• The purpose of the review was to determine the efficacy of any form of enteral or parenteral nutrition support for patients receiving bone marrow transplantation
  • Efficacy is defined in terms of:
    • Time in hospital
    • Complications
    • Change in nutritional status (i.e., body weight)
    • Survival.

• A randomized or quasi-randomized controlled trial
• Comparison of one type of nutrition support (enteral or parenteral) with another or with an intravenous solution of glucose/saline
• Subjects had to be bone marrow transplant (BMT) patients of any age.

• Not a randomized or quasi-randomized study
• No comparison of one type of nutrition support with another or with a glucose/saline solution.

Recruitment

• 35 possible RCTs were identified
• 24 of the 35 met all inclusion criteria
• 11 of 35 studies were excluded
  • Eight excluded because not RCTs
  • Two excluded because no form of nutrition support was used as intervention
  • One excluded because subject was not a BMT patient.

Design

• Search strategy 
  • A computerized search strategy (with no RCT filter) was implemented to identify trials from the following databases:
    • The Cochrane Library (Issue 1, 2000)
    • MEDLINE (1966-2000)
    • Embase (1988-2000)
    • CINAHL (1982-2000).
  • Hand searching included:
    • Nutrition and BMT conference proceedings
    • Reference lists of papers found through electronic search
    • Consultation with experts.
• Study selection
  • Identified studies were scanned by the lead reviewer and relevant studies retained
    • Relevant studies were independently scanned by the lead and co-reviewers and included or excluded according to pre-determined inclusion criteria.
• Methodological quality of studies was evaluated for
  • Allocation concealment
  • Blinding
    • Not considered crucial since outcomes were objective measures.
  • Loss to follow-up.

Statistical Analysis

• For outcomes measured as continuous data
  • Means, mean differences, standard deviation and standard error were used
  • 95% confidence interval.
• For outcomes measured as dichotomous data
  • Odds ratios were used
  • 95% confidence interval.
• For meta-analyses (where possible)
  • Fixed effect model was used
  • Summary estimate of outcomes with 95% confidence interval.
• To detect statistical heterogeneity
  • Chi square test
  • P=0.05, confidence interval 95%
  • If statistical heterogeneity was detected, a random effects model was used to determine a summary statistic
  • 95% confidence interval.
• Statistical limitations
  • Planned sub-group analysis of adults vs. children, disease type and transplant type could not be performed due to insufficient data
  • Effect of type of allocation concealment, loss to follow-up and 'number needed to treat' could not be assessed due to insufficient data.

Data collection

• A data extraction form was designed
  • The form was used to record data on participants, interventions, and outcomes.

Outcome measures assessed

• Primary outcomes
  • Hospitalization duration (Day Zero to discharge)
  • Mucositis (mean days mucositis from start to end of study)
  • Graft vs. host disease [GVHD] (number of patients who developed >grade two GVHD)
  • Nutritional status (% change in body weight from start to end of study)
  • Duration of nutritional intervention/time to achieve adequate oral intake
  • Neutropenia (mean days to achieve normal neutrophil level from Day Zero of BMT)
  • Line infections (number of patients who developed line infections from start to end of study)
  • Number of positive blood cultures
  • Survival to 100 days
  • Survival beyond 100 days (number of patients who completed study and survived +100 days or two years.)
• Secondary outcomes
  • Vomiting (mean number patients who had =3 vomits/day from start to end of study)
  • Diarrhea (mean number patients who had =3 bowel movements/day from start to end of study)
  • Veno occlusive disease [VOD] (actual number patients who developed VOD per group)
  • Liver function disturbance (number of patients with abnormal bilirubin levels from start to end of study)
  • Hepatomegaly (number of patients who developed hepatomegaly from start to end of study)
  • Albumin (mean change in albumin from start to end of study between trial groups)
  • Pre-albumin (mean change in pre-albumin from start to end of study between trial groups)
  • Engraftment (mean duration for each group to achieve engraftment from Day Zero of BMT).

