Precision medicine in sepsis: lessons from the TIGRIS trial and the reawakening of targeted therapy Authors: Toshiaki Iba, Julie Helms, Hideshi Okada, Kunihiko Nagakari, Ricard Ferrer Published: 2026 ABSTRACT AbstractSepsis remains a leading cause of mortality worldwide, in part because it represents a biologically heterogeneous syndrome rather than a single disease entity. Numerous clinical trials targeting inflammation, coagulation, or immune pathways have failed to improve outcomes, largely due to indiscriminate enrollment of biologically diverse patient populations. Precision medicine has therefore emerged as a necessary paradigm shift in sepsis research and care. Endotoxin is a key driver of sepsis pathobiology through Toll-like receptor 4-mediated inflammatory signaling, endothelial dysfunction, and thromboinflammation. The development of the endotoxin activity assay (EAA) enabled real-time functional assessment of circulating endotoxin activity, revealing marked interindividual variability and dynamic changes during illness. These insights provided a rationale for biologically guided endotoxin-targeted therapy. Polymyxin B hemoperfusion, despite its strong mechanistic plausibility and early clinical promise, yielded inconsistent results in trials that relied on syndromic enrollment criteria. The TIGRIS trial was designed to address this limitation by refining patient selection through the integration of EAA-defined endotoxemia and organ dysfunction severity, assessed by the Sequential Organ Failure Assessment score. By defining a therapeutic window in which endotoxin was both biologically relevant and potentially modifiable, TIGRIS provides preliminary evidence suggesting that polymyxin B hemoperfusion may be effective when applied in an appropriate biological context. Beyond endotoxin removal, TIGRIS may offer insights into broader implications for precision medicine in sepsis, including biomarker-guided trial design, rational combination therapy, and reinterpretation of prior “negative” studies. Embracing biological heterogeneity may offer a credible path toward effective, individualized sepsis treatment.Keywords: Lipopolysaccharide, Hemoperfusion, Personalized medicine, Biomarker, Randomized controlled trial Introduction: why precision medicine is no longer optional in sepsis Sepsis continues to impose an unacceptably high burden of mortality and long-term morbidity worldwide, despite substantial advances in antimicrobial therapy, hemodynamic resuscitation, and organ support technologies [1]. Contemporary consensus definitions emphasize that sepsis is defined by infection-triggered, life-threatening organ dysfunction arising from a dysregulated host response, underscoring its biological complexity rather than a single causal pathway [2]. Large epidemiological analyses and mechanistic reviews confirm that patients who meet identical Sepsis-3 or septic-shock criteria may differ profoundly across multiple pathological axes, including inflammatory activation, endothelial injury, coagulation derangement, immune suppression or exhaustion, and metabolic failure, suggesting that sepsis is not a uniform entity [3–5]. Clinical heterogeneity extends to thromboinflammatory changes. Endotoxin bioactivity, tissue factor expression, glycocalyx degradation, and NETosis exhibit marked inter-individual variability and evolve dynamically, often escaping static scoring tools that were not designed to detect compensated, early-phase coagulopathy [6–9]. These observations are consistent with the interpretation that many neutral sepsis randomized controlled trials (RCTs) may reflect challenges in biological alignment and identification of a modifiable therapeutic window, rather than absence of mechanistic relevance [5, 6]. From a methodological perspective, precision medicine has been proposed to require biological enrichment, severity windowing, and longitudinal biomarker assessment to identify responsive, treatable subsets while protecting patients from stage-inappropriate harm [4, 10, 11]. Within this context, the TIGRIS randomized controlled trial examined the effect of polymyxin B-immobilized column hemoperfusion (PMX-HP). This trial has been interpreted as a potentially important step toward operationalizing biomarker-guided enrichment and defining a modifiable endotoxin–severity window. The findings suggest that extracorporeal endotoxin removal may confer clinical benefit when applied to an appropriate biological phenotype at a relevant stage of disease, although this interpretation should be considered in light of the current limitations of the available data [10]. The legacy of failed sepsis trials: a problem of mismatch rather than mechanism First of all, it should be acknowledged that the TIGRIS trial results are primarily derived from ClinicalTrials.gov listings and sponsor communications, and full peer-reviewed publication of the primary data remains limited. The history of sepsis therapeutics is marked by repeated failures of interventions supported by strong biological