez-Chico, JL Mieghem, Van NM Dalen, van BM Tzikas, A 24. J Cheung, A Barbanti, M Toggweiler, S Serra, V Cheema, A Webb, JG Urena, M 23. 6 Ho R 26. D DeLarochellière, R Nombela-Franco, L Dumont, E Serra, V Mok, M Urena, M 25. 7 WMcaln,AvnOseo,adMJanse. M and Oosterom, van A Macfarlane, PW 27. ek.VcocrigahcQSae sanvlpeitro epnet cardiac to response of predictor novel a therapy. as resynchronization area QRS Vectorcardiographic Wecke. L leads. limb and precordial 37:161–186. unipolar by obtained as hypertrophy. ventricular 2010. Media, Business & Science Springer 4. Vol. cut,P ery,P eJeee n LGline rqec fcnuto ab- e Medtronic-CoreValve the conduction the and of with Frequency implantation valve Geleijnse. aortic ML transcatheter after and normalities Jaegere, de PP Serruys, PW Schultz, 7:128–136. (2014) valve. bundle balloon-expandable left valve a aortic persistent transcatheter with new-onset undergoing patients implantation of in outcomes Impact clinical al. late on et block branch DeLarochellière, R Dumont, E Ye, experience. NER 21)107:285–289. (2011) mlnainwt alo-xadbevalve. Cardiology balloon-expandable a valve with aortic transcatheter implantation following block clinical branch long-term bundle and left factors persistent Predictive of al. consequences et Amat-Santos, I Larose, E Igual, A Doyle, etvnrclrltrlwl ciaindtrie yeetontmcmpigi can- in therapy. mapping resynchronization cardiac electroanatomic for by didates determined activation delayed wall identifies lateral area ventricular QRS Vectorcardiographic left Vernooy. K and Prinzen, FW Crijns, trari av mlnaino etvnrclrfunction. ventricular transcath- left after defect on conduction Interventions implantation new a valve of aortic Impact eter Lotfi. S and Marx, N Autschbach, aeae rdcsrsos ocricrsnhoiainteayi ainswt left with patients in therapy block. resynchronization branch cardiac bundle to response predicts area wave ff an eprz ofior ku,KBemr emce,R Lehmacher, W Brehmer, K Aktug, Ö Lotfipour, S Herpertz, R mann, ff c nlf etiua jcinfraction. ejection ventricular left on ect 21)60:1743–1752. (2012) 21)5:1257–1263. (2012) uoenhatjournal heart European adoacElectrophysiol Cardiovasc J mrcnhatjournal heart American Electrocardiol J 21)35:1599–1607. (2014) 21)48:45–52. (2015) er Rhythm Heart AC adoaclrInterventions Cardiovascular JACC: 16)75:752–758. (1968) ora fteAeia olg of College American the of Journal h mrcnjunlo cardiology of journal American The 21)26:176–183. (2015) opeesv electrocardiology Comprehensive mrcnhatjournal heart American 21)13:217–225. (2016) AC Cardiovascular JACC: References (1949) 173 .

9 Chapter 9

33. P Houthuizen, LAFM Van Garsse, TT Poels, P de Jaegere, RMA van der Boon, BM Swinkels, JM Ten Berg, F van der Kley, MJ Schalij, J Baan Jr, et al. Left bundle- branch block induced by transcatheter aortic valve implantation increases risk of death. Circulation (2012) 126:720–728. 34. PW Foley, S Chalil, K Khadjooi, RE Smith, MP Frenneaux, and F Leyva. Long-term effects of cardiac resynchronization therapy in octogenarians: a comparative study with a younger population. Europace (2008) 10:1302–1307. 35. A Achilli, F Turreni, M Gasparini, M Lunati, M Sassara, M Santini, M Landolina, L Padeletti, A Puglisi, M Bocchiardo, et al. Efficacy of cardiac resynchronization therapy in very old patients: the Insync/Insync ICD Italian Registry. Europace (2007) 9:732– 738. 36. J Penn, I Goldenberg, AJ Moss, S McNitt, W Zareba, HU Klein, DS Cannom, SD Solomon, A Barsheshet, DT Huang, and MADIT-CRT Trial investigators. Improved outcome with preventive cardiac resynchronization therapy in the elderly: a MADIT- CRT substudy. J Cardiovasc Electrophysiol (2011) 22:892–897. 37. A Vahanian, O Alfieri, F Andreotti, MJ Antunes, G Barón-Esquivias, H Baumgartner, MA Borger, TP Carrel, M De Bonis, A Evangelista, et al. Guidelines on the management of valvular heart disease (version 2012). European heart journal (2012) 33:2451– 2496. 38. K Vernooy, RNM Cornelussen, XAAM Verbeek, WYR Vanagt, A van Hunnik, M Kuiper, T Arts, HJGM Crijns, and FW Prinzen. Cardiac resynchronization therapy cures dyssynchronopathy in canine left bundle-branch block hearts. Eur Heart J (2007) 28:2148–2155. 39. IE van Geldorp, K Vernooy, T Delhaas, MH Prins, HJ Crijns, FW Prinzen, and B Dijk- man. Beneficial effects of biventricular pacing in chronically right ventricular paced patients with mild cardiomyopathy. Europace (2010) 12:223–229. 40. AB Curtis, SJ Worley, PB Adamson, ES Chung, I Niazi, L Sherfesee, T Shinn, and M St. John Sutton. Biventricular pacing for atrioventricular block and systolic dysfunction. New England Journal of Medicine (2013) 368:1585–1593. 41. LG Tereshchenko, A Cheng, J Park, N Wold, TE Meyer, MR Gold, S Mittal, J Singh, KM Stein, KA Ellenbogen, et al. Novel measure of electrical dyssynchrony predicts response in cardiac resynchronization therapy: Results from the SMART-AV Trial. Heart Rhythm (2015) 12:2402–2410. 42. NHL Kuijpers, E Hermeling, J Lumens, HMM ten Eikelder, T Delhaas, and FW Prin- zen. Mechano-electrical coupling as framework for understanding functional remodeling during LBBB and CRT. Am J Physiol Heart Circ Physiol (2014) 306:H1644– H1659.

