Introduction

Breast cancer-related lymphoedema (BCRL) is one of the most dreaded complications after treatment for breast cancer. The risk factors for BCRL are axillary clearance, radiation therapy, high BMI and post-operative infections [1–3]. The incidence of BCRL is related to the invasiveness of axillary lymph node extirpation, with less BCRL in sentinel node negative patients, and ranges between 12.5 and 49 % [3–6]. The pooled incidence for BCRL, taking into account the larger part of sentinel negative patients, is 16.6 % [3]. BCRL is now recognized as a chronic disease affecting most frequently the upper extremity, followed by the chest wall and breast [7]. This condition can develop directly after surgery or post-radiation therapy, although it can also occur months and even years later [4].

Women with BCRL complain of a reduced quality of life (QOL) [8] and tend to have higher rates of mental health problems [9], while shoulder stiffness and functional limitations in activities of daily living are also reported [10, 11]. Consequently, BCRL has implications on the ability to work, and hence lead to high direct and indirect monetary costs. After breast cancer treatment, women cannot return to work for 10.8 months on average, while in BCRL patients, this period is 12.9 months on average [12]. It is in the interest of the patient, the medical staff, the therapist and the insurance companies, to make the treatment as effective and as acceptable as possible.

The consensus document of the International Society of Lymphology for evaluation and managing peripheral lymphoedema [13] described the following treatment techniques for BCRL reduction: manual lymphatic drainage (MLD), compression bandaging, active exercises, and skin care. In the literature, this consensus treatment is referred to as complex decongestive therapy because the treatment is a combination of the mentioned treatment modalities. Two reviews and one meta-analysis evaluating the effectiveness of different treatment methods are available [14–16]. None of these reviews evaluated precisely the reduction of oedema after a comprehensive treatment or after an exercise intervention without MLD. Therefore, the aim of this present systematic review and meta-analysis was to evaluate the effect of compression and exercise modalities for the management of BCRL. The research question for the study was as follows: What are the effects of compression (bandages) and active exercise during the intensive phase of therapy in the reduction of lymphoedema in breast cancer patients?

Method

Study search

The methods used for this systematic review were based on the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [17, 18]. This systematic literature search was conducted using multiple electronic databases from January 2015 until January 2016. The literature search was performed in CINAHL, Cochrane Register of Controlled Trials, and Physiotherapy Evidence Database (PEDro). The unpublished International Clinical Trials Registry Platform from the World Health Organization (WHO) was also searched. The reference list of all relevant studies was cross-referenced in order to find further literature. This systematic review and meta-analysis were registered at PROSPERO (CRD42014010700).

Study selection

Two independent reviewers (MA, SM) screened titles and abstracts for eligibility. The decision to read the full text was made based upon pre-defined eligibility criteria. Keywords and combination to the PICO-model were used for the search strategy:

Population (P): female or women; Intervention (I): lymphatic drainage or lymphtape or compression bandage or sleeve or intermittent pneumatic compression (ICP) or exercise; Comparator (C): Compression bandage against control intervention or compression bandage against exercise and Outcome (O): volume or oedema reduction.

Afterwards, three independent reviewers (SR, JT, NG) read the full text and selected the studies to include in the systematic review and meta-analysis if they (1) were RCTs, (2) reported mean and SD (or standard error) or mean change and SD (or standard error) or medians and interquartile range (3) were written in English or German language and (4) mentioned one of the following keywords in the title or abstract: lymphoedema, women, mastectomy, axillary dissection or breast cancer.

A study was excluded when the effect of (1) drugs, hormonal, radiation and surgical procedures was examined. The other exclusion criteria were studies with (2) children in the test groups, (3) non-breast cancers, (4) lower extremity oedema, (5) impact on fatigue only, (6) diets or sexually transmitted diseases, (7) cost-analysis only and (8) syndromes that are not carcinogenic nature or (9) investigation of the prevention of breast cancer.

Quality assessment

General study characteristics were extracted by two independent reviewers (MA, SM). The following information was included in this systematic review: study design, participants (N and age), intervention, outcomes and results.

The Cochrane Collaboration’s Risk of Bias (RoB) tool [19] was used to assess the methodological quality of the included studies by two independent reviewers (MA, SM). The RoB criteria list covers six items that represent the aspects of internal validity. Each item was scored with “−” for no, with “+” for yes and with “?” if the information was unclear. A study was defined as having a low risk of bias if all criteria were fulfilled with yes. A study had a moderate risk of bias when one or more items were rated unclear, while a study was coded as high risk of bias if one or more key domains have been rated with no. Where discrepancies existed, a third reviewer (SR) intervened to obtain a consensus.

