I'm presenting this on behalf of the working group, and I have no disclosures. So in terms of background for COVID-19 in children and the recovery trial, the problem that was the Royal College of Pediatrics identified early in 2020 was that children were being given experimental therapies without being part of any formal clinical trials, because the focus naturally was on the larger numbers of adult cases. And the National Institute for Health Research approached two of the large platform trials of which recovery was the one that took on pediatric, and we established a working group to consider interventions and develop guidance to both diagnose and to treat and add treatments as appropriate over the course of the pandemic. It was a pragmatic trial in order to ensure maximal enrollment without a burden on the frontline health care staff. So it was low risk from a clinical trials perspective, because initially repurposed drugs were used. There was no requirement for GCP for patients, for staff recruiting at the sites, and support at a national level to gather data. In terms of the pre-PIMS-TS enrollment of children into recovery, of course, this was initially focused on the acute respiratory presentations, which were relatively low in numbers, but dose adjustments were made in order to randomize children to therapy. That's not what I'm going to report here, because these were smaller numbers. And what obviously became apparent later on was the inflammatory multi-system inflammatory syndrome associated with SARS-CoV-2 called PIMS-TS or MIS-C in a small proportion of children, some of, most of whom would improve, but some of whom became quite unwell because of their multi-organ failure requiring intensive care and immune modulation. In order to have a pathway from diagnosis through to appropriate treatment, a Delphi process was undertaken to establish diagnostic criteria and assess the equipoise for clinical trials interventions. The rationale for how recovery approached PIMS-TS was that initial observational data suggested that symptoms resolved after treatment with IVIG and corticosteroids, but that some children were receiving supportive care and recovered, but those, some of them continued to have refractory inflammation and required intensive care, so were being treated with biologics targeting key pro-inflammatory cytokines, such as interleukin-1, TNF, and IL-6. The primary question from clinicians across the country was whether children and young people given corticosteroids had worse outcomes than those given intravenous immunoglobulin. So the aims for recovery were twofold, to address both the larger number of children requiring first-stage therapy and to assess in children and young people admitted to hospital with mild to moderate PIMS-TS, how does IVIG or corticosteroids compare with usual care on the length of hospital stay? And in children with more severe disease, or those who had ongoing symptoms after having received first-stage randomization, or in those where they were enrolled after having received the drugs included in stage one, how did tocilizumab and anakinra, the first anti-IL-6, the second an IL-1 receptor antagonist, compare with usual care on length of hospital stay? Children were enrolled if they were under 18 and had a criteria, fulfilled criteria for PIMS-TS. Exclusions included if Kawasaki disease were thought likely, and if patients had either hypersensitivity to immunoglobulin, human immunoglobulin, or IgA deficiency, they were excluded from the IVIG treatment arm. And for those patients under a year, tocilizumab was not allowed, and for those under, sorry, anakinra was not allowed, and for those under 10 kilos, tocilizumab. The first-stage interventions were compared in a one-to-one-to-one randomization of standard care, no additional treatment over what they'd already had, if anything, at randomization, and then IVIG and methylpregnizolone. In the second-stage randomization, to ensure, to encourage enrollment, the randomization was adjusted at one-to-two-to-two, so children enrolled had an 80% chance of receiving an active therapy, and these were tocilizumab and anakinra compared to no additional treatment. As I mentioned earlier, it was a pragmatic trial, so investigators were free to use no treatment IVIG or steroids as standard care, if deemed clinically necessary, and through the analysis, the effects of each of these treatments were compared against no initial additional treatments in the presence and absence of the other drug. There was obviously a wide spectrum of disease, because it went from children that admitted to the ward to those requiring critical care because of multi-organ failure, and the baseline levels of any drug use at second randomization was collected. The analysis was undertaken on an intention-to-treat basis, using a Bayesian analysis to look at, in a similar way to the adult recovery trial, to look at the effects of the interventions, both individually and despite the fact that patients had different treatments. I'm not a statistician, but the main point from this slide is to say that the statistical approach was to look at what's called the posterior distribution of difference between the outcome of the active treatment and the outcome of usual care, and the posterior probability was a figure that was calculated to look at the difference in outcome, and the higher the number, the more likely a signal of benefit, so a posterior probability of 95% signified a very strong signal of benefit, and similarly, 70% to 80% and 80% to 95% signaled moderate to strong signals, and on the inverse, a lower score below 50 suggested a possibility of harm, a probability of harm of the intervention. The outcome measures, the primary outcome was duration of hospital stay. Secondary outcomes included the use of inotropes and inflammation as measured by baseline-adjusted CRP at day three. You can scan the QR code directly into the paper, which has just come out, as you've heard. This is the first sort of global RCT, and alongside Lauren Schlappbach's trial in children in Switzerland, we looked at COVID-19. It's the largest one for PIMS-DS, recruiting from May 2020 to January 22. 