J Cancer Surviv DOI 10.1007/s11764-015-0435-1

REVIEW

A systematic review of axillary web syndrome (AWS) W. M. Yeung & S. M. McPhail & S. S. Kuys

Received: 14 July 2014 / Accepted: 27 January 2015 # Springer Science+Business Media New York 2015

Abstract Introduction Axillary web syndrome (AWS) can result in early post-operative and long-term difficulties following lymphadenectomy for cancer and should be recognised by clinicians. This systematic review was conducted to synthesise information on AWS clinical presentation and diagnosis, frequency, natural progression, grading, pathoaetiology, risk factors, symptoms, interventions and outcomes. Methods Electronic searches were conducted using Cochrane, Pubmed, MEDLINE, CINAHL, EMBASE, AMED, PEDro and Google Scholar until June 2013. The methodological quality of included studies was determined using the Downs and Black checklist. Narrative synthesis of results was undertaken.

Electronic supplementary material The online version of this article (doi:10.1007/s11764-015-0435-1) contains supplementary material, which is available to authorized users. W. M. Yeung (*) Department of Physiotherapy, Princess Alexandra Hospital, Metro South Hospital and Health Service, Ipswich Road, Brisbane, QLD 4102, Australia e-mail: [email protected] S. McPhail Principal Research Fellow, School of Public Health and Social Work, Queensland University of Technology, Brisbane, QLD, Australia S. McPhail Centre for Functioning and Health Research, Metro South Hospital and Health Service, Brisbane, QLD, Australia

Results Thirty-seven studies with methodological quality scores ranging from 11 to 26 on a 28-point scale were included. AWS diagnosis relies on inspection and palpation; grading has not been validated. AWS frequency was reported in up to 85.4 % of patients. Biopsies identified venous and lymphatic pathoaetiology with five studies suggesting lymphatic involvement. Twenty-one studies reported AWS occurrence within eight post-operative weeks, but late occurrence of greater than 3 months is possible. Pain was commonly reported with shoulder abduction more restricted than flexion. AWS symptoms usually resolve within 3 months but may persist. Risk factors may include extensiveness of surgery, younger age, lower body mass index, ethnicity and healing complications. Low-quality studies suggest that conservative approaches including analgesics, non-steroidal anti-inflammatory drugs and/or physiotherapy may be safe and effective for early symptom reduction. Conclusions AWS appears common. Current evidence for the treatment of AWS is insufficient to provide clear guidance for clinical practice. Implications for Cancer Survivors Cancer survivors should be informed about AWS. Further investigation is needed into pathoaetiology, long-term outcomes and to determine effective treatment using standardised outcomes. Keywords Axillary web syndrome . Axillary lymphadenectomy . Breast cancer . Melanoma . Post-operative morbidity . Pain

S. Kuys Allied Health Research Collaborative, Metro North Hospital and Health Service, Brisbane, QLD, Australia

Introduction

S. Kuys School of Allied Health Sciences, Griffith Health Institute, Griffith University, Gold Coast, QLD, Australia

Axillary web syndrome (AWS) consists of the appearance of a visible web of axillary skin overlying palpable cords of tissue

