DOI: /01.CCM A9

Transcription

1 Early enteral supplementation with key pharmaconutrients improves Sequential Organ Failure Assessment score in critically ill patients with sepsis: Outcome of a randomized, controlled, double-blind trial* Richard J. Beale, MB, BS; Tony Sherry, RN; Katie Lei, RN, MSc; Laura Campbell-Stephen, RN; Julie McCook, RN, MSc; John Smith, RN; Werner Venetz, PhD; Birgit Alteheld, PhD; Peter Stehle, PhD; Heinz Schneider, PhD Objective: To assess the safety and efficacy of an early enteral pharmaconutrition supplement containing glutamine dipeptides, antioxidative vitamins and trace elements, and butyrate in critically ill, septic patients. Design: A prospective, randomized, controlled, double-blind clinical trial. Setting: Adult intensive care unit in a university hospital. Patients: Fifty-five critically ill, septic patients requiring enteral feeding. Interventions: Patients received either an enteral supplement (500 ml of Intestamin, Fresenius Kabi) containing conditionally essential nutrients or a control solution via the nasogastric route for up to 10 days. Inclusion occurred within 24 hrs of intensive care unit admission. Additionally, patients received enteral feeding with an immunonutrition formula (experimental group) or standard formula (control group) initiated within 48 hrs after enrollment. Measurements and Main Results: Organ dysfunction was assessed by daily total Sequential Organ Failure Assessment (SOFA) score over the 10-day study period in both patient groups. Patients receiving the experimental supplement showed a significantly faster decline in the regression slopes of delta daily total SOFA score over time compared with control. The difference between the regression coefficients of the two slopes was significant irrespective of the level of analysis: intent to treat 0.32 vs. 0.14, p <.0001; per protocol 0.34 vs. 0.14, p <.0001; and completers (patients receiving >80% of the calculated caloric target over a period of 6 days), 0.26 vs. 0.16, p Vitamin C, as a marker of supplement absorption, increased from 10.6 ( ) mol/l (normal range mol/l) on day 1 to 58.7 ( ) mol/l by day 3 (p.002) in the intervention group but remained below the normal range in the control group 17.0 ( ) on day 1 and 14.3 ( ) on day 3. Serum levels of glycine, serine, arginine, ornithine, vitamin E, and -carotene all increased significantly with treatment in the supplementation group. Conclusions: In medical patients with sepsis, early enteral pharmaconutrition with glutamine dipeptides, vitamin C and E, -carotene, selenium, zinc, and butyrate in combination with an immunonutrition formula results in significantly faster recovery of organ function compared with control. (Crit Care Med 2008; 36: ) KEY WORDS: sepsis; early feeding; immune-modulation; organ dysfunction; Sequential Organ Failure Assessment; glutamine; antioxidants; enteral nutrition; enteral pharmaconutrition *See also p From the Department of Adult Critical Care Medicine, Guy s and St. Thomas Hospital, London, United Kingdom (RJB, TS, KL, LCS, JM, JS); Datagen AG, Basel, Switzerland (WV); Department of Nutrition & Food Sciences, University of Bonn, Germany (BA, PS); and HealthEcon AG, Basel, Switzerland (HS). Supported, in part, by a research grant from Fresenius Kabi AG, Germany, which continues to provide research support to Dr. Beale and the Department of Adult Critical Care Medicine at Guy s and St Thomas Hospital. For information regarding this article, richard.beale@gstt.nhs.uk Copyright 2007 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: /01.CCM A9 Nutritional support is vital for critically ill patients, and important developments include early enteral nutrition, immunonutrition, postpyloric feeding, and an increasingly scientific, evidencebased approach. However, there are still many unanswered questions, and new insights about specific nutrients or nutrient combinations are tantalizing rather than definite. Clinical trials of enteral nutrition in the critically ill are difficult, given the size of trial necessary to minimize population heterogeneity and demonstrate efficacy in severe sepsis and acute lung injury (1, 2). A specific problem is the inability to deliver feed effectively in a proportion of subjects due to variations in enteral tolerance with patient and disease type. High-quality clinical care, employing feeding protocols, avoiding excessive sedation, and using aperients and prokinetics early, substantially minimizes the problem, but fullvolume enteral feeding is still difficult or impossible to achieve early in some intensive care unit (ICU) patients (3). An important motive for feeding critically ill patients early is that the gutassociated lymphoid tissue comprises the largest collection of immune cells in the body, and their turnover and the vulnerability of the gut to ischemia/reperfusion Crit Care Med 2008 Vol. 36, No

2 injury in shock states (4, 5) offer potential points of intervention in critical illness. Unfortunately, key components of gut surface defense are impaired by standard interventions, including antibiotics, which modify normal gut microflora (6); dopamine, which impairs gut motility (7); and hemodynamic and respiratory support strategies, which may impair gut perfusion (8, 9). Luminal delivery of key substrates as nutrients for enterocytes and as substances to modify the injury response is therefore an attractive concept (10). This approach was introduced with the immunonutrition products that have been studied in the last 10 yrs. Other than in the planned major surgery population (11), however, immuneenhancing feeds remain controversial, especially arginine-containing feeds in severe sepsis (12 14). Moreover, these formulas contain fixed nutrient doses, so adequate delivery requires fully established enteral feeding, which excludes effective early dosing except with