Cardiac radiotherapy induces electrical conduction reprogramming in the absence of transmural fibrosis

Abstract

Cardiac radiotherapy (RT) may be effective in treating heart failure (HF) patients with refractory ventricular tachycardia (VT). The previously proposed mechanism of radiation-induced fibrosis does not explain the rapidity and magnitude with which VT reduction occurs clinically. Here, we demonstrate in hearts from RT patients that radiation does not achieve transmural fibrosis within the timeframe of VT reduction. Electrophysiologic assessment of irradiated murine hearts reveals a persistent supraphysiologic electrical phenotype, mediated by increases in NaV1.5 and Cx43. By sequencing and transgenic approaches, we identify Notch signaling as a mechanistic contributor to NaV1.5 upregulation after RT. Clinically, RT was associated with increased NaV1.5 expression in 1 of 1 explanted heart. On electrocardiogram (ECG), post-RT QRS durations were shortened in 13 of 19 patients and lengthened in 5 patients. Collectively, this study provides evidence for radiation-induced reprogramming of cardiac conduction as a potential treatment strategy for arrhythmia management in VT patients.

Conflict of interest statement

The authors declare the following competing interests: C.G.R. and P.S.C. have filed two institution-owned patents: WO2017078757 (Noninvasive Imaging and Treatment System for Cardiac Arrhythmias) that relates to overall methods for delivery of cardiac radiation in patients and WO2019118640 (System and Method for Determining Segments for Ablation) that relates to the use of cardiac segments for cardiac radiation targeting. C.G.R. and P.S.C. also provide consulting services to Varian, which produces linear accelerators for radiation treatment delivery. C.G.R. and P.S.C. declare no other competing interests. No other authors declare any competing interests.

© 2021. The Author(s).

Figures

Fig. 1. Cardiac fibrosis alone cannot account for the timing and effect of VT reduction after RT.

a–c Representative Masson’s trichrome stains of regions of remote (a), targeted (b), and failed CA (c) myocardium in the same patient who received 25 Gy ionizing radiation for treatment of refractory VT. Scale bars = 500 μm. Experiment was independently repeated once on the same patient samples and produced similar results. d Percent fibrosis at the time of heart explant from 4 patients who received 25 Gy RT in control (left) and targeted (right) regions of myocardium. Percent fibrosis of the CA region for Patient A is shown in purple. e Defibrillator-recorded episodes of VT in the month before treatment and the 6 months after treatment for each patient. Days post-treatment until specimen collection are shown in the legend. *Patient D’s values are divided by 20 to allow for comparisons on the same scale. f Gadolinium-enhanced MRI of a cardiac RT patient at baseline (left) and 3 months post-treatment (right). Top: the left ventricle with patchy, gadolinium-enhanced scar was transmurally targeted with 25 Gy between 3 and 6 O’clock (red brackets). Nonenhanced, remote myocardium is adjacent to target region (white arrowhead). Bottom: surviving nonenhanced myocardium within the same images is visible in the targeted region at baseline and 3 months post-treatment (yellow outline). Source data are provided as a Source data file.

Fig. 2. Cardiac RT persistently increases adult murine ventricular conduction.

a Representative ECGs from control (top) and IR mice (bottom) highlighting PR and QRS intervals. b, c Effect of radiation on PR (P = 0.39) and QRS (**P = 0.00043) intervals in control (black) versus IR (red) mice (n = 6 biologically independent animals per condition). d Representative ventricular activation maps from control and IR mice. Scale bars = 3 mm. e Effect of RT on ventricular CV (n = 6 biologically independent animals per condition; ***P = 0.00005; **P = 0.0034). f, g APD80 (P = 0.39) and ERP (P = 0.78) in control versus IR mice (n = 6 biologically independent animals per condition). h, i Western blots of NaV1.5 (*P = 0.022) and Cx43 (*P = 0.025) in control and IR ventricles (n = 3 biologically independent samples per condition). j Immunostaining of control (top) and IR (bottom) myocardium for NaV1.5 and k co-stained for Cx43 (red) and N-Cadherin (green). Scale bars = 70 μm. Experiment was replicated three times in biologically independent specimens and produced similar results. l, m Western blots of NaV1.5 (**P = 0.0034) and Cx43 (*P = 0.027) after 42 weeks post-IR (n = 3 biologically independent samples per condition). n Effect of RT on QRS intervals in control versus IR mice after 42 weeks (***P = 0.00027; C, n = 9; IR, n = 6). o Left ventricular CVs in control (n = 9 biologically independent animals) versus IR (n = 6 biologically independent animals) mice at 42 weeks post-IR (***P = 0.00042; **P = 0.0092). All P values determined by two-way unpaired t test. Bar graphs are represented as mean ± SD. Source data are provided as a Source data file.

