Consensus recommendations for patient selection and clinical implementation of polymyxin B hemoadsorption in endotoxic septic shock

Abstract

Background

Despite long-standing clinical use and emerging evidence, no structured framework exists guiding clinicians in identifying septic shock patients for endotoxemia assessment, interpretation of the Endotoxin Activity Assay (EAA), or practical implementation of polymyxin B hemoadsorption (PMX-HA) therapy.

Methods

A modified Delphi consensus process was conducted with ten critical care and nephrology experts, compromising a pre-meeting survey, a one-day in-person consensus conference, and iterative voting on draft statements. Consensus was predefined as ≥ 80% agreement.

Results

Consensus was achieved on 20 statements across three domains: (1) operational criteria to prompt endotoxemia assessment in suspected or presumed septic shock, (2) role and interpretation of the EAA, and (3) logistics and clinical implementation of PMX-HA. Key recommendations included standardized criteria to identify eligible patients, EAA thresholds of ≥ 0.60 and < 0.90 to support PMX‑HA consideration, initiation of therapy as early as possible once eligibility is confirmed, with ≤ 36 h from septic shock onset proposed as a pragmatic implementation benchmark, and standardized treatment protocols consisting of two sessions, 24 h apart, targeting 120 min each. A multidisciplinary team model was endorsed for implementation. Identified research priorities were in treatment effect heterogeneity, EAA kinetics, clinical implementation, and healthcare system integration.

Conclusions

These consensus recommendations provide a structured, implementation-focused framework for PMX-HA use in suspected endotoxic shock. The framework emphasizes stepwise operational criteria, biomarker-guided decision-making, multidisciplinary team-based implementation, and standardized protocols. It is intended to support institutions currently using PMX-HA and to assist U.S. clinicians and health systems in preparing for potential adoption pending regulatory approval.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13054-026-06182-8.

Introduction

Sepsis remains a leading cause of morbidity and mortality worldwide, with septic shock representing its most severe and life-threatening form, defined by profound circulatory and cellular/metabolic dysfunction [1]. Despite significant advances in supportive care and standardized implementation of Surviving Sepsis Campaign guidelines [2], mortality in septic shock remains unacceptably high. Endotoxin, a potent component of gram-negative bacterial cell walls, plays a central role in the pathogenesis of septic shock by driving a dysregulated host inflammatory response and organ failure. Adjunctive extracorporeal blood purification techniques have been explored to mitigate this burden in select patients with severe organ dysfunction [3, 4]. Polymyxin B hemoadsorption (PMX-HA) is one such therapy, which removes circulating endotoxin through direct binding to immobilized polymyxin B on a hemoadsorption cartridge [5, 6].

The clinical evidence evaluating PMX-HA in sepsis and septic shock has evolved over the past two decades. Observational studies and smaller clinical trials have reported improvements in hemodynamic parameters and organ dysfunction, although survival outcomes have varied across studies. Early meta-analyses suggested a potential mortality benefit, including those by Cruz et al. [7] and Chang et al. [8]. More recent meta-analyses by Chen et al. [9] and Meco et al. [10], have reported pooled estimates suggesting reduced mortality; however, substantial heterogeneity, differences in trial design, patient selection, and illness severity limit definitive conclusions. In contrast, Fujii et al. [11], in a meta-analysis restricted to randomized trials without enrichment strategies, found no significant reduction in 28-day mortality with PMX-HA and judged the certainty of evidence to be low. Collectively, these data suggest therapeutic potential while underscoring ongoing uncertainty regarding optimal patient selection and implementation strategies [7, 11–17].

The EUPHRATES trial [12], the largest randomized controlled trial of PMX-HA to date, did not demonstrate an overall mortality benefit in septic shock patients with elevated endotoxin levels. However, a post hoc analysis suggested improved outcomes in a targeted subgroup of patients with Endotoxin Activity Assay (EAA) levels ≥ 0.60 and < 0.90. This upper threshold reflects an endotoxin burden within the adsorption capacity of a single PMX cartridge and approximates the range in which the EAA assay reliably correlates with circulating endotoxin concentration. The recently published TIGRIS trial, which used a Bayesian adaptive design to prospectively evaluate this targeted selection strategy, reported a high probability of improved survival within the same EAA-defined subgroup, with 28-day mortality of 39% in the polymyxin B group compared with 45% in the control group (APACHE-II adjusted odds ratio 0.67, 95% credible interval 0.39–1.08), and a posterior probability of benefit of 95.3% [18]. It is important to recognize that the population studied in TIGRIS represents a highly selected subgroup of patients with septic shock. Of a large screened population, only a small proportion met eligibility criteria (of 14,890 patients screened, only 328 were enrolled, and an additional 171 were subsequently excluded), reflecting the specificity of the clinical and biomarker-defined phenotype. Accordingly, the proposed framework is intended for a narrowly defined subset of patients rather than the broader septic shock population.

