Characterization of alpha-amino-n-butyric acid correlations in sepsis

ORIGINAL ARTICLES Characterization of alpha-amino-n-butyric acid correlations in sepsis CARLO CHIARLA, IVO GIOVANNINI, and JOHN H. SIEGEL ROME, ITALY; AND NEWARK, NJ

Little information is available on the patterns of changes and significance of plasma alpha-amino-n-butyric acid (ABA, mmol/L) in various conditions, particularly in sepsis. This study has been performed to assess the patterns of correlation among ABA, other amino acids, and other variables in a group of septic patients with various degrees of illness. More than 400 determinations of ABA, other amino acids, and simultaneously collected blood variables were obtained in 17 patients with sepsis. The distribution of ABA was characterized by the clustering of most measurements within the normal range (,41 mmol/L), with the spreading of abnormally increased values up to 151 mmol/L. Abnormal increases in ABA were related directly to alanine, serine, tyrosine, histidine, proline, threonine, glycine, glutamine, cysteine, lysine, cystathionine, leucine, valine, phenylalanine, arginine, and citrulline (r2 from 0.86 to 0.32) and related inversely to aspartate, taurine, and phosphoethanolamine (r2 from 0.62 to 0.50) (P , 0.001 for all). Furthermore, increased ABA was correlated with increasing total aminoacidemia, lactate, neutrophil concentration, creatinine, ammonia, osmolarity, glucose, and bilirubin and with decreasing AA Fischer ratio and peripheral O2 extraction (r2 from 0.87 to 0.16) (P , 0.001 for all). High ABA was also associated with low cholesterol, taurine, and platelet count, and with high 3-methylhistidine (partly anticipating the increase), high blood urea nitrogen, and pulmonary shunt (P , 0.001 for all). Finally, high ABA was related to the worsening of sepsis-related organ failure assessment score (SOFA score) and of most plasma AA clearances (P , 0.001 for all). Abnormally increased ABA may signal and partly anticipate the transition to an extreme derangement of septic metabolic patterns, characterized by the worsening of protein hypercatabolism with hyperaminoacidemia and by signs of impaired hepatic amino acid metabolism and oxidative metabolism. Increased ABA may represent an additional landmark of transition to extreme illness, compelling the need for the aggressive resolution of sepsis. (Translational Research 2011;158:328–333) Abbreviations: 3MH ¼ 3-methylhistidine; AA ¼ amino acid; ABA ¼ alpha amino-n-butyric acid; MODS ¼ multiple organ dysfunction syndrome; SOFA ¼ sequential organ failure assessment

From the Department of Surgical Sciences, CNR-IASI Center for the Pathophysiology of Shock, Catholic University of the Sacred Heart School of Medicine, Rome, Italy; New Jersey Medical School, UMDNJ, Newark, NJ. Submitted for publication March 21, 2011; revision submitted June 14, 2011; accepted for publication June 14, 2011. Reprint requests: Carlo Chiarla, MD, CNR-IASI Center for Pathophysiology of Shock, Catholic University School of Medicine, Largo

328

Agostino Gemelli 8, I-00168 Rome, Italy; e-mail: carlo.chiarla@rm. unicatt.it. 1931-5244/$ - see front matter Ó 2011 Mosby, Inc. All rights reserved. doi:10.1016/j.trsl.2011.06.005

Translational Research Volume 158, Number 6

AT A GLANCE COMMENTARY Chiarla C, et al. Background

This article is an extensive analytical work on alpha-n-amino butyric acid (ABA) in septic patients, with an evaluation of patterns of correlations with severity of the septic state and a broad series of different parameters. Relatively little is known about ABA and no studies address ABA correlations in the clinical setting, particularly in critically ill septic patients. Translational Significance

Apart from the originality of the assessed correlations, our study suggests potentially relevant clinical implications, providing evidence that increased ABA may signal (and partly anticipate) the transition to an extreme derangement of septic metabolic patterns.

