Skip to main content

Foundational Trials

CLIP-II

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Cryopreserved vs liquid-stored platelets for the treatment of surgical bleeding: the CLIP-II randomized noninferiority clinical trial
  • Acronym: CLIP-II
  • Year: 2025
  • Journal published in: JAMA
  • Citation: Reade MC, Marks DC, Howe BD, Bailey MJ, Bannon PG, Eastwood GM, et al; CLIP-II Investigators. Cryopreserved vs liquid-stored platelets for the treatment of surgical bleeding: the CLIP-II randomized noninferiority clinical trial. JAMA. Published online December 8, 2025.

Context & Rationale

  • Background
    • Liquid-stored platelets require room-temperature storage with agitation and have a short shelf-life (typically 5–7 days), contributing to supply fragility, wastage, and “platelet deserts” in low-volume, rural, and expeditionary settings.
    • Dimethyl sulfoxide (DMSO) cryopreservation at −80 °C can extend storage to years, but cryopreservation alters platelet phenotype (activation/procoagulant changes) and may affect haemostatic efficacy and thrombosis risk.
    • Existing clinical evidence for cryopreserved platelets in major bleeding and cardiac surgery was limited and heterogeneous; modern, blinded comparative effectiveness data were lacking.
  • Research Question/Hypothesis
    • In patients at high risk of platelet transfusion undergoing cardiac surgery, are cryopreserved platelets noninferior to liquid-stored platelets for haemostatic effectiveness when platelet transfusion is clinically indicated?
    • Does use of cryopreserved platelets result in comparable safety outcomes (including thromboembolic events and other major postoperative complications)?
  • Why This Matters
    • If noninferior, cryopreserved platelets could materially improve resilience of platelet supply chains (reduced wastage; improved access during shortages and for remote hospitals).
    • If less effective, routine substitution could increase bleeding, transfusion exposure, and downstream morbidity—offsetting logistical advantages and potentially worsening outcomes.
    • The trial tests a pragmatic real-world question: whether a long-shelf-life platelet product can safely replace standard platelets for surgical bleeding management.

Design & Methods

  • Research Question: Among high-risk cardiac surgery patients who receive platelet transfusion intraoperatively or within 24 hours of ICU admission, are DMSO cryopreserved platelets noninferior to liquid-stored platelets in haemostatic effectiveness (24-hour postsurgical chest drain blood loss in ICU)?
  • Study Type: Multicentre, randomised, double-blind, parallel-group, noninferiority clinical trial conducted at 11 Australian tertiary hospitals (enrolment August 2021 to April 2024; follow-up to day 90; follow-up completed July 2024).
  • Population:
    • Setting: Adult cardiac surgery (cardiopulmonary bypass and sternotomy common) with postoperative ICU care.
    • Inclusion (risk enrichment): Adults at high risk of platelet transfusion (Australian and New Zealand Society of Cardiac and Thoracic Surgeons Adult Cardiac Surgery Preoperative Transfusion [ACSePT] score ≥1, or clinician judgement) and not previously transfused platelets during the current hospital admission.
    • Key exclusions (published summary): Females aged 18–55 years with RhD-negative or unknown RhD status; history of deep vein thrombosis or pulmonary embolism; coagulopathic conditions; other protocol-defined exclusions.
    • Pragmatic trigger: Study product administered only if platelet transfusion was clinically indicated during surgery or within 24 hours after ICU admission.
  • Intervention:
    • Product: DMSO cryopreserved group O apheresis platelets (reconstituted in ABO-matched plasma or AB plasma) prepared for long-term storage at −80 °C.
    • Dose and timing: Up to 3 units, initiated intraoperatively or within 24 hours after ICU admission when platelet transfusion indicated.
    • Rescue: Additional liquid-stored “open-label” platelets permitted if needed (including after 3 study units; and per protocol-defined withholding criteria).
  • Comparison:
    • Product: Standard liquid-stored platelets (buffy coat–derived pooled platelets or apheresis platelets; ABO-matched as per usual practice).
    • Dose and timing: Up to 3 units, initiated intraoperatively or within 24 hours after ICU admission when platelet transfusion indicated.
    • Rescue: Open-label platelets and other haemostatic products allowed as clinically required.
  • Blinding: Double-blind (patients and treating clinicians blinded); allocation known to blood bank and designated unblinded staff; risk of inadvertent unblinding increased where rescue (open-label platelets) was required.
  • Statistics: Noninferiority margin set at a 20% increase in mean 24-hour ICU bleeding (equivalent to 188 mL, based on a mean of 943 mL); 202 transfused patients (101/group) required for 90% power with a one-sided 95% CI approach; primary analysis performed in a modified intention-to-treat cohort including patients who received platelet transfusion (log-transformed bleeding; effect as ratio of geometric means; adjusted analyses included hospital and EuroSCORE II).
  • Follow-Up Period: Postoperative outcomes assessed through ICU and hospital stay; mortality and selected complications captured to day 90.

