Transient swelling versus lymphoedema in the first year following surgery for breast cancer
The aim was to better understand the incidence, time course and risk factors for swelling in the arm on the side of surgery over the first year following surgery for breast cancer.
Women (n = 160) were assessed 1 month following surgery and then randomised to the exercise or control group. Reassessment occurred 3, 9 and 15 months following surgery. Potential risk factors for swelling included age, body mass index, side of surgery and surgical and medical treatments for their breast cancer, physical measures of shoulder range of motion and strength, inter-limb arm circumference difference and the group to which they were randomised. Swelling was determined using bioimpedance spectroscopy with reference to previously established cut-offs for lymphoedema.
The number of women with swelling at 3, 9 and 15 months was 15, 15 and 13, respectively; however, at 15 months only 5/13 presented with swelling in either of the preceding assessments. The risk of swelling increased at 3, 9 and 15 months for each centimetre increase in the baseline inter-limb difference in sum of arm circumferences by 1.30, 1.17 and 1.14. In addition, risk of swelling at 3 months was 2.6 times greater for women in the control group; at 9 months, 7 times greater for women who had taxane-based chemotherapy; and at 15 months, the risk increased 1.16 times for each day the drain was in situ.
Swelling in the first year is likely to be transient, and factors including exercise and taxane chemotherapy affect the risk of developing swelling.
KeywordsExtracellular fluidRisk factorPredictorLymphoedema
Lymphoedema in the arm on the side of surgery following treatment of breast cancer is well documented. It arises from disruption of the lymphatic system which is critical to regulation of fluid homeostasis. Swelling occurs when homeostasis is disrupted, such that there is an imbalance between the rate of interstitial tissue fluid generation (increase in the ‘lymphatic load’) and the degree to which the lymphatic vasculature is underdeveloped or damaged (reduction in the ‘lymphatic transport capacity’) . In women treated for breast cancer, it is postulated that impairment to lymph transport leads to lymphoedema, more likely in women who are predisposed to this condition [1, 2]. Women who normally have high lymph flow through their axilla are already near to their maximum capacity for transport of lymph; it is hypothesised that an additional insult to these women such as through axillary surgery further predisposes these women to lymphoedema .
While lymphoedema is described as ‘chronic’, for some women, the swelling in the arm may resolve. For example, Norman and colleagues  noted that mild swelling resolved in 30 women (23 %) followed up for 3 years. Similarly, Hayes et al.  found that lymphoedema had resolved by 18 months following surgery for 60 % of a group of 62 women diagnosed with lymphoedema between 6 and 12 months following surgery. This form of swelling may differ from that attributable to chronic lymphoedema, or there may be other intervening factors contributing to its resolution yet to be identified. Identification of underlying causes and factors which are associated with transient as well as established swelling will facilitate better prevention and treatment strategies.
One factor which may contribute to resolution of an increase in interstitial fluid is exercise . Lymph moves through the body via both intrinsic and extrinsic mechanisms . The extrinsic lymph pump relies on the cyclical compression/expansion from the action of the surrounding muscle tissue [7, 8]. Evidence that exercise may play an important role comes from a study of women at high risk for lymphoedema . Women treated for breast cancer 1 to 5 years previously were randomised to a resistance training programme or control group. Among women with five or more lymph nodes removed, the proportion who experienced onset of arm swelling was 7 % (3 of 45) in the weight-lifting intervention group and 22 % (11 of 49) in the control group. For women at high risk of arm swelling, resistance training was protective. When determining risk factors for swelling, often arm usage and exercise have not been considered [10–13] even though it may play an important role.
Treatment factors not previously considered may also contribute to the presence or absence of swelling and explain any transient changes. For example, the use of taxane-based chemotherapy is now common. One of its side effects is generalised oedema characterised by an increase in the size of the interstitial compartment of extracellular fluid (ECF) [14, 15]. Swelling as a consequence of increase in interstitial fluid, compounded by impaired transport of the fluid, may preferentially contribute to swelling in the at-risk arm .