Alpha-amino-n-butyric acid (ABA) is an amino acid (AA) that does not take part in protein synthesis. It is considered primarily a product of the metabolism of methionine, threonine, serine, and glycine, finally deriving from alpha-ketobutyrate.1-4 The alternative fate of alpha-ketobutyrate is decarboxylation to propionate, with subsequent formation of propionyl-CoA, succinyl-CoA, and entry into the Krebs cycle. Increased ABA is generally considered a nonspecific marker of liver dysfunction, malnutrition, increased protein catabolism, or a combination of these. At least to our knowledge, no studies describe the patterns of correlation of ABA with other AAs and with metabolic and clinical variables in sepsis, and our study was performed to assess this specific aspect. METHODS

The assessment was based on the analysis of 435 determinations of ABA and of a large series of complementary variables, which was performed prospectively on 17 patients who developed sepsis after trauma. The patients included 13 men and 4 women; the age was 30 6 15 years (mean 6 standard deviation [SD]), the weight was 71.7 6 13.9 kg, and the height was 173 6 7 cm. They had a combination of abdominal, chest, and head injuries, and the subsequent cause of sepsis was intra-abdominal, pulmonary, or extensive soft-tissue infection. The diagnosis of sepsis was based on the simultaneous occurrence of a temperature .38.3 C, white blood cell count .12 3 109/L or ,3 3 109/L, and clear evidence of infection confirmed by positive cultures from blood, surgical drainage of infected

Chiarla et al

329

areas, or sputum in the case of pulmonary sepsis.5 The median sepsis-related organ failure assessment (SOFA) score for all measurements was 7 (range, 2–16).6 Two patients died from sepsis, and the other patients survived. Serial determinations were performed every 8 to 12 h until criteria for the diagnosis of sepsis persisted or death occurred. All patients underwent total parenteral nutrition at 34 6 14 kcal/kg/day (75% glucose and 25% fat) and 1.5 6 0.6 g/kg/day amino acids. Each determination included the measurement of plasma ABA (normal value ,41 mmol/L), the full amino-acidogram with estimation of AA clearances (alanine, arginine, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tyrosine and valine clearances),7 and of glucose, creatinine, 3-methylhistidine (3MH), albumin, cholesterol, triglycerides, ammonia, lactate, blood cell count, arterial blood gases, and other necessary variables with obtainment of the SOFA score. The Fischer plasma molar amino acid ratio (leucine 1 isoleucine 1 valine)/(phenylalanine 1 tyrosine)8 was calculated. Blood urea nitrogen (BUN), plasma osmolarity, peripheral O2 extraction, and pulmonary shunt were obtained less consistently. The percent pulmonary shunt (the sum of pulmonary shunt and venous admixture components) was assessed based on the arterial and mixed venous blood gas data, according to a common application of the Sackur and Berggren method.9-11 The instrument used for the analysis of amino acids and derivatives was a Beckman 6300 high-performance amino acid analyzer (Beckman Instruments, Inc., Palo Alto, Calif) designed for ion-exchange separation followed by postcolumn ninhydrin derivatization. This instrument operates at a subnanomole sensitivity, with a detection limit for each component of 50–100 pmoles.12-14 The obtained determinations provided a continuous distribution of observations, which was well suited to assessing the correlates of ABA over a wide range of pathophysiologic abnormalities, from moderate to extreme septic illness. The statistical analysis was performed by least-square regression and covariance analysis, evaluating confidence intervals of regression coefficients and Mallows’ Cp criteria15 to assess the hierarchy of correlations and the regressions yielding the best control of variability. Regressions with a coefficient of determination (r2) lower than 0.15 were not taken into account. The Statgraphics Plus (Manugistics, Rockville, Md) and a noncommercial software program were used. The protocol complied with the Helsinki declaration of 1975 as revised in 1996 and was approved by the Institutional Review Board. RESULTS

The distribution of ABA was characterized by the clustering of most measurements within the normal

