This is the first reported clinical study, to the authors’ knowledge, which confirms myocardial PD-L1 expression in the human heart using non-invasive imaging in vivo. We used a novel SPECT tracer, [99mTc]NM-01, which is highly sensitive and specific for PD-L1, for the assessment of PD-L1 expression in patients with NSCLC [8, 9]. We demonstrated the ability to non-invasively quantify PD-L1 activity in the myocardium, that heterogeneity of expression exists within and between individuals and that there was no significant difference in measured PD-L1 expression on serial imaging during ICI therapy in this cohort. Moreover, we confirmed PD-L1 expression of the myocardium was greater than skeletal muscle, thus lending further support for the specificity of this tracer for PD-L1 expression.

Previous studies have reported that positron emission tomography (PET) tracers such as [18F]BMS-986192 and [89Zr]atezolizumab can be used to non-invasively quantify PD-L1 expression in NSCLC tumours in humans [16, 17]. However, these studies have not reported on cardiac PD-L1 expression. Mouse models have shown that PD-L1 expression has an important role in cardiac disease. One study reported that mice with a genetic deletion of PD-1 ligands that were treated with PD-L1 antibody therapy resulted in death from induced myocarditis [18]. Additionally, PD-1 expression has shown to have a protective effect as PD-1 deficient mice had increased myocardial inflammation and inflammatory cell infiltration in a model of experimental myocarditis [19]. PD-1 knockout mice developed severely impaired biventricular systolic function as measured with echocardiography and were found to have immunoglobulin deposition on the cardiomyocytes [20]. This later study suggested that PD-1 may have a protective role in the development of dilated cardiomyopathy from autoimmune disease. More recently, PD-L1 expression has been implicated in acute cellular rejection following heart transplantation [21]. Thus, PD-L1 expression is likely to have an important role in a range of cardiac disease, but these studies have been confined to preclinical experimental animal studies or those requiring invasive myocardial biopsy. Our study is the first to confirm the presence and expression of PD-L1 in myocardium non-invasively in the human heart in vivo. [99mTc]NM-01 has favourable imaging properties as it binds to a different domain of PD-L1 to therapeutic monoclonal antibodies, which means it should not be blocked by anti-PD-L1 immunotherapy agents [9]. Our study provides an opportunity to apply the novel imaging biomarker [99mTc]NM-01 SPECT/CT to assess the biological activity of PD-L1 non-invasively, which may be applied to obtain mechanistic understanding of a wide range of cardiac diseases.

Immune checkpoint inhibitors such as pembrolizumab block the PD-1 pathway; a potential mechanism for subsequent myocarditis may relate to blocking these protective pathways on the cardiomyocytes. [99mTc]NM-01 may be used to understand the mechanism of ICI-associated myocarditis, which is associated with T cell-mediated infiltration of the myocardium [22]. ICI-associated myocarditis is a challenging clinical diagnosis and requires the integration of symptoms, blood biomarkers such as troponin and cardiovascular magnetic resonance (CMR) or 18F-FDG PET imaging [23]. Endomyocardial biopsy is recommended in cases when there are uncertain CMR or PET findings or the patient cannot undergo non-invasive assessment due to hemodynamic instability [24]. However, sampling error is a well-known limitation of endomyocardial biopsy due to the heterogeneous myocardial pattern of various cardiac diseases, and therefore, it is recommended that at least five samples should be taken from different sites [25]. The issue of heterogeneity may potentially be overcome with non-invasive imaging, which provides an assessment of complete myocardial coverage and thus may allow global myocardial assessment of PD-L1 expression and heterogeneity compared to endomyocardial biopsy.

Importantly, there is no validated risk prediction tool, or biomarker that can accurately predict which patients are at highest risk of developing ICI-associated myocarditis. It may be plausible that the non-invasive quantification of PD-L1 expression in the myocardium may be used as an imaging biomarker to predict which patients are at highest risk of myocarditis. For instance, based on the importance of PD-1 in mouse models, it may be possible that patients with low PD-L1 expression at the time of cancer diagnosis and before ICI treatment are at increased risk of ICI-related myocarditis. However, future clinical studies are required to understand this association. This has potential clinical application as it may aid in the identification of patients at highest risk of developing myocarditis in the future, and guide diagnosis and frequency of surveillance for cardiovascular toxicity in patients undergoing immunotherapy.

Another potential application for the assessment of PD-L1 myocardium expression is in patients with dilated cardiomyopathy. There are various causes of dilated cardiomyopathy, which include genetic, toxins, idiopathic, infective, inflammatory, infiltrative and autoimmune conditions [26]. In the extensive workup of such patients, often no clear underlying aetiology is demonstrated and thus considered to be idiopathic. Knockout PD-1 mice have been shown to develop biventricular systolic dysfunction [20]. One potential avenue to consider is whether patients with dilated cardiomyopathy may have reduced PD-L1 activity, which may be investigated in a well-characterized cohort of patients. This may aid understanding of the pathophysiological mechanism in dilated cardiomyopathy and thus guide protective medical therapy to prevent the development of adverse left ventricular remodelling.

There are some important limitations to acknowledge from this study. Firstly, this was a single-centre study with a small number of patients, but nevertheless consistently demonstrates the first in man feasibility of this imaging biomarker to measure myocardial PD-L1 expression. Secondly, patients with established NSCLC were considered, rather than a wider range of cancers. Thirdly, there was no control group of healthy volunteers to determine the potential effect of active malignancy on myocardial PD-L1 expression. Finally, an invasive biopsy of the myocardium was not obtained, which would have been useful to confirm human myocardial PD-L1 expression at the gene and protein levels and the cells where expression is observed. However, previous preclinical data confirmed high specificity of [99mTc]NM-01 for PD-L1 [8], our previous study demonstrated good correlation with immunohistochemistry PD-L1 in lung cancer patients [9] and this current study showed high myocardial activity compared to striated muscle, which all provide strong indirect evidence of PD-L1 specificity in the myocardium for [99mTc]NM-01. The physical limitations of SPECT compared to PET with regard to spatial resolution are acknowledged, but nevertheless our image data confirmed good myocardial to background contrast and was therefore readily quantifiable and reproducible. Potential further improvements could be made with cardiac and respiratory gating.