174 1 HBkr2d ceze3d eu SGer otred eo,SGreen- S Fedor, M Porterfield, J Greer, GS Beau, S 3rd, McKenzie J 2nd, Baker JH 51. 0 RGl,INai idc,R ea,J triat HKm u Ding, J Yu, Y Kim, MH Sturdivant, JL Leman, RB Giudici, M Niazi, I Gold, MR 50. 9 IWint,J ais ilo,A hw AFae WDve,A Hughes, AD Davies, DW Foale, RA Chow, AW Willson, K Davies, JE Whinnett, ZI 49. 8 yicu APbr,Z hnet ri ilo,RBra,BSeean J Stegemann, B Baruah, R Willson, K Arri, S Whinnett, ZI Pabari, PA Kyriacou, A 48. 7 IWint,D rni,ADns ilo,PPsae a edr,MD Guil- De M Geldorp, van I Pascale, P Willson, K Denis, A Francis, DP Whinnett, ZI 47. 6 AVrek uici,YY,JDn,TPce,KVroy rmr Spinelli, J Kramer, A Vernooy, K Pochet, T Ding, J Yu, Y Auricchio, A Verbeek, XA 46. 5 AKs,C hn ur,MTlo,RBre,BFtc,adENv.Ipoe left Improved Nevo. E and Fetics, B Berger, R Talbot, M Curry, C Chen, CH Kass, DA 45. 4 uici,CSelrn,MBok ak ot akr li,AKramer, A Klein, H Bakker, P Vogt, J Sack, S Block, M Stellbrink, C Auricchio, A 44. 3 J a ere,LWce MvnEedne,MSåleg H ase,F Janssen, MHG Ståhlberg, M Everdingen, van WM Wecke, L Deursen, van CJM 43. eg GDod obseo RBie,adLPrefil.Aueeauto of evaluation Acute Porterfield. L and Bailey, JR AV Corbisiero, programmer-guided R Daoud, EG berg, n coadormfrcricrsnhoiainteayi er alr ainsand patients failure implants. heart ICD in dual-chamber therapy resynchronization cardiac for echocardiogram and eoyai piiaindrn ada eycrnzto therapy. resynchronization Electrophysiol for cardiac methods programming during delay AV optimization of hemodynamic comparison prospective A Waggoner. AD and 8:358–366. PFacs n ae.Dtriaino pia tivnrclrdlyfrcardiac pressure. for blood delay non-invasive atrioventricular acute optimal using of therapy Determination resynchronization Mayet. J and Francis, DP aiesml ltymgahcotmzto:pofo ocp n oeta o im- for potential and nonin- concept reproducible, of plantability. proof automatable, Fully optimization: al. plethysmographic et simple vasive Hughes, AD Kanagaratnam, P Mayet, ainteay mlctosfrciia ra einadciia practice. cardiology clinical of and journal design trial clinical for resynchroni- implications cardiac therapy: with zation patients in optimization delay AV for methods hemodynamic eo,SPox lebgn asaure ta.Cmaio fdi of Comparison al. et Haïssaguerre, M Ellenbogen, K Ploux, S lebon, n WPizn alrn ada eycrnzto hrp sn interventricular Physiology Circulatory using model. therapy simple resynchronization a of cardiac Validation asynchrony. Tailoring Prinzen. FW and n etiua odcindelay. cardiomyopathy dilated conduction with ventricular patients and in pacing VDD acute from mechanics ventricular alr.TePcn hrpe o ogsieHatFiueSuyGop h Guidant The Group. Group. Research Study Failure Failure Heart Heart Congestive Congestive for heart Therapies congestive Pacing with The patients paced failure. of function systolic acute on delay atrioventricular ig ao oka,TPce,adJSiel.E Spinelli. J and Pochet, T Tockman, B Salo, R Ding, J therapy. Res resynchronization Transl cardiac Cardiovasc in J Vectorcardi- intervals stimulation Prinzen. of FW optimization and Vernooy, for ography K Crijns, HJGM Bergfeldt, L Braunschweig, aigadCiia Electrophysiology Clinical and Pacing 20)18:490–496. (2007) 21)168:2228–2237. (2013) 20)290:H968–H977. (2006) / VadV ea piiaincmaiga EMmethod IEGM an comparing optimization delay VV and PV 21)8:128–137. (2015) adoacElectrophysiol Cardiovasc J Circulation mrcnJunlo hsooyHatand Physiology-Heart of Journal American Circulation 19)99:1567–1573. (1999) 21)35:948–960. (2012) 19)99:2993–3001. (1999) ff 20)18:185–191. (2007) c fpcn hme and chamber pacing of ect Europace ff rn invasive erent Cardiovasc J International References (2006) 175

9 Chapter 9

52. H Krum, B Lemke, D Birnie, KLF Lee, K Aonuma, RC Starling, M Gasparini, J Gorcsan, T Rogers, A Sambelashvili, A Kalmes, and D Martin. A novel algorithm for individualized cardiac resynchronization therapy: rationale and design of the adaptive cardiac resynchronization therapy trial. Am Heart J (2012) 163:747–752.e1.

176

Summary

n increasing number of people suﬀer from heart failure (HF). In about one quarter of HF patients the electrical conduction pattern is disturbed. One of the most severe A disturbances is left bundle branch block (LBBB) which causes a delayed electrical impulse conduction through the left ventricle (LV). In LBBB, the electrical activation starts in the septum and right ventricle (RV) and proceeds through the slowly conducting septum to activate the LV. Cardiac resynchronization therapy (CRT) is an established treatment for HF patients with LBBB and comparable conduction disturbances. Since initial approval of the therapy 15 years ago, there have been hundreds of thousands of implants worldwide. On the group level, CRT has been shown to improve LV systolic function, reduce the amount of HF hospitalizations, and improve survival rates. However, 30-50% of the patients do not respond signiﬁcantly to the therapy. Such lack of response is not desirable for a treatment which involves a relatively expensive therapy and that requires essentially irreversible life- long implantation of a device.