330

Chiarla et al

range (,41 mmol/L) with the spreading of abnormally increased values up to 151 mmol/L. The main patterns associated with normal and abnormally increased ABA are shown in Table I. Graphic displays and regression analysis showed apparently strong overall correlations between ABA and several other variables. In reality, within the normal ABA range, most variables were unrelated or weakly related to it. The exceptions were direct relationships with alanine (r2 5 0.55), proline (0.50), lactate (0.34), total aminoacidemia (0.33), and arginine, valine, serine, leucine, lysine, histidine, threonine, and asparagine (r2 between 0.25 and 0.15) (P , 0.001 for all). At abnormally high ABA (.41 mmol/L), however, several variables became strongly related to it. Direct correlations were found with alanine (r2 5 0.86), serine (0.81), tyrosine (0.81), histidine (0.78), proline (0.76), threonine (0.76), glycine (0.73), glutamine (0.68), cysteine (0.66), lysine (0.64), cystathionine (0.58), leucine (0.50), valine (0.49), phenylalanine (0.37), arginine (0.34), and citrulline (0.32), as well as inverse relationships with aspartate (0.62), taurine (0.51), and phosphoethanolamine (0.50) (P , 0.001 for all). Furthermore, increases in ABA above 41 mmol/L were correlated with increasing total aminoacidemia (r2 5 0.87), lactate (r2 5 0.76), neutrophil concentration (0.59), creatinine (0.49), ammonia (0.46), osmolarity (0.31), glucose (0.27), and bilirubin (0.18), and with decreasing AA Fischer ratio (0.42) and peripheral O2 extraction (0.16). Increases in ABA were also associated with low cholesterol and taurine (Table I), with the maintenance of lower taurine for any given phosphoethanolamine level (r2 5 0.85), and with decreasing platelet count and increasing BUN and pulmonary shunt (P , 0.001 for all). The covariation of ABA with citrulline was explained by the simultaneous relationship of both variables with increases in creatinine or BUN (r2 between 0.50 and 0.40, P , 0.001 for all). The percent pulmonary shunt was 50.2 6 5.4 (n 5 20) for the cases with ABA . 41 mmol/L and 43.8 6 6.9 (n 5 216) for the remainder. Finally, increases in ABA above 41 mmol/L were related significantly to increasing SOFA score (with r2 values up to 0.80 in individual patient measurements) and to decreases in all the calculated plasma AA clearances (P , 0.001 for all) except for isoleucine, as reflected previously by changes in the corresponding plasma AA levels. Conversely, within normal ABA values, a looser relationship was found with the SOFA score. When assessing all measurements pooled together, the apparently poor correlation between ABA and methionine was explained by independent changes in

Translational Research December 2011

methionine, which were related directly to changes in 3MH. Indeed, among all the AAs, methionine was the best correlate of 3MH (r2 5 0.39, P , 0.001) and when both 3MH and ABA were included as independent variables in an overall regression on methionine, the r2 rose from 0.39 to 0.66 (P , 0.001). This is shown in Table II, regression 1, by accounting also for the impact of renal dysfunction, if present. A time pattern analysis further showed that changes in ABA often preceded by 24–48 h similar changes (increases or decreases) in 3MH, and the correlation between ABA and 3MH improved significantly when accounting for this time shift (P , 0.001), implicating some predictive value of increasing ABA with regard to the severity of muscle hypercatabolism. The relationship between high ABA and hypercatabolism was substantiated by the maintenance of higher BUN for any given creatinine level in cases with high ABA (r2 5 0.70, P , 0.001). The relationship between ABA and the branched chain and aromatic AAs was also assessed in detail. Both leucine and valine were related directly to ABA and increased together with it, whereas isoleucine did not. Leucine and valine were interrelated tightly and linearly (r2 5 0.92, P , 0.001), with a valine:leucine ratio of 1.97 6 0.19 (mean 6 SD), whereas their correlation with isoleucine was less tight (r2 5 0.59 and 0.61, respectively, P , 0.001 for both). This was because leucine and valine sometimes diverged from a linear relationship with isoleucine, spreading toward relatively higher values. The leucine–isoleucine difference was related strongly to increasing ABA (r2 5 0.85, P , 0.001) (regression 2, Table II) and a similar finding regarded the valine–isoleucine difference. The best fit when substituting other AAs for ABA in these regressions was obtained with phenylalanine and tyrosine (r2 5 0.88, P , 0.001) (regression 3, Table II). The highest leucine–isoleucine and valine–isoleucine differences were associated with high ABA, high total aminoacidemia, high lactate, and worsening of hyperglycemia and SOFA score, as well as decreasing AA clearances, Fischer ratio, and platelet concentration (P , 0.001 for all). Finally, with regard to mortality, the highest ABA values (151 and 106 mmol/L, respectively) were observed in a patient who died rapidly of multiple organ dysfunction syndrome (MODS) and in a patient who also developed severe MODS but slowly recovered after the peaking of ABA. Another patient died of MODS, however with normal ABA (25 mmol/L); she had mild respiratory insufficiency, selective liver dysfunction (rapidly worsening nonobstructive hyperbilirubinemia), and other evolving abnormalities (hypocholesterolemia, hypertriglyceridemia, hyperlactatemia and final hypotension),