Taken together, this work suggests that the first year of treatment for breast cancer comprises treatments and events which can lead to transient or chronic swelling. Understanding the timing and duration of swelling, particularly in the first year, will enable better advice to women and, potentially, targeted treatments to address the swelling. The aim of this study was to report on the incidence of swelling, particularly extracellular swelling, at 3, 9 and 15 months following surgery, and to examine risk factors at each time.
The study was a secondary analysis of data collected in a randomised controlled trial designed to investigate the effect of stretching and resistance training in women following early treatment for breast cancer [17, 18]. Women (n = 160) were recruited and underwent baseline measures 4 to 6 weeks post-operatively after which they were randomised to either an exercise group or the control group. Randomisation was balanced according to level of axillary surgery (sentinel node biopsy (SNB) versus axillary node dissection (AND)). The intervention period was 8 weeks, and women were assessed at baseline, following completion of the intervention period and at 6 and 12 months of follow-up by a researcher blinded to group allocation. These time-points coincided with 1, 3, 9 and 15 months post-surgery. Women in the exercise group attended weekly for upper limb strengthening and stretching exercises as well as undertaking a home exercise programme. Women in the control group attended fortnightly sessions in which their arm was assessed for presence of swelling; however, they were not provided any additional exercises. The protocol is described in detail elsewhere [17, 18].
Women diagnosed with early breast cancer were recruited post-operatively to the trial. Inclusion criteria included either SNB or AND, able to attend weekly sessions and able to communicate in English. Women were excluded if they presented with distal metastases or had upper-limb impairments prior to diagnosis of breast cancer. Ethics approval was obtained from the research institution as well as the hospital sites from which participants were recruited.
Risk factor variables
Nineteen potential risk factors were measured at baseline, 1 month following surgery. These included age, body mass index (BMI), dominance of the affected side and whether a woman was randomised to the exercise or control group. Breast cancer-related variables were extracted from participants’ medical records and included type of breast and axillary surgery, number of days the drain was in situ, number of nodes affected, and whether women had received radiotherapy, chemotherapy and, specifically, taxane-based chemotherapy. Physical measurements included inter-limb ECF ratio, arm circumference, range of motion and shoulder muscle strength. Self-reported measures of pain, fatigue, and overall general health were derived from the Breast European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Version 3 (QLQ-C30) and Breast Cancer module (Br23) [19, 20].
At 1 month following surgery, the inter-limb ECF ratio was determined by bioimpedance spectroscopy (BIS) using an SFB7 BIS instrument (Impedimed, Brisbane). The impedance of each arm was measured using the principle of equipotentials, thereby standardising limb length . Electrodes were placed on the dorsum of the wrists in line with the ulnar styloid, on the dorsum of the hands below the third digit and on the dorsum of the right foot. Data are presented as a ratio of impedance between arms (not at risk/at risk) . In addition, the circumference of each arm was measured at standardised locations, commencing at the ulna styloid and progressing proximally by 10-cm intervals to 40 cm. The measurements for each arm were summed, and the inter-limb difference in the sum of the circumferences was determined .
Shoulder passive range of motion and maximal isometric shoulder strength were assessed bilaterally. The protocol for range of motion was designed to minimise compensatory movements. Women were positioned in supine for measurement of forward flexion and in sitting for abduction range of motion; an inclinometer (Base Evaluation Instrument Model 12–1507, White Plains, NY) was used to measure the range obtained. For the assessment of shoulder muscle strength, women were seated, with their arm extended at 90° elevation. Women pushed up against the hand-held dynamometer (Chatillon Medical Dynamometers, Ametek, Largo, FL), positioned 20 mm proximal to the elbow. For each direction of range of motion and strength, three measurements were taken, and the highest was recorded. The inter-limb differences in range and strength for forward flexion and abduction were determined and used as risk factors.
For the quality-of-life measures of fatigue, pain and overall health, the raw scores for each domain were summed and subjected to linear transformation to provide an overall score, with a maximum of 100. A high score indicated a high level of symptoms or high level of function. The number of nodes affected ranged from 0 to 32, so we categorised this variable as 0, 1–2 or 3+ and treated it as ordinal.
