Lessons From the LIPid Intensive Drug therapy for Sepsis Pilot (LIPIDS-P) Trial*

LIPIDS IN SEPSIS Water-based multicellular metazoans (all of us) have evolved mechanism for handling endogenous lipids that comprise cell membranes and many other cell components. Lipoprotein particles are central to these lipid-handling mechanisms and include high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), very-low-density lipoprotein cholesterol (VLDL-C), and interacting molecules such as cholesteryl ester transfer protein and phospholipid transfer protein (PLTP). Not surprisingly, these mechanisms also play a central role in the handling of hydrophobic lipid molecules arising from pathogens (1). Bacterial pathogens contain distinct lipid moieties within their cell membranes. For example, Gram-negative bacteria cell membranes contain lipopolysaccharide (LPS) and Gram-positive bacteria cell membranes contain lipoteichoic acid (1). Fungal pathogens also contain lipid moieties such as phospholipomannan and even viral particle coats may contain lipid-related molecules. Transfer proteins such as PLTP and the molecularly similar LPS-binding protein and bactericidal permeability-increasing protein avidly bind LPS and other toxic pathogen lipids and initially transfer toxic pathogen lipids to HDL-C (2). HDL-C binding sequesters toxic pathogen lipids thereby reducing their ability to induce an inflammatory response. HDL-C carries toxic pathogen lipids to the liver for elimination or, with the help of transfer proteins, HDL-C transfers toxic pathogen lipids to LDL-C and VLDL-C (2). LDL-C binds low-density lipoprotein (LDL) receptors expressed on the liver resulting in internalization, degradation, and elimination of toxic pathogen lipids into bile by the liver (1,3). Interestingly, newly clinically available proprotein convertase subtilisin/kexin type 9 (PCSK9) inhibitors or genetically low PCSK9 function increase hepatic expression of LDL receptors resulting in increased clearance of LDL-C, the effect used to treat hypercholesterolemia (4). Increased clearance of LDL-C also increases clearance of toxic pathogen lipids within LDL-C, thereby decreasing the severity of the septic inflammatory response in adult mice and humans (1,3). Toxic pathogen lipids within VLDL-C can be transferred to adipose tissue, and this effect may account for part of the obesity paradox—obese septic patients have higher survival (5). Low HDL-C and LDL-C levels occur in sepsis and are associated with adverse clinical outcomes, possibly because the reduction of these lipoprotein fractions limits the sequestration and elimination of toxic pathogen lipids (6). Thus, the intricate network that handles endogenous cholesterol is also involved in binding, transport, and elimination of toxic pathogen lipids. RATIONALE FOR THE LIPID INTENSIVE DRUG THERAPY FOR SEPSIS PILOT TRIAL Binding of toxic pathogen lipids to lipoprotein particles results in sequestration of these toxic moieties and, therefore, reduces the septic inflammatory response. HDL-C and LDL-C levels are markedly decreased early in sepsis (6). Thus, it is reasonable to postulate that increasing the number of circulating lipid particles to bind and sequester toxic pathogen lipids may be beneficial in patients with sepsis. An important study, the LIPid Intensive Drug therapy for Sepsis Pilot (LIPIDS-P) trial, published in this issue of Critical Care Medicine starts to address this issue (7). The investigators had recently conducted a phase I trial of a specific lipid emulsion containing soybean oil, medium-chain triglycerides, olive oil, and fish oil with promising results (8). The study by Guirgis et al (7) is the phase II extension. The LIPIDS-P investigators postulated that this particular combination would promote de novo cholesterol synthesis and thereby increase, or prevent further decreases in, total cholesterol levels, HDL-C, and LDL-C. STUDY DESIGN AND EXECUTION The LIPIDS-P investigators conducted a randomized clinical trial of infused lipid emulsion in sepsis patients (7). This was a phase II trial powered to test for a change in total cholesterol levels 48 hours after initiation of a lipid emulsion infusion in sepsis patients. Patients were recruited within the first 24 hours of sepsis recognition, had at least moderate organ dysfunction, and had low total cholesterol or a low sum of LDL-C and HDL-C. Patients receiving lipid formulation such as total parenteral nutrition, propofol, and clevidipine were excluded. Other inclusion and exclusion criteria were sensible and rigorously applied, as reflected by an appropriate ratio of screened to enrolled patients. Two doses of lipid emulsion were administered to 23 patients while 24 control patients received standard care without lipid emulsion infusion. At the start of the trial three different amounts of lipid emulsion infusion were considered, but in this trial adaptive design, this was decreased to two different amounts part way through the trial. At the final data analysis, all lipid emulsion patients