Clinical trial: efficacy of a low or modified fat diet for the prevention of gastrointestinal toxicity in patients receiving radiotherapy treatment for pelvic malignancies


Jervoise 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


How to cite this article: Wedlake L.J., McGough C., Shaw C., Klopper T., Thomas K., Lalji A., Dearnaley D.P., Blake P., Tait D., Khoo V.S. & Andreyev H.J.N. (2012) Clinical trial: efficacy of a low or modified fat diet for the prevention of gastrointestinal toxicity in patients receiving radiotherapy treatment for pelvic malignancies. J Hum Nutr Diet.25, 247–259


Background:  Inflammatory responses to pelvic radiotherapy can result in severe changes to normal gastrointestinal function with potentially severe long-term effects. Reduced or modified fat diets may confer benefit.

Methods:  This randomised controlled trial recruited patients with gynaecological, urological or lower gastrointestinal malignancy due to receive radical radiotherapy. Patients were randomised to a low fat (20% total energy from long chain triglycerides), modified fat (20% from long chain triglycerides and 20% from medium chain triglycerides) or normal fat diet (40% total energy from long chain triglycerides). The primary outcome was a difference in change in Inflammatory Bowel Disease Questionnaire – Bowel (IBDQ-B) score, from the start to end of radiotherapy.

Results:  A total of 117 patients with pelvic tumours (48% urological; 32% gastrointestinal; 20% gynaecological), with mean (SD) age: 65 (11.0) years, male : female ratio: 79 : 38, were randomised. The mean (SE) fall in paired IBDQ-B score was −7.3 (0.9) points, indicating a worsening toxicity. Differences between groups were not significant: = 0.914 (low versus modified fat), P = 0.793 (low versus normal fat) and P = 0.890 (modified versus normal fat). The difference in fat intake between low and normal fat groups was 29.5 g [1109 kJ (265 kcal)] amounting to 11% (of total energy intake) compared to the planned 20% differential. Full compliance with fat prescription was only 9% in the normal fat group compared to 93% in the low fat group.

Conclusions:  A low or modified fat diet during pelvic radiotherapy did not improve gastrointestinal symptom scores compared to a normal fat intake. An inadequate differential in fat intake between the groups may have confounded the results.


During therapeutic radiotherapy to the pelvis for gastrointestinal, urological and gynaecological tumours, up to 90% of patients experience a change in bowel habit (Khalid et al., 2006). In half of all patients, chronic symptoms emerge post-treatment that detrimentally affect quality of life (Gami et al., 2003; Olopade et al., 2005). As the number of survivors of pelvic radiotherapy treatment continues to grow (Hauer-Jensen et al., 2003), strategies to limit acute and hence chronic gastrointestinal toxicity assume increasing importance. Histological evidence (Larsen et al., 2007) points to an essentially inflammatory process that is initiated and exacerbated during the acute phase of treatment (Denham & Hauer-Jensen, 2002) and may then progress to chronic ischaemia and fibrosis months or years later, resulting in functional impairment to normal gastrointestinal physiology and a spectrum of clinical outcomes recently defined as ‘pelvic radiation disease’ (Andreyev et al., 2010).

Modified fat diets [i.e. those in which medium-chain triglycerides (‘MCT’) replace long-chain triglycerides (‘LCT’)] (Middleton et al., 1995; Verma et al., 2000; Sakurai et al., 2002), low fat diets (e.g. <35 g fat as LCT per day) (Gaiffer et al., 1990) or a combination of both (Bamba et al., 2003) have shown promise in the treatment of active Crohn’s disease, although not all studies have demonstrated a benefit (Rigaud et al., 1991; Royall et al., 1994; Khoshoo et al., 1996; Leiper et al., 2001). Two meta-analyses conducted over 10 years apart (Griffiths et al., 1995; Zachos et al., 2007) found only limited evidence for the potential efficacy of such diets, although both recommended that larger-scale studies be conducted. In patients receiving radiotherapy for pelvic tumours, the small and large bowel is inevitably exposed to some radiation. In these patients, an MCT-predominant low fat diet reduced bowel frequency (Karlson et al., 1989), whilst a low fat diet in combination with lactose restriction (Bye et al., 1992) resulted in a significant reduction in stool frequency and the use of anti-diarrhoeal medication at week 6 of treatment.