Translational Research Volume 158, Number 6

Chiarla et al

331

Table I. Plasma AAs and other relevant variables

ABA Alanine Arginine Asparagine Aspartate Citrulline Cysteine Cystathionine Glutamate Glutamine Glycine Histidine Hydroxyproline Isoleucine Leucine Lysine Methionine Phosphoethanolamine Phenylalanine Phosphoserine Proline Serine Taurine Threonine Tryptophan Tyrosine Valine Fischer ratio Prothrombin time Cholesterol 3-methylhistidine Lactate Creatinine SOFA score

ABA # 41

ABA . 41

12.9 6 6.5; 12 297.1 6 105.1; 274 94.3 6 32.2; 90 46.0 6 28.4; 39 9.0 6 8.8; 7.0 10.9 6 4.0; 10 47.4 6 14.7; 45 5.4 6 4.3; 4 76.2 6 55.6; 61 452.0 6 88.7; 444 259.2 6 95.5; 235 73.5 6 16.8; 71 12.4 6 6.1; 11 72.3 6 26.4; 64 125.7 6 32.6; 116 176.1 6 55.5; 168 49.0 6 45.4; 39 11.6 6 9.0; 9 121.0 6 34.7; 114 12.3 6 4.2, 12 225.2 6 161.7; 184 99.3 6 30.2; 100 99.3 6 76.9; 77 114.7 6 49.2; 104 55.5 6 13.9; 55 59.2 6 17.1; 56 247.1 6 69.6; 225 2.9 6 1.3; 2.5 11.9 6 1.3; 11.5 98.6 6 34.4; 93 6.3 6 5.4; 4 14.0 6 7.2; 12.1 1.0 6 0.8, 0.7 7.1 6 2.9; 7

86.6 6 29.2; 83.5 834.8 6 366.4; 808.5 197.4 6 49.3; 193.5 266.4 6 120.8; 240.5 13.1 6 7.9; 10.5 59.1 6 22.5; 57 84.0 6 40.2; 74.5 33.2 6 37.3; 20 53.7 6 62.5; 38.5 1744.8 6 575.5; 1773 808.4 6 270.5; 814 476.0 6 323.9; 359.5 44.5 6 24.2; 42 94.3 6 20.6; 96.5 242.6 6 61.2; 241 372.5 6 92.1; 358.5 205.6 6 114.0; 170 9.1 6 6.6; 7 286.4 6 122.0; 260 24.9 6 13.3, 21.5 1011.2 6 324.1; 1045.5 239.3 6 107.3; 222.5 36.1 6 30.5; 27.5 383.1 6 172.0; 365.5 56.9 6 10.5; 54.5 154.6 6 65.5; 127.5 478.6 6 156.8; 438 1.9 6 0.3; 1.9 12.9 6 1.1; 13 41.3 6 17.1; 33 17.0 6 9.9; 14 29.5 6 4.9; 27.9 1.8 6 0.9; 1.7 13.1 6 2.0; 13

Notes: Mean 6 standard deviation and median values for plasma AAs and other relevant variables (normal range for ABA: ,41 mmol/L; 407 measurements had ABA # 41 and 28 measurements had ABA . 41). Prothrombin time expressed in sec; cholesterol, lactate and creatinine in mg/dL.

however, without relevant signs of hypercatabolism or hyperaminoacidemia. DISCUSSION

