Discussion

This is the first international and multidisciplinary Delphi consensus project among experts on clozapine or myocarditis to synthesise evidence generated from a systematic literature search and evaluation and real-world-experience of clozapine initiation and rechallenge. Previous monitoring or rechallenge protocols for CAM have mostly been derived from case reports, case series and expert opinions, ^{Reference Ronaldson, Fitzgerald, Taylor, Topliss and McNeil28,Reference Griffin, Woznica, Gilotra and Nucifora34,Reference Shivakumar, Thomas, Sollychin, Takács, Kolamunna and Melgar37} rather than robust, consensus-based frameworks, and only three CAM protocols have been developed involving both psychiatrists and cardiologists. ^{Reference Qubad, Dupont, Hahn, Martin, Puntmann and Nagel13,Reference Griffin, Woznica, Gilotra and Nucifora34,Reference Shivakumar, Thomas, Sollychin, Takács, Kolamunna and Melgar37} Furthermore, the variability in recommendations (e.g. with respect to timing and duration of monitoring) derived from current CAM protocols can lead to inconsistency in clinical practice and complicate efforts to identify myocarditis early, potentially leading to either under- or overreporting of CAM.

In a meta-analysis by Siskind et al, an event rate of 0.7% was reported for CAM. ^{Reference Siskind, Sidhu, Cross, Chua, Myles and Cohen8} Recently an incidence of 8.5% CAM risk was reported in a part of Australia employing a mandatory screening programme for CAM. ^{Reference Nachmani Major, Dawson and Clark32} Reported mortality rates after CAM vary between 21 and 50%. ^{Reference Bellissima, Tingle, Cicović, Alawami and Kenedi11,Reference Merrill, Dec and Goff38} There may be significant underreporting of cases in countries where CAM is not actively monitored, but overreporting is also a consideration, as an Australian chart review found that 13 of 20 (65%) cases did not meet the ‘Ronaldson criteria’ for a clinical diagnosis of CAM. ^{Reference Winckel, Siskind, Hollingworth and Wheeler39} Here, although elevated troponin was present in 17 of 20 (85%) cases, other parameters were either not assessed or not indicative of myocarditis. ^{Reference Winckel, Siskind, Hollingworth and Wheeler39} In a retrospective record database analysis from the UK, among 228 patients with suspected CAM, only 11.4% were confirmed as having probable CAM; this highlights the need for clear CAM criteria. ^{Reference Segev, Iqbal, McDonagh, Casetta, Oloyede and Piper40} As a central outcome of our project, we established clear criteria for CAM based on guidelines for myocarditis related to immune checkpoint inhibitor treatment. This could help to reduce overreporting, as, according to expert consensus, confirmation of CAM now requires mandatory use of at least one cardiac imaging technique, such as echocardiography or cardiac MRI. Transthoracic echocardiography is far more available globally, making it a more pragmatic choice in many settings, even though it may lack the detailed diagnostic precision of cardiac MRI. Another significant finding of our consensus Delphi was the critical importance of slow clozapine titration; this is supported by the fact that countries where slower titration in out-patient settings is widely practised, including Denmark, The Netherlands and the UK,have reported low rates of CAM. ^{Reference Rohde, Polcwiartek, Kragholm, Ebdrup, Siskind and Nielsen41–Reference Schulte, Bogers, Bond-Veerman and Cohen44} This suggests that the titration scheme proposed by the clozapine manufacturer may be too rapid. Our consensus also confirmed the diagnostic value of hs-troponin and CRP testing in screening for CAM, whereas certain clinical symptoms including sinus tachycardia were considered to be of minor diagnostic value. ^{Reference Segev, Iqbal, McDonagh, Casetta, Oloyede and Piper40} Nevertheless, sinus tachycardia may be both a transient phenomenon during clozapine up-titration and a non-specific marker of potential CAM. ^{Reference Ronaldson, Fitzgerald, Taylor, Topliss and McNeil28} Guidelines from Denmark and The Netherlands do not consider CRP or hs-troponin screening to be mandatory ^{Reference Nielsen, Young, Ifteni, Kishimoto, Xiang and Schulte45} and only recommend testing these parameters as part of a diagnostic evaluation when CAM is suspected. ^{Reference Nielsen, Young, Ifteni, Kishimoto, Xiang and Schulte45} There is no mandatory screening or monitoring guideline for CAM in the SmPC; this means that the protocol of Ronaldson et al, which does specify monitoring parameters but not clozapine titration speed, is adopted in most cases. ^{Reference Ronaldson, Fitzgerald, Taylor, Topliss and McNeil28}