Together, these findings highlight the importance of patient selection, treatment timing, and standardized implementation in optimizing PMX-HA therapy. However, in countries where PMX-HA is clinically available, substantial variability persists regarding which patients with established septic shock should undergo endotoxemia assessment, how EAA results should be interpreted in clinical context, and how PMX-HA programs should be operationalized across diverse healthcare settings.

Prior consensus initiatives, including those from the Acute Disease Quality Initiative (ADQI), have discussed the role of extracorporeal therapies in critical illness and sepsis-associated AKI [19]. However, these efforts did not provide a structured framework for the clinical implementation of PMX-HA. Consequently, uncertainty remains regarding how PMX-HA should be integrated into routine sepsis workflows, particularly in the setting of evolving sepsis definitions and growing interest in targeted therapies for mechanistically distinct sepsis subtypes characterized by endotoxemia. Recent Surviving Sepsis Campaign guidelines suggest against the use of polymyxin B hemoperfusion and other blood purification techniques in adults with sepsis or septic shock because of low certainty of evidence and inconsistent results across clinical studies [20]. While these guidelines address the level of evidence available at the time of their development, they do not provide practical recommendations regarding endotoxemia assessment, patient selection, interpretation of endotoxin activity assay results, or operational implementation in centers where PMX-HA is available. The present consensus was developed to address these implementation challenges and provide implementation-focused guidance for centers utilizing or evaluating PMX-HA within the context of existing sepsis guidelines.

To address this unmet need for implementation focused guidance, a multidisciplinary expert panel convened to develop consensus recommendations for the clinical application of PMX-HA in patients with septic shock and suspected endotoxemia. Using a modified Delphi methodology, the panel focused on three key domains: [1] operational criteria to prompt endotoxemia assessment and patient selection, [2] interpretation and clinical application of EAA results, and [3] logistical and implementation strategies, including team structure, timing, treatment protocols, and quality metrics. The resulting twenty consensus statements offer a structured, implementation-focused roadmap to guide institutions currently using PMX-HA and to support health systems preparing for future adoption pending regulatory approval.

Methods

Consensus process design

This modified Delphi consensus process was conducted and reported in accordance with the ACCORD (A Comprehensive Checklist for Consensus Reports Development) and DELPHISTAR (Delphi Study Reporting) guidelines to ensure methodological rigor and transparency.

Panel selection and governance

The consensus panel comprised ten experts selected to represent multidisciplinary and geographically diverse perspectives relevant to the clinical scope of this work. Panelists included specialists in pulmonary and critical care medicine, nephrology, anesthesiology and critical care, and critical care nursing, drawn from academic medical centers across the United States, including institutions in Alabama, California, Georgia, Colorado, Louisiana, Maryland, Michigan, and South Carolina. Practice settings encompassed large academic tertiary care centers and university-affiliated hospitals with active extracorporeal therapy programs. Several panelists hold leadership roles in CRRT, ECMO, and renal replacement therapy programs, and multiple members have direct investigator-level experience with the EUPHRATES and TIGRIS trials. Panelists were identified and invited based on subject matter expertise, publication record, and direct clinical experience relevant to sepsis management and extracorporeal blood purification. The panel was not formally constituted under the auspices of a professional society; however, the modified Delphi methodology was employed to minimize individual bias and generate structured, reproducible consensus independent of institutional or commercial influence.

Panel size was determined pragmatically to balance breadth of expert perspective with the logistical requirements of an in-person modified Delphi process, consistent with accepted norms for expert consensus panels in critical care medicine. All panelists received honoraria in recognition of the time committed to attendance at the in-person consensus meeting as well as the preparatory work required, including structured review of the supporting literature on sepsis, endotoxemia, and extracorporeal blood purification. Travel and accommodation expenses were also reimbursed. The consensus meeting was supported by Vantive; however, the sponsor had no role in the design of consensus questions, the deliberative process, voting outcomes, or the content of resulting recommendations. All consensus statements reflect the independent clinical and scientific judgment of the panel members.