These results shed some light on still unknown features of ABA in sepsis, providing an insight into its clinical and biochemical correlations. As mentioned previously, ABA derives from the metabolism of methionine, threonine, serine, and glycine, finally deriving from alpha-ketobutyrate through transamination. The alternative fate of alpha-ketobutyrate is decarboxylation to propionate with formation of propionyl-CoA, succinyl-CoA, and entry into the Krebs cycle.1-4 In our study, increasing ABA was associated with a progressively greater metabolic decompensation, signaling and partly anticipating massive protein catabolism. Evidence of impaired

oxidative metabolism with increased lactate was found, suggesting that impaired entry of alphaketobutyrate into the Krebs cycle contributed to increasing ABA. The simultaneously developing hyperaminoacidemia likely depended on protein hypercatabolism with enhanced intravascular release of endogenous AAs and on liver dysfunction. In fact, apart from increased citrulline (likely related to renal dysfunction),16 the largest increases in AAs regarded system transport A amino acids and the most abundant AAs of muscle,17 as well as those metabolized predominantly in the liver such as phenylalanine, tyrosine, methionine, cystathionine, and proline.18-20 With regard to the branched chain AAs, this was also consistent with the smaller increase in isoleucine compared with leucine and valine, which are more abundant in muscle protein.17 The largest differences in these AAs were associated with patterns

332

Translational Research December 2011

Chiarla et al

Table II. Regressions 1. Methionine 5 1.3 (ABA) 1 6.8 (3MH) – 23.2 (creatinine) 1 13.3 2. Leucine 5 1.1(isoleucine) 11.2 (ABA) 1 31.0 3. Leucine 5 1.1(isoleucine) 1 0.5 (tyrosine) 1 0.2 (phenylalanine) – 8.0

r2 5 0.71 r2 5 0.85 r2 5 0.88

Notes: Overall regressions. The coefficient for creatinine (mg/dL) corrects for the effect of renal dysfunction, if present, on plasma 3MH level. Methionine, leucine, isoleucine, tyrosine, and phenylalanine expressed in mmol/L. Total n 5 435. P , 0.001 for each regression and each coefficient in the regressions. See text.

similar to those associated with high ABA. Regression 3 in Table II provides another view of the same phenomenon, with increasing tyrosine and phenylalanine reflecting protein hypercatabolism and liver dysfunction.18-20 In the past increased ABA was initially considered as a marker of liver dysfunction21,22 and then of alcoholic liver disease. This was long an object of debate regarding both ABA and the ABA/leucine ratio (used to account for nutritional status). It was found that increased ABA and ABA/leucine could not discriminate among acute or chronic alcoholic or nonalcoholic liver disease, and it should be considered as nonspecific markers of hepatocellular dysfunction.2,23-26 Regardless of the uncertainties that characterized the debate, this was not relevant to our study. None of our patients had evidence of alcohol toxicity, and a major impact of this factor was excluded by initially low ABA levels that subsequently increased even above 41 mmol/L with worsening of the clinical condition. Moreover, in most cases with ABA . 41 mmol/L, the ABA and ABA/ leucine values were higher than those considered in previous studies. In our study, the quantification of the observed phenomena by overall regressions including all measurements (Table II) indicates that we are dealing with continuous distributions and sequences of septic metabolic changes, ranging from balanced to totally disrupted patterns. Therefore, the sharp cut-off established at ABA 5 41 mmol/L was not totally correct, but it was only a convenient means to highlight the features associated with normal or high ABA. The features generally associated with high ABA, apart from the already mentioned abnormalities, included other signs of MODS (liver dysfunction with low Fischer ratio and high ammonia, pulmonary and kidney dysfunction, and increased SOFA score), less controllable hyperglycemia, and severe decreases in cholesterol and taurine (Table I), which implicate impaired antioxidant protection.27,28 The occurrence of moderately high cysteine and low taurine could reflect insufficient flux of cysteine to support taurine requirements because an

