Introduction
Immune checkpoint inhibitors (ICIs) have revolutionized cancer therapy over the last decade [
43], since the approval of the first ICI, ipilimumab, in 2011 [
6]. Currently, antibodies targeting four immune checkpoints, namely cytotoxic T lymphocyte-associated antigen 4 (CTLA-4), programmed death 1 (PD-1) and its ligand (PD-L1), and lymphocyte activation gene 3 (LAG-3), are approved by the United States and European regulatory authorities as anticancer agents, either as monotherapy or as adjuvant therapy. Near 50% of all patients with metastatic malignancies are under ICI therapy [
25]. ICIs maintain a long-lasting antitumor potential, whereas their combination therapies present increased efficacy [
58]. Nevertheless, ICI-induced immune-related adverse events are observed. These are triggered by the dysregulation of T-cell immunologic self-tolerance, which might affect multiple organs, including the myocardium [
21,
58,
82]. Although infrequent, cardiotoxicity may be life-threatening. The molecular basis of these immune-related adverse events remains marginally understood, but immune mechanisms are highly implicated [
80].
Current European Society of Cardiology (ESC) Cardio-Oncology guidelines emphasize the cardiovascular complications of anticancer therapies. Cardiovascular diseases and cancer share common confounders and seem to be cross-linked through cardiovascular toxicities [
30]. Regarding ICI-related cardiovascular adverse events (CVAEs), there is an unmet clinical need for efficient management [
48]. The largest observational, retrospective, pharmacovigilance study of 122 patients with ICI-associated myocarditis presented an early onset of symptoms, which resulted in 50% mortality in affected patients [
72]. Long-term CVAEs (> 90 days) are less well-characterized but are generally manifested in the form of noninflammatory heart failure (HF), accelerated atherosclerosis, and hypertension, resulting in increased mortality rates [
14]. Prompt diagnosis and initiation of high-dose corticosteroids within 24 h are important mitigation strategies to improve the outcomes of affected patients [
86]. However, due to the shortage of evidence-based recommendations, the monitoring and management of ICI therapy-related CVAEs remain elusive [
48].
Up-to-date, all established in vivo models of anti-PD-1 cardiotoxicity have used anti-PD-1 antibodies that are not clinically applicable, but are reactive specifically with the murine PD-1 [
22,
54,
84,
87]. Besides the limitation in translation, regarding the use of nonclinically relevant anti-PD-1 antibodies, in vivo studies have also employed an aggressive dose regimen, that of 200 μg/animal, correlating to a human equivalent dose of 8 mg/kg, which is 4 times higher and near 2 times higher than the approved dose for pembrolizumab (Pem) and nivolumab, respectively [
23]. Consequently, basic science lacks appropriate preclinical models, to investigate ICI-induced CVAEs. Moreover, despite the fact that some underlying cardiotoxicity mechanisms have been proposed, including the imputation of ICI-related endothelial dysfunction [
54], the exact pathomechanism of ICI-induced cardiotoxicity remains elusive.
The scope of the current study was to establish a translational approach, elucidating ICI-induced cardiotoxicity, seeking to i) investigate the drug or class effect of ICIs on cardiac homeostasis, ii) establish translational in vitro and in vivo models, with clinically used ICIs, by performing in vitro experiments on isolated primary adult murine cardiomyocytes (pAVCs) and splenocytes and in vivo experiments, implementing state-of-the-art functional analyses such as echocardiography, cardiac magnetic resonance imaging (cMRI) and Doppler coronary blood flow velocity (BFV) mapping, on nontumor-bearing mice, iii) verify the cross-reactivity of Pem with the respective murine epitopes, by biotechnological production of both mouse and human epitopes and circular dichroism (CD) and in silico analyses, with the human epitopes to be used as a positive binding control and to confirm it in vivo by flow cytometry experiments, iv) scrutinize the underlying molecular mechanisms of ICI-induced cardiotoxicity, in a time- and dose-dependent manner and establish strong causal relations between molecular signaling and the observed phenotype, by immunoblotting, confocal microscopy and flow cytometry experiments, v) confirm the mechanistic findings in human cell-based in vitro studies on human peripheral mononuclear cells (PBMCs) and human endothelial EA.hy926 cells and vi) discover an evidence-based translational therapy against ICI-induced cardiotoxicity, which shall not hamper their antitumor potency, but concomitantly prevent cardiovascular complications.
