Randomized controlled trial comparing a low-fat diet with a weight-reduction diet in breast cancer-related lymphedema




Obesity is considered a risk factor for lymphedema of the arm resulting from breast cancer treatment (BCRL) as well as a poor prognostic factor in response to lymphedema treatment. The aim of the study was to compare the effect of 2 dietary interventions on excess arm volume in BCRL.


A total of 64 women with BCRL were randomized to 1 of 3 groups for 24 weeks: 1) weight reduction through reduced energy intake; 2) low-fat diet with no change in energy intake (isoenergetic); 3) control group with no dietary change from habitual intake.


The primary outcome measure was arm volume at 24 weeks. Results showed significant reductions in body weight (P = .006), body mass index (P = .008), and skinfold thickness measured at 4 sites (P = .044) in the weight-reduction and low-fat groups compared with controls. A slightly greater, but not significant (P = .605), fall in excess arm volume occurred in the dietary groups but a significant correlation between weight loss (irrespective of group) and a reduction in excess arm volume (r: 0.423; P = .002) was demonstrated.


Weight loss, whether through reduced energy intake or low-fat diet, appears to be helpful in the treatment of breast cancer-related lymphedema. Cancer 2007. © 2007 American Cancer Society.

Lymphedema secondary to treatment for breast cancer (BCRL) may develop early after treatment or to up to 30 years later.1 BCRL remains a common management problem after curative breast cancer treatment.2–4 Recent reports have suggested estimates of incidence ranging from 20% to 42%.5

Many studies have suggested that obesity or being overweight may predispose women to developing lymphedema after treatment for breast cancer.6–13 Some studies have indicated that the degree of lymphedema was positively correlated with the level of obesity.10, 11 A study of 251 women who had undergone surgical treatment for breast cancer showed that at 3 years posttreatment the risk of lymphedema was related to hospital skin puncture of the limb, mastectomy rather than wide local excision, and a body mass index greater than 26 kg/m2.5

Obesity is a risk factor for postmenopausal breast cancer and is therefore present in a high proportion of this group of patients. It is unclear as to how obesity may influence the development of lymphedema but proposed mechanisms include an increased risk of postoperative complications including infection, reduced muscle pumping efficiency within loose tissues, additional fat deposition contributing to arm volume, and separation of deep lymphatic channels by additional subcutaneous fat.6, 14, 15

To date there is little published literature on dietary intervention in women with breast cancer and lymphedema, although it has been suggested by some that weight reduction may be of benefit in these patients.11 Evidence to date suggests that weight reduction is unlikely to have any adverse effect on the breast cancer patient and may have beneficial effects in terms of prognosis16 and some clinical practice guidelines refer to the maintenance of ideal body weight in the management of lymphedema.17

Although the majority of references to diet and lymphedema refer to body weight, there are also some limited references to the potential influence of low-fat diets and lymphedema. Two studies have suggested a possible benefit for patients with lymphedema following a low-fat diet, although the studies did not show clearly whether this was due to weight reduction or solely due to changes in fat intake.18, 19

Current treatment for BCRL is aimed at control of arm volume by compression exercise and massage with maintenance of good skin condition for avoidance of infection.20 The aim of the current study was to evaluate whether dietary intervention, using either a low-energy diet to induce weight reduction or an isoenergetic low-fat diet, can be beneficial in the treatment of arm lymphedema in patients who have had treatment for breast cancer. A low-fat diet was compared with a weight-reduction diet because of previous suggestions that this diet alone was beneficial. The expectation was that an isoenergetic low-fat diet would not change weight, whereas a reduced-energy diet would. The hypothesis was that weight reduction not fat reduction alone would help reduce excess arm volume in BCRL. The study compared a reduced-energy intake group with a low-fat group and a control group that did not receive dietary intervention but continued with usual care for their lymphedematous arm.


Participants and Study Design

Women with arm lymphedema secondary to treatment for breast cancer were recruited from the lymphedema clinic at the Royal Marsden Hospital.

Inclusion criteria for the study were that the subjects were: 1) swollen arm 20% or greater excess volume when compared with the contralateral unaffected arm and 2) remission from cancer and no chemotherapy or radiotherapy in the previous 12 months. They may or may not have been receiving hormonal treatments such as Tamoxifen. Participants would receive either multilayer bandaging for 19 days followed by compression hosiery or hosiery alone as part of a parallel study.

The study was a randomized control trial with a control group having standard treatment and 2 dietary intervention groups.