The therapeutic rationale for advocating low or MCT-based fat diets during radiotherapy is four-fold. First, radiotherapy has a direct adverse effect on the gastrointestinal brush border (Hauer-Jensen et al., 2003), potentially reducing its ability to absorb LCTs. Second, the provision of diets with a reduced LCT-based fat source may have direct anti-inflammatory effects (Middleton et al., 1995). Third, the reduced secretion of potentially irritant bile acids (required for the emulsification of LCT fats before absorption) may confer a direct benefit. Two studies (Bosaeus et al., 1979; Chary & Thomson, 1984) have demonstrated the anti-diarrhoeal effects of reduced bile acids in the acute (Chary & Thomson, 1984) and post-treatment (Bosaeus et al., 1979) radiotherapy settings. Physiological studies in healthy volunteers have demonstrated that MCTs do not induce cholecystokinin (CCK) release (Hopman et al., 1984; Isaacs et al., 1987; Vu et al., 1999; Costarelli & Sanders, 2001; Symersky et al., 2002) and subsequent gall bladder contraction (Hopman et al., 1984; Symersky et al., 2002), thus limiting the volume of bile acids secreted into the intestinal lumen. Fourth, MCTs fail to stimulate exocrine pancreatic secretions (Symersky et al., 2002), specifically amylase and lipase, thus sparing gastrointestinal mucosa from the proteolytic effects of these enzymes. Data supporting the benefits of manipulating and reducing biliary and pancreatic secretions in terms of a reduction in severe radiotherapy-induced toxicity are further supported by convincing preclinical studies (Sullivan, 1962; Sokol et al., 1967; Morgenstern et al., 1970; Hauer-Jensen et al., 1985; Wang et al., 1999, 2001; Onal et al., 2011).

Over the last two decades, radiotherapy techniques have become increasingly sophisticated. However, normal (i.e. noncancerous) tissue tolerance remains the critical determinant of prescription dose and thus tumour control. Sustained (even modest) attenuation of gastrointestinal toxicity in the acute setting has the potential to confer benefit against late effects (Wedlake et al., 2010). The present study aimed to explore the efficacy of low or modified fat diets on gastrointestinal toxicity in the contemporary radiotherapy setting.

Materials and methods

Study design

This three-arm, nonblinded, randomised controlled trial was approved by the Research and Development and Ethics committees of two participating centres. The independent Institute of Cancer Research, Randomisation Office, randomised patients (using permuted blocks) following provision of written informed consent.

Eligibility criteria

Patients with a histologically-proven gynaecological, urological or lower gastrointestinal malignancy due to receive radical (long course) external beam radiotherapy were eligible.

Interventional rationale and dietary fat prescription

The study design was based on randomisation to one of three possible groups with the interventional period coincident with the first 4 weeks of radical pelvic radiotherapy, which is typically of 5–6 weeks duration. The intervention during the initial 4 weeks of treatment was based on previous observations (Hovdenak et al., 2000, 2003; Larsen et al., 2007) of the typical time course of inflammatory processes, which are maximal at 2 weeks and followed by later emerging gastrointestinal symptoms.

Group 1 (low fat) was prescribed a low fat diet with LCT dietary fats calculated to comprise 20% of total energy intake, with the aim being to reduce the volume of potentially pro-inflammatory fat substrates and minimise the stimulation of bile and pancreatic secretions. Patients in this group were counselled to maintain a stable diet during treatment and were advised to maintain their total energy intake from alternative carbohydrate and protein-based sources.

Group 2 (modified fat) was prescribed a diet with fats calculated to comprise 40% of total energy intake. However, 50% was to be derived from LCT dietary fats and 50% as the MCT-based fat emulsion ‘Liquigen’ (SHS International, Liverpool, UK) providing 1883 kJ (450 kcal) per 100 mL. The aim was to reduce the percentage of LCT fat substrates at the same time as reducing risk of attendant weight loss through the provision of a fat substrate not requiring emulsification or hydrolysis for absorption. Group 3 (normal fat), the control arm was prescribed a normal fat diet with LCT dietary fats calculated to comprise 40% of total energy.

Each patient (all groups) received a ‘fat points’ prescription detailing the number of fat points to be consumed per day [1 fat point = 5 g of (LCT-derived) fat] based on their age/gender grouping as detailed in the UK government guidelines for total energy requirements for the UK population (Department of Health, 1991) Male patients aged between 50 and 59 years in the low fat arm received a fat prescription of 11 fat points per day (equivalent to 55 g of LCT-derived fat), amounting to 20% of their estimated total daily energy intake of 10 670 kJ (2550 kcal). An individualised preprinted fat point diary was provided and patients were encouraged to complete their fat point consumption after each meal. An eight-page booklet listing the fat content in points of typical portions (Food Standards Agency, 1998) of over 150 commonly-consumed foods was provided, which contained additional instructions on how to calculate the fat point content of any nonlisted foods, together with advice on reading food labels. In addition, patients in the modified fat group were given a Liquigen prescription defined in terms of the number of 30 mL ‘shots’ of Liquigen to be consumed per day. Liquigen consumption was titrated from 10 mL tds on day one to 30 mL tds or qds on days 3/4 of treatment to avoid supplement intolerance.

Measurement of gastrointestinal toxicity

Gastrointestinal toxicity was measured using the modified Inflammatory Bowel Disease Questionnaire – Bowel subset (IBDQ-B) (Cheung et al., 2000), the Vaizey incontinence questionnaire (Vaizey et al., 1999) and the Radiation Therapy Oncology Group (RTOG)/European Organisation for Research and Treatment of Cancer acute radiation morbidity scoring criteria for ‘lower GI including pelvis’ (Cox et al., 1995). Maximum and minimum scores for each tool are: IBDQ-B maximum (best) score: 70; minimum (worst) score: 10, Vaizey minimum (best) score: 0; maximum (worst) score: 24, RTOG minimum (best) score: 0; maximum (worst) score: 5. Absolute changes in score from baseline are recorded for all methods.