were considered together. The study was not blinded after initiation of study infusion. A power calculation influenced by the phase I trial results led to a sample size of 24 patients in the combined treatment arm and 24 control patients, testing for a difference between the groups in “change in total cholesterol” of more than 2 mg/dL. DISAPPOINTING RESULTS The absolute difference in “change in total cholesterol” between the groups was numerically greater than the target 2 mg/dL (change in total cholesterol was 5 ± 20 mg/dL in the lipid emulsion group and 2 ± 18 mg/dL in controls); however, the sd in “change in total cholesterol” was surprisingly high so that no statistically significant difference, or even trend, was evident (p = 0.62). No differences were observed in the cholesterol subsets of HDL-C, LDL-C, and triglycerides or change in these subsets. Similarly, no secondary clinical endpoints differed between treated and control patients. No difference in adverse events was observed. POTENTIAL EXPLANATIONS AND LESSONS LEARNED It was reasonable to postulate that this particular lipid emulsion would have an effect of cholesterol levels, but this was not observed (9). Change in HDL-C or LDL-C as a result of de novo synthesis may take longer to occur than the 48-hour duration of the study by Guirgis et al (7). Note that the mean change in triglycerides in the study by Guirgis et al (7), although numerically different between groups, was not statistically significant. Since triglycerides were part of the lipid infusion, triglyceride levels should have changed, effectively a positive control. The nonsignificant effect suggests that it is possible that the lipid emulsion infusion dose was insufficient. Possibly the role of lipids in sepsis is not as important as suggested by a number of previous studies or possibly, of the lipoprotein fractions, triglycerides play a minor role. For example, Phillip Dellinger et al (10) did not observe a difference in clinical outcomes in confirmed or suspected Gram-negative sepsis between 598 patients who received a phospholipid emulsion infusion and 599 patients who did not. Yet a recent clinical trial provides hope that the role of lipids in sepsis is important. Specifically, PCSK9 inhibition during the inflammatory stage of COVID-19 resulted in a statistically significant reduction in mortality (11). It is also possible that the absolute level of total cholesterol, HDL-C, or LDL-C was not the biologically important feature. Rather, the important feature may be the rate of clearance of the toxic pathogen lipid-containing lipoproteins. That is, the concentration of any measured analyte in the blood (or in any compartment) is set by the rate of input compared with the rate of clearance. As postulated in the LIPIDS-P trial, it may be that the absolute level of HDL-C or LDL-C (set by input versus clearance) is important because this absolute level is available to bind and therefore biologically sequester toxic pathogen lipids and thereby diminish the subsequent septic inflammatory response. Alternatively, it may be that the rate of clearance is more important. That is, clearance of toxic pathogen lipids into the liver or other organs, where the next steps in elimination or sequestration take place, may be the key biologically important feature. In that case, the absolute value of HDL-C or LDL-C would be less important than the rate of clearance of HDL-C and/or LDL-C (and the contained toxic pathogen lipids). For example, while LDL-C levels are very low in sepsis and associated with adverse outcomes, the rate of clearance can be increased by PCSK9 inhibition. Increasing the rate of LDL-C clearance would presumably decrease LDL-C concentrations and, by inference, be bad. But increasing the rate of clearance of LDL-C in this way reduces blood endotoxin concentrations and improves outcomes in mice and, using genetic observations, appears to improve outcomes in adult humans (12). Thus, it appears that absolute LDL-C levels are not the dominant biologically important issue in adult humans; rather it is the rate of LDL-C clearance that is important. Notably this is not the case in pediatric sepsis where absolute LDL-C levels appear to be most important. That is, low LDL-C levels are associated with adverse outcomes in children (as they are in adults) while increasing LDL-C clearance appears to be detrimental (the opposite of adults), based on genetic observations (13). Thus, when one considers the importance of absolute analyte concentrations, as is done in the current investigation for total cholesterol, HDL-C, and LDL-C, it is important to consider the factors that set absolute concentrations (input and clearance) to elucidate the biologically important mechanisms. CONCLUSIONS While the LIPDS-P trial result is disappointing, the trial was expertly conducted so it provided valuable and reliable information and, importantly, it suggests that reflecting on underlying mechanistic pathways is important.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.