The presence or absence of swelling was determined again at 3, 9 and 15 months following surgery. All measurements were made with bioimpedance spectroscopy as described above. Previously established cut-offs for the presence of swelling were used: a ratio ≥1.139 for women in whom the surgery was on their dominant side and a ratio ≥1.066 for those in whom the surgery was on the non-dominant side . BIS was preferred for this study as it quantifies only ECF ; the derived ratio is not influenced by change in body weight, and the cut-offs used to determine the presence of lymphoedema take into consideration arm dominance .
The number of women with swelling at each measurement occasion was identified. At each of the three time-points, those with and without swelling were compared using Wilcoxon rank-sum tests for continuous data and chi-square tests for categorical data, including a test for trend in proportions for the ordinal variable and number of nodes affected. Any potential risk factor with P < 0.2 was included in the initial multivariate logistic regression model for that outcome, up to a maximum of five variables (those with the lowest P values) because there were no more than 15 women with swelling at each time-point. Variables that were not statistically significant at P < 0.05 using the likelihood ratio test were then progressively eliminated, starting with the least significant, prior to checking the fit of the final model including checking for over-fitting. The best model with only two variables was, thus, chosen as the final model at each time-point. A log-binomial model that included only the variables in the final model was then fitted to obtain risk ratios instead of odds ratios . Statistical analyses were performed using Stata 11.2 (StataCorp, Texas).
Participant characteristics, cancer treatment and baseline physical measures of participants available at 3, 9 and 15 months after surgery
3 Months (n = 140)
9 Months (n = 141)
15 Months (n = 143)
Body mass index (kg/m2)a
Dominant limb affected n (%)
Mastectomy n (%)
Axillary node dissection n (%)
Number of nodes affectedbn (%)
Chemotherapy n (%)
Chemotherapy without taxane
Chemotherapy with taxane
Radiotherapy n (%)
Exercise group n (%)
Inter-limb difference in ROM (deg)a
Inter-limb difference in strength (N)a
Inter-limb difference in sum of arm circumferences (cm)a
Inter-limb ECF ratio a
Risk factors for elevated ECF
Bivariate relationships between patient characteristics and lymphoedema at 3, 9 and 15 months after surgery
3 Months following surgery (n = 140)
9 Months following surgery (n = 141)
15 Months following surgery (n = 143)
Swelling (n = 15)
No swelling (n = 125)
Swelling (n = 15)
No swelling (n = 126)
Swelling (n = 13)
No swelling (n = 130)
Side of surgeryb
Exercise group allocationb
Arm exercise group
Wide local excision
Axillary node biopsy
Sentinel node biopsy
Number of days drain in situa
Number of nodes affectedb
6 (11 %)
47 (89 %)
2 (4 %)
49 (96 %)
2 (4 %)
50 (96 %)
3 (6 %)
50 (94 %)
5 (9 %)
48 (91 %)
5 (9 %)
49 (91 %)
6 (18 %)
27 (82 %)
8 (22 %)
28 (78 %)
5 (14 %)
31 (86 %)
Inter-limb difference in sum of circumference measures (cm)a,e
Inter-limb difference in forward flexion ROM (degrees)a,e
Inter-limb difference in abduction ROM (degrees)a,e
Inter-limb difference in shoulder forward flexion strength (N)a,e
Inter-limb difference in shoulder abduction strength (N)a,e
Multivariable logistic regression
Risk ratios for prognostic factors for swelling at 3, 9 and 15 months following surgery
Time of assessment (months)
Unadjusted risk ratioa
Adjusted risk ratioa
95 % CI
0.90 to 7.40
Inter-limb difference in sum of arm circumferences (per cm)
1.24 to 1.36
1.73 to 31.6
Inter-limb difference in sum of arm circumferences (per cm)
1.12 to 1.22
Number of days drain in situ (per day)
1.04 to 1.25
Inter-limb difference in sum of arm circumferences (per cm)
1.05 to 1.24
At 9 months, the only significant variables retained were whether women had undergone taxane-based chemotherapy (aRR = 7.4, 95 % CI 1.7 to 31.6) and the inter-limb difference in sum of arm circumference measures at 1 month following surgery (aRR = 1.17, 95 % CI 1.12 to 1.22 per cm). In addition, all 15 women with swelling had had AND, but this could not be included in the regression model because it meant that SNB was a perfect predictor of not having swelling.
