Publication

• Title: Prothrombin Complex Concentrate vs Frozen Plasma for Coagulopathic Bleeding in Cardiac Surgery: The FARES-II Multicenter Randomized Clinical Trial
• Acronym: FARES-II (Factor Replacement in Surgery II)
• Year: 2025
• Journal published in: JAMA
• Citation: Karkouti K, Callum JL, Bartoszko J, et al; FARES-II Study Group. Prothrombin Complex Concentrate vs Frozen Plasma for Coagulopathic Bleeding in Cardiac Surgery: The FARES-II Multicenter Randomized Clinical Trial. JAMA. 2025;333(20):1781-1792.

Context & Rationale

• Background

  • Post-cardiopulmonary bypass (CPB) bleeding is common, clinically consequential, and typically multifactorial (surgical bleeding, platelet dysfunction, factor dilution/consumption, hypothermia, fibrinolysis, residual heparin effect).
  • Frozen plasma has been the dominant “coagulation factor replacement” strategy in many centres, despite practical limitations (time to thaw, large infusion volumes) and biologic variability in factor content.
  • Four-factor prothrombin complex concentrate (4F-PCC) offers rapid delivery of vitamin K–dependent factors in a small volume, but with historical concern about thromboembolic complications in settings where factor deficiency is not confirmed.
  • Pre-FARES-II RCT evidence in cardiac surgery was limited and heterogeneous, including a pilot feasibility RCT and other small RCTs with mixed results and variable PCC dosing strategies.¹²³
  • Systematic review evidence before FARES-II suggested potential reductions in transfusion and/or re-exploration with PCC, but certainty was limited by small trials, clinical heterogeneity, and inconsistent dosing/selection criteria.⁴
• Research Question/Hypothesis

  • In adults undergoing cardiac surgery with CPB who develop coagulopathic bleeding requiring coagulation factor replacement, is 4F-PCC non-inferior to frozen plasma in achieving an effective haemostatic response?
  • If non-inferiority is demonstrated, does 4F-PCC improve haemostatic efficacy and reduce bleeding/transfusion, without excess harm?
• Why This Matters

  • Replacing plasma with 4F-PCC could materially alter operating theatre and ICU bleeding management by reducing time-to-treatment and exposure to large-volume plasma transfusion.
  • Given the off-label nature of PCC use in many cardiac-surgical settings, robust efficacy and safety data are essential for policy, stewardship, and guideline development.
  • The intervention is mechanistically attractive but clinically high-stakes (thromboembolism, graft thrombosis, stroke, myocardial infarction), making trial design and patient selection pivotal.

Design & Methods

• Research Question: Among adults with post-CPB coagulopathic bleeding requiring coagulation factor replacement, is 4F-PCC non-inferior to frozen plasma for effective haemostatic response, and (if non-inferior) is it superior?
• Study Type: Randomised, multicentre (12 hospitals in Canada and the United States), active-controlled, investigator-initiated trial using a non-inferiority framework with hierarchical superiority testing; treating teams unblinded.
• Population:
  • Setting: Cardiac operating theatres (intervention delivered after CPB termination) with subsequent ICU care.
  • Key inclusion features: Adults (≥18 years) undergoing cardiac surgery using CPB; coagulation factor replacement ordered intraoperatively for management of bleeding or anticipated bleeding; known or suspected coagulation factor deficiency (e.g., elevated INR and/or viscoelastic clotting time abnormalities).
  • Operational administration criteria (intraoperative): Final decision to administer based on at least moderate bleeding severity and INR ≥1.5 measured by point-of-care testing after heparin neutralisation; INR confirmation could be waived if bleeding was severe.
  • Key exclusions: Heart transplantation; ventricular assist device insertion/removal; thoracoabdominal aneurysm repair; critical preoperative state with high probability of death within 24 hours; severe right heart failure; requirement for PCC for anticoagulant reversal within 3 days; thromboembolic event within 3 months; severe allergy/IgA deficiency with anti-IgA; refusal of allogeneic blood products; pregnancy; concurrent participation in another interventional trial.
  • Consent model: Preoperative consent at US sites; delayed consent postoperatively at Canadian sites (and some treated patients subsequently declined consent and were excluded from analyses).
• Intervention:
  • Intravenous 4F-PCC (Octaplex) given as first-line coagulation factor replacement.
  • Dose by body weight: 1500 IU if ≤60 kg; 2000 IU if >60 kg (median 2000 IU; median 23.9 IU/kg in the analysis cohort).
  • Second dose permitted within 24 hours if a further order for coagulation factor replacement was received.
  • Maximum permitted cumulative PCC dose approximately 50 IU/kg (after which frozen plasma was used for any additional dosing).
• Comparison:
  • Frozen plasma as first-line coagulation factor replacement.
  • Dose by body weight: 3 units if ≤60 kg; 4 units if >60 kg (median 4 units; median 11.8 mL/kg in the analysis cohort).
  • Second dose permitted within 24 hours if a further order for coagulation factor replacement was received.
  • Protocol allowed non-investigational PCC (commercial PCC) in the frozen plasma group only if ongoing bleeding persisted after 2 doses of investigational frozen plasma.
• Blinding: Treating clinicians were unblinded (products differ in volume and handling); primary outcome was adjudicated by blinded assessors using prespecified criteria.
• Statistics: Non-inferiority margin 10% (absolute) for the risk of ineffective haemostatic response; one-sided α=0.025; 90% power; 410 patients required for the primary analysis set (minimum 205 per group); primary analysis used a modified intention-to-treat approach (patients who received study product and provided consent) with prespecified per-protocol analysis; hierarchical superiority testing followed if non-inferiority met.⁵
• Follow-Up Period: Primary haemostatic assessment from 60 minutes after initiation through 24 hours; transfusion/bleeding outcomes reported through 24 hours and 7 days; adverse events and mortality through 30 days.