prefeeding before planned major surgery, where the strongest benefit is seen (11). These practical difficulties, and the importance of the gut as a potential point of intervention, suggest scope for alternative approaches. One of these is pharmaconutrition, which involves the crucially different step of separating the provision of key nutrients from routine nutritional support by administering a low-volume enteral supplement containing high doses of the key nutrients as early as possible, to increase the likelihood of effective delivery even in patients with limited enteral volume tolerance. Standard enteral or parenteral feed can then be given separately as required. Antioxidant vitamins and trace elements, glutamine, and short-chain fatty acids, which all modulate gut ischemia-reperfusion injury and mucosal immune cell turnover and function, have therefore been combined into the pharmaconutrition enteral supplement Intestamin (Fresenius Kabi, Bad Homburg, Germany). Another important consideration is the optimal trial end point. Mortality and morbidity assessments require large study populations and have proven largely unachievable on intent-to-treat analyses in feeding studies. Length of stay also requires a large population and is affected by organizational factors beyond the influence of the primary intervention. Descriptive organ failure scores that relate to mortality and morbidity end points (via their derivation databases) therefore offer an attractive alternative. The most widely used is the Sequential Organ Failure Assessment (SOFA) (15 17), recommended for this purpose by a recent European roundtable (18). We have therefore tested the effects of an enteral supplement containing conditionally essential nutrients, together with an immune-enhancing feed, on the pattern of change in SOFA score in critically ill patients with sepsis over a 10-day period. MATERIALS AND METHODS This was a prospective, randomized, controlled, double-blind, single-center study performed in the adult general ICU of Guy s and St. Thomas Hospital, London, UK. After Ethics Committee approval of the protocol, written consent was obtained from each patient or next of kin, following procedures according to the Helsinki Declaration. Enrolled in the study were patients with a high probability or signs of infection, two criteria for systemic inflammatory response syndrome, and at least one organ dysfunction: a) pulmonary dysfunction, PaO 2 /FIO mm Hg, or 200 mm Hg with pneumonia or other localized lung disease; b) metabolic acidosis, ph 7.30 or base deficit 5.0 meq/l ( 5.0 mmol/l) with plasma lactate concentration 2.0 mmol/l; c) oliguria, urine output 0.5 ml/kg/hr for two consecutive hours after adequate fluid resuscitation; d) thrombocytopenia, unexplained thrombocytopenia (platelet count 100,000 cells/mm 3 ); e) hypotension, dopamine 5 g/ kg/min or any dose of epinephrine or norepinephrine to maintain systolic blood pressure 90 mm Hg for two consecutive hours occurring within 24 hrs of ICU admission. Additional inclusion criteria were age 18, Acute Physiology and Chronic Health Evaluation (APACHE) II score 10, precipitating injury (surgery, trauma, hypovolemia, episode of infection/sepsis) within 24 hrs preceding ICU Intestamin control supplement day 1 Intestamin / Reconvan / Glucose 20% Control supplement / Fresubin original/ Glucose 20% 20% entry, expected length of stay in the ICU 3 days, and likely indication for enteral nutrition for 5 days. Major exclusion criteria were cardiogenic shock or severe congestive heart failure (New York Heart Association class IV); severe, preexisting, parenchymal liver disease with clinically significant portal hypertension (Childs C); documented chronic obstructive pulmonary disease; chronic renal failure requiring dialysis; pregnancy; acquired immunodeficiency syndrome; and immunosuppression (chronic treatment with high-dose steroids, e.g., 0.3 mg/kg/day methyl prednisolone, active radiotherapy or chemotherapy within preceding year, or, at any time, lymphoma or documented immunohumoral or cellular immune deficiency state). Patients were randomized to receive either the enteral pharmaconutrition supplement (Intestamin) or a control supplement (Fig. 1, feed compositions shown in Table 1) within 24 hrs after enrollment at 21 ml/hr (500 ml/day; 250 kcal) via a nasogastric tube, for up to 10 days. From day 2 patients also received a complete enteral formula according to the ICU s feeding regimen. Patients randomized to Intestamin received immunonutrient-containing Reconvan (Fresenius Kabi, Bad Homburg, Germany Table 1); patients randomized to control supplement received Fresubin original (also Fresenius Kabi), starting at 20 ml/hr and increasing to 25 kcal/kg within 2 3 days. To offset any initial caloric deficit and to allow tight blood sugar control ( mg/l or mmol/l), patients received intravenous 20% glucose until their feeding target was met. Routine laboratory measurements were monitored daily. SOFA and APACHE II scores were assessed on the day of admission (day 0) or the next day (day 1) according to the published methodology (15, 19), and daily thereafter. The worst measurement values within each 24-hr period (from 8 am to 8 am) were retrieved from the patient data management system and entered into an electronic case Reconvan day 1 day 2-10 day 11 - discharge Figure 1. Study schedule. day 2-10 Fresubin original day 11 - discharge 132 Crit Care Med 2008 Vol. 36, No. 1