Improvement of response to CRT

The main aim of this thesis is to improve the response to CRT using vectorcardiographic (VCG) analysis. To this purpose, the underlying disease LBBB is investigated in more detail as well as the role of improved patient selection and better therapy application. In the past, already a great deal of research has been performed towards improved selection of the appropriate candidates for CRT. An extensive overview regarding this matter is presented in chapter 2. There it is reviewed that the currently used electrocardiogram (ECG) selection criteria, QRS duration and morphology, both have their shortcomings: QRS duration may not be speciﬁc enough, while LBBB criteria may be too complex and/or user- dependent. The VCG derived QRSarea might be a good alternative measure to improve patient selection, because it provides an objectively determined index with continuous values and because QRSarea reﬂects LV activation delays, the primary substrate for CRT. Small studies showed better performance of prediction of CRT response by QRSarea compared to QRS duration. Of note, in these studies QRSarea was determined with a dedicated system to measure the (‘Frank-’)VCG. In order to further investigate the role of the VCG in improving the response to CRT, it would be easier to synthesize the VCG from the often employed 12-lead ECG. A few studies had already demonstrated in healthy subjects that a VCG can be calculated from the 12-lead ECG using the Kors matrix. In chapter 3, we showed that also in CRT candidates the Kors- VCG nicely resembles the Frank-VCG. This enables retrospective as well as prospective VCG analysis of routinely recorded 12-lead ECGs, which is used in all other chapters in the present thesis.

Electrical remodeling in patients with LBBB

It is known that patients with LBBB have a high response rate to CRT. The way the heart adapts electrically to a new situation of LBBB is unknown mainly because in the general population the onset of LBBB is often unknown. In chapter 4 we investigated the electrical adaptations to LBBB in patients who developed LBBB during a transcatheter aortic valve implantation (TAVI) procedure. In this procedure LBBB is an undesired complication, occurring in approximately one third of all TAVI patients, presumably caused by the valve pressing against the left bundle branch. For our purposes, the advantage of the developed LBBB is that the exact onset time of LBBB is known. Therefore, these patients can be used as a ‘human model’ to study the time course of electrical remodeling with LBBB. In the majority of patients (67 out of 107 patients) who developed LBBB during a TAVI procedure, electrical remodeling occurs in the ﬁrst month of LBBB, as evidenced by a reduction in various repolarization variables (like JTc interval, T-wave amplitude and Tarea). However, the extent of electrical remodeling was highly variable between patients. The role of the extent of electrical remodeling for CRT response and patient prognosis in general requires further investigation.

The role of repolarization in patient selection for CRT

In chapter 5 and 6 the potential role of repolarization in the patient selection for CRT was investigated. In a study population of 244 CRT recipients, retrospective VCG analysis revealed that the responders to CRT have a larger area under the curve for the T-wave (Tarea; chapter 5) and the predictive value of the Tarea was even better than for QRSarea. The Tarea was especially of added value in the patients diagnosed with LBBB. In a larger patient population consisting of 335 CRT recipients, again the patients with a large Tarea and LBBB less frequently reached the combined endpoint of HF hospitalization, heart transplantation, left ventricular assist device implantation or death after 3 years compared to the other patients

(chapter 6). The predictive power of Tarea may be due to a combination of a reﬂection of good intrinsic myocardial properties as well as a reﬂection of a substrate for CRT, i.e., in the

LBBB patients with a large Tarea, females and patients with non-ischemic HF were overrep-

180 Summary

resented. The combined analysis of QRS morphology and Tarea provides an easy and widely applicable approach to improve selection of CRT candidates.

Tailor-CRT: optimizing stimulation intervals

Once a patient is selected for CRT and the implantation has been performed, the timing of atrial to RV and LV stimulation could be optimized to achieve maximal benefit from CRT. The different timings affect the amount of ventricular filling and timing of contraction (atrioventricular [AV] interval during biventricular [BiV] pacing) and the amount of resynchronization (AV-interval during LV only pacing, and interventricular [VV] interval during BiV pacing). Optimization of these timings can be performed using different measures such as echocardiography or hemodynamic measures. However, all measurements are often time-consuming and are limited in their accuracy, contributing to the scarce evidence of long-term benefit of CRT optimization. Therefore, most physicians leave the timings at the ‘out-of-the-box’ settings. VCG has been shown to be more reliable in finding the optimal timing settings than the tra- ditional hemodynamic measurements. However, these measurements can only be performed during in-office visits. While a single optimization is probably valuable, regular optimization, preferably in an automated fashion, may be more desirable. Therefore, in chapter 7 the possibility to extract a VCG from the electrograms (EGMs) derived from unpaced implanted electrodes (EGM-based VCG, EGM-VCG) was investigated in dogs. In these dogs the proximal left bundle branch was ablated, creating a typical LBBB activation pattern. In eight dogs with chronic LBBB the EGM-VCG derived QRSarea predicted hemodynamic response as measured by LV dP/dtmax and identified optimal settings accurately. A minimal

QRSarea, indicating maximal cancelation of opposing electrical forces, coincided with the optimal settings. These data support the potency of EGM-VCG as a non-invasive and easy tool to individually and continuously optimize the AV- and VV-intervals in CRT. Subsequently, these results were validated in patients receiving CRT (chapter 8). The study population consisted of 25 patients whom represented a typical CRT population with mostly males and half of the patients having ischemic cardiomyopathy. However, the results regarding the optimization guided by EGM-VCG derived QRSarea in these patients were more complicated than those observed in the dogs. A minimal QRSarea did not correspond to a maximal hemodynamic response as deﬁned by LV dP/dtmax, but a better match was found when LV systolic pressure was taken as index of hemodynamic response. Possibly the most practically applicable ﬁnding was that the onset of contribution of intrinsic RV activation could be determined from both the VCG and EGM-VCG. This point in time, is required when optimizing CRT in the ‘LV fusion pacing’ mode: using RV activation from the intact RV conduction system in combination with LV only pacing.

181 Conclusions

The studies presented provide evidence that the relatively old method of VCG can be used to improve the response of the new technique of CRT. The studies presented provide evidence that the relatively simple and widely available measures of VCG-derived QRSarea can be used to improve the patient selection, and that besides the importance of depolarization, repolarization also plays a signiﬁcant role. Furthermore, an EGM-VCG can be extracted from unpaced electrodes which could be used for continuous and individual optimization of the stimulation intervals, especially in the setting of LV fusion pacing.

