Simple nutritional intervention in patients with advanced cancers of the gastrointestinal tract, non-small cell lung cancers or mesothelioma and weight loss receiving chemotherapy: a randomised controlled trial


J. Andreyev, The GI Unit, Department of Medicine, Royal Marsden Hospital, Fulham Road, London SW3 6JJ, UK.
Tel.: +44 (0) 207 811 8216
Fax: +44 (0) 207 811 8107


Background:  Weight loss in patients with cancer is common and associated with a poorer survival and quality of life. Benefits from nutritional interventions are unclear. The present study assessed the effect of dietary advice and/or oral nutritional supplements on survival, nutritional endpoints and quality of life in patients with weight loss receiving palliative chemotherapy for gastrointestinal and non-small cell lung cancers or mesothelioma.

Methods:  Participants were randomly assigned to receive no intervention, dietary advice, a nutritional supplement or dietary advice plus supplement before the start of chemotherapy. Patients were followed for 1 year. Survival, nutritional status and quality of life were assessed.

Results:  In total, 256 men and 102 women (median age, 66 years; range 24–88 years) with gastrointestinal (n = 277) and lung (n = 81) cancers were recruited. Median (range) follow-up was 6 (0–49) months. One-year survival was 38.6% (95% confidence interval 33.3–43.9). No differences in survival, weight or quality of life between groups were seen. Patients surviving beyond 26 weeks experienced significant weight gain from baseline to 12 weeks, although this was independent of nutritional intervention.

Conclusions:  Simple nutritional interventions did not improve clinical or nutritional outcomes or quality of life. Weight gain predicted a longer survival but occurred independently of nutritional intervention.


Adequately powered, randomised, controlled clinical trials are lacking to inform current international guidelines for the nutritional management of patients with cancer (Arends et al., 2006).

Patients with cancer frequently have weight loss at the time of diagnosis and, in patients receiving palliative chemotherapy, weight loss predicts a significantly shorter survival and poorer quality of life (DeWys et al., 1980). Amongst the known reasons for poorer outcomes in patients with cancers of the gastrointestinal tract, lung (non-small cell) cancers or mesothelioma who have lost weight are a reduced response to chemotherapy and the fact that patients experience increased toxicity from treatment, as a result of which a smaller total amount of chemotherapy can be given (Andreyev et al., 1998) (Ross et al., 2004). This provides a rationale to investigate whether improving the nutritional status of patients might lead to improved clinical outcomes.

Weight loss in patients with cancer may be managed by dietetic interventions, which aim to increase oral nutritional intake. These interventions include nutritional counselling to increase food intake, modification of the energy density of meals and the prescription of oral nutritional supplements. Oral nutritional supplements are available on prescription and, theoretically, are easy to use. In the UK, the cost of prescribed nutritional supplements is well in excess of £100 million per annum and studies of patient usage of prescribed supplements have highlighted potential problems with compliance (Fearon et al., 2003). An alternative to the prescription of expensive nutritional supplements is to help patients modify their normal food intake. Manipulation of food intake has potential advantages in that it may offer greater variety, be better tailored to individual needs and, if effective, may reduce healthcare costs.

The small number of studies that have examined the efficacy of the dietary management of weight loss in cancer have suggested that oral nutritional interventions result in an increase in energy and nutrient intake and may be associated with benefits to quality of life in some groups (Arnold & Richter, 1989; Ravasco et al., 2005a,b; Isenring et al., 2007), although they have consistently failed to demonstrate an impact on clinical and functional outcomes. However, all the studies of simple nutritional interventions reported to date have been significantly underpowered to demonstrate whether they alter clinical outcomes (Brown, 2002; Arends et al., 2006; Elia et al., 2006). Despite this, it has been argued that simple nutritional interventions may be futile in the management of weight loss in patients with cancer who are receiving chemotherapy (Koretz et al., 2007) and that efforts should be focused on the use of agents that modulate the inflammatory processes associated with cachexia (Skipworth & Fearon, 2007; Fearon, 2008).

