Glutamine and Antioxidants in Critically Ill Patients

The

n e w e ng l a n d j o u r na l

of

m e dic i n e

Glutamine and Antioxidants in Critically Ill Patients To the Editor: Heyland et al. (April 18 issue)1 have shown that glutamine supplementation (approximately 65 g daily) of a diet that does not supply adequate energy (910 kcal) or protein (45 g) did not affect clinical outcome and increased mortality. However, the addition of a supplement that constituted 60% of total dietary protein introduced an amino acid imbalance with the potential for toxicity. Other, similar examples, such as the lack of arginine as part of dietary protein2 or the use of essential amino acids only3 in otherwise complete diets, have been shown to produce hyperammonemia in humans. Furthermore, glutamine itself is known to exacerbate defects in ammonia metabolism.4 Low-protein diets that are also low in sulfur-containing amino acids alter methionine metabolism, increasing homocysteine levels, which may affect the long-term health of human infants.5 Thus, manipulation of the amino acid composition of the diet, whether by eliminating one (arginine) or a group (all nonessential amino acids) or providing inadequate amounts of key amino acids (e.g., methionine or cysteine), or by grossly distorting the balance (such that glutamine constitutes more than 50%), may have adverse clinical consequences, despite a seemingly modest difference in the amounts of the various amino acids. Bruce R. Bistrian, M.D., Ph.D.

To the Editor: The publication of the trial Reducing Deaths Due to Oxidative Stress (REDOXS), reported by Heyland et al., leads us to make two suggestions for critical care physicians. First, we suggest that outcomes of meta-analyses should not be used in clinical practice guidelines1 because innumerable confounders are introduced when many studies are lumped together. The glutamine saga is an excellent example of the harm inflicted by the positive outcomes of metaanalyses.2,3 Second, those in the field of intensive care medicine are often disappointed by trials of interventions that are initially reported as positive but that are later shown to be ineffective.4 The most important thing we can do to improve the outcome for our patients is to do the basic things well — for instance, by providing adequate resuscitation and antibiotics and carrying out source control in sepsis. Aside from improving patient care by conducting clinical trials, clinicians should be focusing their time and energy on improving the basics of patient care rather than on tweaking the margins with fancy supplements. It is the mark of a good clinician to follow the true evidence after it is available and not before. Michael G.G. Rodgers, M.D. Matijs van Meurs, M.D., Ph.D. Jan G. Zijlstra, M.D., Ph.D.

Beth Israel Deaconess Medical Center Boston, MA [email protected]

University of Groningen Groningen, the Netherlands [email protected] No potential conflict of interest relevant to this letter was reported.

Dr. Bistrian reports receiving consulting fees from Nestle, which makes enteral products for the critically ill. No other potential conflict of interest relevant to this letter was reported. 1. Heyland D, Muscedere J, Wischmeyer PE, et al. A random-

ized trial of glutamine and antioxidants in critically ill patients. N Engl J Med 2013;368:1489-97. [Erratum, N Engl J Med 2013;368:1853.] 2. Kapila S, Saba M, Lin CH, Bawle EV. Arginine deficiencyinduced hyperammonemia in a home total parenteral nutritiondependent patient: a case report. JPEN J Parenter Enteral Nutr 2001;25:286-8. 3. Lee JC, Lai HS, Huang SM, Chang CJ, Wang ST, Chen WJ. Hyperammonemic encephalopathy due to essential amino acid hyperalimentation. J Formos Med Assoc 1994;93:486-91. 4. Bachmann C. Mechanisms of hyperammonemia. Clin Chem Lab Med 2002;40:653-62. 5. Rees WD. Manipulating the sulfur amino acid content of the early diet and its implications for long-term health. Proc Nutr Soc 2002;61:71-7. DOI: 10.1056/NEJMc1306658

482

1. McClave SA, Martindale RG, Vanek VW, et al. Guidelines for

the provision and assessment of nutrition support therapy in the adult critically ill patient. JPEN J Parenter Enteral Nutr 2009;33:277-316. 2. Manzanares W, Dhaliwal R, Jiang X, Murch L, Heyland DK. Antioxidant micronutrients in the critically ill: a systematic review and meta-analysis. Crit Care 2012;16:R66. 3. Novak F, Heyland DK, Avenell A, Drover JW, Su X. Glutamine supplementation in serious illness: a systematic review of the evidence. Crit Care Med 2002;30:2022-9. 4. Ranieri VM, Thompson BT, Barie PS, et al. Drotrecogin alfa (activated) in adults with septic shock. N Engl J Med 2012;366: 2055-64. DOI: 10.1056/NEJMc1306658