Discussion
Herein, we investigated the class and drug effect of ICIs on cardiac function and we have established, for the first time, a translational in vivo model of Pem-induced cardiotoxicity, successfully identifying an evidence-based and clinically applicable prophylactic therapy. Our in vitro findings suggest that among the anti-PD-1, anti-PD-L1 and anti-CTLA-4 antibodies, only Pem exerted notable IC-mediated toxicity in murine pAVCs. The rationale for the use of primary splenocytes relies on the fact that we sought to prove that Pem does not lead to direct cytotoxicity on primary adult cardiomyocytes, while treatment of pAVCs with Pembrolizumab-conditioned media from splenocytes, consisting primarily of immature B and T cells [
13], led to significant cytotoxicity in the cardiomyocytes in vitro
, confirming the IC-mediated nature of Pem’s cardiotoxicity. This finding is in line with clinical observations, as anti-PD-1 therapy presents the highest incidence of CVAEs compared to the other ICIs classes [
70]. Since we observed that Pem’s pharmacological effect on primary splenocytes was clinically relevant to the T-cell activation observed in humans [
67], we investigated whether the observed effect was on- or off-target.
In vitro studies have previously shown that Pem does not cross-react with the murine PD-1, which up-to-now hindered the establishment of translational in vivo models [
44]. However, in all the aforementioned studies, enzyme-linked immunosorbent assays have been employed to test the cross-reactivity of the drug [
44]. Herein, we biotechnologically engineered both the human and murine PD-1-EDs, to investigate the cross-reactivity of the antibody in a temperature- and conformational-dependent manner by CD. Initially, we investigated the similarity of the murine and human PD-1-EDs, concerning the estimated secondary structure content %. We found that both proteins have similar secondary structure content %. Taking into account that human PD-1-ED deranges its secondary structure in order to bind to Pem, similar secondary structures of the human and murine PD-1-EDs facilitate the putative cross-reactivity of the antibody [
85]. Nevertheless, we observed that murine PD-1-ED undergoes conformational changes at 37 °C, which greatly affects the binding of Pem on the murine epitope. Taking into consideration that enzyme-linked immunosorbent assays often require long incubation time at room temperature or 37 °C [
44], our findings on the temperature stability of the protein might explain the reported absence of cross-reactivity of Pem. Previous studies have shown that Pem’s stability and its antigen–antibody binding with the PD-1-ED are affected by stressors such as temperature (≥ 40 °C) [
79]. However, the epitope’s temperature stability is not yet investigated and constitutes a novelty of the current study. Therefore, the up-to-now limitations in Pem’s cross-reactivity should be further scrutinized. Additionally, Pem’s binding to murine PD-1-ED, similar to the human PD-1-ED, was confirmed by in silico protein–protein docking experiments, which additionally supported our previous CD findings. Importantly, we sought to confirm Pem’s cross-reactivity in vivo at 5 weeks. We found that in compliance with humans [
11,
78], Pem increased the total T-cell count in the whole blood, compared to the IgG4 isotype control, without affecting the B- and NK-cell population. The latter solidifies Pem’s cross-reactivity in our murine model of cardiotoxicity, as Pem induced a clinically relevant immune response in vivo, in line with its pharmacological effect in cancer patients.
Taking into account that
i. Pem induced a Th17-type cell activation in primary splenocytes, which is similar to the T-cell activation observed in humans [
67] and led to IC-mediated cardiotoxicity in the pAVCs,
ii. human and murine PD-1-EDs presented similar estimated secondary structure content % facilitating the pharmacodynamic interaction of both PD-1-EDs with Pem,
iii. the similar CD shift of Pem’s spectrum upon incubation with the human and murine PD-1-EDs,
iv. the in silico confirmation of Pem’s binding with the human and the murine PD-1-EDs originating from low-energy models, and
v. our in vivo confirmation of Pem-induced T-cell expansion at 5 weeks of administration, we can safely suggest the cross-reactivity of Pem with the murine PD-1-ED, regarding its binding capacity to its ligand PD-L1. These results are of utmost translational significance, as they permit, for the first time, the conduction of in vivo experiments with anti-PD-1 therapeutics, using Pem as a prototype. As for the Pem dosage, the previous preclinical studies have used
i) anti-PD-1 antibodies that are not clinically applicable and
ii) 4 times higher doses than the approved dose for Pem. Herein, we established for the first time an in vivo model directly translating the human dose into our murine in vivo model on a bench-to-bedside approach.