182

Samenvatting

teeds meer mensen in de westerse wereld leiden aan hartfalen. In ongeveer een kwart van deze patiënten gaat dit gepaard met een geleidingsstoornis in de hartkamers S (ventrikels). Eén van de meeste voorkomende geleidingsstoornissen is linker bundel- takblok (LBTB). Bij een LBTB start de elektrische activatie in de rechter ventrikel (RV) en het kamertussenchot (septum), waarna de linker ventrikel (LV) via het langzaam geleidende septum vertraagd geactiveerd wordt. Sinds ongeveer 15 jaar is cardiale resynchronisatie therapie (CRT) een belangrijke behandeling voor patiënten die symptomatisch hartfalen hebben in combinatie met een geleidingsstoornis. Inmiddels zijn er wereldwijd honderdduizenden CRT-implantaties uitgevoerd. Gemiddeld zorgt CRT voor een verbeterde LV-pompfunctie, minder opnames voor hartfalen en een verbeterde overlevingskans. Echter, bij 30-50% van de patiënten leidt CRT niet tot duidelijke verbeteringen. Aangezien CRT de implantatie van een pacemaker vereist, die bijkomende kosten en risico’s op complicaties met zich meebrengt, is het wenselijk om de pacemaker alléén te implanteren bij die patiënten die er ook baat bij zullen hebben en om de therapie te verbeteren.

Verbeteren van respons na CRT

De hoofddoelstelling van dit proefschrift is om met behulp van vectorcardiografie de respons na een CRT-implantatie te verbeteren. Om het succespercentage te verbeteren wordt de onderliggende ziekte LBTB in meer detail onderzocht, alsook de mogelijke rol van het vectorcardiogram (VCG) in de selectie van patiënten en in de verbetering van het effect van CRT. In het verleden is reeds veel onderzoek gedaan naar de optimale selectie van CRT- kandidaten. Een uitgebreid overzicht van deze studies en hun bevindingen staat beschreven in hoofdstuk 2. In dit hoofdstuk worden de tekortkomingen van de op dit moment gebruikte criteria, QRS-duur en -morfologie, besproken. QRS-duur is waarschijnlijk niet specifiek genoeg, terwijl de vaststelling van de aanwezige QRS-morfologie subjectief en complex is. Er zijn aanwijzingen dat gebruik van de oppervlakte van het QRS-complex (QRSarea), gemeten met behulp van het VCG, de patiënt selectie zou kunnen verbeteren. QRSarea is een objectieve en continue maat die mogelijke vertraging in LV-activatie weerspiegelt, het primaire substraat voor CRT. Tot nu toe werd de QRSarea bepaald met behulp van het zelden gebruikte (‘Frank-’)VCG. Voor toekomstige studies zou het praktisch zijn als het VCG kon worden berekend uit het standaard 12-afleidingen elektrocardiogram (ECG). In hoofdstuk 3 laten we zien dat in kandidaten voor CRT het Frank-VCG en Kors-VCG erg op elkaar lijken. Hierdoor kunnen zowel retrospectieve als prospectieve VCG-analyses worden uitgevoerd met behulp van de routinematig opgenomen 12-afleidingen ECG. Van deze techniek wordt dan ook gebruikgemaakt in alle volgende hoofdstukken van dit proefschrift.

Elektrische remodellering in patiënten met LBTB

De belangrijkste doelgroep voor CRT zijn patiënten met LBTB. Echter, de manier waarop het hart zich aanpast na het ontstaan van een LBTB is onbekend, omdat het exacte aanvangsmoment van LBTB vaak niet bekend is. In hoofdstuk 4 hebben we onderzocht hoe het hart zich elektrisch aanpast aan de nieuwe LBTB-situatie in patiënten die een LBTB ontwikkelen tijdens een transkatheter aortaklep implantatie (TAVI). LBTB ontstaat in een deel van de patiënten die een TAVI-procedureondergaan. Bij hen drukt de nieuwe aortaklep vermoedelijk tegen de linker bundeltak, waardoor een ‘typisch’ LBTB ontstaat. Voor onze studie hebben deze TAVI-patiënten het voordeel, dat bij hen het exacte aanvangsmoment van LBTB bekend is. Hierdoor kunnen deze patiënten gebruikt worden als een ‘menselijk’ model van LBTB om het tijdsverloop van elektrische remodellering te onderzoeken. In de meerderheid van de patiënten die tijdens een TAVI procedure LBTB ontwikkelt (67 van de 107 patiënten) treedt er binnen één maand elektrisch remodelering op, zoals blijkt uit een afname in meerdere repolarisatie variabelen (JTc-interval, T-golfamplitude, en oppervlakte van de T-golf [Tarea]). De mate van elektrische remodellering was echter zeer variabel tussen patiënten. Daarnaast is het nog onbekend wat de rol van de mate van elektrische remodellering is voor de response na CRT, alsook voor de algemene prognose voor de patiënt.

De rol van de repolarisatie in patiëntenselectie voor CRT

In hoofdstuk 5 en 6 werd de potentiële rol van VCG in de patiëntenselectie voor CRT onderzocht. In 244 CRT-kandidaten werd retrospectief een VCG-analyse uitgevoerd en hieruit bleek dat Tarea groter was in patiënten met echocardiograﬁsche verbetering door CRT dan in patiënten zonder deze verbetering (hoofdstuk 5). Tarea had zelfs een betere voorspellende waarde dan QRSarea en had de best voorspellende waarde in LBTB-patiënten. In een grotere groep van 335 CRT-kandidaten bleek ook dat patiënten met een grotere Tarea in 3 jaar minder vaak het gecombineerde eindpunt van opname voor hartfalen, harttransplantatie, steunhart, of dood bereikten (hoofdstuk 6). Deze voorspellende waarde van Tarea zou verklaard kunnen worden doordat vrouwen en patiënten met niet-ischemisch hartfalen relatief een grotere Tarea

186 Samenvatting lieten zien. Van deze groepen is bekend dat ze in het algemeen goed reageren op CRT. De gecombineerde analyse van QRS-morfologie en Tarea biedt een eenvoudige en breed toepasbare aanpak om de selectie van CRT-kandidaten te verbeteren.