Key Results

This trial was not stopped early. Recruitment concluded after achieving the prespecified sample size of transfused participants (388 randomised; 202 received study platelets and were included in the primary analysis).

Outcome Cryopreserved platelets Liquid-stored platelets Effect p value / 95% CI Notes
Primary: Postsurgical bleeding in first 24 h in ICU (mL; geometric mean [95% CI]) 605 (532–688) 535 (480–596) Ratio 1.13 95% CI 0.96 to 1.34; one-sided P=0.07 Noninferiority margin: 1.20 (20% higher bleeding); noninferiority not established (upper CI 1.34)
Total volume of postsurgical chest drain bleeding (mL; geometric mean [95% CI]) 1273 (1096–1479) 972 (848–1114) Adjusted ratio 1.29 95% CI 1.06 to 1.56 Cryopreserved platelets associated with greater total postsurgical bleeding
Bleeding Academic Research Consortium type 4 bleeding (composite) (No./total [%]) 33/104 (31.7%) 19/98 (19.4%) Adjusted risk difference 9.8% 95% CI −0.6% to 20.3% Composite included >4 RBC units transfused within 48 h and/or reoperation for bleeding
Received RBC transfusion in first 24 h in ICU (No./total [%]) 55/104 (52.9%) 38/98 (38.8%) Adjusted risk difference 13.2% 95% CI 3.8% to 22.5% Higher exposure to RBC transfusion with cryopreserved platelets
Received FFP transfusion at any time (No./total [%]) 78/104 (75.0%) 49/98 (50.0%) Risk difference 25.0% 95% CI 12.1% to 37.9% Greater haemostatic product utilisation in cryopreserved group
Received cryoprecipitate transfusion at any time (No./total [%]) 74/104 (71.2%) 53/98 (54.1%) Risk difference 17.1% 95% CI 3.9% to 30.2% More frequent fibrinogen replacement with cryopreserved platelets
Received open-label (nonstudy) liquid platelets at any time (No./total [%]) 44/104 (42.3%) 12/98 (12.2%) Risk difference 30.1% 95% CI 18.6% to 41.6% Marked rescue platelet use suggests reduced haemostatic adequacy and/or lower clinician confidence
Lowest platelet count in first 24 h in ICU (×109/L; median [IQR]) 103 (81–129) 140 (110–177) Difference −36 95% CI −50 to −22 Physiological separation consistent with lower circulating platelet counts postoperatively
Mortality at day 90 (No./total [%]) 13/104 (12.5%) 5/98 (5.1%) Not reported Not reported Trial not powered for mortality/safety differences; interpret cautiously
ICU length of stay (days; median [IQR]) 3.8 (2.0–6.0) 3.0 (1.9–4.9) Not reported Not reported Longer ICU stay with cryopreserved platelets in reported medians
    • Noninferiority was not established for the primary endpoint: the upper bound of the 95% CI for the bleeding ratio (1.34) exceeded the prespecified noninferiority margin (1.20).
    • Across multiple secondary endpoints, cryopreserved platelets were associated with greater bleeding (total chest drain output 1273 vs 972 mL) and increased utilisation of blood components (eg, RBC transfusion in first 24 h: 52.9% vs 38.8%; FFP exposure: 75.0% vs 50.0%).
    • Safety outcomes were not statistically compared in the main report; several adverse outcomes (including mortality) were numerically higher in the cryopreserved group, but event rates were low and the trial was not powered for safety.