Organization of included studies:

• 16 of 24 included studies were arranged into four main groups:
  1. Oral glutamine vs. placebo (four studies; 343 patients total)
     • Anderson, 1998
     • Coghlin, Dickson, 2000
     • Jebb, 1995
     • Schloerb, 1999.
  2. Parenteral nutrition with glutamine vs. standard parenteral nutrition (seven studies; 108 patients total)
     • Zeigler, 1992 (four of the seven studies were duplicates of Zeigler, 1992)
     • Brown, 1998
     • Schloerb, 1993.
  3. Standard parenteral nutrition vs. intravenous hydration (two studies; 166 patients total)
     • Lough, 1990
     • Weisdorf, 1987.
  4. Parenteral nutrition vs. enteral nutrition (one full report and two abstracts; 144 patients were identified)
     • Cope, 1997
     • Szeluga, 1987 (abstract)
     • Young, 1997 (abstract).
  5. Eight additional trials compared a variety of nutritional interventions and could not be allocated to one of the above groups (Aldamiz, 1996; Charuhas, 1997; Jimenez, 1999; Lenssen, 1987; Lenssen, 1998; Malhotra, 1996; Mulder, 1989; Muscaritoli, 1998).

Findings from the studies

• Group One: Oral glutamine vs. placebo (four studies; 343 patients total)
  • Adequate data were provided only by Jebb, 1995 and Schloerb, 1999
    • The use of an oral placebo mouth wash reduced days to neutrophil recovery significantly [6.82 days, 95% CI (1.67-11.98), P=0.009] when compared to a glutamine mouth wash.
  • No significant differences in hospital duration, percentage change in body weight, number of positive blood cultures, number of days of nutrition intervention and survival at 100 days.
• Group Two: Parenteral nutrition with glutamine vs. standard parenteral nutrition (seven studies; 108 patients total)
  • Data was provided for all outcomes by either two or all three authors
  • Most significant finding: Patients who received PN with glutamine had hospital stays 6.62 days shorter than standard PN patients [weighted mean difference; 95% CI (-9.77-3.47), P=0.00004]
  • PN with glutamine patients were less likely to develop positive blood cultures than standard PN patients (odds ratio 0.23)
• Group Three: Standard parenteral nutrition vs. intravenous hydration (two studies; 166 patients total)
  • The odds of developing a line infection are 21.23 times greater for PN patients than for IV hydration patients [from Lough's data; 95% CI (4.15-108.73), P=0.0002]
  • Mean percentage change in albumin concentration increased for hydration group as compared to PN group [from Lough's data; mean percentage change -5.93, 95% CI (-9.90 to -1.96), P=0.003]
  • PN more effective than hydration at maintaining body weight [from Lough's data; weighted mean difference for percentage change 2.76, 95% CI (1.26-4.26), P=0.0003]
  • No significant difference in survival 200 days post-BMT.
• Group Four: Parenteral nutrition vs. enteral nutrition (one full report and two abstracts; 144 patients were identified)
  • Due to study design and data collection problems, none of the data could be used.
• Group Five: Other studies (eight studies)
  • Due to the low power and inability to group these studies, the results could not be utilized.

• The benefits of oral glutamine mouth wash vs. placebo are unclear
• Parenteral nutrition with glutamine should be routinely considered for bone marrow patients thought likely to experience prolonged gastrointestinal failure
  • BMT patients receiving PN with glutamine seem to leave the hospital significantly sooner than placebo patients
  • BMT patients receiving PN glutamine seem to have significantly fewer positive blood cultures than placebo patients.
• PN should be administered with caution due to a highly increased risk of line infection
• Lack of evidence prevents analysis of effectiveness of enteral nutrition vs. parenteral nutrition.

Suggestions for further research:

• Another trial of oral glutamine vs. placebo since existing data are scant and inconclusive 
• A multi-center, four-armed RCT comparing parenteral to enteral nutrition, with and without added glutamine
• Studies to determine appropriate dose of parenteral glutamine.

• Chi-square statistic has low power to detect heterogeneity when only a small number of studies are analyzed
• Studies favorable to glutamine-enriched PN suggest a protective effect of glutamine against infections, with subsequently shorter hospital stay. However, appropriate dose of parenteral glutamine has not been determined.