Randomization and Interventions

On recruitment to the study and before randomization to the control and dietary intervention groups the participants completed a 7-day dietary diary to assess their habitual dietary intake. The diary used was based on household measures for the assessment of portion size with additional photographs to help determine portion size eaten. The validity of this method has been measured by other investigators and has been shown to produce an accuracy comparable to weighed dietary intakes without the inconvenience of performing a weighed food intake.21, 22

Participants were randomized to the following groups, taking into account excess limb volume and concurrent drug treatment as described in the Statistics section.

Control group

Participants were asked to continue with their habitual diet for the 6-month period.

Weight-reduction group

Dietary advice was given on a weight-reduction diet with the aim of reducing body weight to the acceptable average weight for height. Participants were advised to reduce their intake to 1000–1200 kcal (4.2–5.0 MJ) per day. Individualized advice was based around individual meal patterns and the reduction of energy intake was achieved by reduction of intake of foods high in fat or refined carbohydrates. Exchange lists were given for foods supplying protein, fat, and starchy carbohydrate in order to enable a varied diet to be chosen.

Low-fat diet group

Women were advised to reduce dietary fat intake to 20% of total energy intake. This was achieved by teaching the participants to use a series of 5 g fat exchanges. The subjects were issued a list of foods containing 5 g fat together with information on how many of these exchanges should be consumed in a day. Advice was given to maintain energy intake at the normal level by increasing carbohydrate intake.

Dietary advice and intervention was given by the same registered dietitian (C.S.) for the duration of the study. Exercise and activity were not monitored in the study and no specific advice was given to subjects.

Outcome Measures

Anthropometric measures

Height and weight (to the nearest 100 g) were measured using Seca digital scales. Skinfold thickness was measured at 4 sites (triceps, biceps, subscapular, and suprailiac) using Harpenden skinfold calipers.23 Measurements of skinfold thickness were always taken on the unaffected arm by the same investigator (C.S.) who was trained and practiced in the technique.

Arm volume

Arm volume was measured by 1 of 2 methods. 1) By Perometer, a device designed for the measurement of limb volume. The Perometer emits 2 parallel arrays of infrared light beams at right angles to each other. The shadow cast by the limb is sensed by receivers on the opposite 2 sides of the frame to the light emitters and 2 transverse diameters of a ‘slice’ of the limb are measured. The Perometer has been shown to be a reliable way of measuring limb volume.24 2) Circumference was measured manually using a tape measure at 4-cm intervals up the limb starting at the wrist. The volume of each 4 cm segments was calculated using the equation:

equation image

The latter method has been shown to achieve good reproducibility.25

The 2 methods were not combined for individual subjects during the study. The volume measurements were performed by lymphedema practitioners who were blinded to the randomization of the subject. Results are expressed as the percentage of excess arm volume compared with the unaffected arm.

Dietary intake

Dietary diaries were completed before randomization and at Weeks 12 and 24. Dietary intake data were collected using 7-day dietary diaries with foods recorded in household portions. Photographs of small, medium, and large portion sizes were provided for composite dishes to enable food weights to be determined. This method was chosen as it provides a greater degree of accuracy than records that rely entirely on estimation of food portions without the inconvenience of a weighed food record.21, 22 Dietary diaries were analyzed using the Dietplan5 computer program (Forestfield Software, UK), which is based on McCance and Widdowson's Composition of Foods. Weights were allocated to foods using the known weights of the photographed foods in the diet records and using the MAFF guide to food portion sizes where appropriate.26 All the dietary records were coded by 1 dietitian, entered into the computer program as 1 week, and the figures averaged for a daily reported intake. Data are presented for the diary collected at baseline and at the end of the Week 24 period. Measurements taken during the study are outlined in Table 1.

Table 1. Measurements Taken During the Study
 Day 1Day 19Week 7Week 12Week 24
  • *

    Baseline 7-day record completed before Day 1.

Skinfold thickness+   +
Limb volume+++++
Dietary intake by 7-day dietary diary+*  ++

Some subjects underwent multilayer bandaging during the first 19-day period of the study and participants were stratified for this intervention before randomization.


Limb volume was used as the primary outcome measure in the study. The sample size required was determined using the Medical Research Council sample size calculation program. Based on the assumption that dietary intervention would be of benefit to 50% of the patients in the study group and that 5% of the patients in the control group may benefit, it was calculated that a sample size of 60 participants, 20 in each group, would have a 90% power to detect differences between groups at a 5% significance level using a 2-sided test.