Tailor-CRT: het optimaliseren van stimulatie-intervallen

Als een patiënt eenmaal geselecteerd is voor CRT en de implantatie is uitgevoerd, kunnen de tijdsintervallen tussen de stimulatie van de rechterboezem (atrium) en de RV of LV worden geoptimaliseerd om maximaal voordeel te bereiken van CRT. De verschillende timings beïn- vloeden de ventriculaire vulling en het moment van ventrikelcontractie, alsmede de mate van resynchronisatie. Voor het optimaliseren van de atrioventriculaire (AV-) en interventriculaire (VV-) intervallen is een maat voor hartfunctie nodig. Deze kan bijvoorbeeld uit echocardio- graﬁe afgeleid worden, of uit hemodynamische kathetermetingen. Alle bestaande metingen zijn echter tijdrovend en beperkt in nauwkeurigheid, waardoor er weinig bewijs is voor de lange termijn voordelen van CRT-optimalisatie. Hierdoor gebruiken de meeste artsen vaak de standaardinstellingen van het CRT-apparaat. Het VCG is een betrouwbaar alternatief voor het optimaliseren van het AV- en VV-interval in vergelijking met hemodynamische metingen. Het nadeel van deze VCG-metingen is dat ze nog steeds alleen kunnen worden uitgevoerd wanneer de patiënt het ziekenhuis bezoekt. Hoewel een eenmalige optimalisatie waarschijnlijk al waardevol is, zou een regelmatige en bij voorkeur automatische optimalisatie wenselijker zijn. Daarom werd er in hoofdstuk 7 in honden onderzocht of de pacemaker-draad elektroden die niet gebruikt worden om te stimuleren, gebruikt kunnen worden om de elektrische activiteit van het hart te meten. Uit deze metingen kan vervolgens ook een soort vectorcardiogram worden geëxtraheerd (het electrogram-VCG, EGM-VCG). In 8 honden werd een typisch LBTB gecreëerd door ablatie van de proximale linker bundeltak. De veranderingen in QRSarea berekend uit het EGM-

VCG en veranderingen in LV dP/dtmax kwamen overeen en QRSarea voorspelde de optimale stimulatie intervallen. De kleinste QRSarea, overeenkomend met maximale opheﬃng van te- genovergestelde elektrische krachten van LV en RV, voorspelde de optimale stimulatie intervallen goed. Deze gegevens ondersteunen het gebruik van EGM-VCG als een niet-invasieve en eenvoudige methode om de AV- en VV-intervallen voor CRT individueel en continu te optimaliseren. Deze bevindingen in honden werden vervolgens gevalideerd in patiënten (hoofdstuk 8). De 25 patiënten geïncludeerd in dit onderzoek representeerden een typische CRT-populatie met veelal mannen en waarin de helft van de patiënten een ischemische cardiomyopathie had.

In deze patiëntenstudie kwam een minimale QRSarea niet overeen met een maximale toename in LV dP/dtmax, maar wel met een maximale toename in de LV systolische druk. Echter, misschien wel de meest praktisch toepasbare bevinding was dat het begin van de bijdrage van

187 de intrinsieke RV activatie aan de totale ventrikel activatie tijdens LV stimulatie kan worden bepaald uit zowel het VCG als het EGM-VCG. Dit begin van RV activatie is van belang als men gebruik wil maken van zogenaamde LV-fusiestimulatie. Hierbij dient de LV simultaan met de intrinsieke RV-activatie te worden gestimuleerd, een methode waarvan steeds meer duidelijk wordt dat dit een goede CRT response geeft.

Conclusies

De studies beschreven in dit proefschrift tonen aan dat de relatief oude VCG-methode kan worden gebruikt om de respons na de relatief nieuwe CRT te verbeteren. De studies laten zien dat de eenvoudige en ruim beschikbare QRSarea als maat kan worden gebruikt om de selectie van patiënten voor CRT te verbeteren en dat naast depolarisatie ook repolarisatie een belangrijke rol speelt. Verder kan een EGM-VCG worden geëxtraheerd uit geïmplanteerde elektroden die niet gebruikt worden voor het stimuleren, en kan EGM-VCG gebruikt worden voor continue en individuele optimalisatie van stimulatie-intervallen, voornamelijk tijdens LV-fusiestimulatie.

188

Valorization

n the Netherlands, approximately 150,000 people suffer from heart failure (HF). An- nually approximately 7000 patients die due to HF and there are around 30,000 HF I hospitalizations.1 These hospital admissions and treatment of patients with severe symptoms are associated with high costs. In 2007 the costs for HF amounted 455 million euros, of which 60% was due to hospitalizations. In approximately 25% of all HF patients2, the electrical conduction pattern is disturbed leading to a dyssynchronous electrical activation and contraction of the right and left ventricle (LV), resulting in a reduced pump function. In 2001 a new therapy was approved by the FDA, cardiac resynchronization therapy (CRT), which aims to resynchronize the right and left ventricle leading to a better pump function of the heart. In the last few years, the CRT implantation rate was >2000/year in the Netherlands alone.3 CRT has been shown to improve cardiac pump function and quality of life, and reduce HF symptoms, hospitalization, and mortality at the population level.4 However, benefits at the individual level vary considerably. On the one hand, ∼20% of patients that are implanted according to current guidelines5,6 show complete normalization of LVEF whereas a significant portion (30-50%) of patients benefit little from this therapy.7 Such lack of response is especially undesirable since CRT requires the implantation of a costly device during an invasive procedure. Improved patient selection and optimal programming of the CRT device could lead to a higher response rate to CRT.

Improving patient selection

In this thesis it was shown that the baseline value of QRSarea and Tarea, both indices derived from the vectorcardiogram (VCG), are good predictors of response to CRT. They perform better than the established criteria QRS duration and certain LBBB criteria. Patients with a higher baseline value of QRSarea or Tarea have a higher change of CRT response, making both variables good measures to improve patient selection for CRT. The advantage of these measures is that they are easily obtained and widely applicable. As was pointed out in chapter 3 it is justified to synthesize a VCG from a 12-lead electrocardiogram (ECG) using the Kors matrix. This conversion can be applied to almost any 12-lead ECG, with the only requirement being that either a common running lead or a common reference point, such as a pacing artefact, must be present. Ideally, the ECGs are stored digitally but semi-digital ECGs as stored in a pdf-file or even paper ECGs could also be used to calculate the QRSarea and Tarea. As we described in this thesis, the pdf-files contain vector- graphics that embed the digital information of the displayed ECG. The digital information can thus be extracted, making it possible to synthesize a VCG and calculate the QRSarea and

Tarea. For scanned ECGs, the program ECGscan (AMPS LLC, New York, NY, USA) can be used to digitize the ECGs. This enables retrospective as well as prospective VCG analysis of routinely recorded 12-lead ECGs in large patient populations.