A meta-analysis was performed if two or more studies had measured and reported the same outcome. If more than one outcome variable was reported, the reviewers (SR, JT, NG) will decide, without knowledge of the results, which outcome variable should be pooled [20]. The decision was based on the reviewers’ judgment. The main outcomes were reduction of oedema volume and reduction of arm volume.

The meta-analyses used a random-effects model. The effect sizes were expressed as standardized mean differences (SMDs). To explore the review questions, the following meta-analyses were conducted: (i) compression (bandage, sleeve, intermittent pneumatic pressure) versus control for reduction of oedema volume and (ii) exercise versus control for reduction of oedema volume. Furthermore, (iii) a subgroup one-arm pre–post-intervention effect analysis of compression and exercise on reduction of oedema volume was carried out.

Heterogeneity of treatment effects across the individual study estimates was investigated statistically using the Cochran’s Q statistic and its corresponding degrees of freedom and p value. Higgins’ I 2measure was used to determine how much of the observed variability can be explained by the true between-studies variability. Higgins’ proposed benchmarking was used for the interpretation of these heterogeneity measures. An I 2 around 25 % indicates that the heterogeneity might not be important, while an I2 around 50 % and I 2 around 75 % suggest that heterogeneity is moderate and substantially considerable, respectively [21].

For clinical interpretation of the findings based on the data from the included RCTs, the overall weighted standardized mean difference estimate of the meta-analysis was re-expressed in the original units using the “familiar instrument method” as proposed in the Cochrane handbook for systematic reviews of interventions [22]. For clinical interpretation of the findings based on the data from the included pre–post studies, the overall weighted standardized mean difference estimate of the meta-analysis was re-expressed in the original units using the “rule of thumb for effect sizes method” (i.e. Cohen’s benchmarking of effect sizes) as proposed in the Cochrane handbook for systematic reviews of interventions [22].

Risk for publication bias was assessed by funnel plot inspection and the classic fail-safe N algorithm.

For all analyses, p values less than 0.05 were considered statistically significant. All calculations and plots were conducted using the CMA-2 software (Comprehensive Meta-Analysis 2nd version, Biostat, Englewood, NJ, USA).

Results

Flow of studies through this review

Figure 1 depicts the flow process of studies in this systematic review and meta-analysis. In total, 543 articles were found. After removing duplicates and reviewing 411 titles and abstracts, 121 original articles were read in detail. Overall, 32 studies were selected and included for the systematic review, while nine studies were selected for the RCT-based meta-analyses, and 19 studies were included in the pre–post studies-based meta-analyses.

Fig. 1
figure 1

Flow chart of this study

Full size image

Risk of bias

Table 1 shows the RoB assessment of the included studies. Most studies lacked concealed allocation and blinding and therefore showed a moderate to high risk of BIAS.

Table 1 Overview of Risk of Bias (RoB)
Full size table

Study characteristics

The study characteristics are summarized in Table 2. Haghighat et al. [23] and Schmitz et al. [24] included more than 100 participants in their study. The other studies showed a sample size of less than 100 participants. The intervention method and outcome varied across all included studies.

Table 2 Study characteristics
Full size table

Effect of intervention

Nine RCTs could be used to evaluate the effect of intermittent pneumatic compression (IPC), use of a sleeve or exercise vs. control on reduction of oedema (Fig. 2).

Fig. 2
figure 2

Forest plot presenting the effects of intermittent pneumatic compression (IPC), use of a sleeve and exercise on the reduction of lymphoedema in patients with breast cancer based on the RCT-designed studies. Values on x-axis denote standardized mean differences. The diamond illustrates the 95 % confidence interval of the pooled effects. The horizontal line at the diamond illustrates the 95 % prediction intervals indicating that 95 % of the future studies will lie within this interval

Full size image

The meta-analysis for exercise yielded a SMD of −0.49 [95 % CI −0.86 to −0.11] (p = 0.011). The heterogeneity was low (Cochrane`s Q = 2.53; df = 3; p = 0.470) with I2 of 0 %. After re-expression in its original metric, the overall weighted effect size corresponded with a reduction of oedema volume of about 200 cm3.