In the absence of adequately powered studies assessing the impact of simple nutritional interventions on important clinical endpoints such as survival, quality of life or nutritional status in patients with advanced cancer receiving palliative chemotherapy, this large randomised trial aimed to determine whether such simple interventions are useful.

Materials and methods

Study design

This prospective, randomised, controlled trial was approved by research and development and ethics committees for all participating centres. It was also reviewed and adopted into the portfolio of the National Cancer Research Network, UK. All enrolled patients provided their written informed consent before randomisation.


Adults with histologically proven, metastatic or locally advanced tumours of the gastrointestinal tract (oesophagus, stomach, pancreas, liver and biliary tract, colon and rectum), non-small cell lung (NSCL) cancer or mesothelioma were eligible for recruitment if they had lost any weight in the 3 months before presentation, had agreed to undergo palliative chemotherapy and were fit for chemotherapy according to standard local criteria.

Patients were not eligible if they were unable, or unwilling, to provide informed consent to randomisation, had a clinical condition precluding oral nutrition, were unable to tolerate milk or it was considered that they should receive immediate enteral or parenteral nutrition.


Participating patients were randomly allocated to one of four groups: Group 1 received no dietary intervention. Group 2 received advice to increase food intake by an additional 2510 kJ (600 kcal) per day, which is the amount recommended to achieve weight gain (Todorovic & Micklewright, 2004). Consistency of advice was maintained by providing a booklet, which set out a selection of commonly used snack foods, drinks and desserts in portion sizes each providing 628 kJ (150 kcal). Participants were asked to select four items daily from the booklet, in addition to their usual food intake, and were guided to consider suitable choices. Group 3 received one sachet of nutritional supplement [either ScandiShake (SHS International, Liverpool, UK) or Calshake (Fresenius-Kabi, Runcorn, UK)] per day, which provided 2460 kJ (588 kcal), when prepared in 240 mL of full-cream milk. Patients were provided with a selection of both supplements and with assorted flavours at the time of randomisation and could choose which product to take, according to preference. Patients were guided to fit the supplements into their existing meal pattern. In addition, this group received a daily multivitamin and mineral supplement, either a capsule (Forceval; Unigreg, Morden, UK), tablet or effervescent formulation (Supradyn; Roche, Welwyn, UK), according to preference. Group 4 received advice to increase food intake by 250 kJ (600 kcal) per day and one sachet of the nutritional supplements and vitamin preparations as given to groups 2, 3 and 4.

All patients were assessed, recruited and followed up by trained, trial dietitians. Advice given to patients in the three intervention groups was supported by written information and they were followed-up at subsequent outpatient appointments and via weekly telephone calls for 6 weeks. Encouragement was given to continue with the intervention and patients were encouraged to discuss any problems that they experienced. No dietary information was given during the telephone calls to participants in the no dietary intervention group.

The primary endpoint of the study was mortality. The secondary endpoints were change in nutritional status, energy intake and quality of life.

Timing of intervention and assessments

The initial randomisation and intervention took place as early as possible before the start of chemotherapy. The trial intervention period was for 6 weeks only. After 6 weeks, participants were allowed to continue with their nutritional intervention(s), if desired, but did not continue to receive regular weekly encouragement. Patients were followed for 1 year after randomisation or until death.

Data were collected at baseline, 6 weeks (end of the intervention period) and at 3, 6 and 12 months after randomisation. These data included: height, weight, handgrip strength and quality of life data, using questionnaires from the European Organisation for Research and Treatment of Cancer (EORTC-C30) (Aaronson et al., 1993) and Functional Assessment of Cancer Therapy (FAACT) (Cella et al., 1993) (baseline, 6 weeks and 26 weeks only) and performance status (World Health Organization). The date of death was recorded.

Compliance with dietary interventions and change in energy intake was assessed in patients with gastrointestinal cancers recruited at the Royal Marsden Hospital (n = 236) using 3-day food diaries. Patients were shown how to complete their diaries at the same time as they were given patient information sheets about the study, and they were asked to complete the diary before their next appointment if they decided to participate in this trial. All patients randomised to receive a supplement were also asked to complete a diary of volume of supplement usage daily for 6 weeks. For assessment purposes, foods recorded descriptively in the 3-day diaries were converted to weights using standard portions. Dietary data were analysed using dietplan5 (Forestfield Software Ltd, Horsham, UK).