J Cancer Surviv

that are made taut and painful by shoulder abduction [1]. The cord can feel like a rope [2, 3], a band [4], a tendon [5, 6], a string [7], violin string [5], guitar strings [8] or taut wire [9] under the skin and may be comprised of multiple cords bound together like a web [1, 8]. Earlier observation of string-like bands similar to AWS occurred in 1960 [10] under the name of string phlebitis in 1963 [11]. Superficial lymphatic thrombosis was described in 1996 [12] (as cited in Torres Lacomba et al. [8]), and the term ‘AWS’ was later coined in 2001 [1]. AWS has been called by different names, including axillary variant of Mondor disease [1], cording [2–4, 6, 7, 13–25], webbing [17, 19], cording lymphoedema [21], axillary web cord [26], lymphatic cord [2, 17, 19, 27–32], vascular string [5], syndrome of the axillary cords/adhesion [21], superficial lymphatic thrombosis [30], fibrous cords [30, 33–35], fibrotic bands [35–37], axillary string [5, 38], aseptic lymphangitis [38], vascular ring [38], lymph vessel fibrosis [39] or fiddlestring phenomenon [39]. AWS typically results from axillary lymphadenectomies for treatment of malignancy such as breast cancer or melanoma and can cause physical and psychological morbidity for patients. AWS has the potential to hinder subsequent treatment by impairing the positioning of patients for radiotherapy (XRT) where the arm must be positioned overhead in abduction and external rotation [19, 39]. A proportion of patients might recover within the first few months of onset, but some continued to develop persistent or recurrent symptoms [38]. AWS remains a poorly defined and a misunderstood syndrome [2, 18]. Information on the clinical presentation/diagnosis, frequency, natural progression, grading, pathoaetiology, risk factors, symptoms, interventions and outcomes of AWS is often based on retrospective, case studies or other forms of low-level evidence, making it difficult to synthesise the available information. This review aims to systematically retrieve, appraise and synthesise findings to help inform the diagnosis and management of AWS as well as provide a summary of current evidence to inform subsequent research in this field.

Methods Literature search A systematic electronic search of the Cochrane Network, Pubmed, MEDLINE, Cumulative Index to Nursing and Allied Health Literature (CINAHL), EMBASE, Allied and Complementary Medicine Database (AMED) and Physiotherapy Evidence Database (PEDro) databases (from beginning of databases to June 2013) was undertaken using the search algorithm: (breast cancer OR breast tumour OR breast

neoplasm OR breast malignancy) OR (skin cancer OR skin tumour OR skin neoplasm OR skin malignancy OR melanoma) AND (axilla* web OR axilla* cord* OR axilla* band* OR axilla* string*) OR axillary web). Google Scholar was also searched to identify relevant studies not included in the aforementioned databases (e.g. peer-reviewed academic dissertation). Google Scholar search terms used were the following: ‘axillary web’ OR ‘axillary cord’ AND ‘breast cancer’ OR ‘breast tumour’ OR ‘breast neoplasm’ OR ‘breast malignancy’ OR ‘skin cancer’ OR ‘skin tumour’ OR ‘skin neoplasm’ OR ‘skin malignancy’ OR ‘melanoma’. Reference lists of relevant studies selected for the review were manually searched for further studies that may meet the study selection criteria. Where it was unclear, the full texts were obtained for review. No restrictions were applied for language, types or year of publication in the search strategy. The search was conducted by one author (WY).

Study selection After removing duplicate titles, all authors independently screened titles and abstracts to remove manuscripts that did not meet the inclusion criteria. Eligible studies included data on AWS frequency, pathoaetiology, diagnostic characteristics, risk factors, interventions and/or outcomes. Studies were included if written in English or accessible and able to be translated into English. Studies that did not make reference to a previously reported definition of AWS in the manuscript text, or describe diagnostic criteria consistent with previous AWS definitions, were excluded from the review due to the potential for uncertainty as to whether or not study participants definitively had AWS. For studies that had used terminology consistent with AWS but did not make specific reference to a previous definition or describe their diagnostic criteria, corresponding authors were contacted via email for opportunity to provide further information regarding the AWS diagnostic criteria applied in their investigation. Studies were not excluded on the basis of study design or methodological quality; however, quality was taken into account when interpreting results. Studies considered ineligible for inclusion were animal or cadaveric studies, book chapters, narrative reviews and conference proceedings. Studies associated with Mondor’s disease were excluded as they were beyond the scope of this review. Full text of the remaining studies were then retrieved and reviewed independently by all authors to confirm eligibility for inclusion. Justifications for excluding studies were noted and discrepancies between authors discussed. Authors