Internal Validity

    • Randomisation and allocation concealment: Central, computer-generated randomisation with permuted blocks (sizes 2–4) stratified by hospital; allocation concealed from treating clinicians and patients.
    • Post-randomisation exclusions: 388 randomised, but the primary analysis included only transfused participants (202 total: 104 cryopreserved vs 98 liquid); 92 and 94 participants (respectively) underwent surgery but did not require platelet transfusion, and 1 participant in the liquid group did not undergo surgery—creating a modified intention-to-treat cohort that is more vulnerable to selection effects than a full intention-to-treat analysis.
    • Blinding and performance bias: The trial was designed as double-blind; however, differential need for open-label platelets (42.3% vs 12.2%) plausibly increased the risk of unblinding and could have influenced subsequent transfusion and haemostatic management decisions.
    • Protocol adherence and crossover: Protocol separation was incomplete due to rescue; open-label platelet exposure differed substantially (44/104 vs 12/98), and early open-label use before completion of study platelets occurred in 25.0% vs 3.1%.
    • Baseline comparability: Baseline characteristics among transfused participants were broadly similar (eg, age 66.9 vs 67.2 years; EuroSCORE II 4.5 vs 4.5; cardiopulmonary bypass time 137 vs 141 minutes), supporting comparability within the analysed cohort.
    • Heterogeneity: Multicentre design (11 hospitals) with stratification by site; recruitment was concentrated in a minority of sites (72.8% from 3 hospitals), which may amplify centre-level practice effects despite stratification.
    • Timing: First platelet transfusion occurred in the operating room for most patients (84.6% vs 88.8%); delay from platelet order to infusion was similar (median 60 vs 57 minutes), reducing differential timing bias between groups.
    • Dose: Cryopreserved recipients received more study platelets (median 2 vs 1 units) and more total platelets (study + open-label: median 2 vs 1 units), suggesting weaker haemostatic effect per unit and potential dose confounding in pragmatic care.
    • Separation of the variable of interest: Despite intended equivalence, there was measurable physiological separation (lowest platelet count first 24 h in ICU: 103 vs 140 ×109/L; difference −36 [95% CI −50 to −22]) and substantial differences in downstream transfusion behaviour (eg, open-label platelet exposure: 42.3% vs 12.2%).
    • Adjunctive therapy use: Cryopreserved group had greater exposure to haemostatic adjuncts (FFP 75.0% vs 50.0%; cryoprecipitate 71.2% vs 54.1%), which may represent compensatory care and complicate attribution of outcomes to platelet type alone.
    • Outcome assessment: Primary endpoint (chest drain volume) is objective and routinely collected, reducing detection bias; however, thresholds for transfusion and decisions to administer rescue products are clinician-driven and potentially susceptible to unblinding and practice variation.
    • Statistical rigour: The noninferiority framework and analysis methods were prespecified, with effect expressed as ratio of geometric means and one-sided CI; target transfused sample size was achieved, but noninferiority was not demonstrated and several key secondary outcomes suggested harm.

Conclusion on Internal Validity: Overall, internal validity is moderate: randomisation and objective outcome measurement were strong, but the modified intention-to-treat (transfused-only) analysis, substantial differential rescue platelet use, and likely partial unblinding reduce confidence that observed differences purely reflect platelet product efficacy.

External Validity

    • Population representativeness: High-risk adult cardiac surgery patients at Australian tertiary centres, many undergoing complex operations (eg, isolated CABG 30.3% vs 24.5%; valve surgery with CABG and/or other procedures common); findings most directly apply to similar cardiothoracic surgical populations.
    • Intervention specificity: Cryopreserved platelets were group O apheresis units reconstituted in plasma; performance may differ with other cryopreservation methods, storage durations, resuspension media, or ABO strategies.
    • System and resource context: Trial sites were well-resourced tertiary centres with established transfusion services; generalisability to low-resource hospitals, rural “platelet deserts”, and prehospital/military settings is uncertain, despite these being key target use-cases for long-shelf-life platelets.
    • Applicability across bleeding syndromes: This was cardiac surgery bleeding; extrapolation to trauma, obstetric haemorrhage, and medical causes of thrombocytopenia should not be assumed without direct evidence.

Conclusion on External Validity: External validity is moderate for high-risk cardiac surgery in high-income settings using similar cryopreserved platelet manufacture and transfusion workflows, but limited for other haemorrhage contexts and the remote settings most motivated by platelet shelf-life constraints.

