A Genetic Mouse Model Recapitulates Immune Checkpoint Inhibitor-Associated Myocarditis and Supports a Mechanism-Based Therapeutic Intervention

Abstract

Immune checkpoint inhibitors (ICI) targeting CTLA4 or PD-1/PD-L1 have transformed cancer therapy but are associated with immune-related adverse events, including myocarditis. Here, we report a robust preclinical mouse model of ICI-associated myocarditis in which monoallelic loss of Ctla4 in the context of complete genetic absence of Pdcd1 leads to premature death in approximately half of mice. Premature death results from myocardial infiltration by T cells and macrophages and severe ECG abnormalities, closely recapitulating the clinical and pathologic hallmarks of ICI-associated myocarditis observed in patients. Using this model, we show that Ctla4 and Pdcd1 functionally interact in a gene dosage-dependent manner, providing a mechanism by which myocarditis arises with increased frequency in the setting of combination ICI therapy. We demonstrate that intervention with CTLA4-Ig (abatacept) is sufficient to ameliorate disease progression and additionally provide a case series of patients in which abatacept mitigates the fulminant course of ICI myocarditis. SIGNIFICANCE: We provide a preclinical model of ICI-associated myocarditis which recapitulates this clinical syndrome. Using this model, we demonstrate that CTLA4 and PD-1 (ICI targets) functionally interact for myocarditis development and that intervention with CTLA4-Ig (abatacept) attenuates myocarditis, providing mechanistic rationale and preclinical support for therapeutic clinical studies.See related commentary by Young and Bluestone, p. 537.This article is highlighted in the In This Issue feature, p. 521.

Figure 1:. Lethal haploinsufficiency of Ctla4 in the genetic absence of Pdcd1.

A) Kaplan-Meier survival curve of transgenic C57BL6/J mice harboring Ctla4 and Pdcd1 knockout alleles. Mice were derived from an intercross of Ctla4^+/− Pdcd1^+/− mice in which Pdcd1 and Ctla4 loss of function alleles are in trans. Individual mice were censored if used for breeding or alive at the time of data analysis. Death events were defined as mice found dead or identified by veterinary staff as requiring euthanasia. P-value represents the result of the Mantel-Cox Log-rank test. B) Kaplan-Meier survival curve of Ctla4^+/− Pdcd1^−/− (n = 350) and littermate Ctla4^+/+ Pdcd1^−/− (n = 400) mice derived from a Ctla4^+/− Pdcd1^−/− by Ctla4^+/+ Pdcd1^−/− breeding cross performed at the Vanderbilt University Medical Center (VUMC) vivarium. P-value represents the result of the Mantel-Cox Log-rank test. C) Kaplan-Meier survival curve of Ctla4^+/− Pdcd1^−/− and littermate Ctla4^+/+ Pdcd1^−/− mice stratified by sex. Data are pooled from studies performed at MDACC and VUMC. P-value represents the result of the Mantel-Cox Log-rank test.

Figure 2:. Ctla4^+/− Pdcd1^−/− mice present with cardiac immune infiltration.