Computer-generated randomization was undertaken by the data managers at the Institute of Cancer Research (Sutton, UK). Before randomization into control or dietary intervention groups subjects were stratified according the volume of their arm and their lymphedema treatment and each group was randomized independently. This was to ensure that there was no group that had a disproportionate number of large limbs that might be expected to respond to treatment differently. Limbs were stratified into the following 2 groups: 1) swollen limb 20% to 50% greater than unaffected limb and 2) swollen limb greater than 50% bigger than unaffected limb. Patients were also stratified for lymphedema treatment, either bandaging and hosiery or hosiery alone.

Comparisons were made before dietary intervention to ensure that the groups were comparable, then at 24 weeks to examine whether there were any differences in anthropometric measurements and excess limb volume. The Kruskal-Wallis test, a nonparametric test designed to detect differences between 3 or more independent groups, was used. The data were analyzed on an intention-to-treat basis with the Kruskal-Wallis test being used to test for any difference between the 3 groups.

Correlations were calculated using the Spearman rank correlation coefficient. The chi-square test was used to test any differences between the treatment groups before and after treatment. Post-hoc analysis was undertaken to examine correlations between weight loss and changes in arm volume using the Spearman rank correlation coefficient.

Ethical Approval

Signed consent was received from all subjects who entered the study. The study was approved by the Clinical Committee for Research and the Ethics Committee of the Royal Marsden Hospital.


A total of 64 women were recruited to the study and a total of 51 women completed the 24 weeks of treatment for lymphedema and dietary intervention. The 13 women who failed to complete the 24-week period of the study were excluded for the following reasons: did not attend (3 women), recurrence of breast cancer (2 women), requested to stop study (6 women), hospital admission (1 woman), and venous thrombosis in arm (1 woman).

Before dietary intervention there were no significant differences between the treatment or control groups, either in number of patients allocated to bandaging or hosiery alone or in characteristics such as height, weight, or percentage body fat (Table 2). There was no difference in the number of women taking hormonal treatment in each group. Dietary intake of energy and nutrients, as calculated from the 7-day dietary records, were similar in all 3 groups before dietary intervention.

Table 2. Characteristics of Treatment Groups Before Dietary Intervention
 ControlWeight reductionLow fatP
  1. Figures are mean (± standard deviation).

N = 5115 (30%)19 (37%)17 (33%) 
Median age, y696759.166
Percentage difference in arm volume444243.913
Mean height, cm159160163.199
Mean weight, kg73.868.471.9.515
Mean body mass index, kg/m2292627.293
Skinfold thickness measurements, mm (sum of 4 sites)655966.40
Bandaging and hosiery7 (14%)6 (12%)6 (12%) 
Hosiery alone8 (16%)12 (22%)11 (22%) 

After the 24-week period of dietary intervention there were significant differences in the mean body weight (P = .006), body mass index (P = .008), skinfold thickness (P = .044), and percentage body fat (P = .017) between the control group and both the weight-reduction and low-fat groups. There was a reduction in excess arm volume over the 24-week period but there was no significant difference between the 3 groups (Table 3). However, there was a significant correlation between weight loss and a reduction in excess arm volume irrespective of the dietary group (P < .002, Fig. 1). There were small changes in arm volume of the unaffected arm over the 24-week period compared with changes in the lymphedematous arm (Table 4).

Figure 1.

The correlation between weight change and percentage excess volume of arm for all study subjects over the 24-week intervention period is shown. Amount of weight lost (kg); weight Day 1 to weight Week 24; correlation coefficient (Spearman rank), 0.423; P = .002.

Table 3. Changes in Anthropometric Measurements of Participants Between Day 1 and Week 24
 Control n = 15Weight reduction n = 19Low fat n = 17P
  • NS indicates not significant.

  • A Kruskal-Wallis test was used to determine whether there was any significant difference between the 3 dietary groups.

  • Figures are mean (± standard deviation).

  • *

    Significant difference between the 3 groups P < .05.

Weight, kg
Day 173.8 (15.1)68.4 (10.3)71.9 (10.5) 
Week 2473.0 (16.0)64 (11.0)69.0 (11.0) 
Difference0.6 (3.0)*4.0 (2.7)*2.6 (3.0)*.006
Body mass index kg/m2
Day 128.9 (4.9)26.4 (3.9)27.3 (4.3) 
Week 2428.8 (4.9)24.8 (3.9)26.5 (4.0) 
Difference0.1 (1.3)*1.5 (1.2)*0.9 (1.2)*.008*
Skinfold thickness (combined for 4 sites), mm
Day 169.5 (18.3)59.3 (15.4)66.2 (12.8) 
Week 2466.8 (19.4)49.8 (16.4)60.0 (13.2) 
Difference2.7 (11.3)*9.5 (9.8)*6.2 (5.5)*.044*
Percentage Body fat
Day 131.9 (5.3)29.9 (4.0)31.8 (2.9) 
Week 2431.5 (4.4)27.1 (4.2)30.3 (3.4) 
Difference0.4 (2.3)*2.8 (2.7)*1.4 (1.3)*.017*
Percentage excess arm volume Day 144 (23)42 (22)43 (18).913
Percentage excess arm volume Week 2432 (17)28 (21)28 (15).605
Difference %12 (16)14 (13)15 (16)NS
Table 4. Changes in Arm Measurements Between Day 1 and the End of Week 24
 Control unaffected limbControl swollen limbWeight reduction unaffected limbWeight reduction swollen limbLow-fat group unaffected limbLow-fat group swollen limb
  1. Figures are mean (± standard deviation).

