- Email: [email protected]
Severe hypocholesterolemia in surgical patients, sepsis, and critical illness
Journal of Critical Care (2010) 25, 361.e7–361.e12
Severe hypocholesterolemia in surgical patients, sepsis, and critical illness Carlo Chiarla MDa , Ivo Giovannini MDa,⁎, Felice Giuliante MDa , Zdenek Zadak MDb , Maria Vellone MDa , Francesco Ardito MDa , Gennaro Clemente MDa , Marino Murazio MDa , Gennaro Nuzzo MDa a
Department of Surgery, Hepatobiliary Unit and CNR-IASI Shock Center, Catholic University of the Sacred Heart School of Medicine, 00168 Rome, Italy b Department of Gerontology and Metabolic Care, Medical Faculty Charles University, 500 05 Hradec Kralove, Czech Republic
Keywords: Plasma cholesterol; Hypocholesterolemia; Sepsis; Critical illness; Surgery; Mortality
Abstract After surgery, in sepsis and various critical illnesses, factors such as severity of the acute phase response, liver dysfunction, and hemodilution from blood loss have cumulative impacts in decreasing cholesterol; therefore, degree of hypocholesterolemia often reflects severity of illness. The direct correlation between cholesterol and several plasma proteins is mediated by the parallel impact of commonly shared determinants. Cholestasis is associated with a moderation of the degree of hypocholesterolemia. In human sepsis, the poor implications of hypocholesterolemia seem to be aggravated by the simultaneous development of hypertriglyceridemia. Cholesterol and triglyceride levels reflect altered lipoprotein patterns, and the issue is too complex and too poorly understood to be reduced to simple concepts; nevertheless, these simple measurements often represent helpful adjunctive clinical tools. © 2010 Elsevier Inc. All rights reserved.
1. Background Although hypocholesterolemia is anecdotally considered a marker of malnutrition, other factors become prominent causes of hypocholesterolemia after surgery and trauma, in sepsis, and in other critical illnesses. Because there is synergism of clinically adverse components in decreasing cholesterol, hypocholesterolemia often becomes a cumulative marker of severity of illness and poor prognosis. At the same time, the multifactoriality of changes and the undefined ⁎ Corresponding author. Department of Surgery, Hepatobiliary Unit, Catholic University, 00168 Rome, Italy. Tel.: +39 06 3015 4082, +39 06 3015 4967; fax: +39 06 305 1343. E-mail address: [email protected] (I. Giovannini). 0883-9441/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.jcrc.2009.08.006
thresholds of risk limit the information which is obtainable from absolute cholesterol levels; the trend toward decrease or increase or the persistence of major decrease are more significant. This short review addresses the main practical aspects involved in the evaluation of hypocholesterolemia, also including the opposite effect of cholestasis; the relationship with plasma proteins, triglycerides, and outcomes in severe illness; and the particular impact of nutritional substrates.
2. Methods The review was based on relevant data from the literature, describing most of the involved aspects together with
361.e8 controversial issues, and on locally performed patient measurements. These were derived from a prospectively collected data bank including 1313 laboratory measurements obtained in 190 patients (108 males, 82 females) undergoing resective surgery of the liver (89 patients), stomach, colon, or pancreas and other abdominal procedures (61 patients), and/ or with major complications or illnesses such as sepsis, liver failure, hemorrhage, cholestasis, or multiple organ dysfunction syndrome. This provided a continuous distribution of observations from normal preoperative to moderate or extreme and preterminal illness, which was well suited to assess the correlates of plasma cholesterol over a wide range of pathophysiologic abnormalities. Patient characteristics (mean ± SD) were as follows: age, 58.9 ± 13.2 years; body weight, 69.5 ± 10.5 kg; height, 168.1 ± 6.9 cm; and body mass index, 24.6 ± 3.6 kg/m2. Twenty-two patients had liver cirrhosis. Forty-five patients developed sepsis, which was diagnosed according to previously defined criteria [1]. There were 10 deaths, all from multiple organ dysfunction syndrome, mostly related to sepsis. Venous blood measurements were performed according to the normal clinical routine in elective surgical cases without complications and according to clinical indications until recovery or death in complicated cases, without the need for consent. Because of this, some of the blood chemistries were inconsistently determined, as specified for each relevant case in the text. This data bank was built up as a frame of reference to characterize and quantify patterns of biochemical abnormalities in acute illness with or without complications, correlations among variables, and trajectories toward recovery or worsening of illness. It was based on commonly determined blood measurements consistently including cholesterol (which has long been a standard component of
Fig. 1 A 35-year-old male patient referred with peritonitis after multiple abdominal procedures. After surgical drainage, there is progressive reversal of hypocholesterolemia, transiently interrupted by recurrent sepsis. With reoperation for bowel occlusion from adhesions and the subsequent development of an abdominal abscess, there is a relapse of profound hypocholesterolemia, which is finally reversed after percutaneous drainage of the abscess.
C. Chiarla et al. the automatic blood chemistry profile in our hospital) [2]. The correlations between plasma cholesterol and other blood variables were explored in detail in graphic displays of the measurements and then assessed using least square regression analysis [1,3] to confirm significance and to quantify the percentage of variability of cholesterol accounted for by each correlation, on the basis of the r2 value. For these purposes, the Statgraphics package (Statgraphics Plus; Manugistics, Rockville, Md) and noncommercial software were used. Finally, sequential measurements from 2 recently observed patient cases, not included in the basic data set, were used to draw Figs. 1 and 4.
3. Multifactoriality of hypocholesterolemia Plasma cholesterol behaves as a negative acute phase reactant. It decreases after surgery and trauma, in sepsis, in liver dysfunction, and after acute hemorrhage; the main mechanisms involved include the cholesterol-lowering effect of inflammatory mediators, impaired cholesterol synthesis, and hemodilution from blood loss [2,4-10]. The dynamics of these decreases are complex; however, the effect is cumulative, so that the degree of hypocholesterolemia may cumulatively reflect the severity of illness and the importance of the “adverse” factors that are simultaneously involved [2,7,11]. A model for this multifactoriality is offered by liver resection patients, in whom the degree of postoperative hypocholesterolemia is cumulatively related to the magnitude of surgical trauma, sepsis, liver dysfunction, and the fall in hematocrit, with an evident relationship with outcomes [12]. Indeed, the cumulative impact of sepsis and liver dysfunction in lowering cholesterol is also part of their synergism in negatively affecting outcomes [13]. Multifactoriality also includes the level of the baseline preillness cholesterol, the factors affecting the enterohepatic recirculation of cholesterol, and other variables. This prevents the definition of exact threshold of abnormality and the obtainment of very precise clinical information only from hypocholesterolemia. However, in postoperative patients, values of cholesterol persistently less than 60 to 50 mg/dL (1.55-1.29 mmol/L) are often associated with severe illness and/or sepsis and poor outcomes. The lowest value observed by us was 11 mg/dL (0.28 mmol/L) in a case with severe intra-abdominal sepsis, who, however, rapidly recovered after surgery [14]. Therefore, in practice, cholesterol cannot be used as a precise diagnostic tool, but serial changes in the degree of hypocholesterolemia over time are a useful adjunct in evaluating progressive recovery or the occurrence of complications and persistent illness in surgical and intensive care patients [2,7,9,15]. Fig. 1 provides the example of a patient in whom severe falls in cholesterol were associated with initial peritonitis and, thereafter, with recurrent sepsis and with new surgical trauma again complicated by sepsis
Hypocholesterolemia in surgery and sepsis
361.e9
while cholesterol increased during interposed intervals of recovery and during final recovery.