3 Table 1. Product composition Intestamin 500 ml Control Supplement 500 ml Intestamin Reconvan 500 ml 1000 ml Control Supplement Fresubin Original 500 ml 1000 ml Energy kcal kj Caloric density, kcal/ml Caloric distribution, % Protein Carbohydrate Fat Osmolarity, mosm/l Protein, g Glutamine Arginine Glycine Total nitrogen, g Fat, g 1 (as tributyrin) Tributyrin 1 1 EPA DHA n6/n3 fatty acids 2:1 2:1 Carbohydrate Dietary fiber Key vitamins and minerals -carotene, mg Vitamin E, mg TE Vitamin C, mg Sodium, mg Potassium, mg Zinc, mg Selenium, g EPA, eicosapentaenoic acid; DHA, docosahexaenoic acid; Mg TE, total equivalent. Other vitamins and minerals Other vitamins and minerals report form. Random weekly checks were performed to eliminate data errors, and inconsistencies were documented and corrected. Organ failure was calculated for all six SOFA components during the study period of (a maximum) 11 consecutive days (as of day 0). The aggregate score, total maximum SOFA, was calculated by summing the worst scores for each organ system (15). ICU/hospital length of stay and ICU/hospital mortality were assessed at ICU/hospital discharge. The 28-day mortality and 6-month mortality were assessed by separate follow-up. Gastrointestinal tolerance was assessed daily (aspiration, nausea/vomiting, hiccups, bloating, flatulence, constipation, diarrhea, abdominal cramping, bowel movements). Nutrient absorption was assessed by measurement of plasma levels of glutamine, vitamins C and E, -carotene, zinc, and selenium. Plasma amino acids were determined by reverse-phase high-performance liquid chromatography and fluorescence detection (20). Vitamins C and E and -carotene were analyzed by highperformance liquid chromatography (21). Immunologic profiling was undertaken through solid-phase enzyme-labeled chemiluminescent immunometric assay of interleukin (IL)-6 (IMMULITE 2000-IL-6, EURO/DPC Ltd, Caernarfon, Gwynedd, UK) and IL-10 (IMMULITE 1000 IL-10, EURO/DPC Ltd, Caernarfon, Gwynedd, UK) levels. Monocyte human leukocyte antigen (HLA)-DR expression was assessed using two-color flow cytometry (Beckman Coulter, High Wycombe, UK) to generate a mean fluorescence intensity value. Antioxidative status was assessed by measurement of thiobarbituric acid reactive substance (TBARS) (22), thiols, and glutathione (23), all on predefined study days. Safety was evaluated via (serious) adverse event collection and standard laboratory and clinical assessments. Statistics. The primary end point was organ dysfunction evolution assessed by daily total SOFA and delta daily total SOFA score over a study period of maximum 10 days. Our prospectively defined primary analysis included all randomized patients who received 0 ml of the supplement, with patients analyzed according to treatment group allocation (intent to treat). The trial was designed to enroll 344 patients, with one planned interim analysis after enrolling 50 patients to confirm statistical assumptions and for safety monitoring. Statistical analysis was performed independently by HealthEcon AG/Datagen AG (Basel, Switzerland). The primary analysis was based on the evolution (slopes) of the daily total SOFA score and delta daily total SOFA score (change in daily total SOFA score compared with day 0) over 10 study days by comparing the regression coefficients using Student s t-test. Simple Student s t-tests were also used for the difference between mean values of parametric data. Repeated-measures analysis of variance was used to evaluate treatment-related effects. Chi-Square and Fisher s exact test were used for comparison of frequency data between groups. All reported p values are two-sided, and values.05 are considered as significant; there are no corrections for multiple comparisons. Analyses were also performed on two predefined subgroups: the per protocol group, which excluded patients who were randomized but met a study exclusion criteria; and the 80% group, comprising patients who received 80% of the calculated cumulative daily caloric target over six consecutive days. RESULTS After the interim analysis of data from 50 patients, enrollment was suspended because the difference in the decline of the delta daily total SOFA score was statistically significant with a benefit in favor of the patients in the Intestamin/Reconvan group. Crit Care Med 2008 Vol. 36, No

4 Table 2. Baseline characteristics of intent-to-treat patient population Characteristic Results presented here include data from an additional five patients enrolled during the interim analysis period. Of 55 patients who underwent randomization, 27 received Intestamin/ Reconvan and 28 the control feeds. Five patients in the Intestamin/Reconvan group and four patients in the control group met at least one exclusion criteria following randomization and were ruled dropouts. Seventeen Intestamin/Reconvan patients and 18 control patients received 80% of the calculated caloric Intestamin Reconvan Control Supplement Fresubin Original Patients, n Male gender, % Mean (SD) age, yrs 57.4 (19.0) 64.3 (16.8) Mean (SD) weight, kg 71.7 (13.7) 73.8 (14.6) Mean (SD) BMI 25.3 (3.9) 26.1 (4.2) Precipitating injury, % Medical 16 (59) 16 (57) Surgical 5 (19) 6 (22) Trauma 6 (21) 6 (22) APACHE II score Mean (SD) 14.0 (5.5) 16.0 (5.6) Median (range) 15.0 (4 25) 16.0 (6 30) SOFA score on day 0 Mean (SD) Median (range) 6.5 (2 13) 6 (2 15) Requiring mechanical ventilation, n (%) 26 (96) 28 (100) Requiring renal replacement therapy during 16 (60) 17 (61) study period, n (%) Serum creatinine, mol/l Serum bilirubin, mol/l Patients with SIRS, n (%) 27 (100) 28 (100) Patients with sepsis, n (%) 27 (100) 28 (100) BMI, body mass index; APACHE, Acute Physiology and Chronic Health Evaluation; SOFA, Sequential Organ Failure Assessment; SIRS, systemic inflammatory response syndrome. Table 3. Feed delivery information on a per-patient basis Characteristic Intestamin Reconvan Control Supplement Fresubin Original Numbers Intent to treat Per protocol Completers Total supplement received per patient, mean SD in ml Intent to treat Per protocol Completers Total enteral formula received per patient, mean SD in ml Intent to treat Per protocol Completers Total 20% dextrose received per patient, mean SD in ml Intent to treat Per protocol Completers target over six consecutive