At 15 months, the only significant variables retained were the number of days the drain was in situ (aRR = 1.16, 95 % CI 1.04 to 1.25) and the inter-limb difference in sum of arm circumference measures at 1 month following surgery (aRR = 1.16, 95 % CI 1.05 to 1.24 per cm).
Swelling that occurred in the arm on the side of surgery in the first year is likely transient. Although it is extracellular, we question whether it is ‘lymphoedema’ or even ‘latent lymphoedema’. Typically, lymphoedema is reported as a chronic debilitating condition in which there is persistent fluid-based oedema in concert with changes to cellular tissue architecture . In our study, the number of women with elevated ECF in their arm on the side of surgery from 1 month up to and including 9 months following surgery was 22; only five of these women (23 %) still had elevated ECF in their ‘at-risk’ arm at 15 months. This pattern of resolution of swelling, particularly in the first year, has been noted by others [3, 4].
Inclusion of women with transient swelling in the first year inflates the reported incidence of lymphoedema, as the women for whom it resolves are not necessarily removed from further analysis. For example, Armer et al. , using survival analysis, identified up to 60 % of women presenting with lymphoedema by 60 months following surgery. In this analysis, anyone with swelling, regardless of whether it had resolved, was included. The effect that this inclusion of women with transient swelling has on incidence is illustrated by the data of Hayes et al. . The point prevalence of lymphoedema in their study at 18 months was 15 % compared to the cumulative prevalence of swelling, which was 34 %. For more than half the women, the swelling and likely associated symptoms had resolved. The corresponding figures for our study are a point prevalence of 9 % (13/143) at 15 months compared with a cumulative prevalence of 21 % (30/145).
Currently, diagnosis of lymphoedema is made on the presence of ‘swelling’ regardless of duration. If the definition of lymphoedema is chronic swelling, it would suggest that the condition should be present for some duration prior to labelling the swelling as lymphoedema. As we and others [3, 4] have shown that transient swelling appears to be particularly noticeable in the first year following treatment for breast cancer, it may be appropriate not to classify any swelling present in the first year as lymphoedema, unless it has been present for at least 6 months. Refining the definition may resolve the high variability in reported incidence of breast cancer-related lymphoedema. Also, as women are fearful of developing lymphoedema, a message that the incidence is relatively low but transient swelling is common may reduce some of the anxiety that lymphoedema generates.
Findings from this study highlight the importance of arm usage early after surgery and complement those by others who found that resistance training commencing later was not associated with lymphoedema [27–30]. In our randomised controlled trial [17, 18], women in the exercise group commenced 1 month following surgery with a light resistance, either from Theraband™ or free weight. However, by the second week of the 8-week intervention, women were working at a level equivalent to ‘hard’ on the Borg Effort Scale . This programme was protective against development of swelling when assessed at the conclusion of the intervention period, 3 months following surgery. The findings are supported physiologically as resistance training would assist with lymph transport. The tight tethering of the lymphatic wall to the surrounding interstitium helps transmit the cyclical compression/expansion action of the muscle to the lymphatics [27, 28]. We hypothesise that women discontinued the exercises when the intervention period finished, losing the beneficial effects at 9 and 15 months following surgery. It has been believed that moderate to vigorous use of the at-risk arm leads to lymphoedema [32, 33], and this belief may have been perpetuated among health-care workers and resources available to the women .
