Oral nutritional support in patients with cancer of the gastrointestinal tract
Different opinions exist about the value of enteral nutritional support in patients with cancer, including oral nutritional supplements (ONS), dietary advice and/or tube feeding. One view is that such support in cancer is effective and deserves to have a high priority in routine clinical practice, whereas another view is that it is ineffective, with no evidence to justify its use in a range of conditions, including nonsurgical cancer (Koretz et al., 2007). An intermediate position was expressed in a systematic review of ONS in cancer (Elia et al., 2006), which also indicated that certain studies lacked the necessary sample size to resolve some of the controversies. These different views exist partly because a variable emphasis has been placed on the balance between clinical experience and evidence-based practice, which is by no means comprehensive. They also exist because conclusions about nutritional support in patients receiving one type of treatment (surgery, chemotherapy, radiotherapy or palliation) for a particular stage and type of cancer do not necessarily apply to those receiving a different form of treatment for another type and stage of cancer.
Four papers recently published in the Journal of Human Nutrition and Dietetics have rekindled interest in this important area (Burden et al., 2010, 2011; Garth et al., 2010; Baldwin et al., 2011). All of them involved patients suffering exclusively from gastrointestinal cancers, with the exception of one study, which also included patients with lung cancer (Baldwin et al., 2011). Two of these studies were observational (Burden et al., 2010; Garth et al., 2010), and the other two, both published in the current issue, were interventional (Baldwin et al., 2011; Burden et al., 2011).
The first observational study analysed data collected retrospectively from 95 patients who were treated in an Australian hospital for upper gastrointestinal and colorectal cancers (Garth et al., 2010). The length of hospital stay was found to be related to preoperative weight loss, post-operative weight loss, time to achieve adequate nutrition (>75% of requirements) and the presence of malnutrition, which was determined prospectively in a separate sample (n = 25) using Subjective Global Assessment (SGA). A key finding was that patients who had been counselled by a dietitian post-operatively had a higher energy intake on the second day of a soft ward diet compared to patients who had not been counselled by the dietitian. The study raises the possibility that oral nutrition support instigated by a dietitian may prevent or attenuate weight loss, improve nutritional status and reduce the length of hospital stay. However, definitive conclusions about such benefits cannot be made from observational studies alone. They need to be supported by interventional studies.
The second observational study prospectively studied the nutritional status of 87 preoperative colorectal cancer patients in the UK (Burden et al., 2010). Malnutrition (weight loss >10%) was associated with significantly lower grip strength (19.4 kg versus 27.4 kg) and lower fat-free mass (39.7 kg versus 51.9 kg) measured using bioelectrical impedance. The study suggests that functional status declines before surgery, raising the possibility that preoperative nutritional support may improve or attenuate such a decline. However, the effects of nutritional support again need to be examined by interventional studies.
The observational study of Burden et al. (2010) was followed by an interventional study of 125 patients with colorectal cancer who were randomised to receive either preoperative dietary advice in the form of an information leaflet (control group) or dietary advice plus ONS (intervention group) (Burden et al., 2011). One of the strengths of the study is that it involved only patients with colorectal cancer, who were treated in one UK centre, and it provided evidence that the subjects were exposed to the intervention. Of those in the intervention group who completed a dietary diary (93% of all such patients), the majority (72%) managed to consume all of the prescribed ONS [2510 kJ (600 kcal) ONS per day]. Their total dietary intake, measured 48 h before surgery using dietary recall and ONS intake, was reported to be approximately 4184 kJ (1000 kcal) more than that of the control group. Despite this, the intervention group (ONS) was reported to have no significant advantage over the control group with respect to the median length of hospital stay [13.5 days (intervention) versus 14.0 days (control)] or the number of patients with total (primary outcome) or specific types of complications. Therefore, Burden et al. (2011) concluded that there was no evidence that preoperative ONS was beneficial.