VCG indices like QRSarea and Tarea can also be calculated in a prospective manner. Most commercially available ECG machines already have algorithms to construct a VCG from standard 12-lead ECGs and the beginning and ending of the QRS complex and T-wave are often indicated. The excellent predictive power of QRSarea and Tarea for CRT response indicates that these parameters deserve to be applied more frequently in clinical practice to identify appropriate candidates for CRT. Furthermore, if the predictive power of these VCG indices is proven in larger clinical trials, they may be included in the oﬃcial guidelines as a selection criterion for CRT implantation. The present thesis provides important supportive data for this introduction.

Optimal programming of the CRT device

Beside optimal patient selection, optimal programming of the CRT device settings can also play an important role in increasing the response rate to CRT. Currently, all available tools to optimize CRT device settings are time-consuming and/or subject to noise, leading to the use of the ‘out-of-the-box’ default settings by a vast majority of cardiologists. Furthermore, the optimization can often only be performed during in-hospital visits, while the optimal setting might change over time or during different levels of activity. CRT is often employed by pacing both the right end left ventricle of the heart (biventricular [BiV] pacing). However, several acute8 and chronic9 studies have demonstrated that in patients with sinus rhythm and intact AV conduction, LV-only pacing can be at least as effective as BiV pacing. CRT using LV-only pacing has been shown to be most effective when the paced LV impulse is properly timed with respect to the intrinsic activation of the right ventricle (RV), ensuring appropriate fusion with intrinsically conducted activation wave fronts. In chapter 8 of this thesis, we showed that using either the normal VCG or an electrogram (EGM) based VCG (EGM-VCG) that can be extracted from the EGM signals measured from the unused implanted electrodes, this exact optimal timing can easily be found. The point at which the maximal QRS vector amplitude changes or flips direction was equal to the onset of intrinsic activation of the RV. A few years ago, the ADAPTIVECRTTM (trademark of Medtronic, Inc.) algorithm has been developed that switches between BiV pacing and LV only pacing depending on con-

192 Valorization duction measurements.10 When LV only pacing is applied, the ADAPTIVECRTTM algorithm aims to perform fusion pacing which estimates the delay between atrial activation and intrinsic RV activation (A-RV) based on an average, general relation. With the VCG or EGM- VCG the A-RV can be determined individually, making it more precise. It does so by obtain- ing information during LV-only pacing at different AV-delays, while the ADAPTIVECRTTM estimates A-RV using measurements performed when CRT is turned off. Furthermore, the VCG and EGM-VCG methods are more robust than the ADAPTIVECRTTM algorithm since they are independent of the exact position of the RV lead and of LV latency. Therefore, the ADAPTIVECRTTM algorithm could be improved by using the VCG or EGM-VCG to find the exact A-RV. The idea of using the VCG or EGM-VCG to find the exact A-RV and using it to individualize the ADAPTIVECRTTM algorithm is already part of a patent application in collaboration with Medtronic, Inc. To do so, there are two options. First, a single determination of A-RV, at time of implant or shortly after, could be performed using the regular ECG from which a VCG and subsequently the maximal QRS vector amplitude can be calculated. The exact difference in timing between A-RVsense, the delay between atrial activation and the moment of sensing of activation on the RV lead, and A-RV can be programmed in the device instead of the estimated difference relation used in the current ADAPTIVECRTTM algorithm. This would only require adding the constant delay to the algorithm. A second option would be to extract the A-RV from the EGM-VCG. The algorithm needed to find A-RV can be embedded in the CRT device, making automatic and continuous optimization possible. Again, the difference between A-RVsense and A-RV can be determined every few days, and the ADAPTIVECRTTM algorithm can perform its usual continuous optimization with this timing difference embedded. The methods proposed here can be used to objectively and easily tailor the ADAPTIVECRTTM algorithm, possibly leading to a further increase in hemodynamic response by CRT. It does so without spending additional current, purely by optimizing fusion pacing, which benefits form the natural activation of the RV. Because of the high potential of the VCG or EGM-VCG to improve the ADAPTIVECRTTM algorithm, this idea was patented together with Medtronic, Inc.

Conclusion

The findings of this thesis provide valuable tools to improve patient selection for CRT and CRT device optimization, which have been derived from extensive basic and clinical research prior to and during this thesis. These tools can be easily embedded in already existing systems, either ECG machines or devices. Therefore, findings from the present study may well have significant practical implications.

193 References

1. M Bots, I van Dis, C Koopman, I Vaartjes, and F Visseren. Hart- en vaatziekten in Nederland. (2013). 2. NM Hawkins, D Wang, JJV McMurray, MA Pfeﬀer, K Swedberg, CB Granger, S Yusuf, SJ Pocock, J Ostergren, EL Michelson, FG Dunn, and CHARM Investigators and Committees. Prevalence and prognostic impact of bundle branch block in patients with heart failure: evidence from the CHARM programme. Eur J Heart Fail (2007) 9:510–517. 3. M Bots, I van Dis, C Koopman, I Vaartjes, and F Visseren. Hart- en vaatziekten in Nederland. (2014). 4. WT Abraham, WG Fisher, AL Smith, DB Delurgio, AR Leon, E Loh, DZ Kocovic, M Packer, AL Clavell, DL Hayes, et al. Cardiac resynchronization in chronic heart failure. N Engl J Med (2002) 346:1845–1853. 5. M Brignole, A Auricchio, G Baron-Esquivias, P Bordachar, G Boriani, OA Breithardt, J Cleland, JC Deharo, V Delgado, PM Elliott, et al. 2013 ESC Guidelines on cardiac pacing and cardiac resynchronization therapy: the Task Force on cardiac pacing and resynchronization therapy of the European Society of Cardiology (ESC). Developed in collaboration with the European Heart Rhythm Association (EHRA). Eur Heart J (2013) 34:2281–2329. 6. CM Tracy, AE Epstein, D Darbar, JP Dimarco, SB Dunbar, NAM Estes 3rd, TB Fer- guson Jr, SC Hammill, PE Karasik, MS Link, et al. 2012 ACCF/AHA/HRS Focused Update of the 2008 Guidelines for Device-Based Therapy of Cardiac Rhythm Abnor- malities: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. Heart Rhythm (2012) 9:1737–1753. 7. European Heart Rhythm Association, European Society of Cardiology, Heart Rhythm Society, Heart Failure Society of America, American Society of Echocardiography, American Heart Association, European Association of Echocardiography, Heart Fail- ure Association, JC Daubert, L Saxon, et al. 2012 EHRA/HRS expert consensus state- ment on cardiac resynchronization therapy in heart failure: implant and follow-up re- commendations and management. Heart Rhythm (2012) 9:1524–1576. 8. BM van Gelder, FA Bracke, A Meijer, and NH Pijls. The hemodynamic eﬀect of intrinsic conduction during left ventricular pacing as compared to biventricular pacing. Valorization

