Big news for the UK this week, with the release of 2 publications in JAMA reflecting novel approaches to the management of the major trauma patient with active haemorrhage. (1,2) I can hear the corks popping in QMUL (and Aberdeen) after 10 years of hard work and rightly so – completion and publication of both projects is a huge testament to the determination and tenacity required from the investigating teams, reflected in the high impact publications. In particular, CRYOSTAT-2 tells a wonderful journey of trauma research in multiple phases; hypothesis generating observational data; PhD theses utilising bench methods to explore how to arrest bleeding; interrogation of novel technology with viscohaemostatic elastic assay testing; and finally development and conduct of a large scale randomised controlled trial on clinical effectiveness. (3-10) Ross, Karim, Nicky, Simon and others should be heartily congratulated, as should Jan and Marion of course. Give them a high five if you see them..
We have talked about UK REBOA previously and I wouldn’t tread on Zaf Qasim’s toes as our resident balloon expert. However, we haven’t really covered the detail of CRYOSTAT2. Simon and I should declare that we were both principal/site investigators for this study at the outset I think, an experience which no doubt makes us read the paper with rose tinted spectacles. But let’s try and park that and get into the detail. Ultimately, as always, we suggest you read the paper (free open access thanks to the NIHR) and accompanying editorial and use your own noggins to draw final conclusions, as well as listening to our thoughts. (11)
What was the research question?
After many years of hard work, the aim of this study is rightly distilled in the final sentence of the introduction. The investigator team wanted to know if early empirical high dose cryoprecipitate given alongside a standard major haemorrhage protocol (MHP), would reduce 28 day mortality after hospital admission following injury.
Tell me about study design?
This was an open label, pragmatic randomised controlled trial with superiority design, conducted across 26 sites, mainly in the UK (Houston was the only US site). The intervention was 3 pools of cryoprecipitate (6g fibrinogen) to be delivered within 90 minutes of randomisation. Eligibility was assessed by the trauma team leader and patients were included through waiver of consent if deemed eligible, the MHP had been activated and they had received at least 1 unit of blood products. These inclusion criteria are interesting – the TTL had to vote with their feet prior to randomisation, which is pragmatic, but would we be accurate? No ‘bleeding’ scores here and a very broad set of inclusion criteria which would likely lead to a fairly heterogenous population. Very few exclusions, involving transfer from another hospital and/or presentation >3h post injury, presumably both relating to the theoretical concept that this intervention is better administered early when clot strength is fragile and people are still bleeding.
Randomisation occurred in variable permuted block sequence with stratification only by centre, facilitated by the good old fashioned sealed and opaque envelope. Many may raise an eyebrow at this, but having randomised multiple patients I can tell you that allocation was very well preserved; firstly there was definitely no time during these cases to get a steamer and magnifying glass out, and secondly a lot of us actually used the lab to randomise – at our site for example we called the lab, activated the MHP, and then asked for randomisation to CRYOSTAT2 in addition. At this point the lab staff (who had no idea about the details of the clinical case) opened the envelope and informed us regarding allocation. This is actually a very good method of randomisation for these types of study as it takes all the heat out of the situation and enables someone relatively independent to allocate.
The primary outcome was selected as 28-day mortality. Again, very pragmatic, and clearly objective, but does run the risk of confounding – major trauma is very heterogenous and as we saw with the recent PATCH trial, a haemostatic therapy which improves survival at 24h through reduction of bleeding may appear ineffective when the outcome is a broad measure later down the line (although that one was 6 months GOSE!). (12) I suppose the worry with these trials is often the proportion of severe TBI that means whatever you do in the first few days, outcomes at 28 days may be bad in both groups. TBD when we look at the results…
The sample size was based on an expected mortality of 26% and an absolute risk reduction of 7% with the intervention, based on their primary feasibility work and supported by other estimates. You can always critique this in hindsight and trials of this nature take so long to deliver that sometimes baseline mortality shifts as a direct result of other interventions over the 5 year period from design to delivery. It is all well explained and logical however. The golden number was 1600 in the end.