J Cancer Surviv

reached consensus without the need for an independent researcher to adjudicate any disagreements. Data extraction and analysis Data were extracted by one author (WY) and confirmed (SK, SM) using a purpose-developed data extraction form. Data extracted included lead author, year, setting, study design, subject characteristics, diagnosis and oncology therapies, AWS frequency and time frame to occurrence, AWS interventions and outcomes. Extraction items and definitions were developed a priori with consensus reached via pilot testing. Any discrepancies were clarified during regular author meetings through consensus. Kin relationships, defined as multiple publications describing the same or overlapping series of patients, were identified to avoid double counting of participants. Level of evidence appraisal for selected studies was undertaken using the National Health and Medical Research Council (NHMRC) guidelines (Supplementary 1) [40]. The NHMRC designation of levels of evidence has been used in Australia since 1999, to rank the body of evidence according to a hierarchy which indicates the level of bias associated with the different study designs that have contributed to the evidence base (Supplementary 1). This hierarchy ranks the body of evidence into four levels—from systematic reviews of randomised trials at the top of the hierarchy, to case series and case reports at the bottom of the hierarchy. Independent reviewers assessed potential risk of bias of individual studies, with consensus, using a quality scoring sheet based on the Downs and Black checklist (Supplementary 1) [41] and tables where complied to assess risk of bias (Table 1, Supplementary 2). The instrument consists of a 27-item checklist that rates randomised and non-randomised studies on the following key areas: reporting, external validity, internal validity, bias, confounding and power. Each element of the 27-item checklist is given a score of 1 (=yes) or 0 (=no or unable to determine), except for question five, which may score 2 points. The score for each trial was summed to give an overall score. The scoring for question 27 dealing with statistical power to detect a clinically important difference was simplified to a choice of awarding either 1 point (when evidence of sufficient statistical power was described) or 0 points (when evidence of sufficient statistical power was lacking) as has been previously reported [42]. Downs and Black score ranges have been reported according to corresponding quality level: excellent (26 to 28), good (20 to 25), fair (15 to 19) and poor (less than 14) [42–44]. Case studies and case series are generally associated with high risk of bias and were excluded from quality assessment. Once again, any minor discrepancies in quality item ratings were settled during discussion, and consensus between the reviewers was reached without the need for an independent arbiter. Heterogeneity between

study designs and outcomes reported did not allow for a quantitative meta-analysis to be conducted. Instead, study findings were synthesised under the key topics of interest to this review including clinical presentation/diagnosis, frequency, natural progression, grading, pathoaetiology, risk factors, symptoms, interventions and outcomes.

Results Search results The initial search retrieved 1243 studies which were screened by title. Title and abstract screening resulted in 983 studies being excluded. Following full-text review, a further 220 studies were excluded. Searching through reference lists identified two additional studies. Forty-two studies were identified. Of those, five studies reported clinical diagnosis of AWS without describing an AWS definition or diagnostic criteria consistent with an established definition [16, 20, 22, 45, 46]. When contacted, authors of the five studies did not provide further information clarifying AWS diagnosis criteria used, resulting in the exclusion of these studies from this review. Thirty-seven studies including three randomised controlled studies (RCT) [14, 29, 38], 11 prospective studies and one academic dissertation [47] (Table 2), eight retrospective studies (Table 3) and 14 case studies (Table 4) were eligible for final review (Fig. 1). This included two non-English studies [34, 39] which were accessed and translated in full for the review. Quality assessment results Figure 1 outlines the flow of studies retrieved. Quality appraisal of included studies were 12 NHMRC level II studies, three level III studies (two level III-1 and one level III-3) and 22 level IV studies. Methodological quality analysis of 15 prospective and eight retrospective studies via the Downs and Black instrument varied from 11 to 26 on a 28-point scale, indicating that studies ranged from poor to excellent (Table 1). One study had excellent quality, scoring 26; 14 studies were rated good, scoring between 20 and 25; seven studies scored between 15 and 19 and one study was rated as poor, scoring below 14. Quality appraisal subscale values for each study are provided in Table 1. Included studies tended to have clear reporting about objectives, outcomes to be measured, interventions of interest and main findings. Reporting of principal confounders, adverse events, estimates of outcome variability and study dropouts was less consistent. Five of the 37 included studies reported or demonstrated having sufficient power for