Anthropometric and quality of life measurements

Body weight was measured to the nearest 0.1 kg after removal of shoes and heavy clothing using Seca electronic column scales (Marsden, Oxon, UK). Height was measured to the nearest 0.1 cm using the stadiometer integral to the scales. Body mass index (BMI; kg m–2) was calculated to one decimal place and interpreted in accordance with standard categories. Grip strength was recorded using a T.K.K. 5401 Grip D hand grip dynamometer (Takei Scientific Instruments Co. Ltd, Tokyo, Japan) taking the mean of four readings, right and left hands in sequence, to the nearest 0.1 kg. The modified Inflammatory Bowel Disease Questionnaire (IBDQ), which is used to monitor both disease activity and quality of life in IBD patients, was used as the quality of life measure. The entire questionnaire (as distinct from the IBDQ-B subset) comprises 32 questions with maximum (best) score: 224 and minimum (worst) score: 32. This questionnaire has previously been shown to be a sensitive measure of acute gastrointestinal toxicity in this patient group (Khalid et al., 2006).

Analysis of compliance with fat and liquigen prescriptions

Diaries were collected on completion of 4 weeks of radiotherapy (following 20 treatment fractions over an elapsed 28-day period). Self-reported mean daily fat point consumption (to the nearest half-point) was calculated for each patient and used to compute the mean daily fat point consumption for the group. Dietitian administered 24-h recalls were completed at baseline, and at 2 and 4 weeks. The difference between baseline fat intake and required fat point prescription was noted. Agreement in terms of fat point consumption between dietitian assessed 24-h recalls and self-reported diary intake for the same day using the same tools (i.e. the fat point guidance booklet) was compared using Bland–Altman plots (Bland & Altman, 1986) at the 2- and 4-week time-points. Compliance with Liquigen prescription was assessed at the 2- and 4-week time-points in consultation with patients and graded into the following bands: <25%, 25–49%, 50–74%, 75–99% and 100% compliance with prescription.

Measurement points

Fat point consumption was recorded daily by patients. Gastrointestinal toxicity, body weight, BMI, grip strength and quality of life were assessed at baseline (start of radiotherapy) and (excluding BMI) after 2 and 4 weeks of radiotherapy and at 1 year post-treatment. All anthropometric and nutritional measurements were undertaken by registered dietitians. When appropriate, gastrointestinal toxicity was additionally assessed by the responsible clinicians.

Radiotherapy treatment characteristics

Patients were treated with a range of pelvic radiotherapy protocols as described previously (Khalid et al., 2006). Treatment characteristics are summarised by study group in Table 1.

Table 1.   Baseline characteristics and patient follow-up
 Group 1
Low fat
Group 2
Low fat + Liquigen
Group 3
Normal fat
n (randomised)403839117
Pelvic site: urological
 Total24 (60%)16 (42%)16 (41%)56 (48%)
Pelvic site: gastrointestinal
 Total9 (23%)15 (40%)14 (36%)38 (32%)
Pelvic site: gynaecological
 Total7 (17%)7 (18%)9 (23%)23 (20%)
Radiotherapy treatment features
 Concomitant chemotherapy18 (45%)24 (63%)17 (44%)59 (50%)
 Median radiotherapy dose64545454
 Range (Gy)44–7436–7445–7436–74
Patient characteristics
 Male : female29 : 1127 : 1123 : 1679 : 38
 Age (years), mean (SD)65 (10.6)64 (11.4)65 (11.3)65 (11.0)
 Weight (kg), mean (SD)79.5 (15.5)78 (16.9)79.7 (16.1)79.1 (16.0)
 ≤19 (underweight)1113 (3%)
 20–25 (normal range)13151442 (37%)
 26–30 (overweight)18151346 (40%)
 31–40 (obese)441018 (16%)
 >40 (morbidly obese)2204 (4%)
 Data missing2114
Number of patients available for follow-up at 4 weeks
 Withdrawn (= 10)17210
 Total at 4 weeks393137107
Reasons for loss to follow-up at 1 year
 Too poorly2316
 No reply53210
Number of patients available for follow-up at 1 year
 Total at 1 year27232575

Statistical analysis


The primary end-point was the difference in the change in mean (paired) IBDQ-B scores, from baseline to week 4 of radiotherapy between any study groups. With an estimated SD of 7.0 points (previous data), 35 patients per arm were required to give 80% power of detecting a difference of 5 points in the change in IBDQ-B score between any two study groups from baseline to 4 weeks. An additional four patients per group (total study population = 117) was planned to allow for a drop-out rate of 10% at 1-year follow-up.