Did they really manage to recruit 1600 major trauma patients with active bleeding?
Yes they (we) did. A remarkable feat really. These are some of the sickest patients we see, requiring all your grey cells and often your full armoury of time critical interventions. The waiver of consent made it possible, as did the simplicity of inclusion criteria and the generic support of the wider major trauma network in the UK. Hearts and minds had already been won over to the idea of this trial prior to the first patient recruited. That makes a tremendous difference when you are seeking to deliver research in a patient population involving multiple specialties, where the evidence base is relatively light, and people often have strong feelings on issues of relative equipoise. Kudos to the investigators but mainly to the trial team, for delivery of a large and complicated study.
Ok, who were the patients?
1604 individuals who were pretty well matched between control and intervention groups. Approximately 80% were men, with a median age around 39 and ISS of 29 (so very sick by TARN standards). Approximately 64% were blunt and 80% had already received TXA which is not surprising given UK protocols. Of interest, Only a third had a systolic BP <90mmHg at randomisation and most patients were awake, with a median GCS of 13-14.
The only baseline imbalance reported that I could see was a 5% difference in severe TBI, with 29% of the standard care group having a Head AIS >/=4 compared to 24% in the CRYO arm. I couldn’t really understand why the denominator changed here also (maybe excluding people with a head AIS of 0?) but I’m sure someone can explain that to me. I suppose when you are looking for a 7% reduction in absolute mortality, a 5% increase in severe TBI in your control group could potentially bias in favour of the intervention?
People arrived to hospital quickly from incident (76mins median), got randomised quickly (15mins) and received their CRYO quickly, in a median of 68 mins from admission if allocated to the intervention arm. Interestingly, a third of the standard care group received CRYO during the first 24h, with a median time to administration of only 120 mins. I have some anxieties here about treatment separation here; If a third of patients in both groups are getting CRYO within the first 2h of admission, this is likely to bias towards the null. It’s also interesting to consider why this might have occurred – does it reflect the acuity of the patients (with people often requiring pack 2?). Does it reflect the comms around trial launch and the conveyed potential importance of early CRYO? Difficult to see what the investigators could have done about this, if anything, but interesting none the less.
What were the results?
After a small amount of loss to follow up or withdrawal, the authors present results data on 1531 patients. All cause 28 day mortality was as predicted, at around 26%. And the headline? No difference in mortality between groups at 28 days (OR 0.96, 95% CI 0.75 to 1.23) but also no difference at 6h (OR 0.82, 95% CI 0.58 to 1.17), 24h (OR 0.91, 95% CI 0.63 to 1.31) or the later timepoints of 6 and 12 months. Massive transfusion (RBC > 10U) occurred in just over 20% of the ITT and per protocol cohorts.
Prespecific analyses threw up a few interesting points – for example, a potential reduction in 28 day mortality when CRYO was administered between 61 to 90 minutes (although it’s a bit hard to posit a plausible biological rationale for this, given an increased mortality either side of the time frame). 28 day mortality was also higher in those patients with penetrating trauma who received CRYO, compared to standard care. In contrast, a trend towards improved mortality is reported with CRYO in the blunt population (OR 0.82, 95% CI 0.62 to 1.09). This is very interesting – maybe the additional volume pops more clots in the penetrating group with discrete injuries? Maybe the problem in blunt trauma is more disseminated consumption from multiple injuries? The authors highlight these results in Figure 2 and figure 3, and it has to be said that the mechanism is the only really notable influencing factor in the forest plot. Good stuff to chew over and hypothesis generating information for future research.
No additional differences in primary or secondary outcomes were highlighted in the per protocol analysis. Specifically, no differences in CCU stay, hospital stay, quality of life (assessed using EQ-5D-5L) or extended measures of functional outcome.