Key Results

This trial was not stopped early. 528 patients were randomised; 420 were included in the primary analysis and safety analyses (4F-PCC n=213; frozen plasma n=207).

• 4F-PCC met non-inferiority and demonstrated superiority for the primary haemostatic composite, driven by fewer “failure” components (notably second dose requirement, platelet transfusion, and rescue haemostatic products).
• Bleeding and transfusion exposure were consistently lower with 4F-PCC across multiple time windows (≤24 h and ≤7 days), with substantially less drainage volume and fewer allogeneic product units.
• Pre-specified subgroup analysis showed broadly consistent reductions in haemostatic failure across clinical strata, with no clear qualitative interaction signal reported.

Internal Validity

• Randomisation and Allocation:
  • Patients were randomised intraoperatively when coagulation factor replacement was ordered, with investigational product prepared by blood bank/pharmacy and released in tamper-proof boxes opened only if administration criteria were met.
  • Allocation concealment was operationally supported up to the point of administration; treatment blinding was not feasible once product was opened and infused.
• Dropout or exclusions (post-randomisation):
  • 528 randomised; 420 included in primary analysis (4F-PCC 213/265; frozen plasma 207/263).
  • Post-randomisation exclusions were substantial and asymmetric in type: “did not receive either treatment” (4F-PCC 27; frozen plasma 33), “did not meet inclusion criteria” (4F-PCC 25; frozen plasma 31), and delayed consent declinations (4F-PCC 24; frozen plasma 22).
  • Protocol-deviation exclusions further reduced per-protocol sets (4F-PCC 209; frozen plasma 200), raising risk of selection bias and diminishing the protective effect of randomisation for some endpoints.
• Performance/Detection Bias:
  • Unblinded clinical teams could influence transfusion decisions and some components of the composite haemostatic endpoint, despite predefined criteria and adjudication.
  • Primary outcome adjudication was performed by blinded assessors, partially mitigating detection bias for the composite classification.
• Protocol Adherence:
  • Clear separation in delivered exposure: first dose volume 4F-PCC 99 mL (IQR 82–119) vs frozen plasma 811 mL (IQR 694–1000); completion time 7 min (IQR 5–10) vs 26 min (IQR 20–34).
  • Median first-dose dosing aligned with protocol: 4F-PCC 2000 IU (IQR 1500–2000) vs frozen plasma 4 units (IQR 3–4).
  • Rescue haemostatic products were used more frequently in the frozen plasma group within the primary composite window (e.g., recombinant activated factor VII: 0 vs 9; non-investigational PCC: 0 vs 14), supporting meaningful treatment separation.
• Baseline Characteristics:
  • Groups were broadly balanced on major prognostic features (age, sex, procedure mix, CPB duration, baseline haemoglobin and INR) with similar median baseline INR at administration (1.7 [IQR 1.5–2.0] vs 1.7 [IQR 1.5–1.9]).
  • Median CPB duration was similar (171 vs 176 minutes), suggesting comparable exposure to CPB-associated coagulopathy risk.
• Heterogeneity:
  • Multicentre design enhances robustness, but between-centre practice variation in transfusion thresholds/adjuncts remains a potential contributor to outcome heterogeneity.
  • Eligibility and administration criteria (bleeding severity + INR confirmation) functioned as a biological enrichment strategy, potentially reducing clinical heterogeneity in “true factor deficiency” at the expense of broader applicability.