Primary end-point

Group mean (paired) IBDQ-B scores were calculated as the mean of paired scores obtained at both baseline and 4 weeks. Following assessment for normality of data, t-tests or Mann–Whitney tests were planned to assess the significance of the difference in change in score between study groups, with P < 0.05 being considered as statistically significant. A per protocol analysis (primary end-point only) was planned in patients who achieved ≥75% compliance with interventional prescription.

Secondary end-points

The number of obtained measurements at each time point was recorded. Summary statistics for all secondary end-points are calculated using obtained data at each time point unless otherwise stated. Missing values are not carried forward. Vaizey, IBDQ, weight and grip strength are presented as mean (SD) scores for the cohort and study groups at baseline, 2 and 4 weeks. Change in weight (baseline to 4 weeks) is additionally presented using paired data values. Change in measurements at 1 year are presented using only paired observations (i.e. data available for the same patient at baseline and 1 year) due to the risk of data not being obtained at the 1-year point and the inappropriateness of group comparisons. RTOG scores are presented as the percentage of patients at each RTOG toxicity grade per time point by study group.


Patient accrual

Between January 2006 and July 2009, 1671 patients were screened as being potentially eligible for the present study, with 535 meeting the inclusion criteria. A total of 117 patients consented to the study, leaving 418 patients, of whom 102 did not meet the inclusion criteria, 257 declined and 59 were missed. Ten patients (low fat; = 1, modified fat; = 7; normal fat; = 2) withdrew during the acute phase of the study (= 6, intolerance to Liquigen; = 3, wish to withdraw; = 1, already entered in mutually exclusive study) making 107 available for follow-up at 2 and 4 weeks. Of these, 75 were available for follow-up at 1 year. The reasons for loss to follow-up at 1 year are summarised in Table 1.

Baseline characteristics

Study groups were well matched at baseline (Table 1) for treatment and demographic characteristics. Urology patients (largely prostate and pelvic nodes) comprised 48% of the cohort, as reflected by the male : female ratio of 79 : 38.

Data accrual

The number of obtained measurements at each time point is summarised in Table 2. The higher number of acute withdrawals (= 7) in the modified fat group resulted in consistently less data for all end-points being available from this study group. There was considerable attrition of data at 1 year (Table 1).

Table 2.   Evaluable data
Time pointGroup 1: low fatGroup 2: LiquigenGroup 3: normal fatAll
  1. IBDQ, Modified Inflammatory Bowel Disease Questionnaire; IBDQ-B, Inflammatory Bowel Disease Questionnaire – Bowel. *Paired scores (baseline + 4 weeks).

IBDQ-B (primary outcome)*
 Baseline scores obtained393838115
 2 week scores obtained382737102
 4 week scores obtained393037106
 Paired scores (baseline + 4 weeks)*393037106
 One year scores obtained22162058
 Baseline scores obtained393838115
 2 week scores obtained382737102
 4 week scores obtained393037106
 One year (paired) scores obtained22162058
 Baseline scores obtained363736109
 2 week scores obtained35253595
 4 week scores obtained34273697
 One year (paired) scores obtained19151549
 Baseline scores obtained403838116
 2 week scores obtained372837102
 4 week scores obtained393137107
 Baseline weight obtained403839117
 2 week weight obtained382837103
 4 week weight obtained393137107
 4 week (paired) weight obtained393137107
 One year (paired) weight obtained22151855
Grip strength
 Baseline score obtained403738115
 2 week score obtained382836102
 4 week score obtained393037106
Fat point diaries
 Completed diaries returned372935101
24-h recall
 Baseline recall completed29242679
 2-week recall and paired diary data30262884
 4-week recall and paired diary data20182260

Gastrointestinal toxicity

Paired IBDQ-B scores at 4 weeks were available for = 106 (Table 2). Mean (SD) of scores for these 106 patients were 66.2 (5.4) at baseline and 58.9 (9.3) at 4 weeks. Mean (SE) change in paired scores is given in Table 3. Mean (SE) score for all (= 106) patients fell (increasing toxicity) by 7.3 (0.9) points between baseline and 4 weeks of treatment. Mean (SE) fall in paired scores by study group (baseline to 4 weeks) was 7.1 (1.6) for low fat, 7.3 (1.7) for modified fat and 7.6 (1.5) for the normal fat group. There were no significant differences in the change in paired scores between groups with P = 0.914 (low fat versus modified fat), P = 0.793 (low fat versus normal fat) and P = 0.890 (modified fat versus normal fat). Scores were available for = 58 patients at 1 year. By 1 year, the mean (SE) fall in IBDQ-B score (versus baseline value) was 2.9 (1.1) points, representing the presence of less severe symptoms at this time point in this reduced group (= 58).