J Cancer Surviv Table 1

Results of the risk of bias assessment using the Downs and Black assessment tool

Lead author, year

Reporting (11)a

Prospective studies (in chronological order) Johansson [5] (2001) 11 Box [14] (2002) 11 Leidenius [50] (2003) 8 Lauridsen [38] (2005) 10 Hase [29] (2006) 10 Bergmann [53] (2007) 10 Torres Lacomba [8, 64, 65] 11 (2009, 2010) Yang [35] (2010) 11 Moreau [55] (2010) 6 de Oliveira [13] (2010) 11 Bergmann [48] (2011) 11 Fabro [56] (2012) 11 Morehead-Gee [3] (2012) 11 Paiva [30] (2013) 11 Koehler [47] (2013) 11 (Academic dissertation) Retrospective studies (in chronological order) Moskovitz [1] (2001) 10 Huh [51] (2005) 8 Wyrick [28] (2006) 9 Severeid [27] (2007) 9 Josenhans [39] (2007) Davies [49] (2010) Bergmann [54] (2011) Wernicke [17] (2013)

8 7 11 11

External validity (3)a

Bias (7)a

Confounding (6)a

Power (1)a

Total (0–28)

Quality rating

3 3 3 3 3 3 3

4 6 5 5 5 5 5

3 5 3 6 5 3 4

0 1 0 1 0 0 1

21 26 19 25 23 21 24

Good Excellent Fair Good Good Good Good

3 1

5 4

4 0

0 0

23 11

Good Poor

3 3 3 3 3 3

5 5 5 5 5 5

4 4 4 2 4 4

0 0 1 0 0 1

23 23 24 21 23 24

Good Good Good Good Good Good

3 3 3 2

3 4 3 4

2 2 3 3

0 0 0 0

18 17 18 18

Fair Fair Fair Fair

3 3 3 3

4 3 5 5

2 2 3 3

0 0 0 0

17 15 22 22

Fair Fair Good Good

Downs and Black score ranges: excellent (26–28), good (20–25), fair (15–19), poor (≤14) a

Maximum number can be scored in that criterion

the analyses conducted. No investigation attempted to blind participants to the intervention they were receiving. Two of the three studies that randomised patients to receive an intervention did not state who performed the randomisation or how it took place [14, 29].

Evidence synthesis Clinical presentation of AWS and diagnosis All studies had their diagnosis of AWS based on the definition of Moskovitz et al. (2001) as described in the ‘Introduction’ of this review [1]. Specific anatomical locations of AWS in patients affected by the condition were reported in 20 studies (two level II and 18 level IV). All studies reported common origin of AWS from the axilla. Two studies reported cases of AWS affecting the axilla alone [9, 28]. Three studies reported AWS reaching to the medial upper arm [7, 34, 37]. Nine studies

observed AWS extending to the antecubital fossa [2, 4, 6, 9, 15, 19, 26, 33, 47]. Four studies reported AWS beyond antecubital fossa into forearm, wrist or base of thumb [21, 42, 47, 48]. Eight studies reported further extension of AWS through lateral chest wall, the breast or abdominal wall [7, 18, 19, 26, 27, 38, 48, 49]. AWS observed after sentinel lymph node biopsy (SLNB) has been reported as less severe and limited to the axilla and medial arm, without extension to the wrist [1, 50, 51]. A level III-1 study reported 4 % of patients in the SLNB group and 17 % of the patients in the axillary lymph node dissection (ALND) group with AWS reaching from the axilla to the arm [50]. Two level IV studies reported the presence of multiple subcutaneous axilla nodule formation in patients affected with AWS [4, 52]. Diagnosis of AWS was primarily made based on reported symptoms, visual inspection and palpation of the axilla, upper extremity and trunk for signs of cording [47]. Three studies postulated that AWS may not always

Johansson [5] (2001)

BCT/mastectomy+SLNB only or SLNB+ALND

Median age (years):

Median age (years):

Bergmann [53] (2007)