Randomisation was performed by the independent randomisation office of the Institute of Cancer Research, after stratifying patients according to the two known prognostic factors for gastrointestinal cancer: disease site and performance status. Eligibility was checked by the independent study monitor and randomisation was performed using random permuted blocks. Group allocation was concealed from both the investigator and patient until consent to enter the trial had been obtained. Study personnel were not blinded to treatment group.

Statistical analysis

Where data were normally distributed, the appropriate parametric test was used. Data on continuous variables, such as weight, handgrip strength and energy intake results, were analysed using appropriate tests including, analysis of variance, two-sample t-tests or the Mann–Whitney U-test depending on the distribution of the data collected. A Friedman test was used to compare mean weight at each time point. Two-by-two factorial analysis was used to compare groups receiving dietary advice with groups receiving no dietary advice and groups receiving nutritional supplements with no nutritional supplement.

Overall survival was derived using the product-limit method of Kaplan–Meier, which allowed patients lost to follow-up (censored observations) to be included in the analysis. The log-rank test was used to evaluate differences in survival curves. Cox multivariate analysis was used to examine survival differences between comparison groups (supplement versus no supplement and dietary advice versus no dietary advice) after correcting for known prognostic factors. The results are reported as hazard ratios.

Power calculation

The primary outcome measure was survival at 1 year. When the trial was conceived, recent results from a large patient dataset from our unit has shown that patients with advanced tumours of the gastrointestinal tract (all sites), presenting for treatment with no weight loss have a 1-year survival rate of 50%, whereas patients who have lost weight at presentation have a 1-year survival rate of 30% (Andreyev et al., 1998). To detect an increase in 1-year survival of 10% in those patients who had lost weight, it was calculated that, if a total of 660 patients were randomised, the study had at least 80% statistical power, with an alpha of 0.05 (two-sided test), to detect a difference between either of the two arms of the stratifications (i.e. supplement versus no supplement and dietary advice versus no dietary advice).

Data monitoring/patient safety

The protocol stipulated that an independent data monitoring committee would review the data after 200 patients had been randomised, and at additional time-points subsequently specified by the committee.


From 2002 to 2004, patients with gastrointestinal malignancy were recruited from a single centre. Subsequently, the trial was extended to patients with NSCL cancer and mesothelioma. After 2005, four additional UK centres (Maidstone, Guilford, Coventry, Birmingham) and one Australian centre (Melbourne) became involved.

The trial was stopped prematurely in 2006 on the advice of the independent data monitoring committee. At that point, 358 patients (256 men and 102 women) had been randomised. The groups were well balanced for all baseline characteristics (Table 1). The median follow-up was 189 days (6 months) with a range of 0–1485 days (0–49 months). Fourteen patients had no follow-up and a further four patients had <1 month follow-up as a result of early death or withdrawal from the trial (Fig. 1).

Table 1.   Baseline characteristics of patients recruited to the trial of nutritional intervention for weight loss in cancer
 TotalSubdivided according to intervention
No intervention
(Group 1)
Dietary advice alone
(Group 2)
Nutritional supplement alone
(Group 3)
Dietary advice + nutritional supplement
(Group 4)
Number35896 (27%)90 (25%)86 (24%)86 (24%)
Age (years)
 Male25669 (27%)66 (26%)64 (25%)57 (22%)
 Female10227 (26%)24 (24%)22 (22%)29 (28%)
Site of primary tumour
 Oesophago-gastric7117 (24%)19 (27%)17 (24%)18 (25%)
 Pancreas7218 (25%)16 (22%)18 (25%)20 (28%)
 Liver and biliary10 4 (40%) 3 (30%) 3 (30%) 0
 Colorectal11329 (26%)30 (26%)27 (24%)27 (24%)
 Lung8125 (31%)19 (23.5%)19 (23.5%)18 (22%)
 Unknown primary7 3 (43%) 1 (14%) 1 (14%) 2 (29%)
 Other4 0 2 (50%) 1 (25%) 1 (25%)
Performance status (World Health Organization)
 023 5 (22%) 4 (17%) 7 (30.5%) 7 (30.5%)
 118447 (26%)50 (27%)39 (21%)48 (26%)
 28025 (31.25%)17 (21.25%)21 (26.25%)17 (21.25%)
 34 1 (25%) 2 (50%) 1 (25%) 0
 Royal Marsden Hospital29976 (25%)78 (26%)75 (25%)70 (24%)
 Guildford2711 (41%) 5 (18%) 3 (11%) 8 (30%)
 Maidstone18 5 (28%) 4 (22%) 3 (17%) 6 (33%)
 Melbourne10 2 (20%) 3 (30%) 3 (30%) 2 (20%)
 Birmingham3 2 (67%) 0 1 (33%) 0
 Walsgrave1 0 0 1 (100%) 0
Figure 1.