• Timing:
  • Administration occurred intraoperatively after heparin reversal, targeting early coagulopathic bleeding when factor replacement is most plausibly beneficial.
  • Time from starting study product to ICU arrival was similar: 1.0 h (IQR 0.6–1.7) vs 1.2 h (IQR 0.7–2.0).
• Dose:
  • Median PCC dosing of 23.9 IU/kg was higher than dosing used in some earlier cardiac-surgery trials, strengthening biological plausibility for haemostatic effect but complicating cross-trial comparisons.
  • Frozen plasma dosing (median 11.8 mL/kg) reflects common clinical dosing but remains subject to unit volume variability and factor-content variability.
• Separation of the Variable of Interest:
  • Factor-delivery speed and volume were markedly different: 7 min vs 26 min to complete first dose; 99 mL vs 811 mL total first-dose volume.
  • Bleeding separation: chest tube drainage 12 h 471 mL vs 642 mL; 24 h 691 mL vs 923 mL.
  • Transfusion separation: RBC+platelets+frozen plasma ≤24 h after CPB end 6.6 vs 13.8 units (LS mean); LS mean ratio 0.48.
• Outcome Assessment:
  • Primary composite includes clinically meaningful components but mixes patient-centred events (reoperation) with clinician-mediated decisions (additional product use), increasing susceptibility to performance bias in an unblinded trial.
  • Harder outcomes (AKI, thromboembolic events, death) were captured but the trial was not powered for these outcomes individually.
• Statistical Rigor:
  • Non-inferiority framework with prespecified margin and hierarchical testing is methodologically appropriate given equipoise and adoption questions.
  • Post-randomisation exclusions and modified intention-to-treat approach may compromise inferential purity of non-inferiority claims, which are particularly sensitive to analysis-set choices.

Conclusion on Internal Validity: Overall, internal validity appears moderate to strong, supported by pragmatic intraoperative randomisation, blinded adjudication, and clear biological separation; however, substantial post-randomisation exclusions, unblinded care, and a clinician-mediated composite endpoint meaningfully limit causal certainty for some components.

External Validity

• Population Representativeness:
  • Applies to adult cardiac-surgery patients with CPB and clinically significant bleeding where clinicians are already considering “coagulation factor replacement”.
  • Exclusions remove several high-risk or distinct populations (e.g., transplant/VAD cases, recent thromboembolism, urgent anticoagulant reversal), limiting applicability to these groups.
• Applicability:
  • Implementation favours centres with ready access to point-of-care INR and established bleeding algorithms; uptake may be challenging where laboratory turnaround is prolonged or blood bank processes differ.
  • Findings are most relevant to settings where frozen plasma administration is logistically slow or volume-sensitive patients are common (e.g., older, impaired ventricular compliance), but these hypotheses require confirmation.
  • Generalisation to resource-limited settings is uncertain, given product availability, cost, and regulatory differences around PCC indications.

Conclusion on External Validity: External validity is good for high-resource cardiac surgery systems using intraoperative point-of-care haemostatic assessment, but limited for excluded populations (transplant/VAD, urgent anticoagulant reversal, recent thromboembolism) and for settings without rapid haemostatic testing and streamlined product delivery.