Day 1, mL2637 (783)3701 (860)2378 (392)3348 (602)2659 (521)3814 (1011)
End of week 24, mL2585 (694)3427 (896)2253 (379)3010 (538)2606 (533)3434 (917)
Differences, mL−52−229−125−338−53−380

There was a significant difference in mean weight loss between the 3 groups at both 12 weeks (P = .001) and 24 weeks (P = .006) (Table 3). Weight loss occurred in all groups, with 9 subjects (60%) losing weight in the control group, 18 (95%) in the weight-reduction group, and 13 (76%) in the low-fat diet group. The range of weight change in each group was −7.9 kg to +5.2 kg in the control group, −9.1 kg to ± 0.6 kg in the weight-reduction group, and −9.5 kg to +2.2 kg in the low-fat group (Fig. 2).

Figure 2.

Weight changes in the 3 groups during the 24-week study period and weight change of each individual over the 24-week period are shown. The data are grouped into intervention groups.

After the 24-week period of dietary intervention there was a significant difference between the control and both weight-reducing diet and low-fat diet groups with respect to reported intakes of all nutrients except carbohydrate (Table 5).

Table 5. Reported Changes in Dietary Intake Between Day 1 (1) and Week 24 (2)
 Control n = 15Weight reduction n = 19Low fat n = 17P
  • Data were tested to see if there were any significant differences between the 3 dietary groups.

  • Figures are mean (± standard deviation).

  • A Kruskal-Wallis test was used to determine whether there was any significant difference between the 3 dietary groups.

  • *

    Significant difference P < .05.

Energy, kcal 11710 (371)1817 (407)1674 (334) 
Energy, kcal 21693 (259)1290 (252)1336 (353) 
Difference18 (247)*527 (369)*339 (387)*.001*
Fat, g 173 (21)76 (27)68 (16) 
Fat, g 275 (17)45 (14)38 (16) 
Difference− 2 (19)*32 (23)*31 (20)*.000*
Protein, g 168 (11)73 (21)66 (12) 
Protein, g 270 (7)59 (8)66 (13) 
Difference− 3 (11)*14 (18)*0 (13)*.001*
Carbohydrate 1204 (48)206 (39)189 (48) 
Carbohydrate 2189 (41)165 (36)178 (45) 
Difference15 (37)*42 (42)*12 (47)*.042*
Percentage energy from fat 135 (5)38 (5)37 (4) 
Percentage energy from fat 239 (6)31 (5)25 (5) 
Difference−2 (8)*6 (5)*12 (7)*.000*


This is the first intervention study examining the role of diet as a possible treatment for patients with BCRL considered to be overweight. The primary aim of the study was to demonstrate whether a dietary intervention in the form of either an isoenergetic diet with a reduction in the proportion of energy from fat or a low-energy diet resulting in weight reduction would influence the extent of arm swelling when compared with standard care alone. The study showed a slightly greater reduction in excess arm volume in both dietary intervention groups compared with the control, although this was not statistically significant. Post-hoc analysis demonstrated a statistically significant correlation between weight loss and loss of swollen arm volume irrespective of the reason for weight loss. Over the period of 24 weeks the women in both intervention groups showed significant changes in their reported dietary intake, body weight, body mass index, and skinfold thickness.

The lack of a significant difference between the groups in reduction in arm volume may have arisen for a number of reasons: first, there may have truly been no difference between the groups. Alternatively, it may have been because the actual difference between the groups could not be detected due to too few study participants, an initial overestimate of the effectiveness of diet or poor adherence to the intervention. Because of the lack of previous studies of this nature it was difficult to estimate the potential impact of dietary intervention and therefore a realistic sample size calculation. In this study the poor adherence to the dietary intervention may have had a major impact on the results.