4. The effect of cholestasis Cholestasis is peculiar among the commonest “adverse” factors altering cholesterol because it is associated with an increase in cholesterol (rather than contributing to its decrease) or, more often, with moderation of the decrease related to other factors [7,14,16]. This is likely related to the stimulated release of cholesterol-rich lipoprotein-X from the liver and/or the impaired cholesterol excretion in bile and explains why critically ill patients with cholestasis may fail to show extreme hypocholesterolemia. A rough estimation is that cholesterol increases by 5 to 9 mg/dL (0.13-0.23 mmol/ L) per 100 U/L increase in alkaline phosphatase (AP) (normal range, 98-279 U/L) [2]. A deeper assessment of the relationships between cholesterol, AP, γ-glutamyl-transpeptidase (GGT), and bilirubin using the locally collected data bank showed that increases in cholesterol were better related to AP or GGT than to bilirubin. The absence of correlation between cholesterol and bilirubin in the whole group of measurements was because the determinants of hyperbilirubinemia included liver dysfunction due to liver resection, liver insufficiency, and/or sepsis (which were associated with low cholesterol) and obstructive cholestasis (which was associated with high cholesterol). However, in this same whole group, cholesterol was directly correlated with AP and GGT (r2 = 0.25, P b .0001 for both). In the measurements performed only in cases with obstructive or extracellular cholestasis, in which bilirubin increased variably together with AP and GGT, cholesterol was slightly better related to AP and GGT (r2 = 0.26 and 0.27, respectively) than to bilirubin (r2 = 0.21) (P b .0001 for all). In multiple regressions combining AP or GGT with bilirubin, the total r2 rose to 0.40 (P b .0001), which might be consistent with the combined impact of impaired biliary excretion (reflected by bilirubin) and of a partly independent mechanism (reflected by AP or GGT) in increasing cholesterol.
Fig. 2 Relationship between plasma cholesterol and albumin in 1313 measurements on 190 patients. The main distribution with regression line and 95% confidence limits (triangles: cholesterol [mmol/L] = 0.12 [albumin] − 1.26, n = 1046, r2 = 0.38, P b .0001) includes measurements in normal preoperative state, after surgery (liver resection or other major abdominal procedures), in cases with or without complications, and in critically ill patients with sepsis, liver insufficiency, and major bleeding. All conditions fell along the same distribution, except for a downward shift for septic patients (lower cholesterol for any given albumin level, P b .0001). The upward spread for cases with cholestasis (squares) may be explained by the increase in cholesterol related to this condition. See text.
synthetic activity, and hemodilution from blood loss [2,5,7,8,17]. Similar correlations can be shown with other proteins or related measurements, such as transferrin, ironbinding capacity, cholinesterase, prealbumin, retinol-binding
5. Correlation with changes in plasma proteins Although it appears peculiar for a fat substance, cholesterol maintains a relatively strong direct relationship with albumin and several other plasma proteins. Fig. 2 shows it for albumin over a large distribution of observations, ranging from the normal condition to extreme preterminal illness. In the normal condition, the main factor supporting the correlation should be the nutritional state. In postoperative and/or critically ill conditions, the underlying factors are various combinations of acute phase response (surgical stress, sepsis, other acute illnesses), impairment of liver
Fig. 3 Measurements on patients after liver resection. Direct relationship between cholesterol and fibrinogen (a “negative” and a “positive” acute phase reactant, respectively) likely reflecting a similar impact on both variables of substrate limitation plus transient liver dysfunction (although fibrinogen may also be partly influenced by abnormal coagulation). The shift in the distribution for patients with sepsis (triangles, shaded area, dotted regression line) is consistent with strongest priority for fibrinogen synthesis in this condition (regression: cholesterol = 0.11 [fibrinogen] − 0.56[SEPSIS] + 1.51, n = 270, r2 = 0.34, P b .0001). The observations performed before postoperative day 3, excluded to avoid confounding from recent intraoperative bleeding, had a similar distribution.
361.e10 protein, and prothrombin activity (in cases in which it reflects hepatic synthetic activity) [1,17,18]. This was reconfirmed in the locally collected data bank, including prothrombin activity, simultaneously obtained with cholesterol in 899 measurements, and in more inconsistent measurements of cholinesterase and iron-binding capacity: all of them were directly related to cholesterol with r2 of 0.42 for prothrombin activity, 0.50 for cholinesterase, and 0.30 for iron-binding capacity (P b .0001 for all). The main interfering factor in these correlations remains cholestasis, which tends to independently increase cholesterol, as shown in Fig. 2. For “positive” acute phase proteins such as C-reactive protein, the correlation with cholesterol may be the opposite (inverse correlation), as observed in sepsis [16,19,20]. It is not so for fibrinogen after major liver resection, when the competition for substrate to support liver regeneration plus transient liver dysfunction may result in a similar impact on fibrinogen and cholesterol levels (Fig. 3).
