Restrictive or Liberal Transfusion Strategy in Myocardial Infarction and Anemia

Carson et al. NEJM 2023 DOI: 10.1056/NEJMoa2307983  

Clinical Question

• In patients with myocardial infarction and anaemia, does a restrictive or liberal transfusion strategy decrease myocardial ischaemia or death at 30 days? 
• Myocardial Ischemia and Transfusion (MINT) Trial 

Background

• Anaemia is common in patients with acute myocardial infarction and may exacerbate ischaemic injury to vulnerable myocardium due to decreased oxygen carriage in blood 
• The optimum haemoglobin target in this population is unknown and needs to balance this risk of worsening ischaemia vs risks of blood transfusion especially transfusion associated circulatory overload (TACO) 
• Three previous randomised controlled trials (REALITY, CRIT, Carson et al.) (total n=820) have compared transfusion thresholds in patients with myocardial infarction with inconsistent results  
• There exists a large body of well conducted randomised controlled trials generally supporting the safety of a restrictive transfusion strategy in critically ill adults including:
  • TRICC (NEJM 1999) n=838 all-comers in ICU excluding cardiac surgery 
    • Restrictive transfusion aiming Hb >70g/L vs liberal transfusion aiming Hb > 90 g/L 
    • In-hospital mortality was significantly lower in the restrictive-strategy group (22.2% vs 28.1%, p=0.05) 
  • TRICS III (NEJM 2017) n=5243 patients post cardiac surgery
    • Restrictive transfusion aiming Hb >75g/L vs liberal transfusion aiming Hb > 85-95 g/L 
    • Restrictive strategy was non-inferior; primary outcomes (death, myocardial infarct, new onset dialysis requiring renal failure) occured in 11.4% in the restrictive group vs to 12.5% in liberal group (absolute risk difference = -1.11%; 95% CI, -2.93 to 0.72, OR 0.90; 95% CI -2.93 to 0.72)
  • TRACS (JAMA 2010) n=502 patients post cardiac surgery 
    • Restrictive transfusion (aiming hematocrit  ≥24%) vs liberal transfusion (aiming hematocrit ≥30%) 
    • Restrictive strategy was non-inferior for mortality or severe morbidity at 30 days -10% liberal vs 11% restrictive; between-group difference, 1% [95% CI, −6% to 4%]; P = .85) 
    • Number of units tranfused was an independent predictor of mortality regardless of transfusion strategy  
  • TRISS (NEJM 2014) n=998 patients with septic shock 
    • Restrictive transfusion aiming Hb >70g/L vs liberal transfusion aiming Hb > 90 g/L 
    • No mortality difference between strategies (43% restrictive vs 45% liberal, RR 0.94 CI 0.78-1.09 p=0.44) 
• However three recently published trials in the field of neurocritical care have demonstrated a signal towards benefit from a liberal transfusion strategy although not meeting statistical significance for their primary outcomes 
  • HEMOTION – TBI 
  • TRAIN – mixed TBI, SAH, ICH  
  • SAHaRA – SAH  

Design

• Open label multi-centre randomised control trial 
• 1:1 randomisation to restrictive or liberal transfusion strategy by web-based system 
• Permuted block design with random block sizes of 4-6 stratified by clinical site 
• Pre-specified subgroups 
  • Type of MI – Type 1 or 2 
  • Revascularisation for index MI 
  • Heart failure 
  • Pre-randomisation haemoglobin 
  • Acute vs chronic anaemia 
  • Renal function  
  • Diabetes 
  • Age <60, 60-69, 70-79 or >80 
  • Race 

Setting

• 144 sites across USA, Canada, France, Brazil, New Zealand and Australia  
• Recruitment occurred from April 2017 – April 2023 