rationale. Activated protein C initially demonstrated a survival benefit in patients with severe sepsis, suggesting that modulation of inflammation and coagulation could improve outcomes [12]. However, subsequent studies failed to reproduce this benefit and revealed an increased risk of serious bleeding, ultimately leading to withdrawal of the drug from clinical use [13]. Similar trajectories have been observed with antithrombin, recombinant tissue factor pathway inhibitor, and various cytokine-targeted strategies [14, 15]. These failures are often interpreted as evidence that sepsis biology is too complex to be therapeutically tractable. Yet critical analyses suggest that the central problem lies not in the mechanisms themselves, but in their indiscriminate application across biologically heterogeneous populations [6, 16, 17]. Therapies targeting coagulation or inflammation are unlikely to benefit patients in whom these pathways are not pathologically activated and may cause harm when applied outside their biologically appropriate context [18, 19]. Precision medicine reframes these failures as problems of mismatch between therapy and biology, rather than definitive refutations of mechanistic validity. Endotoxin as a driver of sepsis pathobiology Endotoxin, such as lipopolysaccharide (LPS), has long been recognized as a potent trigger of the host response in Gram-negative sepsis. Through the activation of Toll-like receptor 4 and its downstream signaling pathways, endotoxin induces the robust release of proinflammatory cytokines, promotes endothelial activation, and drives coagulation through the upregulation of tissue factor expression on monocytes and endothelial cells [20, 21]. This coordinated inflammatory and procoagulant response places endotoxin at the intersection of immune activation, endothelial dysfunction, and thromboinflammation [22]. Experimental models have demonstrated that endotoxin exposure alone is sufficient to reproduce many cardinal features of septic shock, including vasoplegia, increased vascular permeability, and microvascular thrombosis, even in the absence of live bacterial infection [23, 24]. These findings underscore the causal role of endotoxin as a driver of systemic inflammation and organ dysfunction, rather than merely a marker of infection (Fig. 1). Conceptual framework of endotoxin-induced thromboinflammation. Endotoxin (lipopolysaccharide; LPS) activates Toll-like receptor 4 (TLR4) on monocytes, inducing tissue factor (TF) expression and proinflammatory cytokine release. TF initiates the coagulation cascade, leading to thrombin generation and fibrin formation. Endotoxin also promotes endothelial injury and glycocalyx degradation, increasing vascular permeability and releasing von Willebrand factor (vWF), which enhances platelet adhesion and immunothrombus formation. Neutrophil activation results in the formation of neutrophil extracellular traps (NETs), further amplifying coagulation and microvascular thrombosis. Concurrent release of cytokines and damage-associated molecular patterns (DAMPs) exacerbates endothelial activation, vasoplegia, and capillary leak. Impaired hepatic clearance and gut translocation further contribute to the circulating endotoxin burden Clinically, elevated circulating endotoxin activity has been consistently associated with greater disease severity, increased organ failure, and higher mortality in critically ill patients [25, 26]. Importantly, endotoxemia is not restricted to patients with documented Gram-negative bacteremia. Translocation from the gastrointestinal tract, impaired hepatic clearance, and release from lysed bacteria during antimicrobial therapy all contribute to circulating endotoxin burden [25, 27]. These observations emphasize that endotoxemia reflects a dynamic host–pathogen interaction rather than a simple microbiological classification, providing a strong biological rationale for targeting endotoxin in carefully selected patients with sepsis [21]. Together, these observations highlight the need for a rapid and biologically meaningful method to identify patients in whom endotoxemia is a dominant and potentially modifiable driver of disease, thereby setting the stage for the clinical measurement of endotoxin activity. Measuring endotoxemia: from concept to clinical tool For many years, the clinical application of endotoxin-targeted therapies was limited by the lack of a rapid and reliable method to assess circulating endotoxin. The most widely used conventional approach is the Limulus amebocyte lysate (LAL) assay, which measures endotoxin concentration based on activation of the horseshoe crab coagulation cascade [28]. Although highly sensitive under controlled laboratory conditions, LAL assays are susceptible to interference from plasma proteins, lipids, β-glucans, and antimicrobial agents, and they provide limited information regarding biological activity in vivo [29]. As a result, LAL-derived endotoxin levels often correlate poorly with disease severity or clinical outcomes in critically ill patients [30]. Chromogenic and turbidimetric LAL variants enhance quantification but remain limited by similar preanalytical and interpretive challenges [31]. More recently, mass spectrometry–based lipid A detection and recombinant factor C assays have been explored to improve specificity and reduce cross-reactivity [32]. While these methods offer technical advantages and eliminate reliance on animal-derived reagents, they require specialized equipment, extensive sample preparation, and are not readily adaptable to rapid bedside decision-making. Indirect approaches, such as measurement of endotoxin-induced host response markers (e.g., lipopolysaccharide-binding protein or soluble CD14), reflect exposure rather than circulating endotoxin bioactivity [33]. Collectively, these limitations highlight why alternative assays have remained primarily research tools and underscore the clinical appeal of functional, whole-blood approaches for precision-guided endotoxin-targeted therapy [34]. The development of the endotoxin activity assay (EAA) represented a major advance. Rather than measuring endotoxin concentration directly, EAA provides a functional assessment of endotoxin bioactivity in whole blood by quantifying the priming of neutrophils for respiratory burst induced by endotoxin–antiendotoxin immune complexes [23, 35]. By integrating both endotoxin burden and host responsiveness, EAA offers a biologically meaningful readout of circulating endotoxin activity [35]. Clinical studies have shown that elevated EAA levels correlate with greater disease severity, organ dysfunction, and adverse outcomes in sepsis [25]. Importantly, EAA measurements reveal marked interindividual heterogeneity: some patients exhibit minimal endotoxin activity, whereas others demonstrate moderate or extreme elevations, independent of microbiological classification. Moreover, endotoxin activity is dynamic, changing over time in response to source control, antimicrobial therapy, and host factors [25]. These features make EAA a practical tool for biological stratification, enabling identification of patients in whom endotoxin is a relevant and potentially modifiable therapeutic target. Polymyxin B hemoperfusion: from early promise to apparent failure PMX-HP was developed to exploit the strong binding affinity of polymyxin B for the lipid A moiety of endotoxin, enabling selective extracorporeal removal of circulating lipopolysaccharide while avoiding the nephrotoxicity and neurotoxicity associated with systemic polymyxin administration [36, 37]. By attenuating a proximal trigger of inflammation, endothelial activation, and coagulation, this strategy was conceptually positioned to interrupt the thromboinflammatory cascade central to septic shock pathophysiology [38]. Early clinical studies—most notably in patients with abdominal sepsis—reported improvements in mean arterial pressure, reduced vasopressor requirements, and enhanced oxygenation, supporting the biological plausibility of endotoxin removal as an adjunctive therapy [39]. The randomized EUPHAS trial further fueled optimism by demonstrating improved hemodynamics and a survival signal in patients with abdominal septic shock treated early with PMX-HP [39]. However, subsequent larger trials failed to reproduce these benefits. Studies such as ABDOMIX and later EUPHRATES enrolled patients based primarily on clinical criteria, including septic shock or presumed infection, without requiring evidence of endotoxemia, and reported neutral effects on mortality [7, 39]. These results led to the prevailing perception that PMX-HP was ineffective, prompting skepticism and a decline in its clinical uptake [40]. Upon closer examination, it has become apparent that fundamental biological mismatches hindered these trials. Endotoxin levels were neither measured nor used to guide enrollment, resulting in treatment of patients with minimal or absent endotoxemia alongside those with overwhelming exposure. Moreover, patients with extreme disease severity and advanced organ failure, unlikely to benefit from endotoxin removal alone, were often included. Together, these factors diluted potential efficacy signals and obscured subgroup-specific benefits. This realization has reframed PMX-HP not as a failed therapy, but as an intervention prematurely judged without adequate biological stratification, directly motivating the precision-guided approach later adopted in TIGRIS [41] (Table 1). Table 1PMX-HP trials and enrollment strategyTrialPopulationEnrollment strategyEndpointResultInterpretationEUPHASAbdominal septic shockClinical selection only28-day mortalityMortality reduction observedSignal in clinically selected population; limited biological enrichmentABDOMIXSeptic shockNo biomarker selection28-day mortalityNo significant mortality reductionAbsence of enrichment likely diluted effectEUPHRATESSeptic shock + high EAABiomarkerbased but broad severity28-day mortalityNo significant mortality reductionBiomarker selection insufficient without optimal windowTIGRISSeptic shock with EAA 