Taxane-based chemotherapy can cause generalised swelling characterised by an increase in the size of the interstitial compartment of ECF [14, 15]. In healthy adults, ECF comprises approximately 45 % of fluid in normal tissue and is made up of interstitial fluid, plasma, dense connective tissue water, bone water and transcellular fluid. The interstitial component which includes lymph comprises about 44 % of ECF . Both interstitial fluid pressure and volume increase from taxane therapy due to the stimulation of lymph flow . For some women treated for breast cancer, impairment of lymph transport from axillary dissection may be compounded by increase in lymph volume from taxane-based therapy leading to chronic lymphoedema, whereas for others, it may be transient. Identification of strategies to prevent development of this swelling are now needed. One strategy may be resistance training, as there is preliminary evidence from a study by Courneya et al.  that women undertaking resistance training while on chemotherapy therapy have a low incidence of swelling in their at-risk arm.
At 15 months, the relative risk of having a swelling increased with the number of days the drain was ‘in situ’. Examination of the data revealed that of the four women who presented with swelling at each of the three reassessments, two had the drain in for 8 days, one for 7 days and one for 4 days. For women with axillary node dissection, the median number of days the drain was in situ was five which was exceeded by three of the four women. Physiologically, the extended period of requiring a drain may be indicative of a lymphatic system working at near capacity and, thus, at risk for lymphoedema .
The other risk factor, which was identified at all three time-points, was the inter-limb difference in arm circumference at baseline. The mean (standard deviation) inter-limb difference at 1 month following surgery was 0.7 cm (2.7 cm) and ranged from −6.3 to 14.5 cm. The ECF ratio at baseline was significantly related to swelling at 3 months but was not retained as a significant predictor after allowing for taxane-based chemotherapy and inter-limb difference in arm circumference at baseline. For women with chronic lymphoedema, it is possible that the quantity of excess fluid decreased with concomitant increase in cell numbers within all layers of the affected tissues along with profound and progressive collagen deposition within the affected tissues . However, this factor was noted at 3 months, that is, prior to any opportunity for these long-term changes to occur. The cut-offs for ECF are based on three standard deviations above the mean of a cohort of healthy women [21, 24]. While it prevents a type 1 error from being made (i.e. classifying someone as having lymphoedema when they do not), it may result in type 2 errors, misidentifying those who, in fact, do have the condition. The ideal way to prevent these misclassification errors is to undertake measurements pre-operatively and identify whether an increase ≥0.1 occurs [38, 39].
All those women with swelling at 9 months had had axillary dissection, but this could not be included as a predictor in the regression model because it implied that SNB was a perfect predictor of not having swelling. In this study, all women had undergone axillary surgery, with 62 % undergoing axillary node dissection and 38 % undergoing SNB. Radiotherapy to the axilla following treatment for breast cancer has been linked with development of lymphoedema. However, at the centres where our study was conducted, radiotherapy to the axilla had been discontinued as routine practice; instead, radiation to the breast/chest and occasionally the supraclavicular region was used . Radiotherapy to these regions is not strongly associated with lymphoedema . Obesity has also been commonly associated with the occurrence of lymphoedema , and it appeared to be associated with swelling in our cohort at 3, 9 and 15 months but was not retained in the regression models after allowing for stronger risk factors. This may be a risk factor for chronic lymphoedema rather than this transient form seen in the first year.
This study is based on secondary analysis of data and was, therefore, not adequately powered to examine all potential risk factors for swelling at 3, 9 and 15 months post-operatively. A larger sample would allow a more detailed examination of the risk factors because more women would be above the ECF cut-offs used to determine the presence of lymphoedema.
In summary, swelling in the first year is likely to be transient. Taxane-based chemotherapy causes preferential transient swelling in the arm on the side of surgery. Factors that might help prevent or resolve swelling in the first year and prevent its recurrence in the future include resistance exercise.
The Cancer Council NSW provided a grant to support this project. SLK is supported by the National Breast Cancer Foundation, Australia.
Conflict of interest
Author Ward has consulted to ImpediMed Ltd. ImpediMed Ltd. had no involvement in the conception and execution of this study or in the preparation of the manuscript.