 Consort flowchart of participant progression through a randomised controlled trial of nutritional intervention in the management of advanced cancer. RMH, Royal Marsden Hospital.

Nutritional intervention and survival

Overall 1-year survival was 38.6% [95% confidence interval (CI) 33.3–43.9]. There were no significant differences in overall survival between any of the groups (Fig. 2 and Table 2).

Figure 2.

 Percentage survival according to group allocation.

Table 2.   One-year survival rate subdivided by intervention group
GroupHazard ratio adjusted (95% confidence interval)
No advice
Dietary advice
0.840 (0.653–1.08)
No nutritional supplement
Nutritional supplement
1.016 (0.791–1.306)
No intervention
Any intervention (advice and/or supplement)
0.977 (0.733–1.302)

Patients with a performance status of two or more had significantly poorer survival than patients with performance status 0 or 1 (21.4% (95% CI 12.9–31.3) (PS2) versus 43.0% (95% CI 35.6–51.0) (PS1) and 65.5 (95% CI 42.6–81) (PS0) respectively). Patients with colorectal tumours had significantly better 1-year survival (55.8%; 95% CI 46.9–65.5) than patients with oesophago-gastric and pancreatic (29.6%; 95% CI 22.2–37.4) or lung tumours (32.3%; 95% CI 21.6–43.5). No differences in survival were detected from nutritional intervention within these sub groups.

Nutritional intervention and quality of life

Data on quality of life were collected using two questionnaires (EORTC-C30 and FAACT) at baseline, and at 6 and 26 weeks. A small number of the whole range of questions from each questionnaire were not completed by patients. Therefore, data were available from the EORTC-C30 questionnaire from 302–310 patients at baseline, 213–233 patients at 6 weeks and 118–129 patients at 26 weeks. Data were available from the FAACT questionnaire on 308–313 patients at baseline, 229–237 patients at 6 weeks and 132 patients at 26 weeks. Changes in each parameter of quality of life from baseline to 6 and 26 weeks were not significantly different between the groups receiving dietary advice compared to no dietary advice, as well as oral nutritional supplements compared to no oral nutritional supplements, for any parameter (data not shown).

Nutritional intervention and nutritional status

Weight loss at trial entry varied across the whole group (range 1.1–35.0%). The mean (SD) weight loss for patients with gastrointestinal tumours was 11.2% (6.4%) and 9.8% (6.0%) for lung cancer patients.

The mean weight changes from baseline to 6 weeks and baseline to 26 weeks were small and there were no statistically significant differences between groups (Table 3). The mean weight change at 52 weeks for all surviving patients varied in the range 1.36–4.78 kg according to intervention group (Table 3). Patients randomised to receive dietary advice compared to no dietary advice were significantly heavier at 1 year (P = 0.04), although the numbers involved (n = 68) are small and this finding should be interpreted with caution. There was no significant difference in weight at 1 year between those receiving a nutritional supplement compared to those not receiving supplements. No significant difference was found when this analysis was repeated including all patients missing at 1 year (intention-to-treat analysis), assuming zero change since the last available measurement.

