4.1 Research Trends in Sepsis
Based on bibliometric insights derived from the VOSviewer software analysis, there has been an increase in scientific contributions and researchers dedicated to the field of sepsis, with a 1.96-fold increase in publication volume compared with the preceding decade. The United States, China, Germany, and the United Kingdom emerged as the top four contributors to research on sepsis. The predominant role of the United States is evident from its substantial number of publications. American scholars spearheaded the initial conceptualisation of Sepsis 1.0, so the U.S. exhibited early engagement with this issue compared with its global counterparts. The infrastructure for foundational medical research and clinical trials appears to be more robust in the U.S., reinforcing its leadership.
Notably, China ranks third to second, increasing by 7.53 times, much higher than the average growth in published amount. However, the quality of the published articles remains insufficient. The top ten highly cited references and articles have only one Chinese author as the leading researcher on COVID-19. Potential reasons for the contradiction between the quantity and quality of publications in China are as follows: First, the Critical Care Medicine Branch of the Chinese Medical Association was established in 2005, and domestic guidelines for the initial treatment of sepsis rely on international recommendations. Second, the imbalance between economic and scientific research levels in developed and developing countries has led to relatively few high-level studies on evidence-based medicine in the early stages. With the gradual deepening of Chinese studies and the continuous expansion of funding, high-quality research can be upgraded to the same level as similar global research [17, 18].
The journals with the most publications on sepsis research were mainly in Critical Care Medicine. Critical Care Medicine published 1,907 sepsis-related papers, followed by other journals, including PLOS One, Shock, and Critical Care. The American Journal of Respiratory and Critical Care Medicine stands out in its leading position regarding average citation counts. Thus, considering the quantity and quality, these journals could be the primary platforms for future publications in this field.
4.2 Quality and Status of Global Publications in Sepsis
The high-level institutional presence in the field of sepsis has been updated in 2013–2022 compared with 2003–2012. Ten years ago, most research in this field originated from European and American institutions; however, some Chinese institutions have risen to the top 20, including Shanghai Jiao Tong University, Zhejiang University, Capital Medical University, Cent South University, Chang Gung University, and Fudan University. For the past two decades, Harvard University, Washington University, and Pittsburgh University have been ranked in the top three for sepsis cases by volume.
Through co-authorship analysis, we found that Harvard University, Washington University, and Pittsburgh University collaborated with nearly 300 high-yielding institutions. This finding suggests that a close collaboration between institutions can help generate valuable research. Shanghai Jiao Tong University in China, along with Tongji University, Wuhan University, and Zhejiang University, have collaborated on sepsis, illustrating how multidisciplinary and multicentre cooperation can effectively promote research development.
4.3 Research Focus and Hot Topics on Sepsis
Using keyword analysis, we outlined the Research Focus of Sepsis over the past two decades.
4.3.1 Etiology and Treatment of Sepsis
The first version of the SSC guidelines was published in 2004. Since then, updates have been released in a cycle of approximately four years (2008, 2012, 2016, and 2021) [19]. The choice and timing of antibiotics are integral components of the comprehensive management [20]. The 2021 SSC guidelines recommend antibiotic treatment, preferably within 1 h, for patients with septic shock. A 2023 retrospective study examining 104,248 patients revealed that every hour delay in administering antibiotics correlated with a heightened risk of mortality from septic shock. However, for non-shock patients, delays > 6 h correlate with higher mortality [21]. Inappropriate antibiotic therapy in patients with septic shock is associated with a significant increase in mortality. The patient’s medical history, previous antibiotic use, current infection site, comorbidities, and underlying immunocompromise must be considered when selecting an appropriate empiric antibiotic. A large meta-analysis of 13 randomised controlled trials (RCTs) showed no difference between single and combination antibiotic therapies in terms of mortality in patients with severe sepsis [22]. In a retrospective analysis (approximately 200 patients), empirical polymyxin use was not associated with lower mortality in patients with carbapenem-resistant gram-negative infections [23].