Fig. 3. Conduction velocity reprogramming may be achieved at lower doses of radiation.

a Dose-dependent effects of radiation on QRS interval at 6 weeks post-IR (n = 5 biologically independent animals per IR condition; n = 10 biologically independent 0 Gy control animals; ***P = 0.0009, 20 versus 0 Gy; *P = 0.023, 15 versus 0 Gy). b Dose-dependent effects of radiation on longitudinal conduction velocity (n = 3 biologically independent animals per IR condition; n = 6 biologically independent 0 Gy control animals; **P = 0.0023, 20 versus 0 Gy; *P = 0.046, 15 versus 0 Gy). c Dose-dependent effects of radiation on NaV1.5 protein levels by western blot (n = 3 biologically independent sample per condition over 3 independent experiments; **P = 0.0014, 25 versus 0 Gy; **P = 0.0016, 20 versus 0 Gy; *P = 0.023, 15 versus 0 Gy;). d Dose-dependent effects of radiation on Cx43 protein levels by western blot (n = 3 biologically independent per condition over 3 independent experiments; *P = 0.011, 15 versus 0 Gy). Values plotted from western blots were obtained from three independent western blots each denoted by a closed square, circle, and triangle. Twenty-five Gy QRS interval and CV values from Fig. 3 are plotted for reference (open circles) and are not used in statistical comparisons. Statistical analysis of all subpanels consisted of a one-way ANOVA followed by one-sided Tukey post hoc test of multiple comparisons. All bar graphs are represented as mean ± SD. Source data are provided as a Source data file.

Fig. 4. Radiation specifically reprograms the border zone myocardium in a murine model of myocardial infarction.

a Top: representative ventricular activation maps from control myocardial infarction (MI, left) and irradiated MI (MI+IR, right) mice at 8 weeks post-MI and 6 weeks post-IR. Black arrows point to location of left anterior descending (LAD) artery ligation. Black box denotes approximate 2 mm border zone at the scar/myocardium interface. Regions of scar that were epicardially stimulated but did not capture are denoted by gray circles. Magnified isochrones in the border zone region used to calculate transverse conduction velocities. Scale bars = 5 mm. b Left ventricular conduction velocities in MI (black) versus MI+IR (red) hearts (MI n = 9 biologically independent animals; MI+IR n = 8 biologically independent animals; longitudinal ***P = 0.0006; transverse ***P = 0.0009;). c, d Western blot (top) and quantified density (bottom) of NaV1.5 (***P = 0.0007) and Cx43 (*P = 0.041) protein in MI and MI+IR border zone myocardium superior to the LAD ligation (n = 3 biologically independent samples per condition). e, f Western blot (top) and quantified density (bottom) of NaV1.5 (P = 0.42) and Cx43 (P = 0.68) protein in MI control and MI+IR apical scars (n = 3 biologically independent samples per condition). All P values determined by two-way unpaired t test. All bar graphs are represented as mean ± SD. Source data are provided as a Source data file.

Fig. 5. Post-IR states exhibit distinct transcriptional signatures and activation of the Notch signaling pathway.

