Nutritional Approach in Malnourished Surgical Patients:  A Prospective Randomized Study FREE

PROTEIN ENERGY malnutrition is recognized as an important risk factor for the occurrence of postoperative complications.^1- 3 Thus, artificial nutritional support has been proposed as an essential part of perioperative care of malnourished surgical patients.^4- 6 Feeding enterally rather than parenterally may improve outcome, particularly in malnourished patients who undergo major surgery for a neoplasm.^7- 12

Moreover, administration of standard enteral diets supplemented with arginine, ω-3 fatty acids, glutamine, and other key nutritional substrates (immunonutrition) modulates immune and inflammatory responses and gut function.^13- 14

Results of 2 randomized, double-blind trials^15- 16 consistently indicated that perioperative (before and after surgery) administration of immunoenhancing diets could significantly improve several outcome variables, but none of these studies were designed with separate randomization according to the nutritional status of the patients. Results of a post-hoc analysis¹⁶ suggested that the positive effects of immunonutrition were more pronounced in a subgroup of malnourished patients. Another finding was that patients receiving only preoperative immunonutrition, because of noncompliance with postoperative immunonutrition, also had a significant reduction in complications (M.B., L.G., G.R., et al, unpublished data, 1999). These data suggested that the simple preoperative approach could be sufficient to improve outcome.

Therefore, this study was designed with the primary end point of establishing in a prospective fashion whether administration of perioperative immunonutrition could reduce the rate of post-operative complications and the length of postoperative hospital stay compared with administration of standard enteral formulas in a homogeneous group of patients defined a priori as malnourished. Furthermore, we evaluated, in the policy of cost minimization, whether the simple preoperative administration of immunonutrition could be as effective as the perioperative approach.

This was a randomized clinical trial conducted by a single institution (San Raffaele Hospital, Milan, Italy) between September 1, 1998, and December 31, 2000.

Inclusion and exclusion criteria are summarized in Table 1. Patients had to meet all 3 inclusion criteria to be registered. After applying the exclusion criteria, patients were allocated by computer-generated individual random numbers into 3 arms. Eligible patients were required to sign a written informed consent form after the details of the protocol were fully explained. The protocol was approved by the ethical committee of San Raffaele Hospital.

Before surgery, one group drank 1 L of a supplemented liquid diet (Oral Impact; Novartis Consumer Health, Bern, Switzerland) per day for 7 consecutive days. After surgery, patients continued to be fed enterally with the same supplemented formula (perioperative group).

Before surgery, another group drank 1 L of a supplemented liquid diet (Oral Impact) per day for 7 consecutive days. After surgery, patients were given a standard enteral formula (preoperative group).

The third group received only postoperative enteral feeding with a standard diet (control group).

The standard and enriched diets contained the same amounts of energy and nitrogen (Table 2). Regardless the type of diet, in all 3 groups, postoperative enteral nutrition was administered within 12 hours of surgery via a feeding catheter jejunostomy or a nasojejunal feeding tube. Enteral flow was controlled by using a peristaltic infusion pump. The initial rate of 10 mL/h was progressively increased 20 mL/h per day until reaching the full nutritional goal (28 kcal/kg per day). Enteral infusion was continued until patients resumed adequate oral food intake (approximately 50% of the basal energy requirement).

All patients were advised to consume regular food as desired before surgery. In the preoperative and perioperative groups, presurgical oral supplementation with the enriched formula was given as outpatient therapy. Patients were also asked to register the daily amount of the preoperative supplement consumed. Hospital admission was scheduled for all patients 2 days before surgery.

Eight days before surgery, the following baseline variables were determined in all patients: body weight, degree of weight loss (with respect to usual body weight in the previous 6 months), performance status according to Karnofsky score, hemoglobin level, plasma level of total protein, albumin level, retinol binding protein level, prealbumin level, total circulating lymphocyte count, creatinine level, and arginine plasma level. The arginine level was also determined 1 day before surgery. Comorbidity factors, as measured by the American Society of Anesthesiologists score, were also recorded in all patients.

Type of surgery, duration of surgery, operative blood loss, and rate and amount of homologous blood transfused were registered. The decision to give homologous blood was based on the perioperative hemoglobin level (<8 g/dL) or the clinical condition of the patient.