4.3.2 Pathophysiology of Sepsis
The pathophysiology of sepsis is complex and highly heterogeneous, with a mixture of hyperinflammation and immunosuppression [24]. The release of invading pathogens, including bacterial endotoxins and damage-associated molecular patterns (DAMPs), not only causes a sustained immune stimulus, but DAMPs activate cell membrane and intracellular pattern-recognition receptors (PRRs), which are often adept at identifying pathogen-associated molecular patterns (PAMPs) [25, 26]. PRRs activate inflammatory cytokine production through TLR/MyD88/NF-kB and TLR/Trif/IRF3 pathways, referred to as ‘cytokine storms’ in early clinical studies, including tumour necrosis factor (TNF) α, IL-1β, IL-6, IL-8, and anti-inflammatory cytokines, inducing excessive inflammatory and anti-inflammatory responses[27]. These responses include leukocyte chemotaxis to the sites of infection/inflammation, vascular endothelial damage due to capillary leakage, and complement and coagulation activation. Prolonged stimulation due to sepsis can also lead to immunosuppression. The mechanisms include the increased release of anti-inflammatory cytokines, including IL-4, IL-10, and IL-37; apoptosis of T, B, and dendritic cells; and proliferation of immune cells with anti-inflammatory properties, such as regulatory T cells and myeloid-derived suppressor cells. Further enhancement of immunosuppression is the downregulation of human leukocyte antigen D-related (HLA-DR) antigen expression coupled with increased expression of programmed cell death 1 (PD-1) and its associated ligand PD-L1.
Based on the pathophysiology of septic immunity, several clinical trials have attempted to alter the SIRS by selectively or non-selectively targeting PRRs and their downstream inflammatory factors. However, the therapeutic efficacies of most targets remain controversial in clinical trials. Some studies have shown that blocking pro-inflammatory cytokines does not ultimately benefit survival [28, 29]. However, IL-6 inhibitors have been associated with lower 28-day all-cause mortality in COVID-19 patients [30]. In a Mendelian randomised analysis of the UK Biobank cohort, IL-6 receptor blockade reduced the risk of mortality in patients with non-COVID-19 sepsis. Recent studies have identified several novel immunomodulatory targets [31]. Signalling lymphocyte-activating molecule family-7 (SLAMF7) is an immunoglobulin-like receptor that is significantly elevated in monocytes/macrophages of patients and mouse models of sepsis. In animal models, SLAMF7 reduced inflammation-induced organ damage by downregulating proinflammatory cytokine secretion [32]. Cldio et al. proposed that sequential extracorporeal therapy may be a survival modality for removing pathogens, endotoxins, and cytokines [33]. However, host responses to infections inherently exhibit temporal and spatial heterogeneity. Most cytokine interactions are non-linear and lead to unpredictable outcomes [34]. It is challenging to define sepsis as a specific disease, let alone complete its treatment using a single method.
With the pathophysiological mechanisms of immunosuppression in sepsis, several preclinical and clinical studies have demonstrated the promising effects of immune enhancement against sepsis. One approach to immune enhancement is using immunostimulatory factors acting on the organism, such as Interferon-gamma (IFNγ). IFNγ therapy boosts HLA-DR levels in monocytes for those suffering from sepsis triggered by endotoxin, potentially activating immune cells like macrophages, NK cells, and neutrophils, enhancing the immune defence against pathogens [35]. However, elevated levels of IFNγ during the initial stages of sepsis might provoke secondary Candida infections, indicating patient stratification needs [36]. Treatment with Granulocyte-macrophage colony-stimulating factor (GM-CSF) enhanced the production of neutrophils, monocytes, and macrophages [37]. Although its administration to patients with sepsis can restore HLA-DR expression in monocytes, a meta-analysis showed that infection control did not necessarily decrease mortality [38]. A recent multi-RCT found no significant effect of HLA DR-oriented GM-CSF therapy on ICU-acquired infections or 28-day mortality in patients with immunosuppressed sepsis [39]. Targeting immunosuppressive loci with antibodies, such as PD-1/PD-L1, which affect sepsis-induced organ dysfunction, is another potential approach [40, 41]. Nabumab, a PD-1 antagonist, has been shown to increase lymphocyte count and monocyte HLA-DR expression [42–44]. Nevertheless, further research is needed to understand the safety, efficacy, and long-term immune impact of different dosages.