Table 3.   Change in paired Inflammatory Bowel Disease Questionnaire – Bowel (IBDQ-B) scores
Study groupIBDQ-B scores: mean (SE) [95% confidence interval]
Change: baseline to 2 weeks (= 102)Change: baseline to 4 weeks (= 106)Change: baseline to 1 year (= 58)
Group 1
Low fat
−4.9 (1.4)
[−7.7 to −2.0]
−7.1 (1.6)
[−10.3 to −3.8]
−1.0 (1.3)
[−3.6 to 1.7]
Group 2
Low fat + Liquigen
−8.6 (1.8)
[−12.3 to −4.9]
−7.3 (1.7)
[−10.9 to −3.7]
−5.0 (3.1)
[−11.5 to 1.5]
Group 3
Normal fat
−5.8 (1.3)
[−8.4 to −3.2]
−7.6 (1.5)
[−10.6 to −4.6]
−3.4 (1.7)
[−7.0 to 0.2]
All−6.2 (0.8)
[−7.9 to −4.5]
−7.3 (0.9)
[−9.1 to −5.5]
−2.9 (1.1)
[−5.2 to −0.6]

Means (SD) for Vaizey scores are given in Table 4 together with the change in score at 1 year for patients with paired values. Mean (SD) Vaizey scores for the cohort rose from 1.7 (3.4) to 4.8 (4.7) between baseline and 4 weeks, representing an increase of 3.1 points, indicative of increasing toxicity. Mean scores by study groups all rose during the same period by 3.3 points (low fat) 3.3 points (modified fat) and 3.2 points (normal fat). Paired scores were available for 49 patients at 1 year (Table 4). In these patients, the mean change in score at 1 year was an increase in score of 4.1 points.

Table 4.   Summary statistics for secondary end-points
Time pointGroup 1: low fat
Mean (SD)
Group 2: Liquigen
Mean (SD)
Group 3: normal fat
Mean (SD)
mean (SD)
 Baseline202.8 (11.9)198.5 (18.2)202.1 (13.4)201.2 (14.7)
 2 weeks194.2 (22.3)186.1 (22.2)191.8 (23.2)191.2 (22.6)
 4 weeks189.8 (27.2)185.5 (25.9)187.4 (25.8)187.8 (26.2)
 1 year change (= 58)2.5 (11.3)−9.6 (27.5)−4.0 (19.4)−3.1 (19.9)
 Baseline1.1 (3.1)2.4 (4.4)1.4 (2.1)1.7 (3.4)
 2 weeks2.9 (4.1)4.6 (3.8)3.4 (3.6)3.5 (3.9)
 4 weeks4.4 (4.1)5.7 (5.2)4.6 (4.8)4.8 (4.7)
 1 year change (= 49)2.8 (3.6)4.3 (5.2)5.6 (6.4)4.1 (5.1)
 Baseline79.5 (15.5)78.0 (16.9)79.7 (16.1)79.1 (16.0)
 2 weeks79.2 (16.5)77.7 (18.9)79.6 (16.3)78.9 (17.0)
 4 weeks78.5 (16.0)78.2 (18.2)79.4 (16.4)78.7 (16.6)
 4 week change (= 107)−0.9 (1.8)−0.4 (1.4)−0.6 (2.9)−0.6 (2.1)
 1 year change (= 55)1.2 (5.9)−3.0 (9.4)−0.3 (5.2)−0.4 (6.9)
Grip strength
 Baseline33.7 (7.8)31.5 (8.8)30.8 (9.9)32.0 (8.9)
 2 weeks33.3 (8.0)32.2 (9.3)31.8 (9.5)32.5 (8.8)
 4 weeks33.5 (8.3)33.7 (8.8)32.0 (9.6)33.0 (8.9)

Radiation Therapy Oncology Group toxicity scores are shown in Fig. 1a–c. No patient experienced RTOG grade ≥3 at any time-point.

Figure 1.

 Change in Radiation Therapy Oncology Group (RTOG) scores. (a) Group 1: low fat. (b) Group 2: low fat + Liquigen. (c) Normal fat.

Anthropometric and quality of life measurements

The distribution of BMIs for the cohort at baseline (= 113, missing values = 4) is given in Table 1. Mean (SD) for acute weight and grip strength scores are given in Table 4 together with the change in score at 4 weeks and at 1 year for paired weight values. The mean (SD) weight for the cohort fell from 79.1 (16.0) kg to 78.7 (16.6) kg between baseline and 4 weeks, representing a fall of 0.4 kg. Although the acute change in mean weight (all obtained data) varied by study group with a fall of 1 kg (low fat), a rise of 0.2 kg (modified fat) and a fall of 0.3 kg (normal fat), an analysis of paired data (= 107) revealed a drop in mean (SD) weight for all groups: 0.9 kg (1.8) low fat group; 0.4 kg (1.4) modified fat; and 0.6 kg (2.9) normal fat group (Table 4). At 1 year, paired weight data (i.e. data obtained at baseline and 1 year) were available for 55 patients. Mean (SD) weight for these 55 patients had fallen by 0.4 (6.9) kg. By study group, mean (SD) weight at 1 year compared to baseline had risen by 1.2 (5.9) kg in the low fat group, fallen by 3.0 (9.4) kg in the modified fat group and fallen by 0.3 (5.2) kg in the normal fat group (Table 4).