days, forming the 80% group. The baseline demographic characteristics and severity of disease as evaluated by the APACHE II and total SOFA scores were similar in both groups (Table 2). Supplement and Feed Delivery. Supplement, feed, and calorie delivery is shown in Table 3. On day 1, both groups received a median of 336 ml (range ml) of supplement (Intestamin or Control). From days 2 to 9, both groups received the planned 500 ml/day as median amount on all days. On day 10, the median volume was 420 ml (range ml) in the control group and 500 ml (range ml) in the experimental group. Complete enteral formulas (Reconvan or Fresubin original) were introduced on day 2. The median Reconvan volume on day 2 was 840 ml (range ml), increasing to between 1320 and 1480 ml/day (range ml) for the remainder of the period to day 10; the median Fresubin original volume on day 2 was 835 ml (range ml), increasing to between 960 and 1320 ml/day (range ml) for the remainder of the period to day 10. The median 20% dextrose volume dropped from 823 ml (range ) in the control and 855 ml (range ml) in the experimental group on day 1 to 545 ml (range ) in the control and 570 ml (range ml) in the experimental group on day 2 and to 0 ml on day 3 in both groups, indicating strict adherence to the feeding regimen and good feed tolerance. There were no differences in median daily blood sugar or insulin levels (Fig. 2). Only minor gastrointestinal intolerance was reported during the 10-day study period for both groups, with only the incidence of diarrhea being statistically significantly different (experimental group 21 patients vs. control group 12 patients, p.036). No subject had to discontinue enteral feeding due to adverse effects. Absorption of Nutrients. Table 4 shows the serum levels of vitamins C and E and -carotene, and all three showed significant treatment-related increases during the study (p.002, p.001, and p.019, respectively). In the experimental group, median vitamin C blood levels, determined on days 1, 3, 5, 7, 9, and 11, increased from below normal (20 50 mol/l) on day 1 to above normal on all other days but remained below normal in the control group. Differences were statistically significant from day 3 both between groups and from baseline in the treatment group (p.01). Median vitamin E values were below normal (20 40 mol/l) on day 1 in both groups and increased to above normal in the experimental group but not in the control group. Differences were statistically significant between the groups and from baseline in the treatment group from day 3(p.001). Median levels of -carotene were below the normal range ( mol/l) in both groups at baseline and 134 Crit Care Med 2008 Vol. 36, No. 1

5 Figure 2. Daily blood glucose levels and insulin infusion dosages. Max, maximum; Min, minimum; IU, interational units. increased to above normal in the experimental group but not in the control group. The increase from baseline with treatment was significant from day 3, and the difference between groups became significant by day 6 (both p.05). No statistically significant treatmentrelated differences were found for selenium and zinc. Median values were below the normal ranges ( g/l and mol/l, respectively) in the experimental group (45 [11 77] g/l and 5.3 Crit Care Med 2008 Vol. 36, No

6 Table 4. Comparison of vitamin and trace element serum levels Study Day Intestamin Reconvan, Median (Minimum Maximum) Control Supplement Fresubin Original, Median (Minimum Maximum) t-test p Value ANOVA Repeated Measures Vitamin C (normal range mol/l).002 (treatment effect) ( ) ( ).116 vs. day ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) Vitamin E (normal range mol/l).001 (treatment effect) ( ) ( ).771 vs. day ( ) ( ) ( ) ( ) ( ) ( ) Carotene (normal range mol).019 (treatment effect) ( ) 0.11 ( ).449 vs. day ( ) 0.18 ( ) ( ) 0.25 ( ) ( ) 0.13 ( ) Se (normal range g/l).375 (treatment effect) 1 45 (11 77) 48 (10 103) (18 83) 54 (16 117) (33 96) 53 (21 82) (40 112) 64 (27 93).336 Zn (normal range mol/l).641 (treatment effect) ( ) 5.7 ( ) ( ) 6.7 ( ) ( ) 7.2 ( ) ( ) 9.5 ( ).955 ANOVA, analysis of variance. [ ] mol/l) and in the control group (48 [10 103] g/l and 5.7 [ ] mol/l) on the first study day (Table 4). Levels of each trace element increased during the study period, with selenium levels normalizing in the experimental (66 [40 112] g/l) and control (64 [27 93] g/l) groups, but zinc levels remained slightly below normal (9.8 [ ] mol/l and 9.5 [ ] mol/l). Median values of glutamate, glutamine, serine, glycine, arginine, -amino-butyrate, and ornithine showed significant increases compared with control from day 3 (Table 5), but this was only statistically related to treatment (by analysis of variance) for glycine, serine, arginine (all on day 3) and ornithine (throughout the study period). Those amino acids (glutamine, serine, and arginine) below the normal range at study entry showed consistent normalization only in the treatment group. Values for alanine, taurine, asparagine, histidine, threonine, citrulline, tyrosine, valine, methionine, tryptophane, phenylalanine, isoleucine, leucine, and lysine were normal on all days without significant differences between the groups. Antioxidant Status. There were no significant treatment-related differences between the groups for the thiols cysteine, -glutamyl-cysteine, homocysteine, cysteinyl-glycine, and glutathione or for TBARS. Effect on Mortality, Length of Stay, and Organ Dysfunction. There were no significant differences in mortality at ICU or hospital discharge, 28 days, and 6 months between the two study groups or in length of ICU and hospital stay (Table 6). There were small differences in organ dysfunction between the groups measured by the initial SOFA scores (day 0/day 1). After normalization of the data using the delta total SOFA score to correct for these initial imbalances, there