a Top 20 transcriptionally over-represented hallmark pathways at 2 and 6 weeks post-IR. The Notch signaling pathway is present at both timepoints, and Notch pathway-associated genes Cdkn1a, Jag1, Jag2, and Notch2 were drivers of enrichment in other over-represented pathways. b Heatmap of Notch GO term genes. Samples group by hierarchical clustering of Notch GO gene expression. c Cardiac-related Notch pathway members grouped by function. Bar graphs represent log2-fold change at 2 weeks or 6 weeks post-IR relative to sham control state and error bars represent standard error (n = 6 biologically independent samples per treatment group, *indicates comparisons of 2 weeks post-IR versus control, #indicates comparisons of 6 weeks post-IR versus control, */#P < 0.10, **/##P < 0.01, ***/###P < 0.001). Jag1##P = 0.0049; Jag2#P = 0.03; Notch1#P = 0.076; Notch2 **P = 0.0066, ##P = 0.0012; Notch3 *P = 0.014, ##P = 0.0019; Notch4 *P = 0.015, ###P = 7.2E−6; Psen1 *P = 0.051, ###P = 0.00078; Psenen#P = 0.024; Lfng###P = 1.0E−6; Mfng###P = 0.00023; Neurl1a#P = 0.092; Dtx2#P = 0.051; Dtx3l###P = 0.00014; Numb#P = 0.061; Pofut1 *P = 0.019, ##P = 0.0049; Poglut1#P = 0.020 Maml1 *P = 0.016, ###P = 0.00031; Maml2#P = 0.010; Cdkn1a ***P = 0.00033, ###P = 2.5E-6; Nrarp##P = 0.0019. Adjusted P value statistics determined by Benjamini–Hochberg procedure applied to two-sided limma differential expression values to control false discovery from multiple comparisons (“Methods”). Source data are provided as a Source data file.

Fig. 6. Transient iNICD-mediated, cardiomyocyte-specific Notch activation persistently increases conduction velocity and NaV1.5 protein expression.

a Representative activation maps and b conduction velocity measurements of control (CTRL, black) αMHC-rtTA; tetO_NICD (iNICD, blue) murine hearts after 3-week doxycycline induction and 16-week washout (n = 7 biologically independent CTRL mice; n = 9 biologically independent iNICD mice; Longitudinal ***P = 0.00019, Transverse P = 0.18). c, d Western blot and quantified densitometry from CTRL versus iNICD hearts after 3-week doxycycline induction, 52-week washout for NaV1.5 (n = 3 biologically independent samples per condition, *P = 0.026) and Cx43 (n = 6 biologically independent samples per condition, P = 0.47, ladder was run in parallel on a separate gel to allow for quantitative comparisons on the same blot). All P values determined by two-way unpaired t test. All bar graphs are represented as mean ± SD. Source data are provided as a Source data file.

Fig. 7. Loss of cardiomyocyte Notch signaling inhibits radiation-induced conduction reprogramming.

a Representative activation maps and b longitudinal conduction velocity measurements of irradiated littermate control (left, CTRL, white bars) and αMHC-MerCreMer; R26rdnMAML/+ (right, iLOF, gray bars) murine hearts at 6 weeks post-IR. n = 8 biologically independent CTRL mice; n = 7 biologically independent iLOF mice; *P = 0.027. c, d Western blot and quantified densitometry of NaV1.5 (n = 6 biologically independent samples per condition, *P = 0.012) and Cx43 (n = 6 biologically independent samples per condition across two blots run in parallel, P = 0.59,). All P values determined by two-way unpaired t test. All bar graphs are represented as mean ± SD. Source data are provided as a Source data file.

Fig. 8. Reprogramming in VT patients treated with 25 Gy RT.

a Cardiac 17-segment model illustrating 25 Gy targeted (Segment 1, red) and remote nontargeted (Segment 10, gray) myocardium collected from a heart failure patient with nonischemic cardiomyopathy and VT (Patient F), who previously received cardiac RT. The nontargeted Segment 10 received <5 Gy radiation exposure. For comparison, corresponding Segments 1 and 10 were collected from 2 nonfailing, non-RT donor hearts rejected for transplantation. b Western blot and quantified densitometry of NaV1.5 in Segment 10 versus Segment 1 of nonfailing (Donors A and B) and RT-treated (HF+Cardiac RT, red) hearts. c Western blot and quantified densitometry of Cx43 in Segment 10 versus Segment 1 of nonfailing and RT-treated hearts. d Matched QRS durations in the ENCORE-VT patient cohort at baseline (pre-RT, 149 ± 34 ms) versus 6-week post-RT (139 ± 32 ms) timepoints (n = 19 paired individual patients, P = 0.072 by two-tailed Wilcoxon signed-rank test). e Representative 12-lead ECG of robust QRS shortening recorded from Patient G at the time of RT treatment, with a baseline QRS of 165 ms and presence of LBBB, and then after 6 months post-RT, exhibiting shortened QRS (130 ms) and resolution of LBBB. Examples of pre- and post-RT ECGs of the remaining three patients that exhibited robust QRS shortening in the ENCORE-VT patient cohort, as well as an example of an unchanged QRS interval, are provided in the Supplementary Figures. Source data are provided as a Source data file.