0.60–0.89Biomarker + severity enrichment28- and 90-day mortalityMet prespecified endpoint (posterior probability > 95%)Integrated enrichment defining a modifiable therapeutic windowPMX-HP polymyxin B-immobilized fiber column hemoperfusion, EAA endotoxin activity assay PMX-HP trials and enrollment strategy PMX-HP polymyxin B-immobilized fiber column hemoperfusion, EAA endotoxin activity assay TIGRIS: redefining success through biological enrichment The TIGRIS trial may be interpreted as a methodological advance, although its evidentiary basis remains preliminary in endotoxin-targeted therapy by explicitly integrating both biological burden and organ dysfunction severity into patient selection (diagnosis of septic shock with clinical signs consistent with sepsis-associated circulatory and organ dysfunction with EAA level of 0.6–0.89) [41]. Rather than enrolling patients solely on syndromic criteria such as septic shock, TIGRIS prospectively targeted patients with moderate elevations of endotoxin activity, as measured by the EAA (0.6–0.89), while also requiring evidence of significant but potentially reversible organ dysfunction (Multiple Organ Dysfunction Score [MODS] > 9 and/or Sequential Organ Failure Assessment [SOFA] score > 11) [23, 25, 41, 42]. This dual enrichment strategy intended to address a key limitation of earlier trials, in which neither endotoxemia nor disease severity was biologically aligned with the therapeutic mechanism. Importantly, TIGRIS was designed on the premise that both insufficient and excessive endotoxin burden limit therapeutic efficacy. Patients with low EAA values were excluded because endotoxin was unlikely to be a dominant driver of pathology, whereas those with extremely high EAA levels (> 0.9) were also excluded, as such elevations likely reflect overwhelming infection, profound immune dysregulation, or advanced, irreversible organ failure. Similarly, very low SOFA (or MODS) scores suggest limited disease severity with minimal opportunity for benefit, while extremely high SOFA scores may indicate late-stage disease beyond modification by extracorporeal endotoxin removal. By defining a therapeutic window based on EAA and organ failure score, TIGRIS is conceptually aligned with the operationalization of a core principle of precision medicine: matching intervention to a modifiable biological state rather than a diagnostic label [41, 43]. The clinical signal observed in TIGRIS has been interpreted as suggesting that PMX-HP may be effective when applied within this biologically and clinically appropriate window; however, these findings should be interpreted with caution, particularly given that the currently available evidence is largely derived from registry entries and sponsor-reported summaries without full peer-reviewed validation [42]. More broadly, the trial is consistent with the concept that biomarker- and severity-guided enrichment may represent a useful strategy for advancing sepsis therapeutics. In the absence of fully peer-reviewed primary trial data, the magnitude, consistency, and robustness of the reported treatment effect cannot be independently verified, and therefore should be considered hypothesis-generating rather than confirmatory. Current publicly available information is largely based on registry entries (e.g., ClinicalTrials.gov) and sponsor-reported summaries, which may lack the granularity, methodological transparency, and independent validation provided by peer-reviewed publications. Such sources are inherently limited in their ability to fully assess risk of bias, subgroup consistency, and robustness of effect estimates. Broader implications for precision medicine in sepsis The implications of TIGRIS extend well beyond endotoxin removal. Sepsis-induced coagulopathy (SIC) and disseminated intravascular coagulation (DIC) represent parallel examples in which therapies targeting coagulation may be beneficial only in selected patients. DIC is characterized by systemic activation of the coagulation system, microvascular thrombosis, and consumption of coagulation factors [44]. SIC represents an earlier, compensated stage of this process, with prognostic and therapeutic relevance [45]. The pathophysiology of SIC/DIC significantly overlaps with that of the optimal target of PMX-HP, and a similar therapeutic approach is expected [45–47]. Clinical and experimental data suggest that anticoagulant or endothelial-protective therapies may benefit patients with demonstrable coagulopathy and low bleeding risk, while offering little value—or causing harm—in others [45, 48]. Precision medicine, therefore, requires dynamic assessment of coagulation biomarkers rather than reliance on static clinical labels [45]. Precision medicine does not mean monotherapy A common misconception is that precision medicine seeks to identify a single optimal therapy for each patient. In reality, sepsis is a multidimensional syndrome characterized by the simultaneous activation of multiple pathological axes, including dysregulated inflammation, coagulation