Journal of the American College of Cardiology (2005) 46:2305–2310. 9. G Boriani, W Kranig, E Donal, L Calo, M Casella, N Delarche, IF Lozano, G An- salone, M Biﬃ, E Boulogne, C Leclercq, and B-LEFT HF study group. A randomized double-blind comparison of biventricular versus left ventricular stimulation for cardiac resynchronization therapy: the Biventricular versus Left Univentricular Pacing with ICD Back-up in Heart Failure Patients (B-LEFT HF) trial. Am Heart J (2010) 159:1052–1058.e1. 10. H Krum, B Lemke, D Birnie, KLF Lee, K Aonuma, RC Starling, M Gasparini, J Gorcsan, T Rogers, A Sambelashvili, A Kalmes, and D Martin. A novel algorithm for individualized cardiac resynchronization therapy: rationale and design of the adaptive cardiac resynchronization therapy trial. Am Heart J (2012) 163:747–752.e1.

195

Dankwoord

it proefschrift is het resultaat van teamwork. Dit proefschrift zou niet tot stand zijn gekomen zonder de samenwerking met verschillende labs en zonder de steun van D vrienden en familie. Daarom sluit ik dit werk af met een woord van dank. Allereerst wil ik mijn promotor, prof. dr. Frits W. Prinzen bedanken, Beste Frits, in 2011 hebben wij voor het eerst kennisgemaakt. Als masterstudent aan de TU/e was ik op zoek naar een interessant afstudeerproject. Zoals altijd was je erg enthousiast over de verschillende projecten, waardoor ik ook enthousiast werd. Begin 2012 ben ik bij jou begonnen aan mijn afstudeerproject en dit was het begin van een goede en productieve samenwerking, die werd verlengd met een promotie. Ik kan me geen betere promotor voorstellen: je was altijd beschikbaar voor vragen, dacht mee over mogelijke oorzaken van onverwachte resultaten en door je grote netwerk heb ik tijdens mijn promotietraject mogen samenwerken met veel verschillende mensen uit verschillende landen. Daarnaast ben ik naar vele congressen geweest waar ik veel van heb geleerd en waardoor ik ook veel van de wereld heb kunnen zien. Heel erg bedankt voor alles. Naast een goede promotor had ik ook het geluk van een goede co-promotor, dr. Kevin Vernooy. Beste Kevin, als arts keek jij vaak anders naar onderzoeksresultaten dan ik als ingenieur, wat de resultaten vaak een ander en nieuw perspectief gaf. Je hebt mij geïntro- duceerd in de ziekenhuiswereld en me geleerd hoe om te gaan met de gang van zaken in het ziekenhuis: je kunt er nooit vanuit gaan dat alles goed geregeld is; zorg er altijd voor dat je alles zelf gecheckt hebt. Ik heb onder jouw begeleiding een klinische studie mogen uitvoeren waarvan ik heel veel geleerd heb. Mijn dank is heel erg groot voor alles wat jij mij hebt bijgeleerd. Leden van de beoordelingscommissie, prof. dr. Uli Schotten (voorzitter), prof. dr. An- gelo Auricchio, prof. dr. Harry J.G.M. Crijns, prof. dr. Tammo Delhaas, en dr. Mathias Meine, hartelijk bedankt voor de kritische beoordeling van mijn proefschrift. Beste Masih en Twan, jullie waren voor mij altijd een goed aanspreekpunt voor al mijn vragen gerelateerd aan het reilen en zeilen in het ziekenhuis. Masih, heel erg bedankt voor al je hulp tijdens het opzetten van de Tailor-CRT studie. Jij hebt mij wegwijs gemaakt in het ziekenhuis en mij geleerd hoe alles rondom de studie geregeld moest worden. Daarnaast wil ik je bedanken voor alle gezelligheid, ook tijdens onze gezamenlijke congressen. Twan, nadat Masih is begonnen aan zijn opleiding cardiologie heb jij min of meer zijn taken over- genomen. Je was er altijd om mee te helpen waar nodig. Bedankt voor onze fijne en gezellige samenwerking. Tijdens het laatste anderhalf jaar van mijn promotie heb ik de klinische studie Tailor-CRT uitgevoerd waardoor ik van tijd tot tijd ook op de vaatkamer te vinden was. Ik wil al het vaatkamerpersoneel en alle pacemakertechnici van harte bedanken voor alle geleverde bijdragen. I would also like to thank dr. Jagmeet Singh and Eszter for our collaboration, our discus- sions and for sharing their database with us. This enabled me to perform research on a large patient population resulting in two nice articles. Patrick en Thomas, jullie wil ik graag bedanken voor de hulp tijdens mijn TAVI project. Het heeft lang geduurd maar er is nu toch een mooi manuscript uitgekomen. Thomas, ik wil jou ook bedanken voor onze niet-werkgerelateerde gesprekken. Ook iedereen van de afdeling Fysiologie wil ik bedanken voor de fijne samenwerking, de prettige werksfeer en alle gezelligheid. Caroline, Siamack, Chantal, Uyen en Joris: bedankt voor de leuke sfeer op onze kamer. Marc, Lars, Frans, Marion, Rick en Nienke: ik wil jullie ook bedanken voor jullie bijdragen. Daarnaast wil ik graag de secretaresses van de vakgroep Fysiologie, Bianca en Vivian bedanken voor alle hulp gedurende de afgelopen jaren. Verder wil ik ook een aantal mensen bedanken van de afdeling Biomedische Technologie. Lauren, Georgina, and Jeire thank you for the coffee breaks, lunch breaks, and during the summer for the ice cream breaks. I could always tell you all my complaints and you were always able to make me laugh. Lauren, thank you for being my paranymph. John, Niek en Marieke: bedankt voor de gezellige lunches en borrels. Raf en Jeire en Peter en Sabine, bedankt voor de gezellige uitjes buiten het werk om. Peter en Sabine: Bart en ik zien er al naar uit om jullie te mogen ontvangen in Sydney. Elly en Paul, Bart en ik kunnen altijd bij jullie terecht. Of het nou voor de gezelligheid is, voor het vermaken van kleren, of voor klusjes in het huis. Daarnaast ben ik jullie erg dankbaar voor de gezellige avonden toen Bart in Australië zat, dit was altijd een fijne afleiding voor mij. Lieve papa en mama, bedankt voor jullie onvoorwaardelijke steun. Jullie hebben altijd achter mijn keuzes gestaan zolang ik maar dat deed wat ik leuk vond. Ook zorgden jullie voor de altijd welkome afleidingen. Jullie hebben mij geleerd om zoveel mogelijk te genieten van het leven en zoveel mogelijk van de wereld te zien. Zonder jullie steun, voornamelijk in de maanden dat Bart niet thuis was, was het me nooit gelukt om dit boekje tot een goed einde te brengen. Lieve Tim, Bart en Mirthe, wat ben ik blij dat jullie mijn broers en schoonzus zijn. Tim, jij kan altijd alles zo lekker relativeren, alles komt wel goed. Ik wou dat ik soms iets meer van jouw relaxheid had. Bart en Mirthe, wat heb ik genoten van alle gezellige avondjes toen