The meta-analysis for IPC showed a SMD of −0.54 [95 % CI −1.01 to −0.064] (p = 0.026). The heterogeneity was low (Cochrane`s Q = 1.36; df = 1; p = 0.244) with I 2 of 26.3 %. After re-expression in its original metric, the overall weighted effect size corresponded with a reduction of oedema volume of about 400 cm3.

The meta-analysis for the use of a sleeve showed an overall weighted SMD of −0.15 [95 % CI −0.44 to 0.14] (p = 0.314). The heterogeneity was low (Cochrane`s Q = 0.49; df = 2; p = 0.782) with I 2 of 0 %. After re-expression in its original metric, the overall weighted effect size corresponded with a reduction of oedema volume of about 50 cm3.

Nineteen studies could be included in a meta-analysis of effect sizes from pre–post-intervention studies and from multiple-armed RCTs, of which the arm of interest was extracted and used as an individual pre–post study. This allowed for the inclusion of bandage as an extra type of compression (Fig. 3).

Fig. 3
figure 3

Forest plot presenting the effects of bandage, intermittent pneumatic compression (IPC), use of a sleeve and exercise on the reduction of lymphoedema in patients with breast cancer based on the (uncontrolled) pre–post-intervention data. Forest plot of the effects of WBV plus exercise compared to exercise on TUG. Values on x-axis denote standardized mean differences. The diamond illustrates the 95 % confidence interval of the pooled effects. The horizontal line at the diamond illustrates the 95 % prediction intervals indicating that 95 % of the future studies will lie within this interval

Full size image

The meta-analysis for bandage showed a SMD of −0.33 [95 % CI −0.48 to −0.17] (p < 0.0001). The heterogeneity was low (Cochrane`s Q = 6.34; df = 7; p = 0.501) with I 2 of 0 %. Using the rule of thumb for the re-expression of the SMDs, this overall weighted SMD would correspond with a small effect size.

The meta-analysis for exercise showed a SMD of −0.074 [95 % CI −0.28 to 0.13] (p = 0.479). The heterogeneity was low (Cochrane`s Q = 0.93; df = 4; p = 0.920) with I 2 of 0 %. Using the rule of thumb for the re-expression of the SMDs, this overall weighted SMD would correspond with a small effect size.

The meta-analysis for intermittent pneumatic compression showed a SMD of 0.013 [95 % CI −0.25 to 0.28] (p = 0.926). The heterogeneity was low (Cochrane`s Q = 0.13; df = 2; p = 0.938) with I 2 of 0 %.

The meta-analysis for sleeve showed a SMD of −0.26 [95 % CI: −0.519 to 0.001] (p = 0.051). The heterogeneity was low (Cochrane`s Q = 0.74; df = 2; p = 0.690) with I 2 of 0 %. Using the rule of thumb for the re-expression of the SMDs, this overall weighted SMD would correspond with a small effect size.

Risk of publication bias was moderate. Figure 4 depicts the funnel plots for the meta-analyses based on RCTs and based on one-arm pre–post studies. No real critical funnel plot asymmetry was observed.

Fig. 4
figure 4

Funnel plots for the meta-analyses based on RCTs (left) and based on one-arm pre–post studies (right)

Full size image

The “classic fail-safe N” algorithm revealed that 46 and 22 missing non-significant studies would be needed to bring the p value above the alpha level of 5 % in the RCT- and pre–post-based analysis, respectively.

Discussion

This systematic review and meta-analysis aimed at evaluating the effect of different compression modalities (such as the use of bandage, sleeve or intermittent pneumatic compression) and exercise for the management of BCRL. First, the results from RCT’s are discussed; second, the results of the pre–post designs are discussed.

Four RCT’s reported on the effects of exercise [25–28]. Unfortunately, the exercise programs cannot be compared due to the large variation in protocol. Despite the different protocols (Yoga, Nordic Walking, Resistance training), all protocols favoured lymphoedema volume reduction. On recalculating, exercise resulted in a volume reduction of 200 ml. These results add to the knowledge that exercise is beneficial in the treatment of BCRL and does not aggravate lymphoedema [29, 30].

