TRAIN – Transfusion Strategies in Acute Brain Injured Patients
Restrictive vs Liberal Transfusion Strategy in Patients With Acute Brain Injury
@fabio_taccone. JAMA. 2024. PMID: 39382241
Clinical Question
- In patients with acute brain injury, does a liberal, compared to a restrictive strategy of blood transfusion, improve neurological outcomes at 180 days?
Background
- The TRICC study, published over 20 years go, led to the widespread adoption of restrictive thresholds for blood transfusion in critically ill patients
- However, few neurocritically unwell patients were included in that study and observational data has linked haemoglobin levels below 9g/dL with poorer outcomes in patients with traumatic brain injury (TBI) and subarachnoid haemorrhage (SAH)
- Prospective RCTs in this setting have yielded contrasting results, however these had methodological differences including single centre design, higher liberal transfusion thresholds and the definition of unfavourable neurological outcome
- More recently HEMOTION demonstrated a reduction in unfavourable neurological outcome at 6 months in patients with TBI treated according to a liberal transfusion threshold, although this did not reach the predefined criteria for statistical significance
Design
- Multicentre, parallel-group, open-label, randomised clinical trial
- Conducted in 72 ICUs in 22 countries
- Eligible patients were randomly assigned according to a computer-generated sequence with variable block sizes and stratified by center, disease and GCS at the point of randomisation
- ICU personnel were unmasked, but final neurological evaluation was performed by providers who were blinded to treatment allocation
- The sample size was based on estimated mortality and unfavourable neurological outcome rates of 15% and 35% respectively, and calculated that 794 patients would be required to achieve statistical power of 85% at an α of 0.05 to detect a reduction in the rate of poor neurological outcome at 180 days from 50% to 39%
- This was amended twice from the original sample size calculation due to slow enrollment
- Intention-to-treat analysis
- The trial protocol was published a priori and sample size adjustments were approved by the ethics committee
Setting
- 850 patients were randomised between September 2017 and December 2022, of whom 30 patients were excluded after consent was withdrawn resulting in a study population of 820, of whom 397 were assigned to the liberal strategy and 423 to the restrictive strategy
Population
- Inclusion: Adult patients (aged ≥18 years and ≤80 years) admitted to ICU with TBI, SAH or ICH with a GCS ≤13, an expected ICU stay of >72 hours and Hb ≤9 g/dL within 10 days from brain injury
- Exclusions: The main exclusions were:
- Previous neurological disease
- GCS 3 with fixed dilated pupils, brain death or imminent death
- Medical need to correct anaemia
- ICH due to AVM or brain tumour
- Baseline demographics (liberal vs restrictive):
- Age: 52 vs 51 years
- Male sex: 54.9% vs 53.4%
- Medical History:
- Diabetes: 6.3% vs 5.7%
- COPD: 5.3% vs 5.2%
- Cancer: 2.5% vs 3.5%
- Type of brain injury:
- TBI: 60.5% vs 58.2%
- SAH: 21.7% vs 24.6%
- ICH: 17.9% vs 17.3%
- Clinical & Hb scores at randomisation:
- GCS: 6 (3-8) vs 6 (3-8)
- Hb, g/dL: 8.5 vs 8.5
- ICP monitoring within 48 hours: 70.6% vs 69.1%
- Mechanical ventilation: 90.7% vs 92.2%
- Antiplatelet therapy: 14.2% vs 13.9%
- Anticoagulant therapy: 7.6% vs 4.8%
Intervention (liberal)
- Administration of 1 unit of packed red blood cells when haemoglobin levels were <9g/dL
Control (restrictive)
- Administration of 1 unit of packed red blood cells when haemoglobin levels were <7g/dL
Management common to both groups
- Haemoglobin measured daily
- Transfusion thresholds maintained for a maximum of 28 days after randomisation or until hospital discharge or death, whichever occurred first
Outcome
- Primary efficacy endpoint: The proportion of patients with unfavorable neurological outcome at 180 days (GOS-E 1-5) was significantly lower in the liberal group, compared with the restrictive group
- 62.6% vs 72.6% (RR 0.86; 95% CI 0.78 to 0.95)
- Secondary efficacy endpoints: There was no differences in any secondary endpoints between groups
- 28-day mortality: 20.7% vs 22.5%
- Organ failure: 78.6% vs 77.5%
- ICU length of stay: 21.4 days vs 22.5 days
- Safety outcomes:
- There were fewer cerebral ischaemia events in the liberal group, compared with the restrictive group: 8.8% vs 13.5% (RR 0.65; 95% CI 0.44 to 0.97)
- There were no other differences in serious adverse events between groups
Authors’ Conclusions
- “Patients with anemia and acute brain injury randomized to a liberal strategy of red blood cell transfusion at a hemoglobin threshold of 9 g/dL had a lower probability of unfavorable neurological outcome at 180 days than patients randomized to a restrictive strategy of transfusion at a hemoglobin threshold of 7 g/dL.”
Strengths
- Multi-centre, randomised, controlled trial across various countries and continents
- External validity further enhanced by the study’s pragmatic design, including a trigger and intervention that is relatively straight-forward and generalisable to a wide range of healthcare settings including non-tertiary centres and LMICs
- Blinding of outcome assessors to the assigned intervention reduces potential for detection bias
- Central randomisation with variable blocks ensured allocation concealment
- Baseline demographics including mechanism of brain injury, presenting GCS and baseline haemoglobin, were well matched between groups
- Most patients received treatment according to their randomised group, there were few protocol violations and there was a low rate of attrition
- Adequate separation of measured haemoglobin and transfusion volume between groups throughout the trial enhances internal validity
- The liberal threshold of 9g/dl may be crucial in providing adequate brain tissue oxygenation whilst avoiding the theoretical risks of transfusion, which were not encountered in the safety analysis, but were evident in some studies aiming for higher haemoglobin targets
- Capturing reduced rates of cerebral ischaemia in the restrictive group is a theoretical mechanism by which neurological outcomes may be modified by liberal transfusion targets
- The choice of outcome measure was patient centered
Limitations
- Enrollment was slow and necessitated two sample size adjustments during the running of the study
- ICU staff were not blinded to treatment allocation which could introduce performance bias. This is particularly relevant as neuroprognostication was not protocolised and rates of WLST were not captured
- It is not known if patients received transfusions prior to randomisation, although you might expect that this would bias to the null and this was a positive study
- The range of neurocritical illness included in this study and non-specific surrogate for tissue oxygenation (i.e. serum haemoglobin) appear broad and contrasts with a wider trend towards teasing out heterogeneity of treatment effects and precision medicine in general. However, the improvements in neurological outcome across a range of pathology appear consistent in subgroup analysis
- Some caution is advised in the interpretation of reduced cerebral infarction rates due to the lack of protocolisation and potential for detection bias in an unblinded study. The same is true for the detection of VTE, the true incidence of which may have been underestimated
- The mortality rate was higher in this study compared with similar studies on this topic
The Bottom Line
- Large, multi-center randomised study with excellent generalisability, that demonstrates improved neurological outcomes for survivors of a range of acute brain injury
- There are some limitations and unanswered questions, which in view of the scarcity of blood and dogma of restrictive transfusion practices in critically unwell patients, may be a barrier to implementation
- In view of this and HEMOTION, I will target a more liberal transfusion threshold in acute brain injury
External Links
-
- article Restrictive vs Liberal Transfusion Strategy in Patients With Acute Brain Injury
- further reading TBL Review: HEMOTION
Metadata
Summary author: Andrew Achilleos
Summary date: 11/10/24
Peer-review editor: @davidslessor
Image by: Daniel Abadia on Unsplash