was a more rapid fall in delta daily total SOFA over the 10-day study period (Fig. 3), with statistically significant differences in the coefficients of the regression lines for the Intestamin/Reconvan group compared with control in all three study populations: intent to treat, per protocol, and the 80% group completers (Table 7). When a best-fit approach was taken to the delta daily total SOFA score regression analysis, there was a biphasic relationship, describing a steeper fall between days 0 and 5 and then a more shallow change thereafter (days 6 11). The first phase fall was more rapid in the Intestamin/Reconvan group than in the control group (Fig. 4), and this was highly significant for all analytical groups (Table 7). There were no organ-specific effects on analysis of the individual components of the total SOFA score. There was a trend toward lower delta SOFA values (total maximum SOFA score minus total SOFA score on admission, which represents the amount of organ dysfunction/failure occurring during the ICU stay) in the intervention group in all three study populations: intent to treat, experimental vs control (p.1824); per protocol, vs (p.2458); and completers, vs (p.6367). Immunologic Data. Immunologic data are given in Table 8. IL-6 levels were raised in both groups at study entry and fell over time. In the control group, the IL-6 level rose again at day 7, when it was significantly higher than in the treatment group (p.035), following which the level fell once more, but statistically this could not be confirmed as a treatment effect. IL-10 levels were at or above the top of the normal range on study entry and did not change significantly. There were wide interindividual variations in the levels of both cytokines. HLA-DR ex- 136 Crit Care Med 2008 Vol. 36, No. 1

7 Table 5. Comparison of selected serum amino acid levels Study Day Intestamin Reconvan, Median (Minimum Maximum) Control Supplement Fresubin Original, Median (Minimum Maximum) t-test p Value ANOVA Repeated Measures Glutamate (normal range, 7 39 mol/l).615 (treatment effect) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ).125 Glutamine (normal range mol/l).137 (treatment effect) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ).686 Glycine (normal range mol/l).004 (treatment effect) ( ) ( ).600 vs. day ( ) ( ) ( ) ( ) ( ) ( ) Serine (normal range mol/l).017 (treatment effect) ( ) ( ).268 vs. day ( ) ( ) ( ) ( ) ( ) ( ) Arginine (normal range mol/l).020 (treatment effect) ( ) ( ).281 vs. day ( ) ( ) ( ) ( ) ( ) ( ) amino-butyrate (normal range.680 (treatment effect) mol/l) ( ) ( ) ( ) 9.70 ( ) ( ) ( ) ( ) ( ).788 Ornithine (normal range mol/l).019 (treatment effect) ( ) ( ).232 vs. day ( ) ( ) <.001 < ( ) ( ) ( ) ( ) ANOVA, analysis of variance. Table 6. Comparison of secondary outcome parameters: intensive care unit (ICU) and hospital length of stay (LOS); secondary infections; ICU, hospital, 28-day, and 6-month mortality Characteristic pression (as mean fluorescence intensity) was significantly reduced in both groups (Intestamin/Reconvan, day 1 median Intestamin Reconvan Control Supplement Fresubin Original ICU LOS, mean SD Intent to treat Per protocol Completers Hospital LOS, mean SD Intent to treat Per protocol Completers Total no. of secondary infections Total no. of secondary respiratory infections 13 9 Total no. of secondary bacteremias 11 7 ICU mortality (ITT), n (%) 6 (21) 4 (15) Hospital mortality (ITT), n (%) 7 (25) 7 (26) 28-day mortality (ITT), n (%) 5 (18) 3 (11) 6-month mortality (ITT), n (%) 10 (36) 8 (30) ITT, intent to treat. No significant differences for any comparisons. Crit Care Med 2008 Vol. 36, No , range ; control, day 1 median 0.94, range ) compared with 25 normal control subjects (median 4.11, range ) (p.0001 vs. experimental and control groups) and remained depressed throughout the study period (Intestamin/Reconvan, day 11 median 0.60, range ; control, day 11 median 0.82, range ), but there were no differences between the two groups. Secondary Infections, Complications, and Safety. There were no significant differences in the number of secondary infections or complications across any of the populations analyzed (Table 6). There were six serious adverse events in six patients, with no significant difference between the groups (Intestamin/Reconvan, two patients; control, four patients). The only biochemical variable that was different between the two groups was serum urea, which rose from mmol/l on day 1 to mmol/l on day 6 in the Intestamin/Reconvan group (p.045) and fell in the control 137

8 Figure 3. Delta total Sequential Organ Failure Assessment (SOFA) score over time (study days 0 11). ITT, intention to treat; PPP, per protocol; ICU, intensive care unit. 138 Crit Care Med 2008 Vol. 36, No. 1

9 Table 7. Comparison of regression analysis from delta daily total Sequential Organ Failure Assessment (SOFA) score a Characteristic group from mmol/l to mmol/l (p.264). DISCUSSION Intestamin Reconvan Control Supplement Fresubin Original p Value (Slopes) Slopes (days 0 11) Intent to treat Per protocol Completers Slopes (1 phase days 0 5) Intent to treat Per protocol Completers Slopes (2 phase days 6 11) Intent to treat Per protocol Completers a Normalized via delta total SOFA score to correct for initial imbalances between the two groups in the total SOFA score upon admission (day 0/day 1). Crit Care Med 2008 Vol. 36, No. 1 This is the first randomized trial of the enteral pharmaconutrition concept in patients with sepsis, and it demonstrates the validity of this concept in this predominantly medical ICU population. Glutamine dipeptides, antioxidant vitamins and trace elements, and tributyrin, in a low-volume supplement (Intestamin), were successfully absorbed via the nasogastric route and well tolerated. Administration in combination with an immune-enhancing feed (Reconvan) resulted