To the Editor: Heyland et al. provide evidence that glutamine supplementation increases mor-

n engl j med 369;5 nejm.org august 1, 2013

The New England Journal of Medicine Downloaded from nejm.org on September 14, 2015. For personal use only. No other uses without permission. Copyright © 2013 Massachusetts Medical Society. All rights reserved.

correspondence

Trial

Year

Glutamine

Control

Proportion of Weighting

Effect Size (95% CI)

no. of deaths/total no. of patients

%

Single center Beale RJ 2008 Cai G 2008 Carrol PV 2003 Çekmen N 2011 Duska F 2008 Eroglu A 2009 Fuentes-Orozco C 2004 Fuentes-Orozco C 2008 Garrel D 2003 Goeters C 2002 Griffiths R 2002 Hall JC 2003 Houdijk APJ 1998 Jensen JL 1996 Jones C 1999 Kumar S 2007 Luo M 2008 McQuiggian M 2008 Pérez-Bárcena J 2008 Pérez-Bárcena J 2010 Schneider A 2011 Tjader I 2004 Wischmeyer PE 2001 Spindler-Vesel A 2007 Subtotal (I-squared=0.0%, P=0.80)

10/27 17/55 1/12 3/15 2/20 1/20 2/17 2/22 2/19 11/46 18/42 27/179 4/41 1/14 12/26 8/63 1/29 0/10 3/15 4/23 7/30 11/30 2/15 1/32

8/28 20/55 0/7 6/15 0/10 1/20 3/16 5/22 12/22 21/49 28/42 30/184 3/39 1/14 10/24 5/57 0/15 2/10 0/15 3/20 7/30 4/10 5/16 6/81

1.30 (0.60–2.79) 0.85 (0.50–1.44) 1.17 (0.04–30.52) 0.50 (0.15–1.64) 2.00 (0.10–40.35) 1.00 (0.07–14.90) 0.63 (0.12–3.28) 0.40 (0.09–1.85) 0.19 (0.05–0.76) 0.56 (0.30–1.03) 0.64 (0.43–0.97) 0.93 (0.57–1.49) 1.27 (0.30–5.31) 1.00 (0.07–14.45) 1.11 (0.59–2.08) 1.45 (0.50–4.17) 1.03 (0.04–29.11) 0.25 (0.01–4.88) 6.00 (0.33–109.82) 1.16 (0.29–4.57) 1.00 (0.40–2.50) 0.92 (0.38–2.24) 0.43 (0.10–1.88) 0.42 (0.05–3.37) 0.80 (0.66–0.96)

1.47 3.08 0.08 0.61 0.09 0.12 0.31 0.37 0.46 2.32 5.11 3.76 0.42 0.12 2.16 0.77 0.08 0.10 0.10 0.46 1.02 1.08 0.39 0.20 24.66

Multicenter 2011 Andrews PJD 2002 Conejero R 2006 Dechelotte P 2011 Grau T 2013 Heyland D 2011 Wernerman J 1999 Powell-Tuck J Subtotal (I-squared=7.6%, P=0.37)

115/250 14/47 16/58 16/59 259/613 14/205 10/17

106/252 9/37 9/56 23/68 218/610 20/208 9/25

1.09 (0.90–1.33) 1.22 (0.60–2.51) 1.72 (0.83–3.56) 0.80 (0.47–1.37) 1.18 (1.03–1.36) 0.71 (0.37–1.37) 1.63 (0.85–3.15) 1.14 (1.03–1.27)

21.97 1.66 1.61 3.00 43.11 2.00 1.99 75.34

Heterogeneity between groups: P=0.001 Overall (I-squared=12.4%, P=0.272)

100.00 1.05 (0.95–1.15) 0.00911

1

Glutamine Better

110

Control Better

Figure 1. Forest Plot of 31 Trials Showing the Effect of Glutamine Supplementation on Survival. The shaded gray boxes represent the weight given to each study, and the horizontal lines are an index of dispersion.

