Strengths and weaknesses
Only four of the included studies have adjusted for potential confounding variables (10, 11, 16, 17). This means that it is not possible in most studies to exclude other possible causes of the observed association. In the following discussion, we present two examples of possible confounding factors: sex and disease severity.
Only two studies took account of sex, and both found that women experienced a significantly higher degree of toxicity than men (13, 14). The findings are supported by a meta-analysis of colorectal cancer patients (stages I–IV) who received bolus-based 5-fluorouracil in combination with folinate (22). The meta-analysis points to a possible pharmacokinetic explanation, in that the activity of the enzyme that metabolises 5-fluorouracil is generally lower in women than in men, but the documentation is inconsistent. The authors also ask whether there may be other pharmacokinetic or pharmacogenetic sex-related differences of significance.
The severity of the disease should also be taken into account, as this can affect the risk of malnutrition and toxicity. Four of the studies adjusted for cancer stage and/or functional status (10, 11, 13, 17). All found that the association between malnutrition and toxicity persisted. Studies that only included colorectal cancer patients with stage II or III cancer also found this association (10, 11, 14, 16).
All studies included in the review are historical cohort studies (23). As a general rule, one cannot draw any conclusions about causality in this type of study. On the whole, symptoms of malnutrition were measured before toxicity arose, which increases the probability that malnutrition increases the risk of toxic adverse reactions. The studies primarily used data from patient records to define exposure and outcome variables, which results in high study participation and hence a low risk of selection bias (23). A weakness of using patient record data is that we know little about how systematic the records are. In the studies that have looked at malnutrition in the form of BMI and weight loss, there is also some lack of clarity with respect to the time of recording weight and height, degree of weight loss, and whether account has been taken of any weight changes during chemotherapy (16–19).
The number of participants in the different studies varies substantially. Several have a small number of patients, which reduces the statistical robustness. Five of the studies also include patients with metastatic disease (12, 13, 15, 17, 19). In these studies, the whole population is analysed together, and we are therefore unable to say anything about the effects for patients with stage II–III cancer in isolation. However, the studies of Cespedes Feliciano et al. (10) and Jung et al. (11) included a large number of stage II–III colorectal cancer patients (553 and 229 patients, respectively). Both conclude that there is a significant association between low muscle mass and dose-reducing toxicity in connection with treatment with a combination of 5-fluorouracil, folinate and oxaliplatin.
In light of the limitations of observational studies, it is worth noting an ongoing phase 2 study in France among patients with localised colorectal cancer, where the effect of oxaliplatin-based chemotherapy scaled to patients' fat-free mass is compared with the standard dosage regimen (24). The results of this study will hopefully yield better answers to whether dose-calculation based on body composition has a part to play in the chemotherapy of the future for this patient population.