Mean (SD) grip strength remained stable for the cohort during treatment [baseline: 32.0 (8.9) kg versus 33.0 (8.9) kg at 4 weeks] representing a rise of 1.0 kg indicative of preserved lean body mass. By study group, mean (SD) acute grip strength fell by 0.2 kg in the low fat group and rose by 2.2 kg in the modified fat group and 1.2 kg for the normal fat group.

Quality of life (IBDQ) scores for the cohort fell from a mean (SD) of 201.2 (14.7) points to 187.8 (26.2) points by week 4, representing fall of 13.4 points. Mean (SD) change in score by study group was a fall of 13 points for the low fat group, a fall of 13 points for the modified fat group and a fall of 14.7 points for the normal fat group. At 1 year, data available for 58 patients indicated that the mean (SD) change in score was 3.1 (19.9) points, with the greatest fall in score of 9.6 points occurring in the modified fat group (Table 4).

Nutritional outcomes

Nutritional outcomes are summarised in Table 5. Self-reported fat diaries (covering 28 days) were fully completed by 99 patients and part-completed (2 weeks) by two patients, amounting to a total of 2800 recording days. The 16 missing diaries were accounted as: 10 withdrawing and six failing to return completed diaries. At baseline, mean LCT-based fat intake (assessed using the 24-h recall method) was remarkably similar across groups, ranging from 12.9 fat points per day [64.5 g fat equivalent to 2427 kJ (580 kcal)] to 14.9 [74.5 g fat equivalent to 2803 kJ (670 kcal)]. An analysis of the mean change in fat intake required by study group to meet prescription (24-h recall intake versus LCT-based fat intake prescription) indicated a required reduction of −3.2 fat points [16 g fat equivalent to 602 kJ (144 kcal)] for low fat, a reduction of −5.4 fat points [27 g fat equivalent to 1017 kJ (243 kcal)] for modified fat and an increase of 5.7 fat points [28.5 g fat equivalent to 1071 kJ (256 kcal)] in the normal fat group.

Table 5.   Nutritional intervention results
 Pre-radiotherapy – fat intakePeri-radiotherapy – fat intake
Baseline fat intakeFat prescriptionChange required to meet prescriptionSelf-reported daily fat point intake
Study GroupMean (SD) fat points per day
Equivalents in:
grams per day
kJ (kcal) per day
Mean (SD) fat points per day
Equivalents in:
grams per day
kJ (kcal) per day
Mean fat points per day
Equivalents in:
grams per day
kJ (kcal) per day
Mean (SD) fat points per day
Equivalents in:
grams per day; kJ (kcal) per day
% total energy for males and females*
  1. *Based on UK Government total energy requirement for males (65–74 years) of 2330 kcal/day and for women (65–74 years) of 1900 kcal/day.

Group 1: Low fat12.9 (3.14)
64.5 g
2427 kJ (580 kcal)
9.8 (1.04)
49.0 g
1845 kJ (441 kcal)
16.0 g
602 kJ (144 kcal)
6.7 (1.98)
33.5 g; 1259 kJ (301 kcal)
TE: Males: 13% TE Females: 16%
Group 2: Low fat + Liquigen14.9 (3.78)
74.5 g
2803 kJ (670 kcal)
9.6 (1.06)
48.0 g
1807 kJ (432 kcal)
27.0 g
1017 kJ (243 kcal)
8.2 (1.40)
41 g; 1544 kJ (369 kcal)
TE: Males: 16% TE Females: 19%
Group 3: Normal fat13.3 (2.99)
66.5 g
2502 kJ (598 kcal)
19.0 (1.79)
95.0 g
3577 kJ (855 kcal)
28.5 g
1071 kJ (256 kcal)
12.6 (3.28)
63 g; 2372 kJ (567 kcal)
TE: Males: 24% TE Females: 30%

Compliance with fat prescription differed markedly between groups during treatment. Mean (SD) daily self-reported fat point intake for the low fat group was 6.7 (1.98) points, which is well within the group mean prescription of 9.8 points. Similarly, for the modified fat group, mean (SD) self-reported intake for LCT-based fats was 8.2 (1.4) points, which is also in line with a prescription intake of 9.6 points. However, for the normal fat group, mean (SD) of self-reported intake was 12.6 (3.28) points, thus falling short of the group mean prescription (19 points per day) by over 6 points. Fat consumption for this group amounted to just 24% of total energy intake for men and 30% for women of the planned 40% of total energy requirement for this study group (Table 5).