Intestinal washout with an isosmotic solution (3 L) was carried out the day before surgery in patients undergoing colorectal surgery. The evening before and the morning of surgery, patients were also treated with enemas. These patients received antibiotic prophylaxis in a single intravenous dose (2 g of cefotetan disodium) 30 minutes before surgery. Candidates for gastric, esophageal, or pancreatic surgery were treated with intestinal washout (1 L) the day before surgery and with antibiotic prophylaxis in a single intravenous dose (2 g of cefazolin sodium) 30 minutes before surgery. A second dose of the antibiotic was given if the surgery lasted longer than 4 hours. Deep-vein thrombosis prophylaxis was carried out daily using low-molecular-weight heparin (50 IU/kg).

Adverse effects of postoperative enteral feeding, such as abdominal cramping and distension, diarrhea (defined as >3 liquid stools per day), and vomiting, were evaluated in all patients, as was the time from canalization to gas and bowel movement.

Patients were defined as nontolerant to postoperative feeding if they were unable to tolerate a minimum diet infusion of 800 mL/d by postoperative day 4.

Trained members of the surgical staff who were not involved in the study registered the postoperative complications (definitions are given in Table 3). They also decided independently the day of hospital discharge, the indication, and the duration of antibiotic therapy in the postoperative course and the first day of oral food resumption. Total parenteral nutrition was given to patients who were nontolerant of enteral nutrition.

Death, anastomotic leak, relaparotomy, complications requiring transfer of the patient to the intensive care unit, and percutaneous drainage of a deep abscess using interventional radiologic techniques were defined as major complications. Follow-up for infectious and noninfectious complications was carried out for 30 days after hospital discharge via office visits.

It was assumed that postoperative complications would occur in approximately 40% of eligible control patients.^7,16 A reduction in the incidence to 20% would indicate the efficacy of perioperative immunonutrition treatment. With a targeted maximum of 50 individuals in each group, we had approximately 80% power to detect such a reduction at α = .05.

All patients were analyzed on an intent-to-treat basis. Descriptive results are given as mean (SD) or number (percentage) of observations. One-way analysis of variance and the Kruskal-Wallis test (for nonnormally distributed data, ie, operative time, operative blood loss, length of stay, resumption of oral food intake, duration of antibiotic therapy) were used to compare continuous variables among groups. Post-hoc multiple comparisons were performed using the Bonferroni correction. The χ² test (with Yates correction if there were cells with <10 observations) and the Fisher exact test were used to compare discrete variables. All P values are 2-sided, and significance was set at P<.05. A statistical software program (SPSS version 8.0 for Windows; SPSS Inc, Chicago, Ill) was used for statistical analysis.

During the study, 196 patients were observed: 46 were excluded and 150 were eligible and entered by randomization into the control group (n = 50), the preoperative group (n = 50), or the perioperative group (n = 50). All eligible patients were analyzed on an intent-to-treat basis (Figure 1).

The mean preoperative intake of the oral supplementation was 910 mL/d, with no difference between the preoperative and perioperative groups. Table 4 lists the preoperative (day −8) characteristics and the comorbidity factors of the eligible participants. The 3 study arms were well-balanced for all variables. Mean weight loss was 13%. Mean serum levels of total protein, albumin, prealbumin, and retinol binding protein were below the reference range in all groups. The arginine plasma level was similar 8 days before surgery in all groups, whereas 1 day before surgery the level was 1.15 mg/dL in the control group, 1.64 mg/dL in the preoperative group, and 1.69 mg/dL in the perioperative group (P = .03). The 3 groups were comparable for surgical variables (Table 5). Most patients underwent upper gastrointestinal tract operations.

Outcome variables are given in Table 6. The total number of patients who developed postoperative complications was significantly lower in the perioperative group than in the control group (P = .02). Moreover, the number of patients with complications was 9 (18%) of 50 in the perioperative group vs 14 (28%) of 50 in the preoperative group (50% relative reduction), but this difference did not reach statistical significance (P = .34). Also, comparing the preoperative and control groups, a substantial reduction in complications was observed (28% vs 42%; P = .21). Mean length of postoperative hospital stay was significantly shorter in the preoperative (P = .01) and perioperative (P = .001) groups than in the control group.

Table 7 gives the complications in each group. Considering infectious morbidity, respiratory tract and wound infections were the most frequent, with a lower incidence in the groups receiving immunonutrition than in controls. Among major complications, anastomotic leak seemed to be the most frequent, with a similar trend toward reduction in both groups receiving the enriched diet.

The duration of antibiotic therapy needed to treat postoperative infections was 6.1 (1.7) days in the perioperative group vs 5.8 (3.5) days in the preoperative group and 10.2 (6.3) days in the control group.