Participation was voluntary, and all panelists contributed to the consensus development process, which was conducted in three sequential phases:

1. Pre-meeting survey: Panelists completed an electronic survey to identify key priorities and preliminary viewpoints across three predefined domains.

2. In-person consensus meeting: A one-day structured meeting included domain-specific breakout groups and plenary discussions to refine scope, terminology, and content.

3. Iterative voting and refinement: Draft consensus statements were iteratively revised and anonymously voted upon during and following the meeting until predefined consensus criteria were met.

While the initial framework included domains labeled “defining sepsis in the context of endotoxemia,” “role and interpretation of the Endotoxin Activity Assay,” and “clinical implementation of PMX-HA,” the panel refined and reorganized these domains during the consensus meeting to better reflect clinical priorities and discussion themes, without altering the substance of the statements subjected to voting. The final domain structure and consensus statements reflect this iterative, adaptive process and are consistent with established Delphi methodology.

Minor language revisions were made during manuscript preparation to improve clarity and align terminology with contemporary clinical usage. These edits did not modify the content, intent, or voting outcomes of any consensus statement.

Consensus definition

Consensus was defined a priori as ≥ 80% agreement among panelists. Agreement was calculated as the proportion of panelists selecting “agree” or “strongly agree,” with a total panel size of ten experts. If a statement did not reach the predefined consensus threshold, iterative discussion, revision, and repeat voting in accordance with the modified Delphi process were planned. All final statements included in the manuscript ultimately achieved consensus agreement.

Ethical considerations

All participants completed conflict of interest disclosures prior to participation. As this consensus development process did not involve patient-level data or clinical intervention, institutional review board approval was not required.

Results

Pre-meeting survey results

All ten panelists completed the pre-meeting survey. Across domains, responses consistently emphasized three major themes: [1] identification of patients with septic shock most likely to benefit from endotoxin-targeted therapy, [2] contextual interpretation of EAA results alongside clinical trajectory and organ dysfunction, and [3] operational integration of PMX-HA into existing ICU workflows and multidisciplinary care pathways. Detailed pre-meeting survey results are provided in Supplementary Table 1.

Consensus statements and voting results

The panel achieved consensus (≥ 80% agreement) on all 20 statements organized across three domains No statements were formally withdrawn during the consensus process. Consensus voting results and the full text of all consensus statements are presented in Tables 1, 2 and 3, and the proposed clinical workflow is illustrated in Fig. 1.

Table 1.

Domain 1: Endotoxemia assessment and PMX-HA consideration in patients with established septic shock according to Sepsis-3 criteria

Statement No	Consensus statement	Agreement, n/N (%)	Consensus achieved*
1	Endotoxemia assessment should be considered in patients with suspected or documented infection, and persistent vasopressor requirement following adequate fluid resuscitation, with circulatory instability persisting ≥ 2 h. The panel considered a time window of approximately 2 to 4 h after shock onset appropriate for assessment	10/10 (100%)	Yes
2	PMX-HA consideration should include patients with sustained vasopressor requirement, infection with source control achieved or planned, evidence of organ dysfunction, and an EAA level ≥ 0.60 and < 0.90. PMX-HA candidacy should not be restricted based on presumed Gram-negative infection status	9/10 (90%)	Yes
3	Patients who develop a new episode of septic shock following prior PMX-HA treatment should be reassessed using the same clinical and biomarker criteria to guide repeat endotoxemia assessment and therapy consideration	10/10 (100%)	Yes

Consensus was defined a priori as ≥ 80% agreement among panelists. Agreement reflects combined “agree” and “strongly agree” responses

Total panel size N = 10. PMX-HA, polymyxin B hemoadsorption; EAA, endotoxin activity assay

Table 2.