6. Correlation with outcomes Transiently severe hypocholesterolemia is not necessarily relevant. It is severe persistent hypocholesterolemia, which is associated with death, whereas increasing cholesterol (unless related to cholestasis) is associated with recovery [2,7-10,12,15,20]. Again, the absence of precise thresholds of abnormality and the multifactoriality of hypocholesterolemia limit the definition of exact landmarks of risk. In a previous assessment in patients with sepsis and organ failure after partial hepatectomy, none survived among those in whom cholesterol remained for more than 6 days less than 58 mg/dL (1.50 mmol/L) or less than 40% of its preoperative value [12], although this cutoff cannot be extensively generalized. An important issue is that hypocholesterolemia may not only passively reflect severity of illness but perhaps also contribute directly to poor outcomes. Hypocholesterolemia could reflect hypolipoproteinemia, with impairment of the host defense against bacterial products provided by circulating lipoproteins. Cytokines in sepsis may impair lipoprotein production and/or facilitate their degradation, with loss of their protective effect further amplifying the process in a vicious circle [21]. This is not totally clear because it could be only the cholesterol content of the lipoproteins that is altered; furthermore, it is the phospholipid component that seems to be more involved in host protection [22-26]. In addition, a reduction in lipoprotein-carried vitamin E, contributing to impaired antioxidant protection, is also likely to be implicated [1,12,14]. Another issue is the reduced availability of cholesterol in a condition of increased requirement. The issue involves both the availability of basic substrate and the demanding biochemical processes of cholesterol synthesis, which cannot be supported at a sufficiently high rate [17,27]. Cholesterol
C. Chiarla et al. requirement increases to support the synthesis of stress hormones and the production and function of new cells taking part in host defense and tissue repair [15,17,27]. Cholesterol is a main component of cell membranes (simply by weight alone), and increased requirement characterizes various cell expansion processes, including liver regeneration after partial hepatectomy and tumor cell expansion in leukemia [2]. These considerations have grounded the concept of the “essentiality” of cholesterol, with a possible need for exogenous supply in septic and critically ill patients [2,27,28]. A similar issue has been raised for tubercular infection [29], and it is possible to quote historical references for this idea [28,29].
7. Hypocholesterolemia and triglyceride levels in trauma and sepsis Plasma triglycerides generally tend to follow a pattern similar to that of cholesterol, although not strictly related to it. In the locally collected data bank, triglycerides were obtained in almost half of the cases and were directly correlated with cholesterol, although with an r2 of only 0.25 (P b .0001). One factor was that cholestasis, if present, affected cholesterol more than triglycerides. Remarkably, however, within this variability, severe hypocholesterolemia became sometimes associated with hypertriglyceridemia: this occurred in particular in septic patients in the stages of greater metabolic decompensation or preterminal illness. Indeed, in sepsis, changes in plasma triglycerides may largely diverge from those of cholesterol. This is an object of controversy, mainly because animal and experimental studies fail to reproduce clinical sepsis [6,25,30]. In human sepsis, there is hypocholesterolemia, and in our patients, we have seen it more often paralleled by normal or low triglyceride levels. As already mentioned, persistently severe hypocholesterolemia is related to poor prognosis. At the same time, it is often perceived that increasing triglycerides with persistently low cholesterol is an even worse landmark of severity of illness and impending risk of death. This apparent landmark of greater metabolic deterioration and inflammatory response or preterminal illness [1,2,4] was repeatedly confirmed in additional patient cases observed after the collection of the local data bank, such as that shown in Fig. 4. Although the poor implications of the development of hypertriglyceridemia in clinical sepsis were already described many years ago [31], the issue has not yet been sufficiently emphasized. The underlying mechanism is unclear: modified cytokine or cytokine receptor patterns, decreased lipoprotein lipase activity and/or increase in hepatic triglyceride production, and peculiarity of the involved microorganism [2,8,17,25,30]. The related hyperlipoproteinemia may also be an extreme attempt at peripheral delivery of substrate or defense [2,32-34]. However, hypertriglyceridemia in sepsis often remains a worrisome
Hypocholesterolemia in surgery and sepsis
361.e11
Fig. 4 Patient with persistent retroperitoneal sepsis after duodenocephalopancreasectomy. The triglyceride peak (without fat infusion) corresponds to worsening of the septic state with rapid evolution into adult respiratory distress syndrome. Improvement after surgical drainage, early recurrence of sepsis, new surgical drainage, reversal of hypocholesterolemia alongside final recovery. In spite of the improvement achieved in this case by aggressively treating sepsis at the time of hypertriglyceridemia, hypertriglyceridemia may unfortunately often be part of a near-preterminal pattern. ARDS indicates adult respiratory distress syndrome.
landmark of extreme severity of illness [1,2,4]. The implication that hypertriglyceridemia limits parenteral fat supply is more obvious. However, at this stage, nutrition might also lose its efficacy, and the best hope of recovery, if this is still achievable, lies in the aggressive eradication of the septic focus. Although generalizations are only partly allowable, similar events have been observed in burns: hypocholesterolemia with a tendency to reversal without major changes in triglycerides in recovering patients or worsening of hypocholesterolemia with preterminally increasing triglycerides in nonsurviving patients [35].
8. The impact of nutrition (parenteral nutrition) Although it may appear counterintuitive, the nutritional state has little to do with hypocholesterolemia after surgery and trauma, in sepsis, and in critical illness because other driving factors prevail. Furthermore, the mechanisms by which the infusion of nutrients can modify cholesterol are not so obvious. For instance, some increase in cholesterol related to the infusion of fat emulsions depends on the release of cholesterol from cell membranes, in exchange with the components of the emulsions (which contain cholesterol only in trace amounts) [2,5,28]. Cholesterol has been reported to increase with increasing dose of amino acid; this needs to be better evaluated in additional studies; however, the underlying mechanism may be the provision of cholesterol precursors from some amino acids (for instance leucine) or the facilitated synthesis of the apoprotein component of lipoproteins [2]. Of course, the fundamental role of amino acid supply in critical illness largely overrides
these speculations, and nutrition provides basic substrate also for cholesterol synthesis, although this is not so evidently reflected in the plasma levels. The issue differs with plasma triglycerides because, as mentioned already, hypertriglyceridemia may be worsened by infusing fat emulsions (including, to a lesser extent, also those vehiculating medications such as propofol).
9. Conclusion After surgery and trauma, in sepsis, and in critical illness, 2 very simple measurements such as cholesterol and triglycerides often become markers of severity of disease and, therefore, helpful adjunctive clinical tools. Obviously, cholesterol and triglyceride levels do not only reflect their concentrations but also altered lipoprotein patterns with modified lipoprotein compositions, and the issue is far too complex and still too poorly understood to be reduced to simple concepts. A better understanding of the underlying processes will likely improve our capability to rescue these desperately ill patients from their condition. At present, the most concrete implication of severe hypocholesterolemia, alone or associated with hypertriglyceridemia, is the need for rapid resolution of the underlying illness.
Acknowledgment This work was supported by a contribution from the Catholic University School of Medicine and the Italian Ministry for University and Scientific Research (D.1 Funds).