in a significantly faster improvement in organ dysfunction, as assessed by the normalized delta daily total SOFA score. The innovation in this approach is in administering high doses of key nutrients in a low-volume enteral supplement that is separate from the provision of general nutrition. This is conceptually different from traditional enteral nutrition, including immunonutrition, since the full volume of a supplement can be administered much earlier in the patient s ICU stay, soon after initial resuscitation has been performed. Later, as gastrointestinal tolerance improves, standard enteral feeding can be added to provide normal nutritional support or parenteral feeding can be used if necessary. We used nasogastric feeding successfully in all study patients, confirmed by the rapid increases in vitamin levels by study day 3, although nasojejunal tubes were allowed within the protocol. This is encouraging and further emphasizes the importance of a protocolized approach in achieving successful feeding (24, 25). Since mortality is not a feasible end point in a small single-center study, especially in patients with sepsis, we used the SOFA score as a surrogate outcome measure, which was its primary purpose (14 16). Moreover, since the therapeutic concept is to provide key nutrients to the gut to modulate the injury response, encourage recovery, and minimize secondary organ failure, using a daily measure of organ dysfunction is an attractive approach. Given the slight imbalance in the starting SOFA score between the two groups, the data were normalized by using the delta total daily SOFA score: that is, comparing the degree of change on a daily basis to baseline. This showed a strong, consistent difference in the evolution of organ dysfunction favoring the Intestamin/Reconvan group, although this was an overall phenomenon rather than a specific individual organ system effect. Although we combined the pharmaconutrition supplement Intestamin with the immune-enhancing feed Reconvan, the fact that significant increases in key nutrients and amino acids were seen by day 3, allied with the biphasic delta SOFA response, suggests that the effect of the Intestamin predominated, especially given the lack of convincing outcome benefit seen with immune-enhancing feeds in general ICU populations (11, 24 26), with the possible recent exception of the antioxidant and enriched lipid approach (27). It is therefore important to consider how the three main components of Intestamin (high-dose enteral glutamine, high doses of antioxidant vitamins and trace elements, and tributyrin as a short chain fatty acids source as fuel for colonocytes) might have contributed to the greater improvement in delta SOFA score that we observed. Glutamine is increasingly considered to be indispensable in critical illness and appears beneficial in several patient groups, especially those with burns or major trauma, where it received a level A recommendation in recent ESPEN guidelines (14). Its potentially beneficial actions include acting as a metabolic fuel for gut epithelial and immune cells, attenuating cytokine release, acting as an antioxidant by enhancing glutathione levels, and delaying the induction of nitric oxide synthase. Apart from its metabolic pathway effects, possible mechanisms behind these actions include the ability of glutamine to up-regulate heat shock protein 70 and peroxisome proliferator-activated receptor (26, 28 30) and down-regulate activator protein 1 (30). Some or all of these effects combine to protect the gut and reduce the permeability defect associated with critical illness (31), which may help to explain the beneficial effects of glutamine on some measures of gut permeability, infection rates, immune function, and outcome reported in several clinical trials (32 37). Glutamine itself is relatively insoluble in water, difficult to store in solution, and unstable during heat sterilization, which made it difficult to incorporate into commercially available enteral and parenteral feeding products before the quite recent introduction of stable, highly soluble glutamine dipeptides (38). Consequently, there is still relatively little clinical outcome information for glutamine administration, and most data relate to parenteral solutions containing either alanyl-glutamine or glycyl-glutamine dipeptides. Indeed, a recent Canadian meta-analysis of parenteral glutamine administration in critically ill patients showed a significant benefit for both mortality and infectious complications (39) ( but could only find a mortality benefit for enteral glutamine in patients with burns and reduced infectious complications in patients with burns or after trauma (37). Since this analysis a further study of enteral glutamine supplementation in patients in a surgical trauma ICU has failed to identify any outcome benefit (40, 41). Why are studies of enteral glutamine inconsistent? Part of the problem may relate to different study populations. 139

10 Figure 4. Delta total Sequential Organ Failure Assessment (SOFA) score over time biphasic (study days 0 11). ITT, intention to treat; PPP, per protocol; ICU, intensive care unit. 140 Crit Care Med 2008 Vol. 36, No. 1