Domain 2: Role and interpretation of the Endotoxin Activity Assay (EAA)

A. Clinical consensus statements
Statement No	Consensus statement	Agreement, n/N (%)	Consensus achieved*
4	If the initial EAA level is < 0.60, repeat testing may be considered in patients without clinical improvement when PMX-HA therapy remains under consideration	10/10 (100%)	Yes
5	If the EAA level is ≥ 0.90, eligibility for PMX-HA therapy should be determined using clinical judgment incorporating overall prognosis and resource considerations. Repeat EAA testing is not mandatory	10/10 (100%)	Yes

B. Research consensus statements
Statement No	Consensus statement	Agreement, n/N (%)	Consensus achieved*
6	Research should evaluate EAA kinetics to inform treatment eligibility assessment and repeat treatment decisions	10/10 (100%)	Yes
7	Research should examine EAA dynamics across sepsis subphenotypes and clinical trajectories	10/10 (100%)	Yes
8	Research should assess real-world EAA assay performance across laboratories and healthcare settings	10/10 (100%)	Yes
9	Research should explore integration of EAA results with clinical decision-support systems	10/10 (100%)	Yes
10	Research should identify implementation strategies to support timely EAA testing and therapy delivery	10/10 (100%)	Yes

Consensus was defined a priori as ≥ 80% agreement among panelists. Agreement reflects combined “agree” and “strongly agree” responses.

Total panel size N = 10. PMX-HA, polymyxin B hemoadsorption; EAA, endotoxin activity assay

Table 3.

Domain 3: Logistics and clinical implementation of PMX-HA

Statement No	Consensus statement	Agreement, n/N (%)	Consensus achieved*
11	A multidisciplinary team model should be established to coordinate PMX-HA therapy activation, delivery, and monitoring	10/10 (100%)	Yes
12	Formal training and competency maintenance should be established for clinicians and staff involved in PMX-HA delivery	10/10 (100%)	Yes
13	Standalone PMX-HA should be delivered using a standardized protocol consisting of two treatment sessions performed 24 h apart, with target duration of 120 min per session (minimum 90 min if early circuit complications occur) Anticoagulation practices during PMX-HA should align with institutional extracorporeal therapy protocols, bleeding risk assessment, and circuit-specific considerations	10/10 (100%)	Yes
14	Blood flow rates during PMX-HA therapy should typically be maintained between 80–120 mL/min	10/10 (100%)	Yes
15	PMX-HA therapy should be initiated as early as possible following septic shock onset, with ≤ 36 h considered a practical implementation benchmark rather than a rigid cutoff	10/10 (100%)	Yes
16	PMX-HA activation processes should include timely consent, line placement, specialty consultation, and nursing mobilization	10/10 (100%)	Yes
17	Patients receiving PMX-HA therapy should undergo monitoring of hemodynamics, vasopressor requirements, fluid balance, and organ function	10/10 (100%)	Yes
18	Monitoring protocols should align with institutional critical care and CRRT standards	10/10 (100%)	Yes
19	Standardized documentation of PMX-HA treatment parameters, adverse events, and clinical outcomes should be implemented	10/10 (100%)	Yes
20	PMX-HA programs should be integrated into existing ICU sepsis care pathways	10/10 (100%)	Yes

Consensus was defined a priori as ≥ 80% agreement among panelists. Agreement reflects combined “agree” and “strongly agree” responses. Total panel size N = 10

PMX-HA, polymyxin B hemoadsorption; ICU, intensive care unit; CRRT, continuous renal replacement therapy

Fig. 1.

Operational framework for endotoxemia assessment and PMX-HA consideration in established septic shock. This figure depicts a practical, implementation-focused, two-step workflow for patients with established septic shock according to Sepsis-3 criteria. The framework is intended to guide endotoxemia assessment and consideration of PMX-HA after septic shock has been established. The panel agreed that the presence of an elevated lactate level at the time endotoxemia assessment is being considered should not be required to prompt EAA testing and should not serve as an independent criterion for PMX-HA eligibility. Clinical decision making should instead emphasize persistent vasopressor-dependent circulatory failure and evidence of ongoing organ dysfunction despite adequate resuscitation. Abbreviations: EAA, Endotoxin Activity Assay; MAP, mean arterial pressure; PMX-HA, polymyxin B hemoadsorption

Domain 1: Identifying septic shock patients for endotoxemia assessment and PMX-HA consideration