361.e12
References [1] Chiarla C, Giovannini I, Siegel JH. Hypotransferrinemia and changes in plasma lipid and metabolic patterns in sepsis. Amino Acids 2009;36: 327-31 [Epub 2008 Apr 8]. [2] Giovannini I, Chiarla C, Giuliante F, et al. Hypocholesterolemia in surgical trauma, sepsis, other acute conditions and critical illness. In: Kramer MA, editor. Trends in cholesterol research. Hauppauge (NY): Nova Science Publishers Inc.; 2005. p. 137-61. [3] Seber GAF. Linear regression analysis. New York: Wiley; 1977. p. 369-82. [4] Giovannini I, Boldrini G, Castagneto M, et al. Respiratory quotient and patterns of substrate utilization in human sepsis and trauma. JPEN J Parenter Enteral Nutr 1983;7:226-30. [5] Sun X, Oberlander D, Huang J, et al. Fluid resuscitation, nutritional support, and cholesterol in critically ill postsurgical patients. J Clin Anesth 1998;10:302-8. [6] Fraunberger P, Schaefer S, Werdan K, et al. Reduction of circulating cholesterol and apolipoprotein levels during sepsis. Clin Chem Lab Med 1999;37:357-62. [7] Giovannini I, Boldrini G, Chiarla C, et al. Pathophysiologic correlates of hypocholesterolemia in critically ill surgical patients. Intensive Care Med 1999;25:748-51. [8] Gordon BR, Parker TS, Levine DM, et al. Relationship of hypolipidemia to cytokine concentrations and outcomes in critically ill surgical patients. Crit Care Med 2001;29:1563-8. [9] Dunham CM, Fealk MH, Sever III WE. Following severe injury, hypocholesterolemia improves with convalescence but persists with organ failure or onset of infection. Crit Care 2003;7:R145-53 [Epub 2003 Oct 1]. [10] Bonville DA, Parker TS, Levine DM, et al. The relationships of hypocholesterolemia to cytokine concentrations and mortality in critically ill patients with systemic inflammatory response syndrome. Surg Infect (Larchmt) 2004;5:39-49. [11] Guimarães SM, Lima EQ, Cipullo JP, et al. Low insulin-like growth factor-1 and hypocholesterolemia as mortality predictors in acute kidney injury in the intensive care unit. Crit Care Med 2008;36:3165-70. [12] Giovannini I, Chiarla C, Greco F, et al. Characterization of biochemical and clinical correlates of hypocholesterolemia after hepatectomy. Clin Chem 2003;49:317-9. [13] Nuzzo G, Giovannini I, Giuliante F, et al. Sepsis after liver resection: predisposition, clinical relevance and synergism with liver dysfunction. In: Dionigi R, editor. Recent advances in liver surgery. Austin (Tex): Landes Bioscience; 2009. p. 245-59. [Epub 2008 March http:// www.eurekah.com/chapter/3870]. [14] Giovannini I, Chiarla C, Giuliante F, et al. Biochemical and clinical correlates of hypouricemia in surgical and critically ill patients. Clin Chem Lab Med 2007;45:1207-10. [15] Marik PE. Dyslipidemia in the critically ill. Crit Care Clin 2006;22: 151-9. [16] Chiarla C, Giovannini I, Siegel JH. The relationship between plasma cholesterol, amino acids and acute phase proteins in sepsis. Amino Acids 2004;27:97-100.
C. Chiarla et al. [17] Hrnciarikova D, Hyspler R, Vyroubal P, et al. Serum lipids and neopterin in urine as new biomarkers of malnutrition and inflammation in the elderly. Nutrition 2009;25:303-8 [Epub 2008 Nov 18]. [18] Lopez-Martinez J, Sanchez-Castilla M, Garcia-de-Lorenzo A. Hypocholesterolemia in critically ill patients. Intensive Care Med 2000;26: 259-60. [19] Bentz MH, Magnette J. Hypocholestérolémie au cours de la phase aiguë de la réaction inflammatoire d'origine infectieuse. À propos de 120 cas. Rev Med Interne 1998;19:168-72. [20] Memiş D, Gursoy O, Tasdogan M, et al. High C-reactive protein and low cholesterol levels are prognostic markers of survival in severe sepsis. J Clin Anesth 2007;19:186-91. [21] Chien JY, Jerng JS, Yu CJ, et al. Low serum level of high-density lipoprotein cholesterol is a poor prognostic factor for severe sepsis. Crit Care Med 2005;33:1688-93. [22] Carpentier YA, Scruel O. Changes in the concentration and composition of plasma lipoproteins during the acute phase response. Curr Opin Clin Nutr Metab Care 2002;5:153-8. [23] van Leeuwen HJ, Heezius EC, Dallinga GM, et al. Lipoprotein metabolism in patients with severe sepsis. Crit Care Med 2003;31: 1359-66. [24] Kitchens RL, Thompson PA, Munford RS, et al. Acute inflammation and infection maintain circulating phospholipid levels and enhance lipopolysaccharide binding to plasma lipoproteins. J Lipid Res 2003; 44:2339-48. [25] Khovidhunkit W, Kim MS, Memon RA, et al. Effect of infection and inflammation on lipid and lipoprotein metabolism: mechanisms and consequences to the host. J Lipid Res 2004;45:1169-96. [26] Gordon BR, Parker TS, Levine DM, et al. Neutralization of endotoxin by a phospholipid emulsion in healthy volunteers. J Infect Dis 2005; 191:1515-22. [27] Zadak Z, Hyspler R, Bakalar B, et al. The role of cholesterol and intermediate metabolites of its synthesis in intensive care and parenteral nutrition. Vnitr Lek 2000;46:776-81. [28] Druml W. Is cholesterol a conditionally essential nutrient in critically ill patients? Wien Klin Wochenschr 2003;115:740-2. [29] Pérez-Guzmán C, Vargas MH. Hypocholesterolemia: a major risk factor for developing pulmonary tuberculosis? Med Hypotheses 2006; 66:1227-30. [30] Malmendier CL, Lontie JF, Bubois DY. Mechanisms of hypocholesterolemia. In: Malmendier CL, Alaupovic P, Bryan Brewer Jr H, editors. Hypercholesterolemia, hypocholesterolemia, hypertriglyceridemia. New York (NY): Plenum Press; 1990. p. 173-82. [31] Siegel JH, Cerra FB, Coleman B, et al. Physiological and metabolic correlations in human sepsis. Surgery 1979;86:163-93. [32] Samra JS, Summers LK, Frayn KN. Sepsis and fat metabolism. Br J Surg 1996;83:1186-96. [33] Barcia AM, Harris HW. Triglyceride-rich lipoproteins as agents of innate immunity. Clin Infect Dis 2005;41(S7):S498-S503. [34] Spitzer AL, Harris HW. Statins attenuate sepsis. Surgery 2006;139: 283-7. [35] Kamolz LP, Andel H, Mittlböck M, et al. Serum cholesterol and triglycerides: potential role in mortality prediction. Burns 2003;29: 810-5.