NHMRC level II

Japan—Department of Rehabilitation Medicine, Keio University School of Medicine

Hase [29] (2006)

Cohort

N=1004 F

- Treatment group=50.8±7.1

Stage 0–III breast cancer

17 patients had BCT

Mean age (years)±SD: - Control group=51.4±8.2

Breast cancer

N=40 F

- BCT group=54 (range 42–69)

- MRM group=63 (range 32–77)

- MRM+XRT group=51 (range 29–70)

Delayed treatment group:

- BCT group=54 (range 31–79)

- MRM group=60 (range 37–74)

Standard treatment group:

Breast cancer BCT/MRM+level I and II ALND+ XRT±chemotherapy (CEF)± Tamoxifen

139 (N=125 after dropouts)

- ALND group=26 (range 18–34)

- SLNB group=25 (range 16–38)

Median BMI (kg/m2):

- ALND group=52 (range 39–76)

- SLNB group=61 (range 36–82)

Clinical T1–T2 breast cancer

N=85

- Treatment group=24.3 (range 19.2– 35.8)

- Control group=27.2 (range 19.2–48.7)

Median BMI (kg/m2):

- Treatment group=53±9.5

- MRM+XRT group=49 (range 40–70)

Single blind RCT

ALND+BCT/MRM± chemotherapy±hormone therapy±XRT

- Control group=59±11

Breast cancer

Mean age (years)±SD:

ALND+mastectomy or partial mastectomy±chemotherapy (CMF)±Tamoxifen±XRT to breast alone or breast and axilla

Breast cancer

Diagnosis, surgery, oncology treatment

N=65 F

NHMRC level II

Single blind RCT

Cohort

Single blind RCT

Mean age (years)±SD=56±10

N=61 F

Cohort

Denmark—Aarhus University Hospital

Lauridsen [38] (2005)

NHMRC level III-1

Finland—Maria Hospital, the Breast Surgery Unit of Helsinki University Hospital

Leidenius [50] (2003)

NHMRC level II

Australia—Royal Brisbane Hospital and the Wesley Hospital

Box [14] (2002)

NHMRC level III-1

Sweden—Lund University Hospital

Subject characteristic

Study design

Prospective studies for AWS (in chronological order)

Lead author, year, setting, NHMRC level

Table 2

- Treatment group=84±75.5 (range 10–223)

- Control group=87.9±73.4 (range 8–256)

Mean±SD number of days from surgery to study inclusion:

Time frame to occurrence not specified

3 of 20 (15 %) patients in treatment group developed AWS

1 of 20 (5 %) patients in control group and

79 (56.8 %) patients diagnosed with AWS at first post-op assessment at 7 weeks

86 (61.9 %) patients had AWS at one or more follow-up appointments

10 of 49 (20.4 %) SLNB patients and 26 of 36 (72.2 %) ALND patients diagnosed with AWS at first post-op assessment at 2 week

Time frame to occurrence not specified

Treatment group: 12 of 32 (37.5 %) patients developed AWS

Control group: 7 of 33 (21.2 %) patients developed AWS

47 (77 %) developed AWS within 1 month

AWS frequency, time frame to occurrence

–

Treatment group: as above with oral Zaltoprofen

Control group: single physiotherapy session, 20 mins of passive stretching in flexion, abduction and external rotation, active and passive ROM exercise, HEP

Delayed treatment commenced standard treatment at week 26 post-op

Standard treatment: 12 sessions, 60 mins, twice weekly exe program (extension and relaxation, strength training, vein pump therapy, scar stretching)

Light ROM exercises during hospital stay, advice to continue daily activities but avoid strenuous work or sports or 2–3 weeks (SLNB) or 1 month (ALND) postop

Treatment: physiotherapy management care plan which included pre-op assessment, inpatient and outpatient post-op followup to monitor shoulder ROM, HEP progression, lymphoedema awareness education and individualised intervention as required

Control: exe instruction booklet

Daily HEP five repetitions, three times/day: shoulder flexion, abduction, internal and external rotation, pre-op oedema prevention education (elevation, hand pumps). Programs continued for 6 months and then as long as required