11 Table 8. Plasma interleukin (IL)-6 and IL-10 levels Study Day Intestamin Reconvan, Median (Minimum Maximum) Control Supplement Fresubin Original, Median (Minimum Maximum) t-test p Value ANOVA Repeated Measures IL-6 (normal range (0 3.3 ng/l) (7 2279) 109 (4 17,683) (2 596) 43 (3 972) (0 179) 27 (5 3951) (1 383) 79 (9 813) (2 191) 39 (7 429) (3 153) 37 (5 1119).571 IL-10 (normal range 0 9 pg/ml) 1 10 ( 5 60) 11 ( ) ( 5 254) 10 ( 5 35) ( 5 222) 5 ( 5 199) ( 5 172) 10 ( 5 305) ( 5 120) 9 ( 5 120) ( 5 94) 7 ( 5 108) (treatment effect).599 (treatment effect) ANOVA, analysis of variance. Burn, surgical, trauma, and medical ICU patients exhibit differing inflammatory states, and there is increasing evidence that there may be differential effects of nutrition, especially regarding the administration of arginine (42). Indeed, while arginine may induce nitric oxide production and increase gut permeability (43), glutamine may have a protective effect against this action (43). More important, perhaps, is the difficulty of delivering sufficient glutamine to the gut early enough to be effective using standard enteral feeding. In a small study of nine patients with neurologic injury, Preiser and colleagues (44) examined the effects on glutamine levels in plasma and duodenal biopsy samples of a standard feed and two glutamine-enriched solutions, one containing glutamine-rich proteins and the other free glutamine. Plasma glutamine levels did not increase after 7 days in any of the groups, while duodenal mucosal glutamine concentration only increased in the patients fed the glutamine-rich protein feed. In two small pilot studies in which glutamine dipeptide was given jejunally in the form of Intestamin to patients after gastrointestinal cancer surgery, plasma glutamine levels increased in one study (45) but not the other (46). In a recent dose-finding pilot study using glutamine and antioxidants in ventilated critically ill patients with evidence of hypoperfusion, Heyland and colleagues (47) combined intravenous glutamine (Dipeptiven) with enteral glutamine as Intestamin and demonstrated normalization in plasma glutamine levels, but the small size of this study (seven patients per dosage group) makes it difficult to separate the relative contributions of the intravenous and enteral components. In our current study, plasma glutamine levels were just below normal, rather than being markedly depleted, and there was a significantly greater rise in plasma glutamine on direct comparison in the intervention group by day 3, with levels then being maintained in the normal range, although this change did not quite reach significance as a treatment effect by analysis of variance (p.137). This suggests that the glutamine dose administered may have been sufficient to spill over from the gut itself into the systemic pool. This is further supported by the pattern of increase in amino acids, although the glycine in the glycyl-glutamine dipeptide in Intestamin may also have contributed, especially to the increase in glycine levels. Glycine also has specific effects, including protecting against the effects of organ manipulation during liver transplantation in animals (48) and reducing liver injury, tumor necrosis factor- levels, and mortality in an endotoxic shock model (49). Plasma glutamine levels have been reported to rise in other studies of glutamine-enriched feeds, but less rapidly than in our study (37, 50). The extra glutamine, glycine, and arginine administered to the treatment group also provided a substantially higher protein load, and although the control group was fed successfully according to the unit protocol, as in other enteral feeding studies the total protein delivery in the control group was quite low (51). Nevertheless, it is unclear that such underfeeding is harmful, nor are there convincing data to suggest that high-protein feeding alone would account for the difference in SOFA response in the two groups (39) (www. criticalcarenutrition.com). The next possibility is the effect of the antioxidant vitamins and trace elements in Intestamin. Vitamins C and E and -carotene levels rose rapidly in the intervention group, although selenium and zinc levels were not different and remained relatively low throughout. Lipid peroxidation (measured as TBARS) was similar in the two groups, with TBARS levels being similar to or lower than those for ICU patients either with ( mmol/l) or without ( mmol/l) systemic inflammatory response syndrome recorded in a recent study by Motoyama and colleagues (52) but higher than those recorded by Senkal and colleagues (45) in a surgical population (measured in the same laboratory as our samples). In the most recent version of the Canadian nutritional meta-analyses ( intravenous or enteral administration of antioxidants was associated with a reduced mortality, and Heyland and colleagues (47) recently suggested that high doses of antioxidants may minimize mitochondrial damage in critically ill patients. The third component of Intestamin is tributyrin, which is a novel structured lipid composed of three molecules of butyrate esterified with glycerol, with 1 g of tributyrin producing 10 mmol of butyrate. Butyrate is an important energy source for large and small bowel and also exerts anti-inflammatory effects (53, 54), with the usual source being fermented dietary fiber in the colon. In critically ill patients, high doses of dietary fiber may cause bloating, abdominal distension, and bacterial overgrowth (55), but these Crit Care Med 2008 Vol. 36, No

12 problems have not been observed with enteral tributyrin in experimental (56) and human studies (57), demonstrating good tolerance and safety, and our results further support this, although it is unclear how much of the butyrate generated from tributyrin remains in the gut to reach the large bowel and have an effect there. Another important issue is that of arginine, since giving arginine-enriched feeds to patients with severe sepsis is controversial (12, 58), and Reconvan contains 6.7 g/l arginine. In our study, both patient groups initially had plasma arginine concentrations below normal, but levels were restored by day 3 in the intervention group and remained normal throughout, even after target Reconvan administration had been achieved. This suggests that the rise in arginine levels in the early phase of enteral nutritional support (days 1 3) was probably a consequence of the conversion of glutamine into arginine, as shown previously in multiple trauma patients (59, 60). It is notable that studies which