Table 3.   Mean weight (kg) and weight change from baseline to 6 and 26 weeks, in a randomised controlled trial of nutritional intervention for weight loss in cancer, subdivided by intervention group
InterventionMean (SD) weight (kg) at baselineMean (SD) change (kg): as treatedMean (SD) change (kg): intention-to-treat analysis
6 weeks26 weeks52 weeks6 weeks26 weeks52 weeks
  1. *Statistically significant difference between dietary advice and no dietary advice at 52 weeks (P = 0.04).

Dietary advice70.3 (13.0)
(= 167)
−0.20 (3.5)
(= 132)
0.41 (6.0)
(= 115)
4.78 (5.0)*
(= 31)
−0.16 (3.1)0.06 (5.2)0.12 (5.3)
(= 167)
No dietary advice70.0 (13.4)
(= 168)
−0.43 (3.6)
(= 141)
0.11 (6.1)
(= 122)
1.36 (7.5)*
(= 37)
−0.36 (3.3)−0.16 (5.3)−0.29 (5.9)
(= 168)
Nutritional supplement69.4 (13.3)
(= 159)
0.07 (3.7)
(= 127)
0.58 (6.1)
(= 113)
3.41 (5.8)
(= 38)
0.06 (3.3)0.22 (5.3)0.21 (5.4)
(= 159)
No nutritional supplement70.8 (13.2)
(= 176)
−0.66 (3.5)
(= 146)
−0.04 (6.0)
(= 124)
2.30 (8.1)
(= 30)
−0.55 (3.2)−0.29 (5.2)−0.35 (5.8)
(= 176)

Factors predicting weight gain

Mean weight at each time-point according to group allocation and site of primary tumour did not differ between groups, although the SDs were wide. Therefore, mean weight change from baseline to 6 weeks and to 12 weeks were interrogated according to whether patients did or did not survive beyond 26 weeks. Comparison of mean weight change data for survivors with nonsurvivors from baseline to 6 weeks revealed no significant differences between groups (P = 0.434) (Fig. 3), although a trend towards less weight loss in patients who survived beyond 26 weeks was seen. Mean weight change from baseline to 12 weeks was greater in survivors beyond 26 weeks (P = 0.002) (Fig. 3) [mean (SD) weight change from baseline to 12 weeks in survivors 0.58 (2.47) kg and nonsurvivors −0.66 (3.76) kg]. This finding was independent of group allocation, suggesting that weight gain is primarily associated with survival and not nutritional intervention and thus, possibly, is a surrogate marker of tumour response to chemotherapy.

Figure 3.

 The magnitude of weight change from (a) baseline to 6 weeks and (b) baseline to 12 weeks according to survival. 95% CI, 95% confidence interval.

Hand grip strength

Data on handgrip strength were only collected for patients treated at the Royal Marsden Hospital (n = 282). The mean changes at all time-points were small and there were no significant differences between groups (data not shown).

Compliance with nutritional interventions

Food diaries were given to 236 patients, of whom only 60 (25%) completed one at baseline and 40 (17%) completed one at more than one time-point. The reasons for noncompletion of a second food diary were deterioration in health or premature death. Analysis of these was considered futile in view of the sample size.

Diaries of supplement intake were returned by 62 (41%) of 150 patients. Recorded compliance with the prescribed quantity of supplement fell after the first week of study. In the first week, 31% of patients reported taking all of the prescribed supplement and five patients 8% reported being unable to take any supplement. The number of patients able to take the full prescription decreased to 19% by week 6 and the number of patients being unable to take any supplement increased six-fold, to 48% by week 6.


The primary endpoint of this randomised controlled trial was to assess whether a simple, very modest nutritional intervention in patients with weight loss receiving palliative chemotherapy improves survival. The trial was stopped early by the independent data monitoring committee because no benefit was seen after randomisation of 358 patients. Of the secondary endpoints examined, the study demonstrated that oral nutritional interventions based on dietary advice or supplements had no effect on nutritional status or quality of life.

There are no previously published studies of simple nutritional interventions for the management of weight loss in patients receiving palliative chemotherapy that have been powered to examine clinical endpoints. However, the limited data that do exist (Brown, 2002) have suggested that nutritional intervention probably does not improve clinical endpoints or the quality of life in patients receiving chemotherapy. The findings of the present study are in agreement with this conclusion.