Endothelial activation and microvascular coronary endothelial dysfunction were identified as early mediators of Pem’s cardiotoxicity. Early reports indicate that endothelial PD-L1 orchestrates CD8
+ T-cell-mediated injury in the myocardium, demonstrating an important role of the IFN-γ-inducible PD-L1, in protecting the myocardium against ICIs immune-related adverse events. However, this finding was not further investigated in terms of anti-PD-1 therapy-induced CVAEs [
24]. A contemporary study confirmed the expression of PD-L1, mainly in the endothelial compartment of the myocardium and proposed that TNF-α is a key mediator of the early anti-PD-1-related cardiotoxicity [
54]. Despite the fact that endothelial dysfunction is part of the early-on anti-PD-1-related cardiotoxicity, functional and molecular proofs of this mechanism are still elusive. Herein, we have provided novel evidence that Pem after 1 dose increases circulatory Th17-type cytokines’ levels, leading to endothelial activation and microvascular coronary endothelial dysfunction, as proven by cMRI, Doppler coronary BFV mapping, and immunoblotting at 2 weeks, initiating the establishment of cardiac injury. Circulating cytokines and especially IL-17α, IL-2, and TGF-β have high predictive value on immune-related toxicities in melanoma patients receiving anti-PD-1 therapies [
50]. Therefore, a causal correlation of Th17-type cytokines’ acute release, in the first week, can be associated with endothelial homeostasis disruption and can later trigger Pem-induced CVAEs. Moreover, we have shown that endothelial dysfunction, which stands as a predecessor of severe cardiac systolic dysfunction, is dose- and time-dependently aggravated by Pem’s administration and consequently leads to exacerbated coronary endothelial dysregulation and inflammation at 5 weeks. The establishment of cardiotoxicity at 5 weeks was confirmed by the elevated cTnI levels in the circulation, which was significantly increased compared to the controls and baseline at this time point. The elevation of cTnI upon the establishment of cardiotoxicity is in line with the clinical observations [
48,
81]. Importantly, ICAM-1 was identified as a novel biomarker of early endothelial activation in our in vivo model.
It is generally appreciated that various manifestations of HF, including ischemic cardiomyopathy, dilated cardiomyopathy, coronary microembolization drug-related cardiotoxicity and tachyarrhythmias, share microvascular endothelial dysfunction as a common confounder [
27,
29,
40]. The identification and early intervention against acute endothelial dysfunction are regarded as a pivotal modality in maintaining cardiovascular homeostasis, as endothelial cells’ resilience to acute stress factors is crucial for preventing chronic cardiac dysfunction [
77]. Moreover, documentation of molecular pathways, involved in endothelial dysfunction, can enable the identification of novel druggable targets against endothelial-related functional deficits in cardiovascular diseases [
39,
76]. Identification and management of Pem-related early coronary endothelial dysregulation are of great clinical value, as in a recent clinical study on ICI-induced CVAEs, vascular-driven CVAEs, such as vasovagal syncope, acute myocardial infarction and microvascular dysfunction, were observed within the spectrum of ICI-induced cardiovascular complications [
3]. Therefore, the pharmacological management of ICI-derived early coronary endothelial dysregulation might serve as a druggable target for the mitigation of both cardiac dysfunction and vascular-driven CVAEs by anti-PD-1 therapy. Importantly, microvascular endothelial dysfunction is an independent predictor of cancer development and progression, as it is shown that patients with nonobstructive coronary artery diseases have a higher incidence of malignancies [
34,
65]. Targeting microvascular coronary endothelial dysfunction in anti-PD-1 therapy might facilitate the parallel treatment of both cancer progression and CVAEs.