Strengths & Limitations

  • Strengths:
    • Double-blind, randomised multicentre design with stratification by hospital.
    • Objective primary endpoint (24-hour ICU chest drain bleeding) with prespecified noninferiority margin and analysis approach.
    • Pragmatic embedding into real-world perioperative bleeding management (including allowance for rescue products), improving clinical relevance.
    • Directly addresses a key translational question: can a logistics-optimised platelet product replace standard platelets in major surgical bleeding?
  • Limitations:
    • Primary analysis included only transfused participants (202/388), weakening classic intention-to-treat protection against selection bias.
    • Differential rescue/open-label platelet use (42.3% vs 12.2%) likely reduced blinding integrity and complicated causal attribution.
    • Trial was not powered to robustly assess uncommon safety outcomes (eg, thromboembolism) or mortality; numerical imbalances cannot be resolved definitively.
    • Cryopreserved platelets were group O and reconstituted in plasma; results may not generalise to other cryopreserved products, ABO strategies, or storage technologies.

Interpretation & Why It Matters

  • Clinical implications
    • In high-risk cardiac surgery patients requiring platelet transfusion, this cryopreserved platelet strategy did not meet noninferiority for haemostatic effectiveness and was associated with higher bleeding and increased transfusion exposure.
    • Routine replacement of standard liquid-stored platelets with cryopreserved platelets in tertiary cardiac surgery programmes is not supported by these data.
    • The trial reframes cryopreserved platelets as a potential contingency product (eg, during supply disruption) rather than a like-for-like substitute in settings where liquid platelets are readily available.

Controversies & Subsequent Evidence

    • Noninferiority design choices: The trial used a 20% noninferiority margin (188 mL) and a transfused-only primary analysis cohort (modified intention-to-treat), aligning with the pragmatic clinical question but reducing the protective features of a full intention-to-treat framework.2
    • Potential unblinding via rescue: The large between-group difference in open-label platelet use (42.3% vs 12.2%) plausibly compromised blinding and may have contributed to downstream transfusion differences; the accompanying editorial emphasised this as a key interpretive challenge.1
    • Consistency with prior trials: A pilot RCT in perioperative bleeding (CLIP-I) supported feasibility but was underpowered for definitive clinical outcomes, while an older bypass-era trial suggested potential differences in haemostatic performance—underscoring the need for contemporary, adequately powered trials (which CLIP-II provides).39
    • Biology vs effectiveness: Cryopreservation-associated platelet activation and functional change is well described; CLIP-II’s lower postoperative platelet counts and greater haemostatic product needs are consistent with reduced effective platelet replacement in this context.5
    • Economic implications: A subsequent economic evaluation based on CLIP-II data reported that, in an Australian tertiary cardiac surgery context, cryopreserved platelets were not cost-effective relative to liquid-stored platelets under base-case assumptions.4
    • Broader evidence base and guidance: Recent systematic reviews and contemporary transfusion guidelines provide evolving evidence on platelet transfusion practice, but direct evidence to support routine substitution of cryopreserved for liquid-stored platelets in major surgical bleeding remains limited.678

Summary

    • Multicentre, double-blind noninferiority RCT in 11 Australian tertiary hospitals enrolling high-risk cardiac surgery patients (388 randomised; 202 transfused and analysed).
    • Primary endpoint (24-hour ICU chest drain bleeding) did not demonstrate noninferiority: ratio 1.13; 95% CI 0.96 to 1.34; one-sided P=0.07; noninferiority margin 1.20.
    • Cryopreserved platelets were associated with greater total postsurgical bleeding (1273 vs 972 mL) and greater utilisation of blood components (eg, RBC exposure in first 24 h: 52.9% vs 38.8%; FFP: 75.0% vs 50.0%).
    • Rescue/open-label platelets were substantially more frequent with cryopreserved platelets (42.3% vs 12.2%), complicating blinding and causal attribution while also supporting a signal of reduced clinical effectiveness.
    • Safety outcomes were not definitively different, but the trial was not powered for uncommon adverse events; numerical imbalances (including mortality) should be interpreted cautiously.

Overall Takeaway

CLIP-II is a landmark modern evaluation of cryopreserved platelets in a high-risk surgical bleeding population: despite strong logistics appeal, cryopreserved platelets did not meet noninferiority for 24-hour postoperative bleeding and were associated with higher bleeding and greater transfusion exposure. The trial reshapes practice by supporting liquid-stored platelets as the default for tertiary cardiac surgery bleeding, while clarifying that any future role for cryopreserved platelets will require improved products, clearer indications, and stronger safety and economic evidence.

Overall Summary

  • Cryopreserved platelets did not demonstrate noninferiority to liquid-stored platelets for haemostatic effectiveness in high-risk cardiac surgery bleeding and were associated with increased bleeding and transfusion requirements.

Bibliography