activation, endothelial injury, metabolic dysfunction, and immune imbalance [49]. Large-scale clinical and biomarker-driven analyses have consistently demonstrated the existence of distinct sepsis phenotypes with divergent biological signatures, trajectories of organ failure, and clinical outcomes, underscoring the inadequacy of one-size-fits-all therapeutic strategies [5, 50]. Immune dysfunction in sepsis further exemplifies this complexity. While some patients manifest early hyperinflammatory responses with excessive cytokine release and immunothrombosis, others rapidly transition to—or concurrently exhibit—profound immunosuppression, characterized by lymphocyte apoptosis, impaired antigen presentation, and defective pathogen clearance [4, 51, 52]. These opposing immune states may coexist within the same patient over time, highlighting the dynamic nature of sepsis biology. Within this context, precision medicine should be understood not as therapeutic minimalism, but as rational integration of combination therapy. Targeted interventions, such as endotoxin removal, endothelial protection, anticoagulant modulation, or immune-adaptive therapies, may need to be applied concurrently or sequentially, guided by biomarkers, severity indices, and temporal disease phase [3, 27]. Such an approach aligns with emerging systems-biology and adaptive trial frameworks, offering a pragmatic path toward effective, individualized sepsis care. Reinterpreting “negative” trials through a precision lens The success of TIGRIS invites a rigorous re-evaluation of prior “negative” sepsis trials, many of which were judged neutral because biologically heterogeneous patients were enrolled using syndrome-based inclusion criteria rather than biological enrichment [6, 7, 39]. Accumulating evidence now indicates that trial failure was commonly caused by mechanism–population mismatch, where therapies modulating inflammation or coagulation were administered to patients in whom those pathways were not pathologically activated or were already non-modifiable. This reframing aligns with systems biology critiques that emphasize the fallacy of sepsis uniformity [5, 6, 50]. For example, the mortality benefit signal observed with activated protein C in PROWESS [12] could not be reproduced in later, unselected shock-dominant cohorts, raising major bleeding concerns [13, 14], underscoring that background severity and co-interventions strongly influence the observed risk–benefit profile [13, 14]. High-dose antithrombin likewise appeared neutral in broad enrollment, yet biologically enriched analyses revealed subgroup-specific benefit when administered without heparin and clearer safety in DIC-positive phenotypes [11, 53], demonstrating that context determines both efficacy and harm [54]. Recombinant TFPI inhibition in OPTIMIST [15] and recombinant human thrombomodulin (rhTM) in SCARLET [9] also reported neutral primary outcomes in unselected cohorts [9, 15]. However, post hoc stratification in French cohorts suggested benefit signals in coagulopathy-dominant, low-bleeding- risk subsets [55] (Table 2). Table 2Latent efficacy signals from “negative” trialsTherapyPathwayPopulationResultInterpretationActivated protein C (PROWESS) [12]AnticoagulantSevere sepsisMortality reduction observedEffect depends on phenotype and disease stageActivated protein C (PROWESSSHOCK) [13]AnticoagulantSeptic shockNo significant mortality reductionHeterogeneity reduces signalAntithrombin (KyberSept) [14]AnticoagulantSevere sepsisNo overall benefit; increased bleeding with heparinContext and cointerventions influence outcomeTFPI (tifacogin RCT) [15]Coagulation inhibitionSevere sepsisNo significant mortality reductionSimple laboratory criteria insufficientRecombinant thrombomodulin (SCARLET) [9]AnticoagulantCoagulopathyNo significant mortality reductionBroader phenotype too heterogeneousTFPI tissue factor pathway inhibitor, RCT randomized controlled trial Latent efficacy signals from “negative” trials TFPI tissue factor pathway inhibitor, RCT randomized controlled trial More recently, machine-learning-based sub-phenotyping and longitudinal biomarker trajectories (platelets, prothrombin time/international normalized ratio, fibrinogen, D-dimer, antithrombin, endothelial markers) demonstrated that treatment responsiveness is confined to early, modifiable thrombo-inflammatory or immunosuppressive transitions, not to static clinical labels [5, 56, 57]. These insights provide a roadmap for future trials: biological enrichment + severity window + dynamic reassessment, not static syndrome-based enrollment [5–8, 11, 53, 55–57]. Challenges and future directions Despite its promise, the implementation of precision medicine in sepsis faces substantial practical and conceptual challenges. A primary barrier is timely access to reliable diagnostics, as many biologically informative biomarkers, including endotoxin activity, immune phenotypes, and endothelial injury markers, are not yet universally available or integrated into routine critical care workflows. Even when assays