Domain 1 defines pragmatic criteria to prompt endotoxemia assessment and consideration of PMX-HA therapy in patients with septic shock. Given heterogeneous trial evidence and inconsistent mortality benefit across unselected populations, PMX-HA should be regarded as a potential adjunctive therapy rather than a universal intervention. Because the goal of this consensus was to provide implementation guidance rather than replicate clinical trial enrollment criteria, the panel considered both prior trial experience and practical ICU workflow constraints when developing operational thresholds. Patient selection principles were informed by strategies used in the TIGRIS trial, which targeted patients with high-severity circulatory shock. Because lactate is included in the Sepsis-3 definition of septic shock, the panel recognized the importance of clarifying its role within the proposed framework. The panel agreed that once septic shock has been established according to Sepsis-3 criteria, the presence of an elevated lactate level at the time endotoxemia assessment is being considered should not be required to prompt EAA testing and should not serve as an independent criterion for PMX-HA eligibility. This approach is consistent with EUPHRATES and TIGRIS, in which lactate thresholds were not used to guide endotoxemia assessment or determine eligibility for PMX-HA therapy. Clinical decision making should instead emphasize persistent vasopressor-dependent hypotension and evidence of ongoing organ dysfunction despite adequate resuscitation when considering endotoxin testing and PMX-HA therapy.

Statement 1 (Agreement: 10/10, 100%): The panel endorsed operational criteria to prompt endotoxemia assessment in patients with established septic shock according to Sepsis-3 criteria. Recommended components include:

• Definite infection or ongoing concern for suspected infection

• Sustained vasopressor requirement (> 2 h) despite adequate fluid resuscitation to maintain MAP ≥ 65 mmHg

• Norepinephrine-equivalent dose ≥ 0.05 µg/kg/min, interpreted in the context of sedation status and use of vasodilatory or hypotension-prone sedatives (e.g., propofol, midazolam)

The presence of an elevated lactate level was not considered necessary to prompt endotoxemia assessment once septic shock had been established according to Sepsis-3 criteria. PMX-HA clinical trials required a minimum of approximately 2 h of vasopressor dependent shock after adequate resuscitation to confirm sustained circulatory failure. The panel considered a time window of approximately 2 to 4 h of persistent shock after adequate resuscitation appropriate for assessment, supporting early evaluation while permitting confirmation of sustained circulatory failure when clinically appropriate. Earlier assessment remains appropriate when clinically indicated.

Statement 2 (Agreement: 9/10, 90%): The panel endorsed a multi-component selection framework for PMX-HA consideration in patients with septic shock. Eligibility criteria include:

• Septic shock with sustained vasopressor requirement despite adequate fluid resuscitation

• Source control achieved or planned

• Evidence of multiple organ dysfunction, defined using established scoring systems (e.g., SOFA score > 11 or MODS > 9), consistent with prior clinical trials, including TIGRIS

• Endotoxin Activity Assay (EAA) level ≥ 0.60 and < 0.90

The panel emphasized that, in appropriately selected patients, therapy should be initiated as early as feasible following confirmation of eligibility criteria and that candidacy should not be restricted by presumed Gram-negative status.

Statement 3 (Agreement: 10/10, 100%): For patients who develop a new septic episode after prior PMX-HA treatment, the panel recommended reassessment using the same criteria as Statements 1–2.

Domain 1 consensus voting results are summarized in Table 1 and the stepwise workflow is shown in Fig. 1.

Domain 2: Role and interpretation of the Endotoxin Activity Assay (EAA)

Domain 2 focused on how EAA results should be interpreted in clinical context and applied to guide decision-making regarding PMX-HA therapy in patients with established septic shock. Statements addressed initial testing thresholds, repeat testing strategies, and areas where further research is needed.

Statement 4 (Agreement: 10/10, 100%): If the initial EAA is < 0.60, repeat testing may be considered in patients without clinical improvement when PMX-HA remains under consideration.

Statement 5 (Agreement: 10/10, 100%): If the EAA is ≥ 0.90, eligibility for PMX-HA should be determined using clinical judgment that incorporates overall prognosis and resource availability; repeat EAA testing is not mandatory.

Statements 6–10 (Agreement: 10/10, 100% for each statement): The panel identified key research priorities related to EAA use, including evaluation of EAA kinetics for eligibility assessment and repeat treatment decisions, integration of EAA results with clinical parameters and sepsis phenotypes, assessment of real-world assay performance across laboratory settings, and development of implementation strategies to support timely testing and therapy delivery. These statements reflect consensus priorities for both clinical implementation and future investigation.

Domain 2 consensus statements and voting results are summarized in Table 2.

Domain 3: Logistics and clinical implementation of PMX-HA

Domain 3 addressed practical considerations for implementing PMX-HA in clinical practice, including program structure, treatment delivery standards, monitoring requirements, and integration into existing ICU workflows.

Statement 11 (Agreement: 10/10, 100%): The panel recommended establishing a multidisciplinary team to coordinate PMX-HA activation, delivery, and monitoring.