Guidelines in Australia, New Zealand and Germany require mandatory weekly CRP and troponin assessments during the first 4–6 weeks. ^{Reference Hasan, Falkai, Lehmann and Gaebel46} One previously published protocol suggested weekly monitoring of cardiac troponin, NT-proBNP and CRP for a total duration of 8 weeks, ^{Reference Griffin, Woznica, Gilotra and Nucifora34} whereas our core recommendation is for 4–8 weeks with reduced monitoring frequency in the last 4 weeks; this underscores the necessity for close monitoring in the first month, in which most CAM cases occur according to the literature. ^{Reference Polcwiartek, O’Gallagher, Friedman, Correll, Solmi and Jensen9} It is also important to consider potential false-positive elevations in troponin levels, such as those caused by heterophile antibodies in infectious mononucleosis, fibrin clots, renal dysfunction, pulmonary embolism or rheumatoid factor. ^{Reference Chaulin47,Reference Januzzi and McCarthy48} Other causes of elevated troponin, including sepsis and renal impairment, should also be thoroughly evaluated to avoid misdiagnosis. Moreover, there is considerable variability in the sensitivity of troponin platforms, with variations possible even within the same pathology laboratory; thus, using a threshold of twice the upper limit of normal could lead to false-positive troponin results. To reduce false positives, it may be beneficial to implement local cut-offs for (a) acceptable minor elevation, (b) elevation that warrants repeat testing and (c) clear elevation ^{Reference Clark, Wilton, Dawson, Chiew, Jawahar and Toben49} in the context of concurrent causes of elevated troponin. CRP alone has been suggested as an effective screening marker for inflammation in the absence of genetic testing for clozapine poor-metaboliser status, ^{Reference de Leon, Schoretsanitis, Smith, Molden, Solismaa and Seppälä21} as its levels often rise several days before troponin, ^{Reference Nachmani Major, Dawson and Clark32,Reference Mcneil, Ronaldson, Fitzgerald, Taylor, Topliss and Wolfe50} potentially reducing the need for routine troponin assessments. ^{Reference de Leon, Schoretsanitis, Smith, Molden, Solismaa and Seppälä21,Reference Shelton, Ruan, Ertuğrul, Cotes and De Leon51} Moreover, the literature describes cases of CAM in which CRP elevations occur without concurrent troponin elevation, ^{Reference Ronaldson, Fitzgerald, Taylor, Topliss and McNeil28} a notably rare scenario when using hs-troponin. Nevertheless, although CRP elevation indicates inflammation, it does not reliably demonstrate cardiac involvement, making hs-troponin essential for accurate identification of early myocardial changes associated with clozapine use. This can be juxtaposed with a strategy in which troponin testing is only necessary as confirmation of emerging cardiac issues in suspected CAM cases. ^{Reference Nachmani Major, Dawson and Clark32} Hs-troponin should be incorporated into weekly screening protocols during the first 4 weeks of clozapine treatment, a period of heightened myocarditis risk, owing to its specificity in detecting myocardial injury. Unlike CRP, which is susceptible to fluctuations from various non-cardiac factors, such as infections, smoking and systemic inflammatory conditions, hs-troponin specifically indicates cardiac injury, making it a more reliable marker in this context. CRP elevations can be misleading, reflecting general inflammatory responses rather than myocarditis, and thus may obscure early detection of myocardial involvement. Routine weekly hs-troponin assessments in the initial treatment phase provide a targeted and consistent approach to monitoring, enabling prompt intervention if myocarditis develops.

According to our expert panel, hs-troponin as a singular cardiac laboratory marker is sufficient for effective CAM screening, with NT-proBNP testing reserved for an enhanced monitoring scheme. This approach reduces unnecessary barriers to clozapine use by simplifying screening requirements without compromising patient safety. As genetic testing is not anchored in clinical practice, the issue of incorrect self-identification and mixed ancestries must be mentioned when suggesting slower titration schemes or lower target clozapine doses for individuals of a specific ancestry. Nevertheless, on the basis of slow clozapine metabolism in people with Asian ancestry, ^{Reference Kikuchi, Komatsu, Otsuka, Ito, Kanahara and Tanifuji52,Reference Yasui-Furukori, Otsuka and de Leon53} a maximum titration rate of 25 mg per week for individuals of Japanese ancestry has been proposed, ^{Reference Kikuchi, Komatsu, Otsuka, Ito, Kanahara and Tanifuji52,Reference Inada, Oshibuchi, Ishigooka and Nishimura54} and the available evidence regarding slower metabolisers should be taken into account to reduce CAM risk.