The effects of nutritional intervention in these previous small studies on weight change were variable. In the present study, we have shown that weight gain was associated with an improved survival, although this effect was independent of nutritional intervention and the site of disease, suggesting that weight gain may be a marker of treatment success rather than a positive effect of nutritional intervention.

The inability to detect any clinical benefit in the present study may have been a result of the choice of intervention. The supplements used in the present study were chosen after an earlier randomised and blinded evaluation of a variety of supplements and were clearly the preferred choice (Rahemtulla et al., 2005). Despite this and the very limited nature of the dietary prescription requested of participants, fewer than 50% of patients complied in the first week and this number declined to 19% by 6 weeks. A systematic review of dietary advice and nutritional supplements in the management of weight loss has demonstrated benefits to weight gain, with interventions providing as little as an additional 1–1.25 kJ (250–300 calories) per day (Baldwin & Weekes, 2008). Our findings suggest that patients struggled to comply with a modest increase in nutritional intake for 6 weeks. Failure of the patients to take the intervention mean that it is difficult to draw conclusions about the efficacy of nutritional interventions in this population and highlight difficulties inherent in studying the effects of nutritional interventions in this group.

The supplements options were milk-based. Lactose intolerance during 5-fluouracil-based chemotherapy has been reported in a proportion of patients (Parnes et al., 1994). It is possible that this contributed to decreasing compliance over time, although it is very unlikely to have influenced compliance in the early weeks of treatment.

The interventions were supported by regular contact with the trial dietitian for the first 6 weeks. After this time, patients were free to continue with interventions but received less intensive support, which may have accounted for the failure to find clinical benefits. There is insufficient information to determine the optimal length of intervention. The majority of dietetic interventions in a trial setting are given for 3 months or less (Baldwin & Weekes, 2008), and it is unknown whether this represents a reasonable goal in clinical practice. However, the poor compliance recorded by those prescribed supplements (albeit obtained from approximately half the cohort only) strongly suggests that continuing this intervention for longer would not have produced additional benefit. It is not possible to determine whether persevering longer with the food options to increase energy intake, carefully tailored to usual habits and food preferences, would have been beneficial because so few patients in the study group completed food diaries.

There is no gold standard for the assessment of compliance with dietary interventions. A variety of methods used in other studies, including prescription charts for hospital patients (Lawson et al., 2000), supplement consumption records (Fearon et al., 2003) and counting of cans or packs on home visits and wards (Bruce et al., 2003; Gosney, 2003), are all known to be inaccurate. In the present study, it was difficult to retrieve supplement diaries from patients after 6 weeks. Information on why patients were unwilling or unable to comply with the prescription was not collected. Studies of compliance with nutritional regimens have concentrated on whether patients are compliant or not (Lawson et al., 2000; Bruce et al., 2003; Gosney, 2003), rather than exploring the factors influencing this.

The present study had some limitations. It was stopped on the advice of an independent data monitoring committee prematurely. The original power calculation was based on the likelihood of finding an improvement in survival at 1 year increasing from 30% to 40%, which would equate to a hazard ratio of 0.76. The study had 80% power to detect this change. After randomisation of 358 patients, post hoc calculation (conditional power) requested by the data monitoring committee demonstrated that there was only a 6% probability for groups receiving nutritional supplements, and a 1% probability for groups receiving dietary advice, of observing a significant treatment effect, if the trial had been continued to full accrual, assuming that the projected true hazard ratio was 0.76.

At the time the trial was stopped, only half of the predicted number of deaths had occurred. The present study was initially powered using outcomes in 1500 consecutive patients with gastrointestinal cancers treated in our unit over the previous 10 years (Andreyev et al., 1998). The overall survival of patients in the study had improved above the levels used in the power calculation and reflects improvements in the management of gastrointestinal and lung cancers. Independent data suggest that outcomes in our patient cohort are not simply a statistical aberration because they do not differ from those in other cohorts (Katopodis et al., 2004).