Table 6 summarises compliance and weight change by gender, age and pelvic site within study group. Most strikingly, 68% of females (15 of 22) and 91% (51 of 56) of male patients were compliant with prescription in the low and modified groups (combined), whereas, in contrast, only 21% (three of 14) female patients and no male patients were within 0.5 fat points of prescription in the normal fat group. The analysis of weight change across all study groups by gender reveals little difference between the groups in terms of weight change in male versus female patients. Females were younger than males in each study group by a mean of 12 years (low fat), 3.6 years (modified fat) and 10 years (normal fat), although this does not appear to have markedly influenced compliance or weight outcomes. Weight change by pelvic site sub-group reveals no obvious differences between sub-groups, although the number of patients per sub-group is small. However, the five female patients with gastrointestinal tumours in the normal fat group returned a mean weight loss of −3.4 kgs over the 4 weeks representing the greatest change in any gender/pelvic site sub-group. Adherance to fat point prescription in the low and modified fat groups is reflected by the mean weight loss in each of these groups at 4 weeks (range: −0.9 to −0.4 kg). Although this amounts to a relatively small loss, it may indicate that patients experienced some difficulty in making-up the shortfall in energy intake with alternative carbohydrate or protein-based foods. Similarly, lack of compliance (i.e. low fat consumption) in the normal fat group is reflected by the group’s mean weight change of −0.6 kg. Compliance with Liquigen prescription was available for 31 patients; one male and six females failed to report. Of the 31 patients with evaluable data, 19% achieved 0–24% compliance, 10% achieved 25–50% compliance, 13% achieved 51–75% compliance, 16% achieved 76–99% compliance and 42% achieved 100% compliance. Compliance differed markedly by gender. The 13 patients who reported 100% compliance were all male and represented 50% (= 26) of the total male patients prescribed Liquigen. No female patients who reported Liquigen compliance (= 5) were within the 100% or 76–99% bands. Three achieved 51–75% compliance, one fell within the 25–50% band and one within the 0–24% band.

Table 6.   Compliance and weight change
GroupGenderAge: years mean (SD)100% complianceWeight change: kg mean (SD) baseline – 4 weeksWeight change: kg (baseline to 4 weeks)Group mean (SD) weight loss (= 107)*
Pelvic sitenChange: Mean (SD)
  1. Groups 1 and 2: weight change by gender and pelvic site is given for 37 patients (Group 1) and 29 patients (Group 2) returning 100% compliance with prescription.

  2. Group 3: weight change by gender and pelvic site is given for all patients (= 37) with available weight data.

  3. Missing compliance data: (Group 1: females: 0, males: 3) (Group 2: females: 7, males: 2) (Group 3: females: 2, males: 2) total: 16.

  4. Missing weight data: (Group 2: females: 0, males: 2) (Group 3: females: 2, males: 0) total: 4.

  5. *Paired values at baseline and 4 weeks.

1: low fat (≤10 fat points per day)Females (= 11)59.8 (12.2)11/11−1.0 (1.5)Gynaecological7−1.5 (1.6)−0.9 (1.8)
Urological3−0.4 (1.1)
 Males (= 29)71.8 (7.7)26/29−0.8 (2.0)Urological22−0.9 (2.1) 
 Gastrointestinal4−0.4 (1.8) 
2: modified fat (≤10 fat points per day)Females (= 11)64.9 (11.0)4/110 (0.8)Gynaecological3−0.2 (0.4)−0.4 (1.4)
 Males (= 27)68.5 (11.7)25/27−0.5 (1.5)Urological14−0.3 (1.9) 
 Gastrointestinal9−0.8 (1.1) 
3: normal fat (mean: 19 fat points per day)Females (= 16)62.7 (13.4)3/14−0.9 (3.6)Gynaecological8+0.3 (1.4)−0.6 (2.9)
Gastrointestinal5−3.4 (5.1)
 Males (= 23)72.7 (7.4)0/21−0.5 (2.4)Urological15−0.5 (2.2)
 Gastrointestinal8−0.7 (2.8) 

Bland–Altman plots (Fig. 2a,b) showed good agreement for estimated fat point intake between the 24-h (dietitian assessed) dietary recall method and patient self-reported (daily) diary intake for the same day. Paired (method) data were available for 84 patients at 2 weeks and 60 patients at 4 weeks. Agreement is best at lower intakes (5–10 fat points per day) and diverges as intake increases but, in most cases, lies within 2 SD of the mean. A per protocol analysis of those who achieved ≥75% compliance with fat point (and for the modified fat group, Liquigen consumption) was not undertaken because only three patients in the normal fat group achieved this target and thus there was no appropriate comparator group.

Figure 2.

 Bland–Altman plot. (a) 2 weeks; (b) 4 weeks.


The present study did not demonstrate that a low or modified fat diet reduced gastrointestinal symptoms induced by pelvic radiotherapy. Inadequate compliance with fat prescription, specifically the high level of self-imposed fat restriction in the normal fat group, may have confounded the results. The lack of achievement of fat point prescription in the normal fat group resulted in a mean difference between this group and the low fat group of only 5.9 fat points per day, which is equivalent to 29.5 g of fat or 1109 kJ (265 kcal) (i.e. a percentage difference of 11% in fat intake between groups) and was insufficient to demonstrate an effect.