tality in critically ill patients. We have now completed a meta-analysis of 31 randomized trials in which glutamine supplementation was used to treat critically ill patients after nonelective surgery and for which mortality was reported (Fig. 1). Glutamine supplementation is associated with a significant increase in mortality when the trials considered are limited to the 7 multicenter,

n engl j med 369;5

randomized trials: 35% (434 of 1232 patients) for those receiving glutamine versus 31% (385 of 1231 patients) for controls (P = 0.015 for the comparison). In the 24 single-center trials, there was a significant decrease in mortality: 20% (160 of 819) for those receiving glutamine versus 23% (189 of 826 patients) for controls (P = 0.019 for the comparison). No significant difference in

nejm.org

august 1, 2013


483

The

n e w e ng l a n d j o u r na l

of

m e dic i n e

mortality was observed in the overall population: Nikki Buijs, M.D. 29% (594 of 2051) for those receiving glutamine Mechteld A.R. Vermeulen, M.D. and 28% (574 of 2057) for controls (P = 0.34 for Paul A.M. van Leeuwen, M.D., Ph.D. the comparison). Therefore, the contrasting re- VU University Medical Center the Netherlands sults should be attributed to a single-center study Amsterdam, [email protected] 1 bias ; the adequately powered study by Heyland Dr. van Leeuwen reports receiving consulting fees from Freseet al. has finally proven that glutamine supple- nius Medical Care and participating in the development of a mentation increases mortality in critically ill pa- patented method of treating disorders of the animal or human body by administering amino acids (US6001878A, 1995). He retients. ceives no royalties or other payments for the use of this patent, which is now held by Nutricia. No other potential conflict of Laura Pasin, M.D. interest relevant to this letter was reported. Giovanni Landoni, M.D. 1. Rama Rao KV, Jayakumar KR, Norenberg MD. Glutamine in Alberto Zangrillo, M.D. San Raffaele Scientific Institute Milan, Italy [email protected] No potential conflict of interest relevant to this letter was reported. 1. Bellomo R, Warrillow SJ, Reade MC. Why we should be wary

of single-center trials. Crit Care Med 2009;37:3114-9.

DOI: 10.1056/NEJMc1306658

To the Editor: Heyland et al. report a trend toward higher 28-day mortality in critically ill patients who received glutamine. However, we have major concerns about many aspects of the study, including the inadequate nutritional prescription and the statistical adjustment in which the glutamine groups are combined, showing an imbalance in baseline variables. The number of patients with more than two failing organs at baseline was much higher in the two groups receiving glutamine than in the two groups not receiving glutamine (187 vs. 148), which obviously resulted in higher mortality. Neither 6-month mortality nor in-hospital mortality was predefined as a primary or secondary end point. The reported 6-month mortality is trivial and not properly documented, since many patients were lost to follow-up. In addition, no 6-month mortality significance is reported for the original groups. Sepsis-related Organ Failure Assessment (SOFA) scores were not provided for separate organs (including the liver), which would have allowed for comparisons. In all, we suggest that more severely ill patients were allocated to the glutamine groups as a result of randomization error and believe that patients were not adequately fed, which may explain the observed higher mortality in the groups receiving glutamine. Complementary data are needed to support the scientific value of this study.

484

the pathogenesis of acute hepatic encephalopathy. Neurochem Int 2012;61:575-80.

DOI: 10.1056/NEJMc1306658

The Authors Reply: We agree with Bistrian that high-dose glutamine appeared to cause harm but are uncertain as to the mechanism. Ammonia levels were not determined, since patients with severe liver dysfunction at baseline were excluded from this study. In a subset of 69 patients who underwent randomization, there was no significant difference between plasma amino acid levels in the groups receiving glutamine and the groups not receiving glutamine; however, the former had slightly higher levels of glutamine and alanine (which was part of the intravenous supplement) and citrulline (a metabolite of glutamine). All levels were nonetheless within normal ranges. We disagree with the suggestion from Rod gers et al. that meta-analyses should not be used to inform clinical practice guidelines. While we agree that meta-analyses require cautious interpretation, we contend that well-conducted and carefully interpreted systematic reviews provide the highest level of evidence available.1 Furthermore, meta-analyses provide new insights by exploring the presence of heterogeneity among trials. We do not see the results of our study as a “failure of meta-analyses.” The patients enrolled in our trial were quite dissimilar to the patients in earlier studies of glutamine supplementation that were included in the meta-analyses (e.g., our study included patients with renal failure, a group that was excluded in previous past trials of glutamine). We share the concern of Pasin et al. that single-center trials can provide optimistic estimates of treatment effects. However, we question the