Population

• Inclusion:  
  • 18 years of age or older 
  • STEMI or NSTEMI during index hospitalisation 
    • Defined as ≥ 1 troponin above upper limit of normal AND one of 
      • Symptoms of ischaemia 
      • New ST or T wave changes or left bundle branch block 
      • New pathologic Q waves 
      • Imaging evidence of loss of viable myocardium or regional wall motion abnormality 
      • Angiographically proven thrombus 
    • Definition includes 
      • Type 1 MI – spontaneous related to coronary arterial disease 
      • Type 2 MI – secondary ischaemic imbalance 
      • Type 4b MI – stent thrombosis at angiography 
      • Type 4c MI – severe in-stent restenosis without thrombosis  
      • Haemoglobin concentration <100g/L at time of random allocation  
• Exclusion:  
  • Uncontrolled acute bleeding requiring uncrossed or non-type specific blood 
  • Declined blood transfusion 
  • Scheduled for cardiac surgery during the current admission 
  • Receiving only palliative treatment 
  • Follow up at 30-days known to be impossible 
  • Previously enrolled in MINT 
  • Enrolled in a competing clinical study  
• Sample size 
  • Planned enrollment of 3500 patients to provide 80% power to detect a 20% relative reduction in primary outcome of death assuming an event rate of 16.4%  
    • 3506 patients enrolled – 2 did not approve of use of their data leaving 3504 in the primary analysis  
  • Comparing baseline characteristics of restrictive vs liberal strategy 
    • Mean age 72 vs 72 years old  
    • 44% female vs 47% 
    • Approximately 1/3 of patients in both arms received PRBC prior to randomisation 34.2% in restrictive group vs 36.4% in liberal
    • Well matched for type of MI, interventions received, presence of heart failure  
    • Of note a majority (55.8%) of patients had Type 2 MI (demand ischaemia). 41.7% had Type 1 MI  

Intervention

• Restrictive group 
  • Transfusion permitted but not required when haemoglobin was <80g/L 
  • Transfusion strongly recommended when Hb <70g/L or anginal symptoms were not controlled with medication 
  • Mean Hb at day 3 was 8.9
  • Discontinuation of the protocol occurred in 2.6%

Control

• Liberal group 
  • 1 unit of packed red cells administered after randomisation then as needed when Hb <100g/L until hospital discharge or 30 days 
  • Mean Hb at day 3 was 10.5
  • Discontinuation of the protocol occurred in 13.7%

Management common to both groups

• Transfusion administered one unit at a time followed by repeat Hb measurement 
• Protocol paused at clinician discretion if active bleeding requiring transfusion occurred 
• Transfusion could be delayed in patients with volume overload until adequate diuresis or until day of dialysis 

Outcome

• Total number of red cell units transfused 3.5x higher in liberal group 
  • Mean (+/- SD) units per patient 0.7 +/- 1.6 in restrictive group vs 2.5 +/-2.3 in liberal group  
• Primary outcome: myocardial infarction OR death from any cause at 30 days  
  • Occurred in 16.9% in restrictive group vs 14.5% in liberal group 
  • Crude risk ratio 1.16, (95% confidence interval [CI] 1.00-1.35) 
  • Risk ratio modelled with multiple imputation for incomplete follow-up, 1.15; 95% CI 0.99 to 1.34; P = 0.07 
• Secondary outcomes: Comparing restrictive vs liberal group  
  • Significant difference favouring liberal group 
    • Cardiac death – 5.5% vs 3.2% (RR 1.74; CI 1.26-2.40) 
  • No significant difference but numerically favouring liberal transfusion threshold  
    • Death – 9.9% vs 8.3% (RR 1.19; CI 0.96-1.47) 
    • Myocardial infarction 8.5% vs 7.2% (RR 1.19; CI 0.94-1.49) 
    • Unscheduled revascularisation 2.5% vs 2.2% (RR 1.11; CI 0.72-1.70) 
    • Stroke – 1.7% vs 1.5% (RR 1.16; CI 0.69-1.95) 
    • Pneumonia or bacteraeamia 9.5% vs 8.7% (RR 1.09; CI 0.88-1.34) 
  • No significant difference but numerically favouring restrictive transfusion threshold 
    • Heart failure 5.8% vs 6.3% (RR 0.92; CI 0.71-1.20) 
    • PE or DVT 1.5% vs 1.9% (RR 0.77; CI 0.46-1.27)
• Subgroup analysis for primary outcome (all subgroups pre-specified) 
  • Significant difference in primary outcome favouring liberal group for patients with  
    • Type 1 MI –18.2% vs 13.8% (RR 1.32; CI 1.04-1.67) 
    • Congestive heart failure, acute heart failure or low LVEF 19.6% vs 15.7% (RR 1.25; CI 1.02-1.52) 
  • No statistically significant difference between liberal and restrictive strategy for all other subgroups 
  • All other subgroups numerically favoured liberal transfusion strategy except 
    • Renal dialysis at baseline 
    • Hb <80g/L on randomisation 
• Editorial correspondence by other authors (not part of the prespecified statistical plan) demonstrate  
  • Negative fragility index of 2 – ie if 2 additional patients in the liberal group had not had a primary outcome event the headline result would have been statistically significant 
  • Bayesian reanalysis suggests 90.8-99.8% chance of harm with restrictive strategy depending on certainty of priors adopted  