There were no differences reported in the adverse event rates between groups. However, me being me, I can’t help but highlight the thrombotic event rate approaching 13% in both groups by 28 days. Thirteen percent is absolutely massive, compared to the 2% VTE rate for patients in lower limb immobilisation after trauma, 3-5% VTE risk with hospitalisation and high end estimates of 10% VTE in the context of severe TBI. Although a third of these events were arterial, thirteen percent is a very high rate indeed, and reminds us all that bleeding is not the only issue for these patients.
What about limitations?
Well, there are some, but I would encourage you all to take a long hard look at yourselves and ask how you would have done this better, prior to sticking the boot in. You could, for example, highlight the earlier issues raised with treatment separation and cross over. The only real way round this would have been to blind the intervention (almost impossible) and protocolise CRYO administration within the standard care arm (very difficult). You could also highlight the lack of fibrinogen level reporting and moan about the use of fibrinogen supplementation in a heterogenous population, many of whom may have had a perfectly normal level of fibrinogen. However, bleeding trauma patients make us all keen to do things – and the idea of sending a fibrinogen to the lab and only giving CRYO based on this result would make me nervous for many reasons, and thus likely undeliverable within a trial context. We could talk about high levels of prehospital TXA and blood transfusion, perhaps reducing the potential benefit of the intervention. We could also talk about timings, although anyone who reckons they can safely thaw and administer CRYO in less than hour from admission, in addition to rapid conduct of necessary diagnostics, intervention and treatment in a major trauma patient, needs to tell us all what their secret is. The editorial makes the point that perhaps these patients weren’t sick enough – with a reasonable systolic BP on admission and <25% patients requiring massive transfusion. I don’t think these are strong arguments; if asked I suspect most of us would opine that a single NIBP is a terrible test for bleeding in young trauma patients, and the modern changes to trauma care should mean it is actually very rare that people need a massive transfusion. Last to mention I suppose is the issue of only one US centre participating, thus affecting generalisability.
What can we take away from this work?
There are a few key messages. First – multi centre randomised trials in the sickest trauma patients are deliverable at scale in the UK, thanks to the national major trauma network, a cohort of enthusiastic and hard working individual site leads, a pool of great chief investigators and the NIHR (who have the cash and infrastructure). This is excellent news.
Second, this work does not support early empirical administration of additional fibrinogen (alongside the standard MHP) for major trauma patients with active bleeding. This is interesting news and allows us to use standard MHP strategies initially while we focus on the hundred other things we need to co-ordinate as TTL.
Third, this work highlights the potential differences in endotype between trauma patients, in particular with a differing mechanism of injury, and perhaps moves us further towards bespoke replacement strategies based on biological markers. This is relatively convincing for blunt trauma and a hypothesis that would follow recent work such as ITACTIC and others. (13) Is there a new ‘golden half hour (60-90 mins)’ at which to give CRYO? I don’t think so personally – I suspect rapid identification of those with significant coagulopathy is the goal here, to enable personalised replacement of complex products. In a bleeding cohort, this usually has to follow broad initial volume replacement with 1:1 ratios.
There is a nice infographic produced by the trial team that captures some of the results and one in JAMA along with free open access to the manuscript.
You should also read @karimbrohi ‘s series of tweets that give more insight from the lead research group.
Who knows? People will talk about fibrinogen concentrate, the SWIFT trial, TBI coagulopathy and the next big questions in bleeding. All well and good. But I think this trial shows us what is possible for UK trauma research, and we would be missing a trick if we focussed on the next single hypothesis. If we signed up to a platform trial or adaptive design, we could progress all of these questions or rapidly introduce a new one, without waiting 5 years for the next James Lind Alliance PSP. It would also give us momentum, sustainability and direction. In other words, I agree with Prof Coats…
Any negatives to a project like this? Someone has to write the grant application and ask politely for millions of pounds…..
Thanks to the investigators for their amazing efforts on this one and thanks to JAMA for recognising the importance of this work. You should all be very proud of yourselves.