Severe hypocholesterolemia in surgical patients, sepsis, and critical illness Carlo Chiarla MDa , Ivo Giovannini MDa,⁎, Felice Giuliante MDa , Zdenek Zadak MDb , Maria Vellone MDa , Francesco Ardito MDa , Gennaro Clemente MDa , Marino Murazio MDa , Gennaro Nuzzo MDa a
Department of Surgery, Hepatobiliary Unit and CNR-IASI Shock Center, Catholic University of the Sacred Heart School of Medicine, 00168 Rome, Italy b Department of Gerontology and Metabolic Care, Medical Faculty Charles University, 500 05 Hradec Kralove, Czech Republic
Keywords: Plasma cholesterol; Hypocholesterolemia; Sepsis; Critical illness; Surgery; Mortality
Abstract After surgery, in sepsis and various critical illnesses, factors such as severity of the acute phase response, liver dysfunction, and hemodilution from blood loss have cumulative impacts in decreasing cholesterol; therefore, degree of hypocholesterolemia often reflects severity of illness. The direct correlation between cholesterol and several plasma proteins is mediated by the parallel impact of commonly shared determinants. Cholestasis is associated with a moderation of the degree of hypocholesterolemia. In human sepsis, the poor implications of hypocholesterolemia seem to be aggravated by the simultaneous development of hypertriglyceridemia. Cholesterol and triglyceride levels reflect altered lipoprotein patterns, and the issue is too complex and too poorly understood to be reduced to simple concepts; nevertheless, these simple measurements often represent helpful adjunctive clinical tools. © 2010 Elsevier Inc. All rights reserved.
1. Background Although hypocholesterolemia is anecdotally considered a marker of malnutrition, other factors become prominent causes of hypocholesterolemia after surgery and trauma, in sepsis, and in other critical illnesses. Because there is synergism of clinically adverse components in decreasing cholesterol, hypocholesterolemia often becomes a cumulative marker of severity of illness and poor prognosis. At the same time, the multifactoriality of changes and the undefined ⁎ Corresponding author. Department of Surgery, Hepatobiliary Unit, Catholic University, 00168 Rome, Italy. Tel.: +39 06 3015 4082, +39 06 3015 4967; fax: +39 06 305 1343. E-mail address: [email protected] (I. Giovannini). 0883-9441/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.jcrc.2009.08.006
thresholds of risk limit the information which is obtainable from absolute cholesterol levels; the trend toward decrease or increase or the persistence of major decrease are more significant. This short review addresses the main practical aspects involved in the evaluation of hypocholesterolemia, also including the opposite effect of cholestasis; the relationship with plasma proteins, triglycerides, and outcomes in severe illness; and the particular impact of nutritional substrates.
2. Methods The review was based on relevant data from the literature, describing most of the involved aspects together with
361.e8 controversial issues, and on locally performed patient measurements. These were derived from a prospectively collected data bank including 1313 laboratory measurements obtained in 190 patients (108 males, 82 females) undergoing resective surgery of the liver (89 patients), stomach, colon, or pancreas and other abdominal procedures (61 patients), and/ or with major complications or illnesses such as sepsis, liver failure, hemorrhage, cholestasis, or multiple organ dysfunction syndrome. This provided a continuous distribution of observations from normal preoperative to moderate or extreme and preterminal illness, which was well suited to assess the correlates of plasma cholesterol over a wide range of pathophysiologic abnormalities. Patient characteristics (mean ± SD) were as follows: age, 58.9 ± 13.2 years; body weight, 69.5 ± 10.5 kg; height, 168.1 ± 6.9 cm; and body mass index, 24.6 ± 3.6 kg/m2. Twenty-two patients had liver cirrhosis. Forty-five patients developed sepsis, which was diagnosed according to previously defined criteria [1]. There were 10 deaths, all from multiple organ dysfunction syndrome, mostly related to sepsis. Venous blood measurements were performed according to the normal clinical routine in elective surgical cases without complications and according to clinical indications until recovery or death in complicated cases, without the need for consent. Because of this, some of the blood chemistries were inconsistently determined, as specified for each relevant case in the text. This data bank was built up as a frame of reference to characterize and quantify patterns of biochemical abnormalities in acute illness with or without complications, correlations among variables, and trajectories toward recovery or worsening of illness. It was based on commonly determined blood measurements consistently including cholesterol (which has long been a standard component of
Fig. 1 A 35-year-old male patient referred with peritonitis after multiple abdominal procedures. After surgical drainage, there is progressive reversal of hypocholesterolemia, transiently interrupted by recurrent sepsis. With reoperation for bowel occlusion from adhesions and the subsequent development of an abdominal abscess, there is a relapse of profound hypocholesterolemia, which is finally reversed after percutaneous drainage of the abscess.
C. Chiarla et al. the automatic blood chemistry profile in our hospital) [2]. The correlations between plasma cholesterol and other blood variables were explored in detail in graphic displays of the measurements and then assessed using least square regression analysis [1,3] to confirm significance and to quantify the percentage of variability of cholesterol accounted for by each correlation, on the basis of the r2 value. For these purposes, the Statgraphics package (Statgraphics Plus; Manugistics, Rockville, Md) and noncommercial software were used. Finally, sequential measurements from 2 recently observed patient cases, not included in the basic data set, were used to draw Figs. 1 and 4.
3. Multifactoriality of hypocholesterolemia Plasma cholesterol behaves as a negative acute phase reactant. It decreases after surgery and trauma, in sepsis, in liver dysfunction, and after acute hemorrhage; the main mechanisms involved include the cholesterol-lowering effect of inflammatory mediators, impaired cholesterol synthesis, and hemodilution from blood loss [2,4-10]. The dynamics of these decreases are complex; however, the effect is cumulative, so that the degree of hypocholesterolemia may cumulatively reflect the severity of illness and the importance of the “adverse” factors that are simultaneously involved [2,7,11]. A model for this multifactoriality is offered by liver resection patients, in whom the degree of postoperative hypocholesterolemia is cumulatively related to the magnitude of surgical trauma, sepsis, liver dysfunction, and the fall in hematocrit, with an evident relationship with outcomes [12]. Indeed, the cumulative impact of sepsis and liver dysfunction in lowering cholesterol is also part of their synergism in negatively affecting outcomes [13]. Multifactoriality also includes the level of the baseline preillness cholesterol, the factors affecting the enterohepatic recirculation of cholesterol, and other variables. This prevents the definition of exact threshold of abnormality and the obtainment of very precise clinical information only from hypocholesterolemia. However, in postoperative patients, values of cholesterol persistently less than 60 to 50 mg/dL (1.55-1.29 mmol/L) are often associated with severe illness and/or sepsis and poor outcomes. The lowest value observed by us was 11 mg/dL (0.28 mmol/L) in a case with severe intra-abdominal sepsis, who, however, rapidly recovered after surgery [14]. Therefore, in practice, cholesterol cannot be used as a precise diagnostic tool, but serial changes in the degree of hypocholesterolemia over time are a useful adjunct in evaluating progressive recovery or the occurrence of complications and persistent illness in surgical and intensive care patients [2,7,9,15]. Fig. 1 provides the example of a patient in whom severe falls in cholesterol were associated with initial peritonitis and, thereafter, with recurrent sepsis and with new surgical trauma again complicated by sepsis
Hypocholesterolemia in surgery and sepsis
361.e9
while cholesterol increased during interposed intervals of recovery and during final recovery.