Although this may be the case, at least one possibility deserves further consideration. Careful examination of the baseline patient characteristics suggests that, compared to the control group, those in the ONS group were lighter (body mass index 25.0 versus 26.8 kg m−2) had more weight loss (6.2% versus 3.9%) and had a greater prevalence of malnutrition [SGA (B+C categories)] (57% versus 38%). These differences were either significant or borderline significant (using data provided in the paper, P ∼ 0.05; M. Elia, unpublished results). There was also a tendency for the tumour-node-metastasis score to be greater in the ONS group, although the difference was not significant. Patients with malnutrition and preoperative weight loss are expected to have more complications and a longer length of stay than well-nourished patients with no weight loss. The absence of significant differences between the control and intervention groups and, indeed, the tendency for some of the outcomes to favour the ONS group both imply that the intervention could have produced some benefits. For example, the number of patients with wound infections was two-fold lower in the ONS group (significantly lower in those with preoperative weight loss) and the number with chest infections was approximately 1.5-fold lower in the ONS group. Analysis of the outcomes after adjustment for some of these baseline imbalances might have suggested different conclusions from those reached by Burden et al. (2011) using analyses of unadjusted data. Such an alternative interpretation would be in keeping with the benefits reported by certain preoperative nutritional supplementation studies (Waitzberg et al., 2006), and also a meta-analysis (M. Elia, unpublished results), who examined three randomised controlled trials (RCTs) published by the same group of Italian workers (Braga et al., 2002a,b; Gianotti et al., 2002) and another unpublished RCT undertaken in Holland, the results of which were summarised in a review article (Waitzberg et al., 2006). All four of these studies involved the administration of an immunonutrition ONS to patients who were undergoing abdominal surgery for various conditions, including cancer of the colon (exclusively in one study). The meta-analysis revealed a two-fold reduction in infective complications [relative risk, 0.462 (95% confidence interval, 0.317–0.674; P < 0.001; I2 = 0.0%) and an approximately 20% reduction in the length of stay (2.29 days) (95% confidence interval, 1.48–3.09 days), P < 0.001; I2 = 0.0%]. Three other RCTs in which a more standard ONS was administered preoperatively to patients undergoing abdominal surgery, including cancer of the colon (but not exclusively), produced more variable results, with no overall significant differences between the ONS and control groups (Flynn & Leightty, 1987; Macfie et al., 2000; Smedley et al., 2004).
The second interventional study examined the effect of 6 weeks of nutritional support on long-term survival in weight-losing patients who were scheduled to receive chemotherapy for advanced cancer of the gastrointestinal tract (and lung) (Baldwin et al., 2011). Because weight loss is common at presentation and is associated with reduced survival, it was argued that preventing weight loss by increasing dietary intake might improve survival. Patients with any amount of weight loss in the preceding 3 months were randomised into four groups (dietary advice alone, ONS alone, dietary advice plus ONS and no intervention) and were followed up for up for approximately 4–5 years at three UK centres and one in Australia. The overall results showed no significant differences between any of the groups in 1-year (or >1 year) survival, and no significant differences in quality of life, representing a secondary outcome measure that was documented in a proportion of patients. Although oral nutritional interventions may be ineffective in this patient group, several other possible explanations exist.
Although the study by Baldwin et al. (2011)is one of the largest of its kind, the sample size required to address the hypothesis that the nutritional interventions improve survival appears to have been inadequate. Initial calculations indicated the need to recruit 1500 consecutive patients with various types of gastrointestinal cancers, although a case was made that 660 patients would be sufficient to examine the hypothesis (with 80% power and P = 0.05) that nutritional support improves 1-year survival from 30% to 40% (i.e. ONS versus no ONS or dietary advice versus no dietary advice). The recruitment of patients with gastrointestinal cancers was slower than anticipated such that, even after the study was opened to patients with lung cancer, the final sample size was only 358, which is only slightly over half of the target of 660. The study, which required a substantial amount of time and effort, was terminated prematurely after a statistical analysis, which suggested that, even if continued to completion, it would be very unlikely to yield a significant effect. This additional information suggests the need to explore other explanations.