1 – Jansen JO, Hudson J, Cochran C, et al. Emergency Department Resuscitative Endovascular Balloon Occlusion of the Aorta in Trauma Patients With Exsanguinating Hemorrhage: The UK-REBOA Randomized Clinical Trial. JAMA. Published online October 12, 2023. doi:10.1001/jama.2023.20850
2 – Davenport R, Curry N, Fox EE, et al. Early and Empirical High-Dose Cryoprecipitate for Hemorrhage After Traumatic Injury: The CRYOSTAT-2 Randomized Clinical Trial. JAMA. Published online October 12, 2023. doi:10.1001/jama.2023.21019
3 – Curry N, Rourke C, Davenport R, et al. Early cryoprecipitate for major haemorrhage in trauma: a randomised controlled feasibility trial. British Journal of Anaesthesia, 2015, 1–8 doi: 10.1093/bja/aev134
4 – Schlimp, C.J., Voelckel, W., Inaba, K., Maegele, M. and Schochl, H. Impact of fibrinogenconcentrate alone or with prothrombin complex concentrate (+/- fresh frozen plasma) on plasma fibrinogen level and fibrin-based clot strength (FIBTEM) in major trauma: a retrospective study. Scand J Trauma Resusc Emerg Med 2013;21:74.
5 – Nienaber, U., Innerhofer, P., Westermann, I., Schöchl, H., Attal, R., Breitkopf, R. and Maegele, M. The impact of fresh frozen plasma vs. coagulation factor concentrates on morbidity and mortality in trauma-associated haemorrhage and massive transfusion. Injury 2011; 42(7):697- 701.
6 – Innerhofer P, Westermann I, Tauber H, Breitkopf R, Fries D, Kastenberger T, et al. The exclusive use of coagulation factor concentrates enables reversal of coagulopathy and decreases transfusion rates in patients with major blunt trauma. Injury 2013; 44: 209-16
7 – Schöchl, H., Nienaber, U., Hofer, G., Voelckel, W., Jambor, C., Scharbert, G., Kozek- Langenecker, S. and Solomon, C. Goal-directed coagulation management of major trauma patients using thromboelastometry (ROTEM)-guided administration of fibrinogen concentrate and prothrombin complex concentrate. Crit Care 2010;14:R55.
8 – Stinger HK, Spinella PC, Perkins JG, Grathwohl KW, Salinas J, Martini WZ, et al. The ratio of fibrinogen to red cells transfused affects survival in casualties receiving massive transfusions at an army combat support hospital. Journal of Trauma 2008; 64:S79-85.
9 – Dente, C.J., Shaz, B.H., Nicholas, J.M., Harris, R.S., Wyrzykowski, A.D., Patel, S., Shah, A., Vercruysse, G.A., Feliciano, D.V., Rozycki, G.S., Salomone, J.P. and Ingram, W.L. Improvements in early mortality and coagulopathy are sustained better in patients with blunt trauma after institution of a massive transfusion protocol in a civilian level 1 trauma centre. J Trauma 2009; 66(6):1616-24.
10 – Shaz, B.H., Dente, C.J., Nicholas, J., MacLeod, J.B., Young, A.N., Easley, K., Ling, Q., Harris R.S. and Hillyer C.D. Increased number of coagulation products in relationship to red blood cell products transfused improves mortality in trauma patients. Transfusion 2010;50(2):493- 500.
11 – Tisherman SA, Brenner ML. Contemporary Adjuncts to Hemorrhage Control. JAMA. Published online October 12, 2023. doi:10.1001/jama.2023.16135
12 – The PATCH-Trauma Investigators and the ANZICS Clinical Trials Group. Prehospital Tranexamic Acid for Severe Trauma. N Engl J Med 2023; 389:127-13
13 – Baksaas-Aasen, K., Gall, L.S., Stensballe, J. et al. Viscoelastic haemostatic assay augmented protocols for major trauma haemorrhage (ITACTIC): a randomized, controlled trial. Intensive Care Med 47, 49–59 (2021). https://doi.org/10.1007/s00134-020-06266-1