4. The effect of cholestasis Cholestasis is peculiar among the commonest “adverse” factors altering cholesterol because it is associated with an increase in cholesterol (rather than contributing to its decrease) or, more often, with moderation of the decrease related to other factors [7,14,16]. This is likely related to the stimulated release of cholesterol-rich lipoprotein-X from the liver and/or the impaired cholesterol excretion in bile and explains why critically ill patients with cholestasis may fail to show extreme hypocholesterolemia. A rough estimation is that cholesterol increases by 5 to 9 mg/dL (0.13-0.23 mmol/ L) per 100 U/L increase in alkaline phosphatase (AP) (normal range, 98-279 U/L) [2]. A deeper assessment of the relationships between cholesterol, AP, γ-glutamyl-transpeptidase (GGT), and bilirubin using the locally collected data bank showed that increases in cholesterol were better related to AP or GGT than to bilirubin. The absence of correlation between cholesterol and bilirubin in the whole group of measurements was because the determinants of hyperbilirubinemia included liver dysfunction due to liver resection, liver insufficiency, and/or sepsis (which were associated with low cholesterol) and obstructive cholestasis (which was associated with high cholesterol). However, in this same whole group, cholesterol was directly correlated with AP and GGT (r2 = 0.25, P b .0001 for both). In the measurements performed only in cases with obstructive or extracellular cholestasis, in which bilirubin increased variably together with AP and GGT, cholesterol was slightly better related to AP and GGT (r2 = 0.26 and 0.27, respectively) than to bilirubin (r2 = 0.21) (P b .0001 for all). In multiple regressions combining AP or GGT with bilirubin, the total r2 rose to 0.40 (P b .0001), which might be consistent with the combined impact of impaired biliary excretion (reflected by bilirubin) and of a partly independent mechanism (reflected by AP or GGT) in increasing cholesterol.
Fig. 2 Relationship between plasma cholesterol and albumin in 1313 measurements on 190 patients. The main distribution with regression line and 95% confidence limits (triangles: cholesterol [mmol/L] = 0.12 [albumin] − 1.26, n = 1046, r2 = 0.38, P b .0001) includes measurements in normal preoperative state, after surgery (liver resection or other major abdominal procedures), in cases with or without complications, and in critically ill patients with sepsis, liver insufficiency, and major bleeding. All conditions fell along the same distribution, except for a downward shift for septic patients (lower cholesterol for any given albumin level, P b .0001). The upward spread for cases with cholestasis (squares) may be explained by the increase in cholesterol related to this condition. See text.
synthetic activity, and hemodilution from blood loss [2,5,7,8,17]. Similar correlations can be shown with other proteins or related measurements, such as transferrin, ironbinding capacity, cholinesterase, prealbumin, retinol-binding
5. Correlation with changes in plasma proteins Although it appears peculiar for a fat substance, cholesterol maintains a relatively strong direct relationship with albumin and several other plasma proteins. Fig. 2 shows it for albumin over a large distribution of observations, ranging from the normal condition to extreme preterminal illness. In the normal condition, the main factor supporting the correlation should be the nutritional state. In postoperative and/or critically ill conditions, the underlying factors are various combinations of acute phase response (surgical stress, sepsis, other acute illnesses), impairment of liver
Fig. 3 Measurements on patients after liver resection. Direct relationship between cholesterol and fibrinogen (a “negative” and a “positive” acute phase reactant, respectively) likely reflecting a similar impact on both variables of substrate limitation plus transient liver dysfunction (although fibrinogen may also be partly influenced by abnormal coagulation). The shift in the distribution for patients with sepsis (triangles, shaded area, dotted regression line) is consistent with strongest priority for fibrinogen synthesis in this condition (regression: cholesterol = 0.11 [fibrinogen] − 0.56[SEPSIS] + 1.51, n = 270, r2 = 0.34, P b .0001). The observations performed before postoperative day 3, excluded to avoid confounding from recent intraoperative bleeding, had a similar distribution.
361.e10 protein, and prothrombin activity (in cases in which it reflects hepatic synthetic activity) [1,17,18]. This was reconfirmed in the locally collected data bank, including prothrombin activity, simultaneously obtained with cholesterol in 899 measurements, and in more inconsistent measurements of cholinesterase and iron-binding capacity: all of them were directly related to cholesterol with r2 of 0.42 for prothrombin activity, 0.50 for cholinesterase, and 0.30 for iron-binding capacity (P b .0001 for all). The main interfering factor in these correlations remains cholestasis, which tends to independently increase cholesterol, as shown in Fig. 2. For “positive” acute phase proteins such as C-reactive protein, the correlation with cholesterol may be the opposite (inverse correlation), as observed in sepsis [16,19,20]. It is not so for fibrinogen after major liver resection, when the competition for substrate to support liver regeneration plus transient liver dysfunction may result in a similar impact on fibrinogen and cholesterol levels (Fig. 3).