Therefore, subsequently, we challenged the prophylactic potential of high-intensity statins against Pem-induced cardiotoxicity. Our rationale for selecting statins was supported by a recent clinical study, exhibiting that statins are associated with improved antitumor efficacy of anti-PD-1 therapy in malignant pleural mesothelioma and advanced nonsmall-cell lung cancer patients [
7]. We employed two statins with proven endothelial protective potential [
83] and different intensities as potential cardioprotective candidates. Our in vitro human-based studies deduced that only Atorv, at a high dose, mitigated Pem-induced endothelial toxicity, while it also prevented early and late cardiac histological, functional, and molecular deficits induced by Pem in vivo. The lack of Prav’s cardioprotective potential can be accredited to the lower lipophilicity and intensity compared to Atorv leading to differential potency on the endothelium [
83]. Though the putative antitumor potential of statins is already studied and might be attributed to their direct effect ontumor cells, downregulating PD-L1 [
45] and suppressing tumor escape by inhibiting PD-L1 trafficking [
10], statins’ effect on anti-PD-1-related CVAEs is not yet investigated. This is of great interest, considering the proven prophylactic effect, especially of Atorv, against anthracycline-induced cardiotoxicity [
63]. The potential prophylactic effect of Atorv against doxorubicin-induced cardiotoxicity is previously revealed by the STOP-CA clinical trial on 300 patients with lymphoma. In this study, Atorv 40 mg reduced the incidence of cardiac dysfunction, as evaluated by left ventricular EF % decline [
59]. The latter clinical trial reinforces the high prophylactic potential of Atorv in the cardio-oncology setting, regarding the anthracycline-induced cardiotoxicity. However, its impact on the cardiovascular function of ICI-treated patients is not yet investigated. Taking under consideration that anthracycline- and ICI-induced cardiotoxicity present differences in the mechanism and manifestation of CVAEs [
42], our study is merited with novel findings on the prophylactic potential of Atorv also in ICI-induced cardiotoxicity, besides its already supported beneficial effect against anthracycline-induced cardiomyopathy.
Approximately 30% of patients receive statins at the start of their cancer therapy [
7,
15]. In a recent study of 14,902 patients with breast cancer, it was presented that, compared with nonusers, patients receiving statins had a significantly lower risk of cancer-related mortality. The ratios of patients who experienced CVAEs, including cardiovascular death, HF, and arterial or venous events, were similar between statin users and nonusers [
8]. However, it should be noted that patients receiving statins in the aforementioned study were relatively older and had a higher incidence of coronary artery disease, hypertension, and diabetes. Additionally, in the statin-receiving patients, a higher co-medication frequency with angiotensin-converting enzyme inhibitors/angiotensin receptor blockers and antiplatelet agents was observed. Therefore, a linear conclusion on the protection of cancer patients by statin therapy cannot be drawn per se and the presence of cardiovascular comorbidities and comedications seems to complicate their cardiovascular benefits. Data on the effect of different statins on CVAEs in cancer patients are scarce. Among the studies investigating the effect of statins on cancer progression, lipophilic/high-intensity statins seem to have a favorable effect [
4,
46,
47]. Specifically for anti-PD-1 therapy, it is shown that only high-intensity statins improve its clinical potential in the clinical setting [
7]. However, regarding their prophylactic value in patients manifesting anticancer therapy-related cardiovascular complications, both hydrophilic and lipophilic statins may also be cardioprotective during cancer therapy [
26]. Since cardioprotection in cardio-oncology has raised a critical concern on the selection of cardioprotective modalities in the presence of malignancies, it can be assumed that lipophilic/high-intensity statins, such as Atorv, exhibiting concurrent cardioprotective and anticancer potential should be preferred in the cardio-oncology setting.