blame enteral arginine administration for putative adverse effects do not provide any amino acid profiling data to justify such an interpretation (13, 61). So while these two important amino acids may have different effects in various animal models, especially under conditions of ischemia and reperfusion (28, 29), we have shown here that it is possible to feed septic patients with both a high-dose enteral glutamine dipeptide and an arginine-enriched feed without causing either an overshoot in arginine production or, since plasma citrulline levels did not increase, evidence of increased nitric oxide production. These real-life data challenge the theoretical notion that arginine supplementation is necessarily harmful in sepsis (62), and subsequent studies in this controversial area must provide amino acid profiling data to be plausible. Several previous trials using immuneenhancing feeds or enteral or parenteral glutamine supplementation (63, 64) in surgical and trauma patients have demonstrated presumed beneficial effects on immune status, evidenced by changes in interleukin levels and HLA-DR activation, with suppressed expression being a poor prognostic sign (65 67). In our septic patients, initial IL-6 levels were high and fell as the study progressed and most patients recovered, without clear differences in interleukin patterns between the two groups. HLA-DR expression was extremely low from the outset and remained depressed throughout in spite of the relatively low mortality in our study population, with no difference between the groups. In view of the consistency of this response, the meaning of this variable may need to be better defined. There were no significant differences in length of stay or numbers and types of secondary infections between the two study groups, in spite of the more rapid fall in the delta daily total SOFA in the intervention group. In a small study of predominantly medical patients with sepsis, this is not surprising, since this is a more heterogeneous group than would be the case with trauma or major gastrointestinal cancer surgery patients. Blood sugar control was not a factor, since a tight blood sugar regimen (68) was used, giving similar blood sugar levels and insulin dosage in both groups. Our study has several important limitations. We had originally planned to enroll 344 patients, but it became apparent at our planned interim analysis that a highly significant difference in the rate of decline in the delta total daily SOFA score already existed. In view of this and the pilot nature of the study, we decided to cease enrollment, even though this necessarily had the effect of limiting the strength of the analysis, especially for some of the secondary outcomes. The other major limitation of our study design relates to the use of Intestamin and Reconvan in combination, rather than using Intestamin or placebo followed by a single standard feed, or even the scientific ideal of studying specific feed components individually. This makes it more difficult to disentangle the effects of Intestamin from those of Reconvan, although the pattern in changes of vitamins, amino acids, and the biphasic delta total daily SOFA response all suggest that it was probably the effect of Intestamin that predominated and also make it impossible to know which individual components are most important. Additionally, the patient group we enrolled had medium severity of illness, and the sepsis was predominantly medical in origin, so it is unclear whether similar results are achievable in other patient populations. CONCLUSIONS We believe that our results provide encouraging early data on the use of the pharmaconutrition concept with Intestamin, demonstrating its feasibility in a population of patients with predominantly medical sepsis. The new supplement was safely and effectively absorbed when administered by the nasogastric route and, in combination with the immune-enhancing feed Reconvan, was associated with a significantly more rapid fall in the delta total daily SOFA score compared with control, indicating a significantly faster recovery of organ function. Its use should be investigated further in other populations of critically ill patients. ACKNOWLEDGMENTS We thank the staff and patients of the general adult ICU at St Thomas Hospital, London. REFERENCES 1. Bernard GR, Vincent JL, Laterre PF, et al; Recombinant human protein C Worldwide Evaluation in Severe Sepsis (PROWESS) study group: Efficacy and safety of recombinant human activated protein C for severe sepsis. N Engl J Med 2001; 344: The Acute Respiratory Distress Syndrome Network: Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342: Montejo JC, Nutritional and Metabolic Working Group of the Spanish Society of Intensive Care Medicine and Coronary Units: Enteralnutrition-related gastrointestinal complications in critically ill patients: A multicenter study. Crit Care Med 1999; 27: Kudsk K: Enteral versus parenteral feeding in critical illness. In: From Nutritional Support to Pharmacological Nutrition in the ICU. Vol. 34. Pichard C, Kudsk K (Eds) Berlin, Springer, 2000, pp Kong SE, Blennerhassett LR, Heel KA, et al: Ischaemia-reperfusion injury to the intestine. Aust N Z J Surg 1998; 68: Wilcox MH: Gastrointestinal disorders and the critically ill: Clostridium difficile infection and pseudomembranous colitis. Best Pract Res Clin Gastroenterology 2003; 17: Lisbon A: Dopexamine, dobutamine, and dopamine increase splanchnic blood flow What is the evidence. Chest 2003; 123: 460S 463S 8. Putensen C, Wrigge H, Hering R: The effects of mechanical ventilation on the gut and abdomen. Curr Opin Crit Care 2006; 12: Taylor DE: Revving the motor of multiple organ dysfunction syndrome: Gut dysfunction in ARDS and multiorgan failure. Respir Care Clin N Am 1998; 4: Ali S, Roberts PR: Nutrients with immune- 142 Crit Care Med 2008 Vol. 36, No. 1