Does this difficulty in achieving fat prescription call into question the study rationale? With respect to normal fat intake, previous data suggests that we were not seeking to change fat intake by an unachievable amount. An earlier study by our group (Armitage et al., 2008) indicated a mean daily fat intake for a mixed pelvic cohort of 77 g day−1 (SD: 27.8) at the start of radiotherapy, which is only 18 g less than the normal fat group’s ‘target’. In addition, a series of low fat interventional studies in breast cancer patients between 1987 and 2006 (Chelbowski et al., 1987, 1993, 2006; Crighton et al., 1992; Djuric et al., 1999) clearly demonstrated the ability of (women) patients to reduce fat intake from a mean of 69 g day−1 (5 studies; = 1153 patients) to a mean of 34 g day−1 over interventional periods ranging from 4 weeks (Crighton et al., 1992), 12 weeks (Chelbowski et al., 1987; Djuric et al., 1999), 6 months (Chelbowski et al., 1993) and 1 year (Chelbowski et al., 2006). Compared to these studies, we deemed our low fat prescription for the low and modified fat groups of 49 and 48 g day−1, respectively, to be readily achievable. This assumption is supported by our results.

It is possible that patient self-reporting of fat intake was flawed and that patients in the normal fat group under-estimated or under-reported their fat intake. In the previously identified studies in which a successful reduction of fat intake was clearly demonstrated, interventional tools were based on fat counting, guidance booklets and included dietary counselling, a similar strategy to that employed in the present study. Thus, it is likely that patients’ reported intakes are as truthful and accurate as the interventional tools allowed. This is supported by the level of agreement between self-reported fat intake diaries and dietitian assessed 24-h recalls. Patients showed remarkable diligence in completing and returning diaries and were mostly motivated to comply with study requirements. Their inability to meet the normal fat prescription may have been influenced by current healthy eating messages and a perception that efficacy was associated with lower fat intake.

Despite the low fat intake, the change in body weight differed little by study group in the acute phase of treatment (range: −0.9 low fat to −0.4 kg modified fat group). However, there was considerably greater individual variation in weight change than these group means imply. The subgroup of female patients with gastrointestinal tumours experiencing a mean weight loss of 3.4 kg emphasises the importance of individual monitoring and nutritional care. Our rationale for incorporating Liquigen was to ensure against possible weight loss from reduced fat-based energy and to make-up for this with fats derived from an MCT-based substrate. Interestingly, the modified fat group experienced the least change in weight acutely. Group comparisons at 1 year are not appropriate given the high attrition of data resulting in noncomparable study group populations.

Although there was no significant difference between groups in the primary outcome measure, IBDQ-B, the mean acute fall in score for the cohort was −7.3 points. This compares favourably with previous mixed pelvic cohorts where the mean acute change in score (four studies, = 409) was −9.0 points (Khalid et al., 2006; McGough et al., 2008; Wedlake et al., 2008, 2010). Because most patients in the normal fat group restricted their fat intake to levels in line with the low and modified fat groups, it is tempting to speculate that this reduced fall in score may have been influenced by dietary change resulting in reduced potentially irritant exocrine pancreatic secretions. Verification of this hypothesis would require data collection on the change (i.e. fall) in a relevant biological marker such as faecal elastase (‘FE-1’).

There were limitations inherent to our interventional study design. First, we lacked specific data on energy requirements in this patient group and based our fat point prescriptions (as a percentage of total energy intake) on estimated energy requirements for a healthy population (Department of Health, 1991). It is probable that the energy requirements of the present study population differed from healthy individuals and that the true energy requirements of these patients were less than for healthy individuals. Second, for practical reasons, our guidance booklet (counting fat points) did not list the fat content of all foods, especially those with only a small quantity of fat per portion, and thus self-reported fat point intake may have under-estimated true dietary fat intake. These two limitations taken together may have artificially enlarged the discrepancy between prescription and reported fat point intake.

In summary, the present study failed to show the advantage of a low or modified fat diet during pelvic radiotherapy treatment, with a lack of efficacy most likely the result of an inability to achieve the required fat intake differential between study groups. In view of an increasingly well informed study population and the well publicised hazards of dietary fat, future studies may be limited to nonblinded interventional trials of different low fat interventions with the employment of a biological marker to underpin results. Despite being unable to counter self-imposed fat restriction, the validation of our apparently successful interventional guidance and self-reporting dietary intake strategy may deserve further attention.


We would like to thank Louise Henry for helpful discussions regarding nutritional data analysis.

Conflict of interest, source of funding and authorship

The authors declare that there are no conflicts of interest.

SHS International (Liverpool, UK) provided the Liquigen supplements and an unrestricted educational grant to cover partial funding of the present study. We acknowledge NHS funding to the NIHR Biomedical Research Centre.

HJNA, CMG and LJW designed the study. CMG, LJW and TK undertook data collection. DPD, PB, DT and VSK approved patient inclusion in the study. KT undertook statistical analysis. AL managed the study database and undertook data entry. KT, LJW, CS and HJNA were involved in data analysis and interpretation. LJW wrote the manuscript. All authors critically reviewed the manuscript and agreed the final version submitted for publication.