6. Correlation with outcomes Transiently severe hypocholesterolemia is not necessarily relevant. It is severe persistent hypocholesterolemia, which is associated with death, whereas increasing cholesterol (unless related to cholestasis) is associated with recovery [2,7-10,12,15,20]. Again, the absence of precise thresholds of abnormality and the multifactoriality of hypocholesterolemia limit the definition of exact landmarks of risk. In a previous assessment in patients with sepsis and organ failure after partial hepatectomy, none survived among those in whom cholesterol remained for more than 6 days less than 58 mg/dL (1.50 mmol/L) or less than 40% of its preoperative value [12], although this cutoff cannot be extensively generalized. An important issue is that hypocholesterolemia may not only passively reflect severity of illness but perhaps also contribute directly to poor outcomes. Hypocholesterolemia could reflect hypolipoproteinemia, with impairment of the host defense against bacterial products provided by circulating lipoproteins. Cytokines in sepsis may impair lipoprotein production and/or facilitate their degradation, with loss of their protective effect further amplifying the process in a vicious circle [21]. This is not totally clear because it could be only the cholesterol content of the lipoproteins that is altered; furthermore, it is the phospholipid component that seems to be more involved in host protection [22-26]. In addition, a reduction in lipoprotein-carried vitamin E, contributing to impaired antioxidant protection, is also likely to be implicated [1,12,14]. Another issue is the reduced availability of cholesterol in a condition of increased requirement. The issue involves both the availability of basic substrate and the demanding biochemical processes of cholesterol synthesis, which cannot be supported at a sufficiently high rate [17,27]. Cholesterol
C. Chiarla et al. requirement increases to support the synthesis of stress hormones and the production and function of new cells taking part in host defense and tissue repair [15,17,27]. Cholesterol is a main component of cell membranes (simply by weight alone), and increased requirement characterizes various cell expansion processes, including liver regeneration after partial hepatectomy and tumor cell expansion in leukemia [2]. These considerations have grounded the concept of the “essentiality” of cholesterol, with a possible need for exogenous supply in septic and critically ill patients [2,27,28]. A similar issue has been raised for tubercular infection [29], and it is possible to quote historical references for this idea [28,29].
7. Hypocholesterolemia and triglyceride levels in trauma and sepsis Plasma triglycerides generally tend to follow a pattern similar to that of cholesterol, although not strictly related to it. In the locally collected data bank, triglycerides were obtained in almost half of the cases and were directly correlated with cholesterol, although with an r2 of only 0.25 (P b .0001). One factor was that cholestasis, if present, affected cholesterol more than triglycerides. Remarkably, however, within this variability, severe hypocholesterolemia became sometimes associated with hypertriglyceridemia: this occurred in particular in septic patients in the stages of greater metabolic decompensation or preterminal illness. Indeed, in sepsis, changes in plasma triglycerides may largely diverge from those of cholesterol. This is an object of controversy, mainly because animal and experimental studies fail to reproduce clinical sepsis [6,25,30]. In human sepsis, there is hypocholesterolemia, and in our patients, we have seen it more often paralleled by normal or low triglyceride levels. As already mentioned, persistently severe hypocholesterolemia is related to poor prognosis. At the same time, it is often perceived that increasing triglycerides with persistently low cholesterol is an even worse landmark of severity of illness and impending risk of death. This apparent landmark of greater metabolic deterioration and inflammatory response or preterminal illness [1,2,4] was repeatedly confirmed in additional patient cases observed after the collection of the local data bank, such as that shown in Fig. 4. Although the poor implications of the development of hypertriglyceridemia in clinical sepsis were already described many years ago [31], the issue has not yet been sufficiently emphasized. The underlying mechanism is unclear: modified cytokine or cytokine receptor patterns, decreased lipoprotein lipase activity and/or increase in hepatic triglyceride production, and peculiarity of the involved microorganism [2,8,17,25,30]. The related hyperlipoproteinemia may also be an extreme attempt at peripheral delivery of substrate or defense [2,32-34]. However, hypertriglyceridemia in sepsis often remains a worrisome
Hypocholesterolemia in surgery and sepsis
361.e11
Fig. 4 Patient with persistent retroperitoneal sepsis after duodenocephalopancreasectomy. The triglyceride peak (without fat infusion) corresponds to worsening of the septic state with rapid evolution into adult respiratory distress syndrome. Improvement after surgical drainage, early recurrence of sepsis, new surgical drainage, reversal of hypocholesterolemia alongside final recovery. In spite of the improvement achieved in this case by aggressively treating sepsis at the time of hypertriglyceridemia, hypertriglyceridemia may unfortunately often be part of a near-preterminal pattern. ARDS indicates adult respiratory distress syndrome.
landmark of extreme severity of illness [1,2,4]. The implication that hypertriglyceridemia limits parenteral fat supply is more obvious. However, at this stage, nutrition might also lose its efficacy, and the best hope of recovery, if this is still achievable, lies in the aggressive eradication of the septic focus. Although generalizations are only partly allowable, similar events have been observed in burns: hypocholesterolemia with a tendency to reversal without major changes in triglycerides in recovering patients or worsening of hypocholesterolemia with preterminally increasing triglycerides in nonsurviving patients [35].
8. The impact of nutrition (parenteral nutrition) Although it may appear counterintuitive, the nutritional state has little to do with hypocholesterolemia after surgery and trauma, in sepsis, and in critical illness because other driving factors prevail. Furthermore, the mechanisms by which the infusion of nutrients can modify cholesterol are not so obvious. For instance, some increase in cholesterol related to the infusion of fat emulsions depends on the release of cholesterol from cell membranes, in exchange with the components of the emulsions (which contain cholesterol only in trace amounts) [2,5,28]. Cholesterol has been reported to increase with increasing dose of amino acid; this needs to be better evaluated in additional studies; however, the underlying mechanism may be the provision of cholesterol precursors from some amino acids (for instance leucine) or the facilitated synthesis of the apoprotein component of lipoproteins [2]. Of course, the fundamental role of amino acid supply in critical illness largely overrides
these speculations, and nutrition provides basic substrate also for cholesterol synthesis, although this is not so evidently reflected in the plasma levels. The issue differs with plasma triglycerides because, as mentioned already, hypertriglyceridemia may be worsened by infusing fat emulsions (including, to a lesser extent, also those vehiculating medications such as propofol).
9. Conclusion After surgery and trauma, in sepsis, and in critical illness, 2 very simple measurements such as cholesterol and triglycerides often become markers of severity of disease and, therefore, helpful adjunctive clinical tools. Obviously, cholesterol and triglyceride levels do not only reflect their concentrations but also altered lipoprotein patterns with modified lipoprotein compositions, and the issue is far too complex and still too poorly understood to be reduced to simple concepts. A better understanding of the underlying processes will likely improve our capability to rescue these desperately ill patients from their condition. At present, the most concrete implication of severe hypocholesterolemia, alone or associated with hypertriglyceridemia, is the need for rapid resolution of the underlying illness.
Acknowledgment This work was supported by a contribution from the Catholic University School of Medicine and the Italian Ministry for University and Scientific Research (D.1 Funds).