Although the role of eosinophils in CAM diagnosis has been debated over time, ^{Reference Kilian, Kerr, Lawrence and Celermajer14,Reference Ronaldson, Fitzgerald, Taylor, Topliss and McNeil28} we also included FBC, as it has previously been suggested that monocytosis may be an early marker of CAM, whereas eosinophilia may be a late marker of a hypersensitivity reaction. ^{Reference Shivakumar, Thomas, Sollychin, Takács, Kolamunna and Melgar37} Ultimately, both of these alterations should prompt enhanced CAM monitoring. Eosinophilia typically occurs after troponin elevation, ^{Reference Ronaldson, Taylor, Fitzgerald, Topliss, Elsik and McNeil55} which has led to the exclusion of eosinophils in earlier monitoring protocols ^{Reference Ronaldson, Fitzgerald, Taylor, Topliss and McNeil28} ; however, the assessment does not impose an additional burden in the context of mandatory weekly absolute neutrophil count monitoring during the period of CAM monitoring, and the proportion of CAM survivors with eosinophilia has been reported to be as high as 66%. ^{Reference Ronaldson, Taylor, Fitzgerald, Topliss, Elsik and McNeil55} Repeated ECG monitoring was endorsed, similar to previous CAM protocols developed by multidisciplinary working groups, ^{Reference Griffin, Woznica, Gilotra and Nucifora34} as it is considered to be usually abnormal in myocarditis, although ECG signs are neither specific nor sensitive, ^{Reference Caforio, Pankuweit, Arbustini, Basso, Gimeno-Blanes and Felix25} and it is recommended in all individuals with clinically suspected myocarditis. ^{Reference Caforio, Pankuweit, Arbustini, Basso, Gimeno-Blanes and Felix25}

In the previous Australian CAM monitoring protocol from Ronaldson and colleagues, baseline transthoracic echocardiography before commencing clozapine was presumed to be clinically warranted. This would mean that myocarditis, when suspected, could be diagnosed with a high degree of specificity, and appropriate treatment could be established for any cardiac disease detected. ^{Reference Ronaldson, Fitzgerald, Taylor, Topliss and McNeil28,Reference Ronaldson, Fitzgerald, Taylor, Topliss and McNeil56} Nevertheless, this was not recommended in later CAM protocols, ^{Reference Griffin, Woznica, Gilotra and Nucifora34} as from a global and more pragmatic perspective, implementing mandatory baseline echocardiography (defined as a minimal requirement in our consensus) would probably result in fewer individuals with TRS receiving clozapine in healthcare systems where echocardiography is unavailable. In such contexts, in the trade-off between benefits and risks, our experts prioritised close cardiac monitoring during treatment over achieving highly refined risk stratification at baseline, and so baseline echocardiography was only recommended in the enhanced monitoring investigations. By contrast, baseline echocardiography was considered to be mandatory before potential clozapine rechallenge; this is in line with previous rechallenge protocols ^{Reference Qubad, Dupont, Hahn, Martin, Puntmann and Nagel13,Reference Griffin, Woznica, Gilotra and Nucifora34,Reference Shivakumar, Thomas, Sollychin, Takács, Kolamunna and Melgar37} and acknowledges that rechallenge should not be recommended in cases of cardiomyopathy with significantly reduced left ventricular ejection fraction. Whereas previous guidelines recommended against rechallenge, ^{Reference Manu, Lapitskaya, Shaikh and Nielsen57} this study found that the majority of experts approved rechallenge in selected cases after CAM. It is important to caution that clozapine rechallenge after CAM is a multiprofessional clinical decision involving psychiatry, cardiology and pharmacy that has to be carefully considered in terms of risks versus benefits together with the patient and their family.

A previous CAM protocol recommended a waiting period of at least 6 months after an episode of CAM to allow the inflammation to resolve and the myocardium to recover before rechallenge, ^{Reference Griffin, Woznica, Gilotra and Nucifora34} whereas in our Delphi survey, a minimum waiting period of 8 weeks was considered adequate. The limited evidence regarding rechallenge after myocarditis related to immune checkpoint inhibitor therapy indicates that clozapine rechallenge should be a multidisciplinary decision and considered on an individual basis. For example, patients with a mild CAM index episode might be considered for earlier rechallenge (minimum waiting period of 8 weeks), whereas those with more severe CAM may require a longer waiting period, often at least 6 months, as proposed by Griffin et al. ^{Reference Griffin, Woznica, Gilotra and Nucifora34} According to the expert panel, the shorter timeframe for individuals with a mild index episode allows clinicians to safely monitor resolution of inflammation while minimising the risk of inadequately treated psychosis.

Our algorithms will provide guidance for management of selected cases in which rechallenge is considered to be a viable option. The recommendations in this consensus statement should be validated using prospective CAM registry-based studies ^{Reference McMahon, Wagner, Hasan, Burrage, Amarena and Fox31} with standardised interventions and outcome measurements. The results of our Delphi process can be regarded to represent a consensus guideline, as we used a structured methodology to gather expert opinions and progressively refine them towards a common agreement. This process is particularly valuable in fields such as CAM, in which empirical evidence may be limited, but expert clinical experience and judgement are critical. By incorporating a diverse range of expert insights and iterating through multiple rounds, the Delphi method ensures that the final guidelines represent a broad-based consensus that is both robust and well informed, resulting in a reliable framework for clinical decision-making. In summary, our multidisciplinary Delphi process represents the most representative guideline-related endeavour regarding prevention, screening and monitoring of CAM to date, and its results may inform clinicians and future guidelines.