Authors’ Conclusions

• In patients with acute myocardial infarction and anaemia a liberal transfusion strategy did not significantly reduce the risk of recurrent MI or death at 30 days. Trial endpoints suggest some benefit of a liberal strategy over a restrictive strategy , but additional studies would be needed to confirm this 

Strengths

• Pragmatically conducted large multi-center randomised controlled trial  
• 4x larger sample size than all previous trials related to this question combined  
• Broad inclusion criteria captured a heterogenous population of patients with myocardial infarction improves external validity 
• Separation between groups in mean Hb and number of units transfused improves internal validity and decreases risk of type 2 error 
• Trial transfusion protocols reflect real world practice and were pragmatic eg adjusted protocols for volume overloaded and dialysed patients 
• Excellent follow up (98.3% of patients at 30 days) 
• Blinded central adjudication of outcomes  

Weaknesses

• Unblinded intervention  
• Only primary outcome of myocardial infarction was centrally adjudicated  
• Adherence to liberal transfusion strategy 86.3% at hospital discharge, largely due to clinician discretion which may increase type 2 error 
• No data provided with regards to number of patients screened but not enrolled in the trial  
• No statistical adjustments for multiplicity of outcomes  
• Benefits seen in subgroups are counterintuitive and the mechanism for these benefits are no clear  
  • One might expect patients with Type 1 MI and true coronary occlusion to benefit less from a higher Hb target than those with demand ischaemia (Type 2) however in MINT patients with Type 1 MI benefitted more 
  • Despite patients in the liberal group having a non-statistically significantly increased risk of heart failure, patients with heart failure infact had a lower rate of death or repeat MI with a liberal strategy than those without 

The Bottom Line

• Although the threshold for statistical significance was not met for the primary outcome, from the MINT trial suggests that patients with anaemia following a myocardial infarction may benefit from a higher haemoglobin target of 100g/L and are unlikely to experience harm from a liberal transfusion strategy 
• Based on these data I will adopt a Hb target of 100g/L for patients with anaemia following myocardial infarction, especially those with Type 1 MI, although the decision to transfuse should remain individualised considering the patient’s clinical context, risk factors for complications, available resources and degree of compensation

External Links

• • Restrictive or Liberal Transfusion Strategy in Myocardial Infarction and Anemia
  • Restrictive versus Liberal Transfusion in Myocardial Infarction — A Patient-Level Meta-Analysis

 

Metadata

Summary author: Daniel Chung
Summary date: 08/05/2025
Peer-review editor: Aidan Burrell and David Slessor

Picture by: icon0 com/pexels