Strengths & Limitations

• Strengths:
  • Largest pragmatic RCT to date comparing first-line 4F-PCC vs frozen plasma for post-CPB coagulopathic bleeding.
  • Biological enrichment (bleeding severity + INR confirmation after heparin reversal) increases likelihood the intervention targets the relevant physiology.
  • Marked and clinically meaningful separation in time-to-delivery and volume exposure (7 vs 26 minutes; 99 vs 811 mL).
  • Blinded adjudication of the primary haemostatic endpoint.
  • Clinically important secondary outcomes captured (bleeding severity, transfusion burden, AKI, thromboembolic events, mortality to 30 days).
• Limitations:
  • Open-label delivery with clinician-mediated composite endpoint, increasing vulnerability to performance bias.
  • Substantial post-randomisation exclusions (including delayed consent declinations) complicate interpretation in a non-inferiority framework.
  • Fixed-dose strategy by weight threshold may not precisely match individual factor deficits; dose equivalence between PCC and plasma remains conceptually challenging.
  • Not powered for mortality or uncommon thromboembolic events; safety conclusions are necessarily probabilistic.
  • Industry involvement (product supply) creates potential for perceived bias despite investigator initiation and independent adjudication.

Interpretation & Why It Matters

• Clinical signal

  In an enriched population with documented/suspected factor deficiency after CPB, 4F-PCC improved haemostatic effectiveness and reduced bleeding/transfusion exposure compared with frozen plasma, while not demonstrating an excess thromboembolic signal.
• Operational significance

  Large-volume and slow-delivery limitations of plasma were directly addressed: first-dose volume 99 mL vs 811 mL and completion time 7 vs 26 minutes, supporting a pragmatic advantage in time-critical haemorrhage management.
• Safety interpretation

  Lower AKI and fewer serious adverse events in the 4F-PCC group are clinically important but mechanistically non-specific; the trial supports targeted PCC use when factor deficiency is established rather than empiric “PCC-first” haemorrhage care.

Controversies & Subsequent Evidence

• Composite endpoint vulnerability in an open-label trial: The editorial emphasised that clinician-driven components (additional blood products/adjuncts) can inflate apparent haemostatic benefit in unblinded settings, even with adjudication, and highlighted post-randomisation exclusions as a key interpretive limitation.⁶
• Dose equivalence and “fair comparison”: Correspondence questioned whether fixed PCC dosing (threshold-based IU) and plasma dosing (units) yield comparable factor replacement across body sizes and baseline INR strata, and whether differing factor profiles (e.g., factor IX content variability in plasma) could confound interpretation of superiority claims.⁷
• Eligibility enrichment vs generalisability: Correspondence raised that reliance on point-of-care INR confirmation and selection for “suspected factor deficiency” may limit external validity for the broader phenotype of post-CPB bleeding (where platelet dysfunction, fibrinolysis, surgical bleeding, or heparin rebound may dominate).⁸
• Mechanisms for the AKI signal: Correspondence argued the AKI difference may plausibly relate to plasma-related volume load and venous congestion (a hypothesis consistent with perioperative haemodynamic physiology), but the trial did not measure central venous pressure or renal congestion markers.⁹
• Author response: Trialists responded that INR confirmation and bleeding-severity criteria were intended to target true coagulation factor deficiency; they also clarified protocol allowances for rescue products and emphasised that the trial compared “coagulation factor replacement strategies” under real-world bleeding-management conditions.¹⁰
• Subsequent evidence synthesis: A recent meta-analysis of PCC vs plasma in cardiac surgery supported reductions in transfusion and bleeding outcomes with PCC, but highlighted ongoing uncertainty driven by trial heterogeneity, dosing differences, and outcome definitions.⁴
• Safety context from outside cardiac surgery: The PROCOAG trial (trauma) reported more thromboembolic events with empiric PCC-first strategies in major haemorrhage, reinforcing the importance of restricting PCC use to settings with confirmed/likely factor deficiency rather than undifferentiated bleeding.¹¹
• Guideline trajectory: Contemporary perioperative guidance supports PCC use in selected surgical bleeding with attention to dosing and thrombotic risk, but formal integration of FARES-II findings into society guidance depends on update cycles and local product/regulatory context.¹²¹³¹⁴

Summary

• In adults with post-CPB coagulopathic bleeding requiring factor replacement, 4F-PCC achieved higher effective haemostatic response than frozen plasma (77.9% vs 60.4%), meeting non-inferiority and demonstrating superiority.
• 4F-PCC substantially reduced transfusion burden and bleeding volume, including lower chest tube drainage at 12 h (471 vs 642 mL) and fewer allogeneic product units within 24 h.
• Serious adverse events and AKI occurred less frequently with 4F-PCC (SAE 36.2% vs 47.3%; AKI 10.3% vs 18.8%), while thromboembolic events and mortality were similar.
• Interpretation is tempered by open-label delivery, a clinician-mediated composite primary endpoint, and substantial post-randomisation exclusions (especially delayed consent declinations).
• FARES-II supports targeted 4F-PCC as first-line factor replacement for selected coagulopathic bleeding after CPB, with ongoing need for implementation science, cost-effectiveness analyses, and guideline integration.