important role of cysteine in this condition is taurine and glutathione production to preserve antioxidant protection.29-32 This state of extreme illness, in which cholesterol and taurine decrease concomitantly, obviously differs from that of stable subjects in whom a reduction in cholesterol may be amenable to increased taurine disposal from exogenous supply.33 In the development of the abnormalities that we observed in our patients, changes in ABA often preceded those of 3MH, implicating some predictive value with regard to a further extreme worsening of sepsis, with patterns which were previously described as a picture of preterminal illness.18,20 Indeed, the more severe increases in ABA (151 and 106 mmol/L, respectively) were observed in a patient who died rapidly of MODS and in a patient who also developed severe MODS but slowly recovered from desperate illness. Another death occurred in a patient with normal ABA. Although she had fewer organ dysfunctions, she had worsening nonobstructive hyperbilirubinemia, other metabolic abnormalities that were typical of severe sepsis,28 and final hypotension, without signs of massive hypercatabolism or hyperaminoacidemia. Therefore, although increased ABA and the associated patterns do seem to signal extreme illness, low ABA in itself does not exclude lethal sepsis. This finding is consistent with the concept that, within the patterns of septic death, there may be a variable expression of features which may prevail or remain relatively undisclosed in different patients. Increased ABA seems to be associated mostly with hypercatabolism and hyperaminoacidemia, and its determination may be useful because the cut-off for abnormal values of other AAs is less easily identifiable. Finally, there is to note that our analytical procedure did not allow to distinguish between D and L amino acids. Although bacteria may have a role in the production of D amino acids, to our knowledge their contribution to plasma amino acid levels, and in particular to ABA levels, has never been assessed in clinical sepsis, which might be an issue for future studies. In conclusion, our findings provide an original insight into the biochemical and clinical correlations of changes in ABA in sepsis. The data suggest that increased ABA may signal (and partly anticipate) the transition to an extreme derangement of septic metabolic patterns, which is characterized by the progressive worsening of protein hypercatabolism, with hyperaminoacidemia, signs of impaired hepatic AA metabolism, and impaired oxidative metabolism. The clinical relevance of high ABA should be assessed in more patients because it might represent an additional landmark of transition to extreme illness, compelling the need for the aggressive resolution of sepsis. This would help also to make definitive conclusions

Translational Research Volume 158, Number 6

on the relationship with mortality, which was not allowed by the limited number of deaths in our patients. REFERENCES

1. McMenamy RH, Vang J, Drapanas T. Amino acid and alpha-keto acid concentrations in plasma and blood of the liverless dog. Am J Physiol 1965;209:1046–52. 2. Yudkoff M, Blazer-Yost B, Cohn R, Segal S. On the clinical significance of the plasma alpha-amino-n-butyric acid:leucine ratio. Am J Clin Nutr 1979;32:282–5. 3. Darling PB, Grunow J, Rafii M, et al. Threonine dehydrogenase is a minor degradative pathway of threonine catabolism in adult humans. Am J Physiol Endocrinol Metab 2000;278:E877–84. 4. Lieber CS. Pathogenesis and treatment of alcoholic liver disease: progress over the last 50 years. Rocz Akad Med Bialymst 2005; 50:7–20. 5. Chiarla C, Giovannini I, Giuliante F, et al. Plasma bilirubin correlations in non-obstructive cholestasis after partial hepatectomy. Clin Chem Lab Med 2008;46:1598–601. 6. Vincent JL. Organ dysfunction in patients with severe sepsis. Surg Infect (Larchmt) 2006;7(suppl 2):S69–72. 7. Clowes GH Jr, Heideman M, Lindberg B, et al. Effects of parenteral alimentation on amino acid metabolism in septic patients. Surgery 1980;88:531–43. 8. Freund H, Atamian S, Holroyde J, Fischer JE. Plasma amino acids as predictors of the severity and outcome of sepsis. Ann Surg 1979;190:571–6. 9. Sackur P. Weiteres zur Lehre von Pneumothorax. Virchows Arch Pathol Anat Physiol 1897;150:151–60. 10. Berggren SM. The oxygen deficit of arterial blood caused by nonventilating parts of the lung. Acta Physiol Scand 1942;4(suppl 11):l–92. 11. Siegel JH, Giovannini I, Coleman B. Ventilation:perfusion maldistribution secondary to the hyperdynamic cardiovascular state as the major cause of increased pulmonary shunting in human sepsis. J Trauma 1979;19:432–60. 12. Arrizon-Lopez V, Biehler R, Cummings J, Harbaugh J. Beckman System 6300 high-performance amino acid analyzer. In: Mant CT, Hodges RS, eds. High-performance liquid chromatography of peptides and proteins: separation, analysis, and conformation. Boca Raton, Fla: CRC Press, 1991:859–63. 13. Mant CT, Zhou NE, Hodges RS. Amino acids and peptides. In: Heftmann E, ed. Chromatography, 5th ed. Part B. Amsterdam, the Netherlands: Elsevier, 2005:B75–150. 14. Tyler MI. Amino acid analysis: an overview. In: Cooper C, Packer N, Williams K, eds. Methods in molecular biology, vol. 159: Amino acid analysis protocols. Totowa, NJ: Humana Press Inc, 2000:1–7. 15. Seber GAF. Linear regression analysis. New York: Wiley, 1977. 369–82.