Statement 12 (Agreement: 10/10, 100%): The panel recommended formal training and competency maintenance for clinicians and staff involved in PMX-HA delivery.

Statement 13 (Agreement: 10/10, 100%): The panel recommended delivery of PMX-HA as stand-alone hemoperfusion using a standardized protocol consisting of two treatment sessions performed 24 h apart, with a target duration of 120 min per session. Anticoagulation practices during PMX-HA should align with institutional extracorporeal therapy protocols, bleeding risk assessment, and circuit-specific considerations.

Statement 14 (Agreement: 10/10, 100%): The panel recommended maintaining blood flow rates during PMX-HA therapy between 80–120 mL/min.

Statement 15 (Agreement: 10/10, 100%): The panel recommended initiating PMX-HA as early as possible following septic shock onset, with ≤ 36 h considered a practical implementation benchmark.

Clinical trials evaluating PMX-HA enrolled patients relatively early in the course of septic shock and generally excluded patients with shock duration exceeding approximately 24–30 h from onset. Recognizing that real-world clinical workflows, including confirmation of eligibility, endotoxin testing, vascular access placement, and treatment mobilization, may introduce unavoidable delays, the panel proposed ≤ 36 h as a pragmatic implementation benchmark rather than strict replication of trial enrollment criteria. Importantly, therapy should be initiated as early as feasible once eligibility criteria are met, and this threshold should not be interpreted as justification for delaying therapy.

Statement 16 (Agreement: 10/10, 100%): The panel recommended establishing structured activation processes including timely consent, vascular access placement, specialty consultation, and nursing mobilization.

Statement 17 (Agreement: 10/10, 100%): The panel recommended monitoring hemodynamics, vasopressor requirements, fluid balance, and organ function during PMX-HA therapy.

Statement 18 (Agreement: 10/10, 100%): The panel recommended monitoring protocols aligned with institutional critical care and CRRT standards.

Statement 19 (Agreement: 10/10, 100%): The panel recommended standardized documentation of PMX-HA treatment parameters, adverse events, and clinical outcomes.

Statement 20 (Agreement: 10/10, 100%): The panel recommended integration of PMX-HA programs into existing ICU sepsis care pathways.

Domain 3 consensus statements and voting results are summarized in Table 3.

Discussion

This expert consensus initiative aimed to address a pragmatic gap in the care of patients with septic shock: how to operationalize endotoxin assessment and PMX-HA therapy in a consistent, implementation-focused, and evidence-informed manner. The panel did not seek to redefine sepsis or septic shock [1, 21] but rather to provide implementation guidance for clinicians managing patients with septic shock in whom endotoxemia may contribute to persistent circulatory failure and organ dysfunction. These recommendations are intended to provide implementation-focused guidance within the context of existing sepsis guidelines and should be applied with clinical judgment in the setting of evolving evidence.

A central focus of domain 1 was distinguishing the diagnosis of septic shock from operational triggers for endotoxemia assessment and potential consideration of PMX-HA. All recommendations were anchored to an established diagnosis of septic shock, acknowledging its biological heterogeneity and multiple mechanistic drivers [1, 19]. The panel endorsed pragmatic hemodynamic criteria centered on persistent vasopressor dependence despite adequate resuscitation to identify patients with sustained shock physiology rather than transient hypotension responsive to early therapy. Sustained vasopressor requirement to maintain mean arterial pressure ≥ 65 mmHg, interpreted alongside norepinephrine equivalent dose thresholds and sedation status, was intended to provide operational clarity. These thresholds were framed as practical implementation guidance rather than pathophysiologic definitions or strict replication of clinical trial enrollment criteria.. The norepinephrine-equivalent threshold of ≥ 0.05 µg/kg/min was selected as a pragmatic operational threshold informed by prior PMX-HA clinical trial experience, including the TIGRIS trial, which enrolled patients with sustained vasopressor-dependent septic shock requiring norepinephrine doses exceeding 0.05 µg/kg/min or equivalent vasopressor support. The panel did not intend this threshold to represent a biologic cutoff but rather a practical criterion identifying clinically meaningful circulatory failure despite adequate resuscitation. Vasopressor requirements exist on a continuum and should always be interpreted within the broader clinical context.

