The cytokine release assay & immunological safety, by Pr. J. Descotes



Even before interleukin-1 was cloned (1984), some adverse effects of early immunomodulatory drugs, such as levamisole or tilorone, were hypothesized to be due to ‘direct macrophage activation’. Subsequently, a variety of vaccines, therapeutic proteins or monoclonal antibodies were shown to induce the release of cytokines, such as IL-1, IL-6 and TNF-a. The commonest reactions involving cytokine release are ‘flu-like reactions’ with moderate fever (<40°C), chills, arthralgias, myalgias, nausea and fatigue. Paracetamol (acetaminophen) can alleviate clinical signs, or accelerate full recovery. Typical reactions are noted after vaccine recall injections, especially in children. After the introduction of potent recombinant cytokines and monoclonal antibodies, severe and often dose- and/or treatment-limiting adverse reactions were reported: ‘acute cytokine release’ syndromes are characterized by marked hyperpyrexia (>40°C), severe clinical signs of a flu-like reaction, together with cardiovascular disturbances (e.g. hypotension or collapse) and/or neurological disorders (e.g. seizures, stupor or coma). Finally, hyperacute cytokine release – the so-called ‘cytokine storm’ – is a major feature of avian flu or septic shock with severe acute respiratory syndrome (SARS) and multi-organ failure. Whatever the severity of the clinical presentation, the key event is the release of pro-inflammatory cytokines and inconsistently IL-2 or IFN-g, the mechanism of which can be a direct activation of certain immunocompetent cells or the complement cascade, among the likeliest targets.

Despite many clinical reports and laboratory findings, the risk of adverse reactions linked to cytokine release induced by biopharmaceuticals has been fairly overlooked for two decades, i.e. until the ‘TGN1412 affair’ where very severe adverse reactions were recorded in healthy volunteers enrolled in the first-in-humans clinical trial of this CD28 agonist monoclonal antibody. The next year (2007), the EMA released a guideline with recommendations to mitigate this risk (revision of this guideline is ongoing). Later, the poor, if any, relevance of in vivo monkey studies was elucidated so that an in vitro cytokine assay using human material became the recommended approach, especially when humanized biopharmaceuticals are concerned. This TGN1412 affair proved to be a major blow from a regulatory perspective, and significant after-effects are indeed still obvious today. It is noteworthy that no guideline has yet been released providing clear recommendations on when and how the cytokine-releasing potential of a drug candidate should be investigated. It is nevertheless beyond doubt that EMA or FDA reviewers do expect the submission of such data as they critically contribute to the determination of the MABEL (Minimal Biological Active Dose Level) or MPAD (Minimally Pharmacologically Active Dose) of most biopharmaceuticals, on which the starting dose in first-in-human clinical trials is based. The cytokine-releasing potential of nanomedicines or small immunostimulatory molecules may also have to be addressed case by case.

No extensive consensus on the most adequate study design has yet been reached. The conclusions of an ongoing international inter-laboratory validation study should be known in 2018. Whatever the pending uncertainties, there is a general agreement that the cytokine-release potential of a drug candidate should be investigated, whenever needed or required, in an in vitro assay (CRA) using human cells. Anyway, some investigators tend to prefer whole blood instead of PBMCs and as many support the opposite position, while a third group recommends using both sources simultaneously. Likewise, there is no general agreement regarding the number of samples to be tested (from n=5 to n=20); the panel of cytokines to be measured (even though the major pro-inflammatory cytokines together with IL-2 and IFN-g have clearly been shown to be the most informative endpoints); the number and occasions of serial cytokine measurements. In order to minimize the impact of this current inter-laboratory variability, it is widely agreed that any CRA performed in one given laboratory should include positive reference compounds, such as muromonab or alemtuzumab to ensure benchmark comparison. Despite unsolved issues, the CRA is today an in vitro regulatory assay that should be included in the immunological safety evaluation of most biopharmaceuticals.