The clinical arsenal lacks specific prophylaxis against anti-PD-1-related CVAEs. Preclinical studies have already proposed various prophylactic therapies against the observed cardiotoxicity, extending from anti-IL-17α and anti-CD4 or anti-CD8 to anti-TNF-α therapies [
22,
54]. However, the interference of the prophylaxes with the antitumor effect of anti-PD-1 therapy seems to limit the safety and efficacy of these interventions. For instance, anti-CD8 therapy abrogates anti-PD-1 antitumor potential in vivo [
54]. Therefore, targeting T-cell populations to combat ICI-related cardiotoxicity appears to increase the risk of cancer relapse. Although anti-TNF-α therapy seems to mitigate the anti-PD-1-related AEs, without interfering with its antitumor effect [
54], and PD-1 + TNF-α dual blockade might additionally reduce tumor resistance in vivo [
5], combination therapy should be considered with caution. In the heart, it seems that TNF-α contributes to ischemia/reperfusion injury, post-myocardial infarction remodeling, and heart failure development making it a favorable target for cardioprotection [
40]. Despite the fact that acute TNF-α blockade might present a short-term beneficial effect on anti-PD-1-related CVAEs, long-term TNF-α inhibition is implicated with CD8 + -T-cell senescence and toxicity which might lead to a possible cancer relapse [
9]. Therefore, extensive short- and long-term clinical trials must be carefully designed and implemented for the establishment of a solid beneficial potential of the aforementioned combination. Additionally, the pharmacokinetics and pharmacodynamics of the drugs in the combination regimens should be scrutinized in future clinical studies. Herein, we provide for the first time solid evidence that Atorv, a widely used drug that does not interfere with the antitumor effect of Pem (as it has been shown in clinical studies) [
7], can prevent anti-PD-1-related cardiotoxicity.
In our study, male mice were used for the conduction experiments. The use of male animals was selected, as male mice do not present the hormonal fluctuations due to the menstrual cycle observed in female mice, leading to difficulties in data interpretation and increased variability of the results. However, according to sex relevance in cardio-oncology studies [
1], this is a limitation of the study and future studies should be conducted to investigate the sex differences in Pem-induced cardiotoxicity. Another shortcoming of the current study is that data were not validated in a tumor-bearing in vivo model. There is an imperative need to understand the pivotal biological crosstalk between cardiovascular morbidities and malignancies, as on the one hand they may enable the development of novel therapeutic and preventive modalities for both diseases [
52], whereas on the other hand they may reveal novel challenges for cardioprotection [
28]. Despite the fact that prophylactic therapies against new-onset HF in cancer patients have been extensively investigated, studies and indications on cancer management in patients with preexisting HF and data on whether guideline-directed medical therapy for HF should be modified upon cancer diagnosis are still obscure and need further investigation [
73]. It is described that endothelial dysfunction and cancer might share common confounders, namely activation of the Wnt signaling pathway and depression of peroxisome proliferator-activated receptor gamma (PPAR gamma) signaling [
53], which might link endothelial microvascular dysfunction and cancer. These links can facilitate the identification of high-risk individuals for developing malignancies and may permit the improved insight from healthcare providers to risk-stratify these patients. Also, they might further support the concept of joint pharmacologic strategies against cardiovascular diseases and cancer [
52]. Besides the interplay of cancer and microvascular endothelial dysfunction, it is well-known that cancer has a direct negative impact on the myocardium. Cancer itself may pose a major burden to cardiovascular homeostasis, with a significant impact on the manifestation of CVAEs, whereas cardiovascular disease may also accelerate tumor progression [
55]. The interplay between cancer and cardiovascular outcomes is also evident in the clinical arena, as lung cancer patients with high tumor burden, receiving ICI therapy, manifested more frequently severe immune-related adverse events, than the low tumor burden individuals [
71]. Therefore, the validation of our data on a murine model of malignancy is of utmost importance, regarding the complex regulatory circuits between the tumor, endothelial cells and cardiac dysfunction, which will be investigated in future preclinical and clinical studies. Finally, to the best of our knowledge, cardio-oncology preclinical and clinical studies have not yet identified the optimal dose regimen for atorvastatin cardioprotection, while only the high doses of statins are investigated in contemporary preclinical and clinical studies [
20,
59,
61]. Additional studies on statins’ dose titration, for achieving prophylaxis against anticancer agent-related cardiotoxicity, should be performed. In our study, only the high translational dose of atorvastatin was investigated according to previous preclinical studies [
20,
61]. However, the lack of atorvastatin dose titration in our in vivo model of Pem-induced cardiotoxicity is a limitation of the study.