4.3.3 Clinical Biomarkers of Sepsis
To date, > 250 sepsis-related biomarkers have been identified and evaluated [45]. We mainly discuss the commonly used clinical infection markers, PCT, the inflammatory marker CRP, and scd14.
PCT, a product of endotoxins or mediators in bacterial infections, is correlated with the severity and extent of infection. Values below 0.1 ng/ml are compassionate for excluding bacteraemia in patients with community-acquired pneumonia (CAP) [46]. However, PCT levels are affected by renal replacement therapy and the glomerular filtration rate. Several studies have explored its use as a benchmark for initiating antibiotic therapy in sepsis; however, outcomes vary owing to patient heterogeneity [47, 48]. PCT-guided treatment may decrease antibiotic use without worsening morbidity and mortality in sepsis. Research, including multiple RCTs, indicates that PCT-guided management in lower respiratory tract infections lessens antibiotic exposure and adverse outcomes [49, 50]. Additionally, an RCT by Kyriazopoulu et al. involving 266 patients with sepsis suggested that PCT-guided therapy may lower the incidence of long-term infection-related adverse events such as secondary de novo infections and multidrug-resistant organism infections [51]. CRP, an acute-phase protein synthesised by the liver and triggered by IL-6, is unaffected by renal failure and exhibits similarities with and without cirrhosis [52]. Rather than single values, dynamic changes in CRP levels indicate sepsis, aiding in the diagnosis and evaluation of antibiotic response [53, 54]. CRP ratio usage delineates four response patterns to antibiotic treatment: 1) rapid response (CRP ratio < 0.4 by day 4), 2) slow response (day 4 CRP ratio > 0.4 but < 0.8), 3) nonresponse (CRP ratio continuously > 0.8), and 4) biphasic response (initial decrease in CRP ratio < 0.8, followed by a secondary rise). In severe CAP, lower mortality rates are observed in patients with rapid or slow response patterns than in those with non-response or biphasic patterns [55].
Some biomarkers have been compared with PCT and CRP, and a few have been shown to have superior diagnostic value, such as Presepsin and CD64 [56, 57]. Before using these biomarkers, clinicians must determine whether an infection has occurred, the location of the infection, or the possible pathogenic microorganisms that cause the infection. It is important to recognise that a single biomarker cannot accurately diagnose sepsis or determine its severity.
4.3.4 Coagulation Disorders of Sepsis
4.3.4.1 Sepsis‑induced Coagulopathy
Sepsis-induced coagulopathy manifests as hypercoagulability and antifibrinolytic reactions [58]. First, hyper-inflammatory cytokines cause vascular endothelial cells (ECs) and monocytes to express tissue factors, initiate coagulation, and producing thrombin [59]. Direct activation of the contact phase by bacterial polyphosphates and inhibition of reactive fibrinolysis amplifies this response, leading to the overproduction of thrombin and defective fibrin degradation [60]. Concomitantly, increased production and activation of fibrinolytic inhibitors (e.g. PAI-1 and TAFI) and a decrease in tPA further hindered fibrinolysis. Additionally, natural anticoagulant factors (such as protein C, protein S, thrombomodulin, and antithrombin) and TFPI have been depleted. [61] In sepsis, immune cells interact with blood clotting to activate, leading to ‘immune thrombosis’. Initially considered as a beneficial host defence mechanism, immune thrombosis plays a different role in sepsis. Uncontrolled immune thrombosis in sepsis can result in increased coagulation activation, culminating in DIC and severe organ function failure [62]. In the septic immune response, activated neutrophils release procoagulant microparticles (MVs) and reticular DNA (NETs), further influencing immune thrombosis [63]. Additionally, the colocalisation of MVs on NETs provides a surface that enhances the potential procoagulant activity of NETs by activating the contact phase (Factor XII).