The present study was opened to patients with lung cancer after 18 months to address difficulties with recruitment. Patients with lung cancer have a different disease from patients with gastrointestinal cancer and it is possible that the inclusion of patients with lung cancer may have influenced the findings of the study. The addition of patients with NSCL cancer and mesothelioma was considered to be reasonable in view of a study that showed almost identical outcomes in these patients with and without weight loss to the gastrointestinal patients (Ross et al., 2004).

Other possible confounding factors are the possibility that patients obtained nutritional information from alternative sources to the study dietitian. Additionally, the trial dietitians were not blinded to treatment group. These are major methodological issues relating to the conduct of randomised controlled trials of dietetic interventions (Baldwin et al., 2008) and issues relating to this have been reported in previous trials (Lundholm et al., 2004). This may have influenced assessment of the subjective outcomes. Third, the EORTC-C30 was one of the questionnaires used to assess quality of life. This questionnaire has been developed and validated for use in patients with cancer, although it is possible that this questionnaire is too generic, such that important quality of life issues specific to individual disease sites might have been missed. We attempted to capture the food-related aspects of quality of life by using FAACT, a quality of life questionnaire focusing on nutritional issues associated with weight loss, although this did not detect differences between groups. In the absence of a questionnaire designed to measure the impact of nutrition on quality of life, it remains a difficult relationship to investigate (Barr & Schumacher, 2003).

The present study represents the largest, randomised controlled trial of nutritional intervention in patients with cancer and weight loss receiving palliative chemotherapy performed to date. The study failed to demonstrate that significant improvements in clinical, nutritional and quality of life end-points could be achieved with simple nutritional interventions.


We would like to thank Jacqui Oates for her help with obtaining and recording data collected in the study and for maintaining all the trial paperwork. We also thank everyone in the GI Trials Unit for the support given to the study. We are very grateful to those who collaborated in recruiting patients at other centres. They include: St Lukes, Guildford: Dr Marianne Illsley, Tanya Klopper, Dr Gary Middleton and Dr Clare Topham; Maidstone Oncology centre: Carol Drabble, Dr Mark Hill and Dr Justin Waters; Austin Hospital, Heidelberg, Australia: Leonie Pearce, Olive Schmid and Dr Neil Tebbutt; Birmingham Heartlands: Dr Joyce Thompson and Fiona Underwood; Coventry: Dr Caroline Humber, Gordon O’Neil, Mrs J Travis and Dr Sharmila Sothi. We are also very grateful to the members of the independent data monitoring committee who spent a great deal of unpaid time reviewing the data on two separate occasions. They include: Dr Roger A’hern, Professor Marinos Elia and Dr Kevin Harrington. We thank Peter Milligan, King’s College London for additional statistical input.

Approval: This study was approved by the R&D and Ethics Committees of all the participating centres. Registration with clinical trials register: ISRCTN86266498.

Conflict of interest, source of funding and authorship

Nutritional supplements and vitamin supplements used in this study were provided free by Nutricia Clinical Care (Trowbridge, UK), Fresenius Kabi Ltd (Runcorn, UK), Unigreg Ltd (Morden, UK) and Roche Products Ltd (Welwyn Garden City, UK).
This study was funded by grants from The Henry Smith Charity and The Special Trustees of the Chelsea & Westminster Hospital, London. We acknowledge NHS funding to the NIHR Biomedical Research Centre.
CB helped design the study, recruited many of the patients, helped with the analysis and wrote many of the drafts of this manuscript. AS recruited all the lung cancer patients and helped with the writing of the drafts. CMcG recruited and followed up many of the patients for the study. ARN designed and powered the study and analysed all the data for the interim analyses. AG was the data manager who handled all the data generated by the study, entered it into the trial data base and coordinated all the multiple site permissions. KT performed all the statistical data analysis at the end of the study. DC was involved in the conceptual design of the study, facilitated the recruitment of patients to the trial and funded the data manager for this trial. MO’B was involved in the conceptual design of the study and facilitated the recruitment of patients to the trial. HJNA conceived and wrote the study protocol, contributed and edited all drafts of this manuscript, funded the additional funds required to pay all the research dietitians and is the guarantor for this study. All authors have critically reviewed the manuscript and approved the final version submitted for publication.