A) 20X H&E images of lymphocytic infiltration in Ctla4^+/− Pdcd1^−/− mouse (left) and human (right; autopsy sample from myocardium of a patient that had complete heart block and ventricular tachycardia following ICI treatment.) B) Quantification of lymphoid infiltrate scores from H&E stained heart tissue (see methods) and frequency of CD3, CD4, and CD8+ cells (IHC) as a fraction of total nucleated cells. P-value represents result of unpaired Student’s T-test with Welch’s correction. C) Representative images of CD3, CD4, and CD8 immunohistochemistry (right) stained heart tissue sections from female Ctla4^+/− Pdcd1^−/− mice. Left panels: IHC; Middle panels: segmentation with red=positive and blue=negative cells; right panels: positive cell density (blue=low, green=intermediate, yellow=high). Heatmap values represent arbitrary density units. D) Representative images of additional immunohistochemistry (CD3, F4/80+ macrophages and Foxp3+ Tregs) stained heart tissue from Ctla4^+/− Pdcd1^−/− mice. E) Flow cytometry analysis of immune populations in murine cardiac tissue. *ANOVA p<0.05; +ANOVA p<0.10. Dashed line at 9% CD45+ cells was used as a stratification factor for ‘high’ versus ‘low’ infiltrated samples in Fig. 3. See also Supplemental Figure 9A for analyses as percentage of total CD45+ immune cells. F) Kaplan-Meier survival curve of Ctla4^+/− Pdcd1^−/− Rag1^−/− (n=10) and littermate RAG1 competent Ctla4^+/− Pdcd1^−/− Rag1^+/+ and Ctla4^+/− Pdcd1^−/− Rag1^+/− mice (n=8 and 16, respectively). Littermate Ctla4^+/− Pdcd1^+/− Rag1^+/− and Ctla4^+/− Pdcd1^+/− Rag1^−/− mice are also displayed (n=99 and 51, respectively).

Figure 3:. Functional cardiologic manifestations of autoimmune myocarditis in Ctla4^+/− Pdcd1^−/− mice are similar to those occurring in patients.

A) Representative echocardiograms from Ctla4^+/+ Pdcd1^−/−, low immune infiltrate (<9% CD45+ cells) Ctla4^+/− Pdcd1^−/−, and high immune infiltrate (≥9% CD45+ cells) Ctla4^+/− Pdcd1^−/− mice. B) Quantification of cardiac properties and output in Ctla4^+/+ Pdcd1^−/−, low immune infiltrate (<9%) Ctla4^+/− Pdcd1^−/−, and high immune infiltrate (≥9%) Ctla4^+/− Pdcd1^−/− mice. LVPWs: Left Ventricular Posterior Wall (systole); IVSd: Interventricular Septum (diastole); HR: Heart Rate; LVEF: Left Ventricular Ejection Fraction *p < 0.05 and ⁺p <0.10 Kruskal-Wallis test with Dunn’s multiple comparison. C) Representative electrocardiography records from wild-type C57BL6/J (left panel) and Ctla4^+/+ Pdcd1^−/− mice demonstrating normal sinus rhythm. D) Representative electrocardiography records from Ctla4^+/− Pdcd1^−/− mice demonstrating sinus arrest and AV block. E) Percent of mice in wild-type C57/BL6, Ctla4^+/+ Pdcd1^−/− , and Ctla4^+/− Pdcd1^−/− genotypes or Ctla4^+/− Pdcd1^−/− treated with abatacept demonstrating observed conduction disorders. F) Representative electrocardiograms (ECG) from 3 patients with ICI-associated myocarditis. All patients were treated with ipilimumab and nivolumab with clinical presentation within 2 weeks of dose with ECG on admission. There is 3:2 conduction block and QRS widening (left panel), third degree heart block with ventricular pacing (middle panel), and ventricular pacing competing with an accelerated ventricular rhythm (right panel).

Figure 4:. Modulation of CTLA-4 and PD-1 T cell negative costimulation leads to functional changes in cardiac pathology.

A) Kaplan-Meier survival curve of female Ctla4^+/− Pdcd1^−/− mice treated with either vehicle (n=51) or abatacept (n = 19). P-value represents the outcome of the log-rank test. B) Infiltration of CD45+ immune cells assessed by flow cytometry analysis of heart tissue from female Ctla4^+/+ Pdcd1^−/− and Ctla4^+/− Pdcd1^−/− mice with and without abatacept treatment. *, p < 0.05 ANOVA with Tukey’s multiple testing correction. C) Heart-weight/body-weight (HW/BW) ratio of female Ctla4^+/+ Pdcd1^−/− and Ctla4^+/− Pdcd1^−/− mice with and without abatacept treatment. ** p < 0.01; *p<0.05 ANOVA with Dunn’s multiple testing correction comparing all groups to Ctla-4^+/− Pdcd-1^−/− control mice.