One hundred four patients (69%) did not experience any adverse effects related to postoperative enteral nutrition. Abdominal cramps or distension were observed in 29 patients (19%), diarrhea in 13 (9%), and vomiting in 4 (3%). The rate of gastrointestinal tract adverse effects was similar in the 3 groups. Because most adverse effects were controlled by the temporary interruption or reduction of the jejunal infusion, overall the nutritional goal was achieved in 138 patients (92%). Twelve patients (8%) had to be switched to total parenteral nutrition for intolerance (5 patients in the control group, 3 in the preoperative group, and 4 in the perioperative group).

Recovery of bowel function as measured by canalization to gas and feces occurred after 3.5 (1.7) and 5.2 (2.1) days, respectively, without any substantial difference among groups. Patients recovered adequate oral food intake 9.8 (5.4) days after surgery in the control group, 8.2 (2.3) days in the perioperative group, and 8.4 (3.6) days in the preoperative group.

Between surgery and hospital discharge, patients lost an additional 3.1% of their body weight. The value was similar among groups.

Malnutrition in patients who are candidates for major surgery is a considerable problem for the surgeon because it represents a risk factor for postoperative morbidity and mortality.^1- 3 Cancer-bearing patients may have additional protein energy depletion due to the occurrence of cachexia.¹⁷ Despite extensive research in the field of clinical nutrition,¹⁸ indisputable results on which to base rational nutritional support in malnourished surgical patients are sparse.

A recent consensus statement¹⁸ concluded that malnourished patients are candidates for preoperative artificial nutritional support. This support could reduce the rate of postoperative complications by approximately 10%. These general conclusions have several potential flaws: (1) in the trials analyzed, malnutrition was defined by a multitude of different scores and threshold values for weight loss and serum protein; (2) the study design was rarely a priori dedicated to malnourished patients; (3) enrolled patients were often heterogeneous for the primary diagnosis; (4) the quantity of calories was, in most cases, excessive compared with current standards; and (5) the type of nutrients and the presence of lipids were frequently uncontrolled. These considerations made the comparison among trial results puzzling and the conclusions difficult to be applied in clinical practice. Moreover, nutritional support, with the suggested modalities, requires preoperative hospitalization for at least 7 to 10 days, with an obvious increase in sanitary costs. Another reason for skepticism is that severe malnutrition (the only condition that benefits from preoperative parenteral feeding)¹⁹ is often an indirect sign of advanced neoplastic disease (unresectable tumor, metastatic spread, peritoneal carcinosis, etc), which usually makes patients unsuitable for a surgical approach. Therefore, the cost-effectiveness of this nutritional treatment may be limited.²⁰

Results of recent studies strongly suggest that in malnourished patients, postoperative enteral feeding with standard feeds should be preferred to parenteral nutrition. In fact, surgical patients fed enterally had an improved outcome compared with patients treated parenterally,^7- 12 and the nutrition-related sanitary costs are substantially reduced when using the enteral route.^7,21- 23 Thus, we believe that postoperative enteral nutrition with a standard diet could be considered the reference treatment. In the present study, we did not include an unfed group because it was considered unethical to restrain such patients from any support.

In the past decade, research in clinical nutrition focused on the addition of key substrates (arginine, ω-3 fatty acids, glutamine, RNA, etc) to standard formulas. Two meta-analyses^24- 25 showed that in surgical patients, postoperative use of enteral immunonutrition seems to improve outcome compared with standard formulas. Nevertheless, by analyzing singularly the major studies,^26- 30 the results were conflicting. In fact, the positive findings reported by Daly et al^26- 27 were partially confirmed by others.^28- 30 Postoperative treatment is conceptually limited because the amount of key substrates given in the first days after surgery is little and therefore may be insufficient for an efficacious modulation of the immune and inflammatory responses after surgery.^31- 34 As any other substance with supposed pharmacological action, immunonutrients should reach suitable tissue and plasma concentrations to be active. Thus, a different approach was tested to anticipate the provision of immunonutrients before surgery to obtain high levels at the time of surgical stress, when the depression of the host response is maximal.

Phase 2 clinical trials^35- 37 showed that perioperative treatment (before and after surgery) with an enteral diet enriched with arginine, ω-3 fatty acids, and RNA significantly prevented early postoperative impairment of the host defense mechanisms, controlled the overwhelming inflammatory reaction, improved intestinal microperfusion and postoperative gut mucosal oxygen metabolism, and modulated the metabolic response favoring the synthesis of constitutive proteins instead of acute-phase proteins.