51 hospitals recruited children to the first randomization, and then the 25 recruited into R2, second-level randomization, so looking at the total numbers, the number of children considered as having a diagnosis of PIMS-DS was 244, of which some entered second randomization directly, 23 of them, and 214 were randomized to either usual care, methalpred, or IVIG, and then in the second randomization, there were 70 children, and as I mentioned earlier, it was a one-to-two-to-two randomization to usual care, tocilizumab, or anakinra. The number, reason for the low number of anakinra is because it was introduced later on during the course of the pandemic, and therefore, obviously, there were smaller numbers of children enrolled to the trial at that point. In terms of baseline characteristics, the average age of patient, children involved in first-line randomization was at around nine years of age. They presented to hospital around six to seven days of symptom onset, and within the first day or so of hospitalization, they were randomized and all had a relatively strong inflammatory response as measured by their CRP. Second-stage randomization, similar age distribution of about nine years of age, and it was interesting to see a higher proportion of black and Asian minority ethnic children in the tocilizumab arm, but that's just an observation. Otherwise, the presentation occurred that these children were randomized at about two days post-hospitalization. Now, this slide looks at the treatment given, and as I mentioned earlier, because it was a trial that allowed clinicians the freedom to treat as they felt clinically necessary at the time, but in order to maximize enrollment to the study, what we did see was a high level of treatment given in patients who had been randomized to usual care. As you can see here, for IVIG, 35% of patients randomized to usual care received IVIG, and similarly, for methylprednisolone, 50% of children randomized to usual care received the steroids regardless. Second-stage randomization numbers were much better. Certainly, for tocilizumab, only those randomized to active treatment received it compared to the usual care arm for that comparison, and for anakinra, a small number did receive it, but otherwise, the majority did not. When we look at the Bayesian analysis for the first randomization, what this shows, so on the left are the numbers, the raw numbers, and on the right, you can see the posterior probability. So what we found was that both of IVIG and methylprednisolone reduced the numbers of days in hospital, particularly methylprednisolone, which you can see has an 87% posterior probability of benefit. In terms of inflammation, methylprednisolone, again, had a reduction that was highly beneficial, 97% posterior probability of reduction in baseline-adjusted CRP at day three. When we looked at second randomization, again, both anakinra and tocilizumab did reduce the number of days in hospital, as you can see here, from 9.9 to 6.6 and 8.5, respectively, but tocilizumab particularly had a very strong beneficial effect of over 99%. Anakinra, too, although, as you remember, it was a very small group of patients, only 14 patients, so we can say with less certainty that there was some benefit with anakinra. Interestingly, although both of these drugs are anti-inflammatory, that we did observe a higher use of inotropes and an increase in the number of days on inotropes for both these. So you can see at the bottom that patients in usual care had 0.2 days on inotropes compared to 0.6 and 1.1 for tocilizumab and anakinra, respectively, so an evidence of posterior probability of harm in terms of that particular outcome. Nevertheless, as expected, CRP was reduced with a moderate effect with tocilizumab. In terms of safety outcomes, there were two deaths which were not as appeared to be any associated with any particular intervention. The main outcome, as you all know, is the risk of cardiovascular sequelae from PEMS-TS and there wasn't any association with any of the interventions. So in summary, methylprednisolone, there was a strong signal for benefit with a reduction in the duration of hospital stay with 87% probability that the number of days in hospital was shorter and a 97% probability of benefit for inflammatory response as measured by baseline adjusted CRP. IVIG was interesting because of the modest, slightly less strong reduction in terms of outcomes, so steroid was clearly superior. In terms of the immunomodulation in randomization 2, tocilizumab had a very high confidence of reduction in hospital stay of over 99%, but the baseline adjusted CRP was improved, but the inotrope use increased with a posterior probability of harm of 96%. In terms of anakinra, again, a reduced length of stay, but more time on inotropes and not as good an anti-inflammatory response. So in terms of our final remarks and limitations, clearly, in order to expedite setup in a pandemic setting, this trial had to be pragmatic and adapted as time went on and new therapies came on board, which is why anakinra was a smaller group. The sample size and trial size and the substantial use of study treatments in usual care group did reduce the ability to detect plausible effects of treatment, but despite this, what we observed was a clear benefit of steroids over standard of care, but no clear benefit of IVIG over standard of care for the initial treatment of PIMS-TS. And tocilizumab was an effective second-line treatment. The anakinra arm was too small to be confident that it's of benefit. It's a huge group of work and led by Peter Horby and Martin Landry, who are the co-chief investigators of the overall recovery trial, and Saul Faust, who my colleague at Southampton is the chair of the Pediatric Recovery Working Group, of which I'm part. Work is funded by NIHR. Thank you very much for listening. Thank you.

PIMS-TS

pediatric trial

intravenous immunoglobulin

methylprednisolone

tocilizumab