198 Dankwoord jullie nog in de Zakstraat woonden. Nu wonen jullie wat verder weg, Tanzania of all places, maar toch zijn jullie nog altijd heel erg betrokken bij alles wat er hier in ons kikkerlandje gebeurt. Bart, ﬁjn dat je mijn paranimf wilt zijn. Ook al wonen we straks allemaal ergens anders op de wereld, ik hoop dat onze broer-en-zus band altijd zo sterk mag blijven. Allerliefste Bart, zonder jou was het me allemaal nooit gelukt. Ik kan altijd mijn hart bij jou luchten en je krijgt me op de een of andere manier altijd rustig. Ook kunnen we inhoudelijke discussies voeren en kan ik je altijd om hulp vragen waar nodig. Daarnaast zorg je ervoor dat ons leven nooit saai is, we reizen heel wat af en we gaan nu voor langere tijd naar het buitenland. Ik zie er naaruit om deze mooie ervaring samen met jou te beleven.

199

About the author Curriculum Vitae

lien Engels was born on the 4th of August 1989 in Helmond, The Netherlands. From 2001 until 2007 she attended secondary education at the Theresialyceum in Tilburg E where she obtained her high school diploma (atheneum). In September 2007 she started her academic education at Eindhoven University of Tech- nology were she obtained her Master of Science degree in Biomedical Engineering in Janu- ary 2013. The Master’s program included an internship at the Hôpital du Sacré-Coeur in Montreal, Canada (September 2011–December 2011) under the supervision of dr. ir. Vincent Jacquemet. She investigated the sensitivity of QT heart-rate correction methods for detecting drug-induced QTc changes. In her ﬁnal Masters project, she developed an algorithm to automatically analyze 12-lead electrocardiograms and convert them to vectorcardiograms. She used both tools to study changes over time in left bundle branch block patients and to study diﬀerent types of left bundle branch blocks. This project was performed under supervision of dr. ir. Nico H.L. Kuijpers and prof. dr. Frits W. Prinzen. From March 2013 until July 2016, she worked as a PhD candidate at the department of Physiology (CARIM School for Cardiovascular Diseases, Maastricht University) under the supervision of dr. Kevin Vernooy and prof. dr. Frits W. Prinzen. During her PhD project, she aimed to investigate vectorcardiographic manners to improve the response rate of cardiac resynchronization therapy. Moreover, she was involved in supervising medical students during this period. In 2015 she received an award for best oral presentation at the annual MALT meeting and won the best poster prize at the local CARIM conference. As of August 2016, she is working as a postdoctoral research fellow at the department of Physiology at Maastricht University.

202 About the author

List of publications

UC Nguyên, M Maﬁ-Rad, JP Aben, MW Smulders, EB Engels, AMW van Stipdonk, JGLM Luermans, SCAM Bekkers, FW Prinzen, K Vernooy. A novel approach for left ventricular lead placement in cardiac resynchronization therapy: intra-procedural integration of coronary venous electro-anatomic mapping with delayed enhancement cardiac magnetic resonance imaging. Heart Rhythm (2016) [Epub ahead of print]

EB Engels∗, M Maﬁ-Rad∗, AMW van Stipdonk, K Vernooy, FW Prinzen. Why QRS duration should be replaced by better measures of electrical activation to improve patient selection for cardiac resynchronization therapy. J Cardiovasc Transl Res (2016) 9(4):257-65 ∗authors contributed equally

EM Végh∗, EB Engels∗, CJM van Deursen, B Merkely, K Vernooy, JP Singh, FW Prinzen. T-wave area as biomarker of clinical response to cardiac resynchronization therapy. Europace (2016) 18(7):1077-85 ∗authors contributed equally

M Maﬁ-Rad, GWM Wijntjens, EB Engels, Y Blaauw, JGLM Luermans, L Pison, HJGM Crijns, FW Prinzen, K Vernooy. Vectorcardiographic QRS area identiﬁes delayed left ventricular lateral wall activation determined by electroanatomical mapping in candidates for cardiac resynchronization therapy. Heart Rhythm (2016) 13(1):217-25

EB Engels, S Alshehri, CJM van Deursen, L Wecke, L Bergfeldt, K Vernooy, FW Prinzen. The synthesized vectorcardiogram resembles the measured vectorcardiogram in patients with dyssynchronous heart failure. J Electrocardiol (2015) 48(4):586-92

EB Engels∗, EM Végh∗, CJM van Deursen, K Vernooy, JP Singh, FW Prinzen. T-wave area predicts response to cardiac resynchronization therapy in patients with left bundle branch block. J Cardiovasc Electrophysiol (2015) 26(2):176-83 ∗authors contributed equally

RC González, EB Engels, B Dubé, R Nadeau, A Vinet, AR LeBlanc, M Sturmer, G Becker, T Kus, V Jacquemet, Assessment of the sensitivity of detecting drug-induced QTc changes using subject-speciﬁc rate correction. J Electrocardiol (2012) 45(6):541-5

203