correspondence

appropriateness of including our trial in the same group as the other randomized glutamine trials performed in the critical care setting, given differences in patient population, administration route, dosage, and approach to feeding. Buijs et al. express several concerns that we believe are overstated. First, we attempted to optimize nutrition delivery through the application of evidence-informed strategies. Our success with feeding is probably consistent with the results in most intensive care units worldwide. The overall degree of imbalance in baseline characteristics is consistent with expected random variation and did not affect our conclusions. Our prespecified analysis plan, which followed best practices for factorial, randomized, controlled trials, did not adjust for the number of organ failures.2-5 Nevertheless, when we adjust for important baseline characteristics (including organ failures), the adjusted odds ratio of 28-day mortality for glutamine as compared with placebo remains consistent with our primary analysis, at 1.4 (95% confidence interval, 1.0 to 2.1; P = 0.05). Finally, our secondary 6-month mortality results were based on survival analysis techniques that remain valid in the presence of random loss to follow-up. Besides, almost the entire treatment effect was observed

within the first 30 days, when there was almost no loss to follow-up. Daren Heyland, M.D. Kingston General Hospital Kingston, ON, Canada [email protected]

Paul E. Wischmeyer, M.D. University of Colorado Aurora, CO

Andrew G. Day, M.Sc. Kingston General Hospital Kingston, ON, Canada

for the Canadian Clinical Care Trials Group Since publication of their article, the authors report no further potential conflict of interest. 1. The 2011 Oxford CEBM levels of evidence. Oxford, United

Kingdom: Oxford Centre for Evidence-Based Medicine, 2013 (http://www.cebm.net/index.aspx?o=5653). 2. Heyland DK, Dhaliwal R, Day AG, et al. REducing Deaths due to OXidative Stress (the REDOXS Study): rationale and study design for a randomized trial of glutamine and antioxidant supplementation in critically ill patients. Proc Nutr Soc 2006; 65:250-63. 3. Raab GM, Day S, Sales J. How to select covariates to include in the analysis of a clinical trial. Control Clin Trials 2000;21:33042. 4. McAlister FA, Straus SE, Sackett DL, Altman DG. Analyzing and reporting of factorial trials: a systematic review. JAMA 2003; 289:2545-53. 5. Montgomery AA, Peters TJ, Little P. Design, analysis, and presentation of factorial randomised controlled trials. BMC Med Res Methodol 2003;3:26. DOI: 10.1056/NEJMc1306658

Weight Loss in Persons with Serious Mental Illness To the Editor: Daumit and colleagues (April 25 issue)1 describe the results of the Randomized Trial of Achieving Healthy Lifestyles in Psychiatric Rehabilitation (ACHIEVE), and they report that an intensive behavioral weight-loss intervention significantly reduced weight in adults with severe mental illness. This comprehensive intervention used existing rehabilitation-program staff in community-based psychiatric rehabilitation settings. Unfortunately, the 2011 Centers for Medicare and Medicaid Services (CMS) policy for weightloss counseling2 would not cover this effective intervention. The CMS policy covers only brief weight-loss counseling delivered by primary care physicians in primary care settings, not the comprehensive counseling by behavioral providers in a

rehabilitation setting that Daumit and colleagues found effective. Current CMS policy will have little effect on health disparities because populations that are disproportionately affected by obesity often use the services of community-based organizations. Further, unlike the offices of most primary care doctors, community clinics employ behavioral counselors and typically have facilities (e.g., kitchens and recreation rooms) that are conducive to behavioral weight-loss programs. The study by Daumit et al. and other community-based obesity trials3 provide support for the case for CMS to cover a wider range of settings and providers. This is not only consistent with the evidence but is also the most feasible approach to reducing weight in underserved populations.



485