Two RCT’s reported from a sample of BCRL patients that additionally received IPC to the consensus treatment [23, 31]. Both IPC protocols were comparable, and a recalculation of the effect of IPC demonstrated that IPC was able to reduce lymphoedema volume to 400 ml in the intensive phase. Unfortunately the effect of IPC cannot be maintained in the maintenance phase as is demonstrated in another meta-analysis [32]. Therefore, these results should be interpreted with precaution. IPC lacks the ability to be a standalone therapy since it only stimulates the lymphatic drainage in working collectors. Therefore, IPC has a limited effect on the resorption of the interstitial oedema fluid.

Three RCT’s reported on the effect of a compression sleeve in the intensive phase [33–35]. In two studies, the compression sleeve was additional to exercises [34, 35], and in one study, the compression was the only treatment provided when arm volume started to increase in comparison to pre-operative volume [33]. The effect on oedema volume reduction was limited to 50 ml. These results were to be expected since a compression sleeve is not a treatment modality to reduce volume but to maintain the leanest volume. Therefore, a sleeve should not be used in the intensive phase unless the sleeve is provided very early after onset of lymphoedema, as was the case in the study of Stout-Gergich [33]. In the treatment of severe lymphoedema, a compression sleeve should be provided by the start of the maintenance phase to limit the risk of volume increase. In a large cohort study, it was demonstrated that patients who adhere to wearing the compression sleeve have the lowest risk for regaining oedema volume [36].

For the pre–post results, we were able to extract data concerning the use of bandages, IPC, compression sleeve and exercises. These results were based on a comparison between baseline measurements and measurements taken at the end of the intervention; therefore, no control group is available. Again, all interventions relate to the intensive phase of BCRL treatment.

Eight samples from five studies were selected to demonstrate the effect of compression bandages [37–41]. Overall, it was shown that bandaging has the ability to decrease the oedema volume in the intensive phase. As demonstrated by the different samples, therapists need to be aware that the pressure provided by the bandages must be optimal [37, 38]. Compression bandaging reduces volume more and faster when compared to wearing a compression sleeve in the intensive phase of the consensus treatment [41].

Continuing with the results concerning compression sleeve, we were able to extract data from three studies [41–43]. Comparable to the results from the RCT’s, compression sleeves had a low effect on volume reduction in the intensive phase. The small reduction of volume by wearing a compression sleeve is due to the increased interstitial pressure, limiting filtration. (ref: http://www.woundsinternational.com/media/issues/212/files/content_177.pdf) As stated before, compression sleeves are more appropriate in the maintenance phase.

In contrast to the results from the RCT’s, IPC [44, 45] as well as exercise [34, 46–48] effect sizes from the pre–post-designed studies showed no benefit on volume reduction. Especially for IPC, the results demonstrated a very low effect size, confirming that IPC is not a standalone therapy. For exercises, however, it is a mixed story. Low effect sizes were found in the study that did not include compression during exercise. [34] The studies that did combine exercise and compression demonstrated a better result: Gautam et al. [47] demonstrated a 122 ml reduction during exercise.

Unfortunately, the research question “what are the effects of compression bandages and active exercise on the reduction of lymphoedema volume in breast cancer patients during the intensive phase?” could not be answered conclusively. This conclusion is based upon the many encountered limitations in the selected papers. Therefore, several limitations of the current systematic review and meta-analysis need to be discussed. Overall, we were confronted with a low number of studies that reported on the outcomes selected for this meta-analysis and unfortunately most of them had but poor to moderate methodological quality. Due to the consensus treatment proposed by the ISL, it is difficult to select studies that scope only one treatment modality. Recently, two Cochrane reviews were published concerning the added value of MLD in the consensus treatment demonstrating likewise difficulties [14, 15]. In studies reporting from the consensus treatment, no information about the separate effects of the different modalities are reported. Many of the selected studies provided a general treatment based upon the consensus treatment and added the treatment modality of interest to the experimental group [23, 25, 43]. Besides the low number of studies, sample sizes of the selected studies were also low (n ranged from 7 to 56 patients). Furthermore, a risk of publication bias cannot be excluded. However, we believe that this risk is limited since a rigorous search was performed in different databases, and no real critical asymmetry was observed in the funnel plots.

Conclusion

This systematic review and meta-analysis showed some evidence that active exercising may reduce oedema volume in BCRL. IPC seems beneficial in helping to reduce the oedema volume in the acute phase of treatment, while compression sleeves do not aid in the volume reduction in the acute phase but they do prevent additional swelling. All conclusions should be taken with precautions because of the insufficient quality of the selected papers.