CORRESPONDENCE LINK TO ORIGINAL ARTICLE LINK TO INITIAL CORRESPONDENCE

thymic emigrants, which do not respond with such effector functions. Furthermore, Response to ‘Storm forecasting: systemic tumour necrosis factor release would have induced apoptosis in the thymo‑ additional lessons from the CD28 cytes of the HIS mice; however, thymocyte numbers actually increased in response to superagonist TGN1412 trial’ the CD28 superagonist. Most importantly, however, Horvath et al. describe these effects as occurring at “as little as ~0.3 mg per kg” Thomas Hünig when, in fact, the doses that were used by Legrand et al. are reported as mg per mouse, In their correspondence piece on my accelerate recovery from lymphopenia7. which translates into a 50‑fold higher value recently published Comment article (The Obviously, an IL‑2 burst such as that expe‑ (15 mg per kg, 150-fold more than during storm has cleared: lessons from the CD28 rienced in the human volunteers was as the human trial). At the lowest dose tested superagonist TGN1412 trial. Nature Rev. unintended as it was unanticipated. Eastwood (0.03 mg per mouse or 1.5 mg per kg), which Immunol. 12, 317–318 (2012))1, Horvath et al.4 and Römer et al.5 have since shown was still 15‑fold higher than the one used et al. (Storm forecasting: additional lessons that the source of this IL‑2 burst is effector during the first‑in‑human trial, no effects from the CD28 superagonist TGN1412 memory CD4+ T cells, which, as discussed were observed at all. Accordingly, Horvath trial. Nature Rev. Immunol. 31 Aug 2012 in the Comment article, are frequent in adult et al. should rather quote this study as (doi:10.1038/nri3192‑c1))2 do not consider humans but rare in ‘clean’ laboratory rodents. predicting a misleading safety margin over the recent evidence I discussed regarding the The absence of CD28 expression on cyno­ the first‑in‑human dose applied. failure of pre-trial experimentation to molgus monkey effector memory CD4+ Coming back to the topic of my Comment predict the TGN1412‑induced cytokine- T cells now provides a convincing reason why article, the identification of the deficiencies release syndrome (CRS). Instead, their an increase in circulating T cell numbers, but of preclinical in vitro and in vivo systems in critique focuses on the interpretation of not a cytokine burst, was observed in this modelling the TGN1412‑triggered CRS not the pre­clinical data available in 2006 for the preclinical model for TGN1412 (REF. 4). only reminds us to more critically interpret design of the ill-fated trial, including With regard to the effects of TGN1412 on results obtained in those specific assays, but the selection of the cynomolgus primate the human immune system itself, Horvath also to be prepared for unexpected pitfalls of model for testing. and colleagues quote a preclinical data set complex biological test systems in general. I wish to clarify that my Comment arti‑ published shortly after the trial from which Thomas Hünig is at the Institute for Virology and cle did not deal with these issues. Rather, I they would have predicted the CRS expe‑ Immunobiology, Versbacher Strasse 7, pointed out that the new results published rienced by the volunteers. In this study by D-97078 Würzburg, Germany. during the past few years on species-specific Legrand et al.8, mice with a human haemato­ e-mail: [email protected] differences in CD28 superagonist responses poietic stem cell-derived immune system doi:10.1038/nri3192-c2 in vivo and on the pitfalls of in vitro assays (HIS mice) were treated with the murine Published online 31 August 2012 have provided a plausible explanation as to TGN1412 precursor 5.11A1. Although no 1. Hunig, T. The storm has cleared: lessons from the why the preclinical data available in 2006 led acute toxicity was observed and regulatory CD28 superagonist TGN1412 trial. Nature Rev. Immunol. 12, 317–318 (2012). to a misprojection of the risk for toxicity with T cells actually transiently accumulated in 2. Horvath, C. et al. Storm forecasting: additional lessons TGN1412 in humans. Indeed, it is now clear the periphery, the depletion of conventional from the CD28 superagonist TGN1412 trial. Nature Rev. Immunol. 31 Aug 2012 (doi:10.1038/nri3192‑c1). why rodents do not respond to CD28 super‑ human T cells in antibody-treated HIS mice 3. Gogishvili, T. et al. Rapid regulatory T‑cell response treatment with systemic cytokine is seen by Horvath et al. as evidence for prevents cytokine storm in CD28 superagonist treated 3 mice. PLoS ONE 4, e4643 (2009). release , why cynomolgus macaques were not ‘superactivation’, indicative of a threatening 4. Eastwood, D. et al. Monoclonal antibody TGN1412 a relevant primate species for toxicity testing4, CRS. The authors of the study discuss three trial failure explained by species differences in CD28 expression on CD4+ effector memory T‑cells. and why conventional human peripheral possibilities for the mechanism of T cell Br. J. Pharmacol. 161, 512–526 (2010). blood mononuclear cell (PBMC) assays do depletion in that system: antibody-dependent 5. Romer, P. S. et al. Preculture of PBMCs at high cell density increases sensitivity of T‑cell responses, not reveal a cytokine burst in response to cell-mediated cyto­toxicity (ADCC) and revealing cytokine release by CD28 superagonist soluble TGN1412, as they do with OKT3, the complement-mediated depletion (because, TGN1412. Blood 118, 6772–6782 (2011). 6. Hunig, T. Manipulation of regulatory T‑cell number and monoclonal antibody used for comparison in contrast to the IgG4 antibody TGN1412, function with CD28‑specific monoclonal antibodies. both by myself 5 and by Horvath et al. 5.11A1 is of the IgG1 subclass), and sen‑ Adv. Immunol. 95, 111–148 (2007). 7. Elflein, K., Rodriguez-Palmero, M., Kerkau, T. & It is correct that T cell activation, in par‑ sitivity of recent thymic emigrants to the Hunig, T. Rapid recovery from T lymphopenia by ticular interleukin‑2 (IL‑2) production, is induction of cell death by activating signals. CD28 superagonist therapy. Blood 102, 1764–1770 (2003). part of the intended of Whatever the mechanism, it was most likely 8. Legrand, N. et al. Transient accumulation of human CD28 superagonists. This mechanism results unrelated to the acute toxicity observed dur‑ mature thymocytes and regulatory T cells with CD28 superagonist in “human immune system” Rag2−/− in the expansion and functional activation of ing the human trial. Thus, as mentioned γc−/− mice. Blood 108, 238–245 (2006). regulatory T cell populations (for which IL‑2 above, the cytokine burst in humans was 6 + Competing interests statement is required in a non-redundant manner)­ , mediated by effector memory CD4 T cells, The author declares competing financial interests: see web and the expansion of T cell populations to whereas the HIS mice contained only recent version for details.

NATURE REVIEWS | www.nature.com/reviews/immunol

© 2012 Macmillan Publishers Limited. All rights reserved