exist, their clinical utility depends on rapid turnaround times, standardized interpretation, and clear therapeutic actionability [58–61]. Integration into established clinical pathways presents an additional challenge. Current international guidelines emphasize early recognition, prompt antimicrobial therapy, and standardized care bundles, approaches that have improved outcomes at a population level but are inherently syndrome-based [40]. Incorporating biologically guided interventions will require the evolution of guidelines toward adaptive frameworks that preserve early standardized care while allowing subsequent stratification based on dynamic biomarker profiles. This transition will also necessitate focused clinician education to ensure appropriate interpretation of biomarker data and avoidance of overcomplexity at the bedside. From a research perspective, future progress will depend on biologically enriched trial designs, incorporation of adaptive or platform methodologies, and acceptance of narrower inclusion criteria to reduce heterogeneity [62–64]. Advances in systems biology, machine learning, and longitudinal phenotyping are expected to further refine patient stratification and therapeutic timing [65]. Ultimately, successful translation of precision medicine in sepsis will require coordinated advances in diagnostics, trial methodology, and implementation science to bridge the gap between biological insight and real-world clinical practice [66, 67]. Despite its conceptual appeal, the implementation of EAA-guided precision strategies raises several practical considerations. First, the availability of the EAA remains variable across institutions, and access may be limited in resource-constrained settings. Although EAA provides a functional, whole-blood assessment of endotoxin bioactivity, questions remain regarding assay reproducibility and standardization, particularly when applied across different laboratories and clinical environments [35]. Inter-center variability in sample handling, timing of measurement, and interpretation thresholds may further influence consistency of patient classification and treatment decisions. Second, the cost-effectiveness of EAA-guided strategies requires careful evaluation. While biologically enriched approaches may improve therapeutic targeting and reduce exposure to ineffective treatments, the added costs of biomarker testing and extracorporeal therapies must be balanced against uncertain outcome benefits in the absence of definitive large-scale evidence. Formal health-economic analyses will therefore be essential to determine the value proposition of such approaches. Finally, implementation feasibility represents a key challenge. Integration of EAA into routine clinical workflows will require rapid turnaround times, clinician familiarity with assay interpretation, and alignment with existing sepsis management protocols [68]. Educational efforts, standardization of testing algorithms, and incorporation into adaptive clinical pathways will be necessary to translate biomarker-guided strategies from trial settings into real-world practice. However, the translation of these approaches into routine practice will depend on considerations of diagnostic accessibility, cost-effectiveness, and health-system integration. Conclusion The TIGRIS trial may be viewed as a potentially informative step in sepsis therapeutics, suggesting that biologically guided therapy can succeed where syndrome-based approaches have failed. By aligning endotoxin removal with measurable endotoxemia, TIGRIS provides preliminary observations that are consistent with the feasibility of precision medicine in critical care. However, these interpretations should be considered in the context of currently available data, which remain largely dependent on registry entries and sponsor-reported summaries and therefore require confirmation through full peer-reviewed analyses. As the field advances, embracing biological heterogeneity rather than ignoring it may present a credible path toward improving outcomes in sepsis, particularly when combined with rigorous biomarker validation, careful patient selection, and adaptive trial design. At present, the available evidence should be interpreted as supportive of a precision-medicine framework rather than definitive proof of therapeutic efficacy. Acknowledgements None. Abbreviations Endotoxin activity assay Polymyxin B-immobilized column hemoperfusion Lipopolysaccharide Limulus amebocyte lysate Multiple organ dysfunction syndrome Sequential organ failure assessment Sepsis-induced coagulopathy Disseminated intravascular coagulation Recombinant human thrombomodulin Authors’ contributions All authors met authorship criteria and participated significantly in the study. T. Iba and J. Helms wrote the draft. H. Okada, K. Nagakari, and R. Ferrer critically reviewed and revised the manuscript. Funding None. Data availability No datasets were generated or analysed during the current study. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.