Relationship between vectorcardiographic QRS area , 1 myocardial scar quantification, and response to 2 Cardiac Resynchronization Therapy

Purpose: To investigate the relationship between vectorcardiography (VCG) and 30 myocardial scar on cardiac magnetic resonance (CMR) imaging, and whether combining 31 these metrics may improve cardiac resynchronization therapy (CRT) response 32 prediction. 33 Methods: Thirty-three CRT patients were included. QRS area , T area and QRST area were 34 derived from the ECG-synthesized VCG. CMR parameters reflecting focal scar core 35 (Scar 2SD , Gray 2SD ) and diffuse fibrosis (pre-T1, extracellular volume [ECV]) were 36 assessed. CRT response was defined as ≥15% reduction in left ventricular end-systolic 37 volume after six months’ follow-up. 38 Results: VCG QRS area , T area and QRST area inversely correlated with focal scar (R=-0.44–- 39 0.58 for Scar 2SD , p≤ 0.010), but not with diffuse fibrosis. Scar 2SD , Gray 2SD and QRS area 40 predicted CRT response with AUCs of 0.692 ( p =0.063), 0.759 ( p =0.012) and 0.737 41 ( p =0.022) respectively. A combined ROC-derived threshold for Scar 2SD and QRS area 42 resulted in 92% CRT response rate for patients with large QRS area and small Scar 2SD or 43 Gray 2SD . 44 Conclusion: Incremental predictive value for CRT response is achieved by a combined 45 CMR-QRS area analysis. 46

Summarizing the above literature, it appears that certain electrical characteristics from 99 the VCG and low myocardial scar burden is favorable for response to CRT. The 100 association between VCG and myocardial scar as measured by CMR is however not 101 known.

102
The purpose of this study was therefore to investigate the association between VCG 103 parameters and myocardial scar (both focal and diffuse) on CMR in HF patients with 104 ventricular conduction disturbance, and whether combining VCG with CMR scar 105 parameters improves prediction to CRT response. loop and baseline in X, Y, and Z direction calculated as QRSarea = (QRSarea,x 2 + QRSarea,y 2 + 126 QRSarea,z 2 ) 1/2 , Tarea = (Tarea,x 2 + Tarea,y 2 + Tarea,z 2 ) 1/2 , and QRSTarea = (QRSTarea,x 2 + 127 QRSTarea,y 2 + QRSTarea,z 2 ) 1/2 .(4) 128 129 Cardiac magnetic resonance imaging 130 Patients underwent CMR prior to their CRT implantation using a 1.5T scanner with a 131 32-channel coil (Philips Healthcare, Best) as described previously.(6) Two independent 132 CMR experts, blinded to CRT outcome, assessed the CMR images. In case of discrepancy, 133 consensus between the reviewers was reached. LV mass was quantified using CMR42 134 (Circle Cardiovascular Imaging Inc, Calgary) software and used to index the delayed 135 enhancement (DE-CMR) quantification of focal scar. The extent of scar core was 136 automatically quantified using the 2-standard deviation (2SD) method, defined as the 137 region with signal intensity (SI) >2SD above reference myocardium (Scar2SD). The    171 Statistical analyses were performed using SPSS 24.0 (SPSS Inc., Chicago, Illinois) and 8 as mean±SD or median and interquartile range (IQR) and dichotomous variables in 174 frequencies and percentages. Spearman correlation analyses were carried out between 175 and within VCG and CMR parameters. Parameter differences between CRT responders 176 vs. non-responders were compared using Mann Whitney U-tests. Receiver operating 177 characteristics (ROC) curves were generated to evaluate the diagnostic accuracy of all parameters in identifying CRT response and to find optimal cut-off values. These cut-off 217 There was no association between pre-T1 or ECV and QRSarea or Tarea (all p>0.142). All 218 VCG parameters inversely correlated with Scar2SD and Gray2SD.The strongest VCG-CMR 219 association was found between QRSarea and focal scar parameter Scar2SD ( Figure 4).

221
Combining VCG and CMR scar parameters 222 The study population was dichotomized using the cut-off values for Scar2SD, Gray2SD and 223 QRSarea derived from the ROC analyses in Table 3. The percentage of CRT responders 224 was significantly higher in patients with low Gray2SD and low Scar2SD versus patients 225 with high focal scar parameters ( Figure 5A). The percentage of CRT response was also 226 higher in patients with high QRSarea as compared to those with low QRSarea ( Figure 5B).

227
Crosstab analyses between QRSarea and Gray2SD/Scar2SD showed that the percentage CRT  The present study is the first to investigate the relationship between VCG parameters 237 and CMR defined scar, and between these parameters and CRT response. The principal 238 findings of this study are that QRSarea significantly correlated inversely with focal scar, 239 suggesting that myocardial scar leads to a smaller QRSarea, Additionally, by combining 240 QRSarea and CMR focal scar assessment, CRT response prediction improves beyond that 241 by either VCG or scar parameters alone. The association between VCG and CMR scar 259 The usefulness of VCG for identification of myocardial scar has been investigated by In the present study, correlation analyses suggested that QRSarea decreased with focal 267 scar burden (encompassing dense scar core), and to a lesser extent scar border zone; 268 but VCG parameters were not significantly associated with measures of diffuse fibrosis.

269
This suggests that scar tissue with higher density affects the VCG 3D loop the most.  (14) found that patients with a high QRSTarea had significantly greater odds of LV 302 reverse remodeling than those with lower QRSTarea. QRSTarea was also associated with 303 ∆LVESV reduction in our data but was not a significant CRT response predictor in the 304 ROC analyses (p = 0.074). Altogether these results indicate that the role for Tarea in CRT 305 response prediction is not fully understood yet. (4,13,14) 306 307 The relevance of myocardial scar regarding CRT response 308 The association between focal scar burden and poor CRT response has been  329 The present study demonstrates that combining parameters reflecting both electrical 330 and tissue substrate for CRT may be an approach to further improve CRT response 331 prediction. Almost all (92%) patients with a low extent of focal scar and a large QRSarea 332 were CRT responders. This finding is important, since myocardial scar burden and 333 QRSarea are inversely related to each other. Apparently, CRT response prediction is 334 better when incorporating focal scar metrics in addition to QRSarea compared to using 335 QRSarea alone. Potential explanations for the negative effect of scar on CRT may be that Baron-Esquivias G, Bordachar P, et al. 2013 Table 3. parameter when dividing the study population using the cut-off value as determined by 504 ROC analyses in Table 3. P-values in A and B are based on Chi-squared tests.

505
3D bar graphs demonstrating CRT response percentage when combining QRSarea with 506 focal scar CMR parameters (C). P-values in each graph are based on Fisher's exact tests.