Because lactate is included in the Sepsis-3 definition of septic shock, the panel recognized the importance of clarifying its role within the proposed framework. Once septic shock has been established according to Sepsis-3 criteria, the presence of an elevated lactate level at the time endotoxemia assessment is being considered should not be required to prompt EAA testing or serve as an independent criterion for PMX-HA eligibility. This approach is consistent with EUPHRATES and TIGRIS clinical trials, in which lactate thresholds were not used to guide endotoxemia assessment or determine eligibility for PMX-HA therapy. Clinical decision-making should instead emphasize persistent vasopressor-dependent hypotension and evidence of ongoing organ dysfunction despite adequate resuscitation when considering endotoxin testing and PMX-HA therapy. Importantly, these operational criteria are intended to support timely recognition of sustained shock physiology and should not be interpreted as a recommendation to delay endotoxemia assessment or therapy.

Domain 2 focused on interpretation of the EAA as a contextual decision-support tool rather than a standalone determinant of therapy. The panel emphasized that EAA results should be interpreted alongside clinical trajectory, organ dysfunction, and response to standard therapies. An EAA range of ≥ 0.60 and < 0.90 emerged as a pragmatic window informed by prior trial experience and subgroup analyses suggesting differential treatment response; while avoiding rigid threshold-based algorithms [12, 17]. Repeat testing is reasonable in patients with initial EAA values below 0.60 who fail to improve clinically when PMX-HA remains under consideration, acknowledging the dynamic nature of endotoxemia. Conversely, extremely high EAA values (≥ 0.90) may reflect overwhelming illness or limited reversibility; therefore, prognosis and resource availability should be incorporated into treatment decisions [12]. Collectively, these recommendations reflect a deliberate balance between biomarker-guided precision and clinical judgment.

Domain 3 addressed the operational realities of PMX-HA implementation, recognizing that variability in program structure and execution likely contributes to inconsistent outcomes reported across studies and regions. The panel emphasized that PMX-HA should be integrated into existing sepsis care pathways rather than deployed as an isolated intervention. PMX-HA should be considered an adjunctive therapy and should not delay initiation or optimization of antimicrobial therapy or source control. The panel emphasized that prompt administration of appropriate antimicrobial therapy remains a cornerstone of septic shock management and should not be delayed for endotoxemia assessment or PMX-HA treatment. Available data regarding antimicrobial adsorption by PMX-HA are limited. In vitro studies have demonstrated minimal adsorption of most commonly used antimicrobial agents, although higher adsorption has been reported for linezolid [22]. Clinical pharmacokinetic data remain sparse, and available evidence does not currently support routine modification of antimicrobial dosing solely because of PMX-HA treatment [23]. Clinicians should continue to follow established antimicrobial stewardship principles and therapeutic drug-monitoring practices when appropriate. Involvement of infectious diseases and pharmacy teams may further support antimicrobial stewardship during PMX-HA implementation.

PMX-HA has been delivered as stand-alone hemoperfusion therapy in major randomized trials evaluating the technology, including EUPHRATES and TIGRIS. In this context, stand-alone delivery refers to PMX-HA being performed as a separate hemoperfusion treatment using a dialysis or CRRT platform rather than being incorporated directly into an active CRRT or ECMO circuit. Importantly, stand-alone delivery does not preclude concurrent CRRT or ECMO support when clinically indicated. Rather, it reflects the manner in which the PMX cartridge is connected and operated. In EUPHRATES and TIGRIS, PMX-HA was delivered as a separate extracorporeal treatment, and the manufacturer’s Instructions for Use similarly describe stand-alone delivery. Evidence for integrated configuration where the PMX cartridge is incorporated directly into an active CRRT circuit or ECMO is currently limited to case reports [24, 25]. Comparative studies evaluating standalone versus integrated configurations are lacking, and standardized protocols for integrated delivery have not been established. Accordingly, the panel favored standalone PMX-HA delivery. When PMX-HA and CRRT share the same vascular access, the timing and sequencing of therapies should be individualized according to clinical priorities and local workflow considerations. Systemic heparin has historically been the most commonly used anticoagulation strategy during PMX-HA and remains the approach most frequently described in clinical trials and manufacturer protocols. However, limited case reports and small case series have demonstrated the feasibility of regional citrate anticoagulation during PMX-HA, particularly in patients at high bleeding risk [26].