Chiarla et al

333

16. Nakakariya M, Shima Y, Shirasaka Y, et al. Organic anion transporter OAT1 is involved in renal handling of citrulline. Am J Physiol Renal Physiol 2009;297:F71–9. 17. Matthews DE. Proteins and amino acids. In: Shils ME, Olson JA, Shike M, Ross AC, eds. Modern nutrition in health and disease, 9th ed. Baltimore, Md: Williams and Wilkins, 1999:11–48. 18. Siegel JH, Cerra FB, Coleman B, et al. Physiological and metabolic correlations in human sepsis. Surgery 1979;86: 163–93. 19. Cerra FB, Caprioli J, Siegel JH, McMenamy RR, Border JR. Proline metabolism in sepsis, cirrhosis and general surgery. The peripheral energy deficit. Ann Surg 1979;190:577–86. 20. Cerra FB, Siegel JH, Coleman B, Border JR, McMenamy RR. Septic autocannibalism. A failure of exogenous nutritional support. Ann Surg 1980;192:570–80. 21. Wu C, Bollmann JR, Butt HR. Changes in free amino acids in the plasma during hepatic coma. J Clin Invest 1955;34:845–9. 22. Iyer GY. Plasma alpha-amino-n-butyric acid in health and disease. Curr Sci 1956;25:224–5. 23. Morgan MY, Milsom JP, Sherlock S. Ratio of plasma alpha amino-n-butyric acid to leucine as an empirical marker of alcoholism: diagnostic value. Science 1977;197:1183–5. 24. Shaw S, Lieber CS. Plasma amino acid abnormalities in the alcoholic. Respective role of alcohol, nutrition, and liver injury. Gastroenterology 1978;74:677–82. 25. Shaw S, Lieber CS. Increased hepatic production of alpha-aminon-butyric acid after chronic alcohol consumption in rats and baboons. Gastroenterology 1980;78:108–13. 26. Musshoff F, Daldrup T. Determination of biological markers for alcohol abuse. J Chromatog B Biomed Sci Appl 1998;713: 245–64. 27. Giovannini I, Chiarla C, Greco F, Boldrini G, Nuzzo G. Characterization of biochemical and clinical correlates of hypocholesterolemia after hepatectomy. Clin Chem 2003;49:317–9. 28. Chiarla C, Giovannini I, Giuliante F, et al. Severe hypocholesterolemia in surgical patients, sepsis, and critical illness. J Crit Care 2010;25:361.e7–12. 29. Malmezat T, Breuille D, Pouyet C, et al. Methionine transsulfuration is increased during sepsis in rats. Am J Physiol Endocrinol Metab 2000;279:E1391–7. 30. Santangelo F. The regulation of sulphurated amino acid junctions. Fact or fiction in the field of inflammation? Amino Acids 2002;23: 359–65. 31. Grimble RF. Nutritional antioxidants and the modulation of inflammation: theory and practice. New Horizons 1994;2:175–85. 32. Biolo G, Antonione R, De Cicco M. Glutathione metabolism in sepsis. Crit Care Med 2007;35:S591–5. 33. Lombardini JB, Militante JD. Effects of taurine supplementation on cholesterol levels with potential ramification in atherosclerosis. Adv Exp Med Biol 2006;583:251–4.