One such explanation is that the patients were inadequately exposed to the interventions. They struggled to comply with increased dietary intake. Data on compliance were scanty: only 25% of patients completed food diaries at baseline, only 17% of patients completed food diaries at more than one more time point, and only 41% of patients provided information on ONS intake. Of those reporting ONS intake, only 31% took all the supplements prescribed during the first week, although this declined to 19% at 6 weeks. At this time, 48% of patients reported taking no ONS at all. A possible reason for the poor compliance is that the disease and chemotherapy may have made the patients too sick to comply with the interventions. Another possibility for those in the ONS groups concerns the inconvenience of having to buy the milk needed to prepare the supplement and then to clean up. A ready-made high energy density ONS may have been associated with better compliance, although other considerations, including taste and texture, are also important.
Another possible explanation for the lack of effect concerns the composition of the powdered supplements, both of which lack vitamins and trace elements, and thus their supply depends on the milk that is used to constitute the supplements. However, the prescribed daily dose prepared using 240 mL of full fat milk [total of 2.3–2.5 MJ (550–600 kcal)] still lacks several nutrients. Even if as much as 8.4–10.5 MJ (2000–2500 kcal) of the constituted supplements were prepared, they would contain nutrients in amounts that are several-fold less than the reference nutrient intake. ONS providing more trace elements and vitamins than those used in the study might have induced different responses and outcomes, especially because nutrient deficiencies are expected to occur commonly in the population studied. Finally, it is possible that the administration of ONS, with or without dietary advice, for longer than 6 weeks could have improved survival over the next 1–5 years.
Such studies are complex and difficult to undertake, especially if considerable effort is required to improve compliance with the intervention. The reported difficulty in increasing dietary intake reflects a real-life situation and represents an important finding.
Practical implications and future directions
Where do all these results leave us and what information needs to be gathered to inform policies on nutritional care in patients with cancer? First, even when only gastrointestinal cancer is considered, cancer represents a heterogeneous group of conditions. Because the response to nutritional support is expected to vary according to the type and stage of cancer, generalisations should be made with great caution. Identification of the specific types of patients who are most likely to benefit from nutritional support remains a challenge. The study of Burden et al. (2011) largely involved ‘all comers’ and the study of Baldwin et al. (2011) involved cancer patients with any degree of weight loss in the previous 3 months, even if this was minimal and within normal variation. Information about the extent to which nutritional support might produce benefits in ‘malnourished’ compared to ‘well nourished’ patients is important. If recent nutritional intake during critical periods of a disease or its treatment can produce clinically relevant benefits in malnourished as well as normal weight/overweight patients, as previous studies suggest (Stratton et al., 2003), then effort should be made to prevent a major and prolonged reduction in dietary intake in a wide range of patient groups. If benefits are much more likely to occur in those who are already thin or have lost a lot of weight, then it would be more appropriate to target such individuals. Controlled trials do not necessarily address some of these issues. For example, many studies are not adequately powered to examine results of subgroups of patients, nor to compare malnourished with well-nourished subgroups, and some are even underpowered to examine the primary outcome using all the patients. Furthermore, study designs may deviate from real-life situations. For example, if, in routine clinical practice, a patient clearly does not respond to a specific treatment, the treatment is withdrawn. If a malnourished patient is perceived to need continued nutritional support, the support is continued. RCTs with fixed periods of treatment may not adequately bridge the gap between the investigative study conditions and the more flexible conditions associated with clinical practice. In the absence of the above information, a judgment about the type and amount of nutritional support is often made using a broad range of considerations, including clinical experience and judgement (Elia et al., 2011). After all, the practice of clinical medicine and clinical nutrition is both an art and a science.