Based on the suspicion that a type II error had occurred, a post-hoc analysis of weight change and reduction in arm volume across all 3 groups was carried out. Figure 1 demonstrates that weight reduction had a significant impact on the reduction of volume in the lymphedematous arm (correlation coefficient: 0.423; P = .002). There was a significant correlation between weight loss and reduction in percentage of excess arm volume. A number of subjects in the low-fat group lost weight due to a fall in their energy intake and they too appeared to benefit in terms of excess arm volume reduction. The standardized effect size for percentage arm reduction between the control group and weight-loss group was 0.28, and between the control group and low-fat group was 0.37.27 This would indicate that in designing future trials, in order to have 90% power to detect a significant difference (alpha = 0.05) between average reduction in arm volume, 270 people would be required in each group if comparing weight reduction and control and 155 patients in each group for the low-fat vs control comparison.

Within each dietary group there are patients whose diet records show that they did not comply with the dietary advice given. This was confirmed with additional analysis of the dietary records that compares recorded intake with an estimated energy requirement.28 In addition, a number of women in the weight-reduction group failed to lose weight. Conversely, there were women in the control and low-fat groups who were advised to maintain weight but who actually lost weight during the study period (Fig. 2). In both the control and low-fat groups there were individual participants who lost up to 9 kg, a weight loss that was comparable to that of some of the participants in the weight-reduction group. These individuals, as part of a relatively small study group, may have affected the statistical significance of the results.

The group of women recruited to the study showed a higher degree of obesity than the normal population. This would perhaps be expected for women with breast cancer, as obesity and weight gain during adult life have been shown to be risk factors for postmenopausal breast cancer.29 In the weight reduction group all subjects except 1 lost weight. As a group they showed a mean loss of 4 kg over 24 weeks, with a 3.2 kg loss at 12 weeks, which compared well with the mean loss of 6 kg in 1 year achieved by a multicenter weight-reduction trial in breast cancer patients30 and is comparable with other studies.31, 32 The weight loss compares poorly, however, with the expected weight loss of approximately 0.7 kg per week that would be achieved if the women had adhered to a 1200-kcal reducing diet. Such a diet would achieve a net energy deficit of 700 kcal assuming a requirement of 1900 kcal daily.33 Indeed, given the strong correlation between weight loss and arm volume reduction (Fig. 2) it is proposed that if all patients adhered to a weight-reduction program dietary manipulation could prove a very effective treatment for lymphedema in overweight breast cancer patients. This study does demonstrate, however, that this group of women are capable of losing weight despite the observation that when body weight is discussed with women with breast cancer they comment on how difficult it is to lose weight, especially when taking endocrine therapies.34

The majority of the low-fat diet group (76%) lost weight, with a mean group loss of 2.9 kg from Day 1 to the end of Week 24. Subjects in this group were given dietary advice to increase their carbohydrate intake with the aim of maintaining their weight. This finding compares with another study undertaken in premenopausal and postmenopausal women at the Royal Marsden Hospital, London, where the participants lost an average of 0.5 kg and 1.4 kg, respectively, when following a low-fat diet over a period of 1 month. These subjects had also been given advice on increasing their intake of carbohydrate foods in order to maintain body weight.35 It was not clear from those results whether the low-fat diet alone was having any beneficial effect or whether this was due to the influence on body weight.

During the period of the study the weight reduction and low-fat dietary groups apparently achieved a significant change in their reported dietary intake as assessed by a 7-day dietary diary, although the records of reported intake indicated that not all individuals within each dietary group managed to comply with the dietary advice given. At Week 24 only 8 of 19 (42%) of the weight reduction group and 10 of 17 (59%) of the low-fat group reported an intake that adhered to the dietary advice given. In addition, there was no correlation between reported energy intake and weight loss in either of the dietary intervention groups. In this study, as in others, dietary intake may have been influenced during the period of recording and may therefore not represent true habitual intake.36 Keeping dietary records may influence subjects either to undereat or to under-report their intake.37 If the study groups had been larger in size then additional post-hoc analysis on a subset of those whose adherence to the diet was shown to be good could have been interesting, but as the study groups were small this was not considered appropriate.

In conclusion, dietary manipulation in patients with lymphedema subsequent to breast cancer treatment appears to assist in the reduction of lymphedematous arm volume if body weight is reduced. A low-fat diet did not appear, in this group of patients, to assist in the reduction of arm volume unless the patient lost weight. Weight reduction, however, achieved either by a preferential reduction in fat intake or by a reduction of energy from dietary fat, protein, and carbohydrate appears to be an effective means of assisting in the reduction of arm volume during treatment of the lymphedematous arm.