Overall Takeaway

FARES-II provides robust evidence that, in a selected post-CPB bleeding population enriched for coagulation factor deficiency, 4F-PCC can outperform frozen plasma on haemostatic efficacy while reducing transfusion and bleeding burden and without an obvious thromboembolic penalty. The trial’s pragmatic design and strong biological separation support practice change in appropriately resourced centres, but open-label delivery and extensive post-randomisation exclusions require careful interpretation and thoughtful implementation.

Bibliography

• 1Karkouti K, Bartoszko J, Grewal D, et al. Comparison of 4-factor prothrombin complex concentrate and frozen plasma for management of bleeding during cardiac surgery: a pilot randomized clinical trial. JAMA Netw Open. 2021;4(4):e213936.
• 2Green L, Roberts N, Cooper J, et al. Prothrombin complex concentrate versus fresh frozen plasma in adult patients undergoing heart surgery (PROPHESY): a pilot randomised controlled trial. Anaesthesia. 2021;76(7):892-901.
• 3Smith MM, Schroeder DR, Nelson JA, et al. Prothrombin complex concentrate vs plasma for postcardiopulmonary bypass coagulopathy and bleeding: a randomized clinical trial. JAMA Surg. 2022;157(9):757-764.
• 4Viana P, Relvas JH, Persson M, et al. Prothrombin complex concentrates vs frozen plasma to control bleeding after cardiac surgery: a systematic review and meta-analysis. J Chest Surg. 2024;57(1):25-35.
• 5Bartoszko J, Callum JL, Karkouti K, et al. Prothrombin complex concentrate versus frozen plasma for coagulopathic bleeding in cardiac surgery (FARES-II): study protocol for a randomised trial. BMJ Open. 2024;14(8):e091381.
• 6Wang R, Bennett-Guerrero E. Is it time to replace plasma with prothrombin complex concentrate in cardiac surgery? JAMA. 2025;333(20):1777-1778.
• 7Mazzeffi MA, Henderson R. Questions about strategies and dosing levels of PCC for coagulopathic bleeding in cardiac surgery. JAMA. 2025;334(9):831.
• 8González-Ruiz FJ, Queb-Pech NM, Portillo-Yañez IE. Questions about strategies and dosing levels of PCC for coagulopathic bleeding in cardiac surgery. JAMA. 2025;334(9):830-831.
• 9Miles LF, Molaee M, Auzinger G. Questions about strategies and dosing levels of PCC for coagulopathic bleeding in cardiac surgery. JAMA. 2025;334(9):831-832.
• 10Bartoszko J, Callum JL, Karkouti K. In Reply. JAMA. 2025;334(9):832-833.
• 11Bouzat P, Barbier O, Ract C, et al. Prothrombin complex concentrate vs fresh frozen plasma for major trauma bleeding (PROCOAG): a randomized clinical trial. JAMA. 2023;329(16):1367-1375.
• 12Raphael J, Mazer CD, Subramani S, et al. Society of Cardiovascular Anesthesiologists clinical practice improvement advisory for management of bleeding and blood conservation in cardiac surgery patients. Anesth Analg. 2019;129(5):1209-1221.
• 13Erdoes G, Koster A, Ortmann E, et al. A European consensus statement on the use of four-factor prothrombin complex concentrate for cardiac and non-cardiac surgical patients. Anaesthesia. 2020;76(3):381-392.
• 14Salenger R, Arora RC, Bracey A, et al. Cardiac surgical bleeding, transfusion, and quality metrics: joint consensus statement by the ERAS Cardiac Society and the Society for the Advancement of Blood Management. Ann Thorac Surg. 2025;119(2):280-295.