4.3.4.2 Coagulopathy in Sepsis and COVID-19—A double-edged sword cuts both ways
Significant elevations in D-dimer levels are correlated with an increased risk of COVID-19, and extensive microvascular thrombosis in the alveoli observed at autopsy revealed a prognostic effect of SARS-CoV-2-induced hypercoagulability [64–66]. Bacteria-induced Sepsis and COVID-19 show many similarities in the molecular mechanisms of inflammation-coagulation coupling. In the PAMPs, the SARS-CoV-2 spike protein (S) has been identified as a ligand for TLR4 in monocytes and macrophages, activating the NF-κB [67]. SARS-CoV-2 RNA activates the TLR3 and TLR7 signalling pathways [68, 69]. The NF-κB pathway can directly regulate the levels of FIII or TF, FVIII, TFPI, u-PA, PAI-1, antithrombin, and thromboregulatory protein [70]. In the DAMPs, COVID-19 infection promotes the release of HMGB1, participating ARDS [71]. HMGB1 activates macrophages and ECs by binding to several PRRs [72]. HMGB1 also modulates fibrinolysis through plasminogen and t-PA interactions, and simultaneously promotes coagulation through exposure to TF and inhibition of protein C [73, 74]. Similarly, in the DAMPs, histones, due of their cytotoxic, pro-inflammatory, and pro-thrombotic properties, have emerged as prognostic significance in COVID-19 and sepsis [75]. As effectors of the immune response, NETs can cause SARS-CoV-2-induced microvascular thrombosis, with enzymes and histones released from NETs promoting endothelial cell death and dysfunction [76–78]. Eslamifar et al. pointed out that monocytes and macrophages can directly release the activated coagulation TF cascade during COVID-19 [79].
The infection route of the virus also exacerbates blood clotting disorders. Entry into susceptible cells occurs as a result of interactions between the spike protein (S) and the specific angiotensin-converting enzyme 2 receptor [80]. The host cell transmembrane protease serine 2 is located on the cell surface, and cathepsin L in the endobody facilitates the internalisation of the virus by cleaving the S protein, thus causing endotheliitis [80, 81]. The cleavage of the S protein by the activated coagulation factors FX and thrombin promotes viral entry and initiates a potential positive feedback loop for inflammatory coagulation [82]. Although the interaction mechanism between SARS-CoV-2 and platelets is not completely understood, and more in-depth research is needed, many studies agree that the interaction is frequent, causing platelet overactivation and thrombosis [83, 84].
4.3.4 Hot Topics of Sepsis
In recent years, hot keywords have included COVID-19, machine learning, NLRP3 inflammasome, autophagy, gut microbiota, and miRNA.
The application of machine learning in the study of sepsis includes diagnosing and identifying sepsis, identifying inflammatory subtypes, and classifying metabolomics [85]. However, it is still commonly used to diagnose sepsis. In 2020, a meta-analysis (24 articles included) suggested that machine learning models could accurately predict the onset of sepsis in retrospective cohorts [86]. At the same time, owing to the differences in the diagnostic criteria for sepsis in the included cases (sepsis1.0-sepsis3.0), there are few prospective validation cohort studies, and external validation data are scarce; therefore, the application value of machine learning models in sepsis diagnosis needs to be demonstrated by more research. The more successful field for identifying inflammatory subtypes is the inflammatory phenotype of ARDS, including the confirmation of highly inflammatory subtypes with high expression of IL-6, IL-8, and sTNFR1, and the prediction of intervention response to treatment [87, 88]. A Few studies have been conducted on machine learning for sepsis metabolomics, which may represent a new research hotspot.