The impact of perioperative administration of immunoenhancing diets was tested subsequently in phase 3 double-blind trials to address whether the advantages listed in the previous paragraph on surrogate end points could translate into true outcome benefits. Senkal and colleagues¹⁵ showed that patients receiving enteral immunonutrition had significantly fewer infections than the control group (14 vs 27). We¹⁶ reported that in the intent-to-treat analysis, the overall rate of postoperative infectious complications was 14% in the supplemented group and 30% in the control group (P = .009), and length of hospital stay was also shorter in the supplemented group (P = .01). Furthermore, our data suggested that perioperative immunonutrition is efficacious regardless of the baseline nutritional status of the patients. In fact, the rate of complication in well-nourished patients was 10% (6/63) in the supplemented group vs 27% (14/68) in the control group (P = .05). In a subgroup of malnourished patients, the rate of complication was 14% (3/22) in the supplemented group vs 39% (7/18) in the control group (P = .05).

Yet, these studies were not originally designed to randomize patients by their nutritional status. To overcome this bias, we designed a prospective study to investigate the effect of perioperative immunonutrition on outcome in patients a priori homogeneous for nutritional status. In the present trial, we also included a group receiving only preoperative immunonutrition because previous results (M.B., L.G., G.R., et al, unpublished data, 1999) suggested that patients receiving only preoperative immunonutrition because noncompliant with postoperative immunonutrition had a significant reduction in complications. This suggested that the simple preoperative approach could be sufficient to improve outcome. Moreover, in the light of cost-minimization policy, the restriction to preoperative treatment only may save approximately $90 to $180 per treated patient.^15,38

The results of the present trial are consistent with those of the previous subset analysis and confirm that perioperative treatment with immunonutrition may significantly decrease the postoperative morbidity rate in malnourished patients. In contrast, the expected equivalence between perioperative and preoperative immunonutrition was not achieved. This may be explained by the fact that malnourished patients, beside having energy and nitrogen needs, have marked impairment of the immune response, and, thus, prolonged and increased administration of key nutrients is required. This speculation is consistent with recent data³⁹ suggesting instead that in well-nourished patients, the simple provision of immunonutrients before surgery was sufficient to improve outcome.

The present data showed a clear trend on the incidence of postoperative complications according to the type of nutrition strategy. In particular, a progressive effect was noted, with perioperative immunonutrition giving the best outcome, postoperative standard formula the worst results, and preoperative immunonutrition intermediate results. Thus, a dose-response effect of key nutrients on the occurrence of complications may be hypothesized.

The 2 groups supplemented with the enriched formula received more energy and proteins than the control group before surgery. However, the better outcome in the groups receiving immunonutrition should not be attributed to the differences in preoperative energy and nitrogen intake compared with the control group because it has been shown previously that when isoenergetic and isonitrogenous preoperative load was used, the results were still in favor of immunonutrition.^15- 16 This allows us to speculate that the key factor for improved outcome was the provision of immune-enhancing substrates rather than the simple provision of energy and proteins.

In the present study, the incidence of infection was not significantly different among groups. Yet, patients receiving the supplemented diet required a shorter duration of antibiotic therapy to treat infectious complications. These data may be interpreted as the result of the therapeutic combination of drugs and a more efficient host response on the process of bacterial clearance. Also, it may be hypothesized that the reduction in the postoperative infectious morbidity rate in the supplemented groups is related to improved gut oxygen metabolism and tension already reported in patients perioperatively supplemented with immunonutrition.³⁵ This may result in advantages on intestinal barrier function and subsequent decreased bacterial translocation that, coupled with malnutrition, may become clinically relevant.⁴⁰

This study was not designed to address treatment and complication costs. Nevertheless, the health care resources saved by the marked reduction in the morbidity rate by using perioperative immunonutrition may largely outnumber the additional costs of the supplemented diet.^15,38

In conclusion, the present trial is the first, to our knowledge, to report that in a selected population of malnourished patients with cancer undergoing major elective surgery, the administration of key substrates was beneficial to outcome. Administration of the supplemented diet before and after surgery seemed to be the best strategy to reduce complications and length of hospital stay.

The diets were provided by Novartis Consumer Health.

Corresponding author and reprints: Luca Gianotti, MD, ScD, Department of Surgery (Chirurgia 3) San Gerardo Hospital, Via Donizetti 106, 20052 Monza (MI), Italy (e-mail: luca.gianotti@unimib.it).