A multidisciplinary team model was endorsed to facilitate timely decision-making, coordination of care, and adherence to standardized protocols [19]. This approach mirrors successful implementation strategies for other time-sensitive, resource-intensive therapies in critical care and reflects the complexity of PMX-HA delivery, which spans clinical, nursing, technical, and laboratory domains. Standardization of treatment protocols, monitoring practices, documentation, and training was viewed as essential for ensuring safety, reproducibility, and quality improvement while allowing adaptation to local resources and regulatory environments [19].

Several limitations of this consensus should be acknowledged. The recommendations have not undergone prospective validation and are based on expert consensus informed by available evidence on PMX-HA and clinical expertise in the management of septic shock, rather than strict adherence to the specific methodologies or operational criteria employed in randomized trials.

The panel size was modest at ten members, and while participants represented diverse specialties and geographically distributed U.S. institutions, several important limitations in representativeness should be acknowledged. The panel was composed exclusively of clinicians practicing in the United States, which limits global generalizability. PMX-HA has been in clinical use in Japan since 1994 and holds CE marking in Europe, and a substantial body of observational and trial evidence originates from Japanese and European centers. The absence of international representation, particularly from Japan and Europe, where clinical experience with PMX-HA is most extensive, is a meaningful gap that future consensus efforts should address.

Importantly, the present consensus was developed within the context of contemporary North American PMX-HA practice, where endotoxemia assessment using the Endotoxin Activity Assay has been incorporated into enrollment strategies in major clinical trials, including EUPHRATES and TIGRIS clinical trials. This differs from much of the Japanese and European PMX-HA experience, which generally did not employ EAA-guided enrollment. Furthermore, the availability of EAA testing may be limited in some settings, which may restrict implementation of this biomarker-guided approach. The use of PMX-HA in the absence of endotoxin measurement remains uncertain and was not addressed in the current consensus, which was developed as an implementation framework for EAA-guided PMX-HA use.

The panel was assembled by the meeting sponsor, Vantive, rather than by an independent professional society. While the sponsor had no role in the deliberative process or voting outcomes, formal endorsement by organizations such as the Society of Critical Care Medicine or the American Society of Nephrology would further strengthen the authority and broader adoption of these recommendations and represents an important future direction.

The panel identified priorities for future investigation including refinement of patient subphenotyping, validation of endotoxemia-guided treatment strategies, characterization of EAA kinetics, and evaluation of real-world implementation across diverse healthcare systems. Prospective studies addressing these areas will be essential to advancing precision approaches to adjunctive therapy in septic shock.

Conclusion

This expert consensus provides an implementation-focused framework to support the use of PMX-HA in patients with established septic shock and suspected endotoxemia. By clearly distinguishing septic shock diagnosis from operational triggers for endotoxemia assessment, emphasizing contextual interpretation of the EAA, and outlining practical strategies for clinical implementation, these recommendations aim to reduce practice variability and support informed clinical decision-making. The statements are intended to complement existing sepsis guidelines and clinician judgment rather than serve as prescriptive mandates, and should be interpreted within the context of evolving clinical evidence. As additional trial data and real-world experience emerge, this framework may evolve to further refine patient selection and implementation strategies.

Abbreviations

ADQI

Acute Disease Quality Initiative

CRRT

Continuous renal replacement therapy

EAA

Endotoxin activity assay

ESS

Endotoxic septic shock

EUPHRATES

Evaluating the Use of Polymyxin B Hemoperfusion in a Randomized controlled trial of Adults Treated for Endotoxemia and Septic shock

EUPHAS

Early Use of Polymyxin B Hemoperfusion in Abdominal Sepsis

ICU

Intensive care unit

LPS

Lipopolysaccharide

PMX-HA

Polymyxin B hemoadsorption

Author contributions

AT conceptualized and led the study and designed and conducted the consensus process. AT and LY jointly drafted the manuscript and contributed equally to manuscript development, data synthesis, and presentation of consensus findings. AT and LY prepared the figures and tables. All authors participated in the consensus process, including survey completion, structured discussions, and voting on consensus statements. All authors critically reviewed the manuscript, provided important intellectual contributions, and approved the final version for submission.

Funding

The consensus meeting was supported by Vantive, which covered travel and accommodation but had no role in the content, voting, or preparation of this manuscript.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests. Some authors are members of Tigis Study (See: Neyra JA, Legrand M, Tidswell MA, Al-Khafaji A, Galphin C, Rains R, et al. Polymyxin B haemoadsorption in endotoxic septic shock (Tigris): a multicentre, open-label, Bayesian, randomised, controlled, phase 3 trial. The Lancet Respiratory Medicine).