NLRP3 is a group of high-molecular-weight cytosolic protein complexes that mediate host immune responses to DAMPs and PAMPs. Considering the critical signalling role of NLRP3 in inflammatory immunity in sepsis, several infectious microorganisms have been found to activate the NLRP3 inflammasome [89–92]. Activation of the NLRP3 inflammasome induces substantial alterations in sepsis. Inhibition of inflammasome activation can prevent the associated inflammatory responses. Further studies are required to elucidate the detailed effects of the NLPR3 inflammasome on sepsis pathophysiology.
Autophagy induction is a primitive innate immune response in which autophagy engulfs pathogens and directs them to lysosomes for degradation, which kills pathogens; the upregulation of autophagy may enhance the clearance of some infectious pathogens [93]. A correlation exists between the NLRP3 inflammasome and autophagy, with NLRP3 activation influencing autophagy induction, and vice versa [94, 95]. This bidirectional regulation establishes negative and positive feedback loops that are essential for balancing host defence and inflammatory responses, and preventing excessive inflammation. The activation of caspase-1 may inhibit the induction of autophagy, promoting an increased inflammatory response necessary for pathogen clearance [96]. However, excessive inflammation can cause organ damage, and autophagy mitigates this damage by removing the inflammatory components. The intricate regulation of the inflammatory response involves pathways such as autophagy and inflammasome activation. Manipulation of these pathways may be a potential treatment for sepsis.
Disruption of the gut microbiota is a risk factor for sepsis [97, 98]. The gut microbiota can affect host susceptibility and response to sepsis through its structure, function, and metabolites [99]. For example, the symbiotic Enterobacteriaceae family inhibits Salmonella colonisation by competing for oxygen [100]. Additionally, commensal bacteria activate host-pattern recognition receptors to enhance AMP and MUC production. Further, they induce IgA secretion by providing moderate immune stimulation, establishing a vital immune adaptation that is crucial for maintaining host and microbial community homeostasis [101].
Regarding metabolites of gut microbes, the latest studies have shown that Akkermansia can produce a new tripeptide Arg-Lys-His, which directly binds to TLR4 to inhibit inflammatory signalling pathways in immune cells, showing significant anti-inflammatory activity [102]. Another study on the metabolites of the gut symbiotic Candida albicans revealed that phenylpyruvate, which directly binds to TLR4, increases the production of dependent reactive oxygen species, enhances the phagocytic capacity of macrophages, thereby inhibiting sepsis-related organ function damage [103]. To treat sepsis-related gut microbiota disturbances, faecal microbiota transplantation may enable more robust microbiome reconstruction and affect sepsis outcomes through multiple mechanisms, such as the production of short-chain fatty acids and immunomodulation [104]. Although it has demonstrated promising clinical outcomes in the treatment of Clostridium difficile infection (CDI), more clinical trials are needed to verify its efficacy and safety for clinical use in patients with non-CDI sepsis.
MiRNA is non-coding RNA composed of 20 to 24 nucleotides that alter cellular responses through synergistic effects on multiple targets under stress conditions. Recent studies have suggested that miRNA is an essential regulator of endotoxin tolerance via the TLR signalling pathway [105]. In terms of diagnosis, miR-16a, as a diagnostic marker for neonatal sepsis, inhibits the mRNA expression of pro-inflammatory factors such as IL-6 and TNF-α, and also promotes the mRNA expression of anti-inflammatory factors such as IL-10 [106]. MiR-193a-5p and miR-542-3p can distinguish non-infectious diseases from infectious diseases (CAP or sepsis) [107]. MIR-146a binds key components downstream of TLR4 signalling, including IRAK1 and TRAF6, as well as other pathways such as Notch-1, attenuating the inflammatory response [108]. Endotoxin-induced upregulation of miR-146a in macrophages protects against endotoxin-induced organ damage. Similarly, during sepsis, the addition of miR-146a has been shown to reduce excessive inflammation and prevent multiorgan failure [109]. Although there is a wealth of research on the role of miRNA in the diagnosis of sepsis, evidence for their protective clinical applications needs to be further explored.