361.e12
References [1] Chiarla C, Giovannini I, Siegel JH. Hypotransferrinemia and changes in plasma lipid and metabolic patterns in sepsis. Amino Acids 2009;36: 327-31 [Epub 2008 Apr 8]. [2] Giovannini I, Chiarla C, Giuliante F, et al. Hypocholesterolemia in surgical trauma, sepsis, other acute conditions and critical illness. In: Kramer MA, editor. Trends in cholesterol research. Hauppauge (NY): Nova Science Publishers Inc.; 2005. p. 137-61. [3] Seber GAF. Linear regression analysis. New York: Wiley; 1977. p. 369-82. [4] Giovannini I, Boldrini G, Castagneto M, et al. Respiratory quotient and patterns of substrate utilization in human sepsis and trauma. JPEN J Parenter Enteral Nutr 1983;7:226-30. [5] Sun X, Oberlander D, Huang J, et al. Fluid resuscitation, nutritional support, and cholesterol in critically ill postsurgical patients. J Clin Anesth 1998;10:302-8. [6] Fraunberger P, Schaefer S, Werdan K, et al. Reduction of circulating cholesterol and apolipoprotein levels during sepsis. Clin Chem Lab Med 1999;37:357-62. [7] Giovannini I, Boldrini G, Chiarla C, et al. Pathophysiologic correlates of hypocholesterolemia in critically ill surgical patients. Intensive Care Med 1999;25:748-51. [8] Gordon BR, Parker TS, Levine DM, et al. Relationship of hypolipidemia to cytokine concentrations and outcomes in critically ill surgical patients. Crit Care Med 2001;29:1563-8. [9] Dunham CM, Fealk MH, Sever III WE. Following severe injury, hypocholesterolemia improves with convalescence but persists with organ failure or onset of infection. Crit Care 2003;7:R145-53 [Epub 2003 Oct 1]. [10] Bonville DA, Parker TS, Levine DM, et al. The relationships of hypocholesterolemia to cytokine concentrations and mortality in critically ill patients with systemic inflammatory response syndrome. Surg Infect (Larchmt) 2004;5:39-49. [11] Guimarães SM, Lima EQ, Cipullo JP, et al. Low insulin-like growth factor-1 and hypocholesterolemia as mortality predictors in acute kidney injury in the intensive care unit. Crit Care Med 2008;36:3165-70. [12] Giovannini I, Chiarla C, Greco F, et al. Characterization of biochemical and clinical correlates of hypocholesterolemia after hepatectomy. Clin Chem 2003;49:317-9. [13] Nuzzo G, Giovannini I, Giuliante F, et al. Sepsis after liver resection: predisposition, clinical relevance and synergism with liver dysfunction. In: Dionigi R, editor. Recent advances in liver surgery. Austin (Tex): Landes Bioscience; 2009. p. 245-59. [Epub 2008 March http:// www.eurekah.com/chapter/3870]. [14] Giovannini I, Chiarla C, Giuliante F, et al. Biochemical and clinical correlates of hypouricemia in surgical and critically ill patients. Clin Chem Lab Med 2007;45:1207-10. [15] Marik PE. Dyslipidemia in the critically ill. Crit Care Clin 2006;22: 151-9. [16] Chiarla C, Giovannini I, Siegel JH. The relationship between plasma cholesterol, amino acids and acute phase proteins in sepsis. Amino Acids 2004;27:97-100.
C. Chiarla et al. [17] Hrnciarikova D, Hyspler R, Vyroubal P, et al. Serum lipids and neopterin in urine as new biomarkers of malnutrition and inflammation in the elderly. Nutrition 2009;25:303-8 [Epub 2008 Nov 18]. [18] Lopez-Martinez J, Sanchez-Castilla M, Garcia-de-Lorenzo A. Hypocholesterolemia in critically ill patients. Intensive Care Med 2000;26: 259-60. [19] Bentz MH, Magnette J. Hypocholestérolémie au cours de la phase aiguë de la réaction inflammatoire d'origine infectieuse. À propos de 120 cas. Rev Med Interne 1998;19:168-72. [20] Memiş D, Gursoy O, Tasdogan M, et al. High C-reactive protein and low cholesterol levels are prognostic markers of survival in severe sepsis. J Clin Anesth 2007;19:186-91. [21] Chien JY, Jerng JS, Yu CJ, et al. Low serum level of high-density lipoprotein cholesterol is a poor prognostic factor for severe sepsis. Crit Care Med 2005;33:1688-93. [22] Carpentier YA, Scruel O. Changes in the concentration and composition of plasma lipoproteins during the acute phase response. Curr Opin Clin Nutr Metab Care 2002;5:153-8. [23] van Leeuwen HJ, Heezius EC, Dallinga GM, et al. Lipoprotein metabolism in patients with severe sepsis. Crit Care Med 2003;31: 1359-66. [24] Kitchens RL, Thompson PA, Munford RS, et al. Acute inflammation and infection maintain circulating phospholipid levels and enhance lipopolysaccharide binding to plasma lipoproteins. J Lipid Res 2003; 44:2339-48. [25] Khovidhunkit W, Kim MS, Memon RA, et al. Effect of infection and inflammation on lipid and lipoprotein metabolism: mechanisms and consequences to the host. J Lipid Res 2004;45:1169-96. [26] Gordon BR, Parker TS, Levine DM, et al. Neutralization of endotoxin by a phospholipid emulsion in healthy volunteers. J Infect Dis 2005; 191:1515-22. [27] Zadak Z, Hyspler R, Bakalar B, et al. The role of cholesterol and intermediate metabolites of its synthesis in intensive care and parenteral nutrition. Vnitr Lek 2000;46:776-81. [28] Druml W. Is cholesterol a conditionally essential nutrient in critically ill patients? Wien Klin Wochenschr 2003;115:740-2. [29] Pérez-Guzmán C, Vargas MH. Hypocholesterolemia: a major risk factor for developing pulmonary tuberculosis? Med Hypotheses 2006; 66:1227-30. [30] Malmendier CL, Lontie JF, Bubois DY. Mechanisms of hypocholesterolemia. In: Malmendier CL, Alaupovic P, Bryan Brewer Jr H, editors. Hypercholesterolemia, hypocholesterolemia, hypertriglyceridemia. New York (NY): Plenum Press; 1990. p. 173-82. [31] Siegel JH, Cerra FB, Coleman B, et al. Physiological and metabolic correlations in human sepsis. Surgery 1979;86:163-93. [32] Samra JS, Summers LK, Frayn KN. Sepsis and fat metabolism. Br J Surg 1996;83:1186-96. [33] Barcia AM, Harris HW. Triglyceride-rich lipoproteins as agents of innate immunity. Clin Infect Dis 2005;41(S7):S498-S503. [34] Spitzer AL, Harris HW. Statins attenuate sepsis. Surgery 2006;139: 283-7. [35] Kamolz LP, Andel H, Mittlböck M, et al. Serum cholesterol and triglycerides: potential role in mortality prediction. Burns 2003;29: 810-5.