Interventions

–

–

43 (31 %) patients had AWS at 13–15 weeks. 21 (15.1 %) patients had AWS at 6 months. All AWS cases resolved by 12 months

One patient in both ALND and SLNB groups had residual signs of AWS at 3 months

Patients with ROM restriction secondary to AWS gradually recovered over 3–6 months; although the proportion of cases with complete symptom resolution was not reported

- 4/47 (8.5 %) at 5 and 6 months

- 2/47 (4.3 %) at 3 and 4 months

- 21/47 (44.7 %) at 2 months

Persisting AWS reported:

Outcomes, time frame to AWS resolution

J Cancer Surviv

Morehead-Gee [3] (2012)

NHMRC level II

Brazil—Cancer Hospital III/ National Cancer Institute

Fabro [56] (2012)

NHMRC level II

Brazil—Cancer Hospital III of the National Cancer Institute

Bergmann [48] (2011)

NHMRC level II

Brazil—Breast Cancer Clinics of the Centro de Atencao Integral a Saude da Mulher in Sao Paulo State

de Oliveira [13] (2010)

NHMRC level IV

Belgium

Observational

Cohort

Cohort

Prospective

Prospective

- 45–64 years (N=96)

Moreau [55] (2010)

- 30 days, 1 (0.9 %) at 150 days, 2 (1.7 %)>200 days

Overall, 56 (48.3 %) patients developed AWS.

Breast cancer

Unilateral BCT or mastectomy± SLNB±ALND±XRT

N=193 F (185 after lost to follow-up)

2 months after surgery

- 17 (41.5 %)≥60 days

- 12 (29.3 %) between 31 and 60 days

- 13 (31.7 %) between 1 and 30 days,

63 (29.4 %) breast cancer patients developed AWS. Of 41 subjects,

17 % cases had an approx. time to onset of ≥1 year

Average time lapsed between most recent surgery and physiotherapy evaluation (weeks): 35.9±66.9 (range 14 days to 5.8 years)

31 of 167 (19 %) documented AWS episodes

Average length of physiotherapy care (weeks±SD)=10.1±9.5: stretching, HEP (shoulder active assisted flexion, active abduction±trunk rotation, horizontal abduction), soft tissue stretching techniques, resistance exe, Airdyne bicycle exe

Physiotherapy two to three times per week, ½hour each time, average 9 sessions after BCT, average 12 sessions after mastectomy: release manipulations, scar mobilisation, myofascial release, muscle stretching,

–

Length of care for patients who attended irregular physiotherapy=18±17.1 weeks

Length of care for patients who attended regular physiotherapy=7.3±3.4 weeks

SLNB=2.8±0.8

Mastectomy: 33/36 (91.7 %) resolved; 3/36 (8.3 %) had residual AWS

BCT: 66/69 (95.7 %) resolved; 3/69 (4.3 %) had residual AWS

AWS resolved in 94 % cases; 6 % had residual AWS post-treatment

–

Patients who received regular physiotherapy had resolution of AWS symptoms within 3 months

ALND=3±0.7

Mean (weeks)±SD:

AWS resolved within 2 to 4 weeks in both groups

ALND=6.8±1.9

Mean (weeks)±SD:

Median age (years): 46 (range 35–59)

All resolved in 2–3 months

Outcomes, time frame to AWS resolution

SLNB=4.4±1.1

10 of 110 (9 %) developed AWS

Clinical T1–T2 breast cancer BCT+SLNB/level I to II ALND

N=110

Huh [51] (2005)

Korea—Breast Surgery Unit of Samsung Medical Center

–

NSAIDs, physiotherapy, ROM exercises

44 (5.9 %) recorded cases of AWS between 1 and 8 weeks

Various breast cancer histologies of stage I–IV

N=750

Moskovitz [1] (2001)

NHMRC level IV

Interventions

AWS frequency, time frame to occurrence

Diagnosis, surgery, oncology treatment

Subject characteristics

Retrospective studies for AWS (in chronological order)

Lead author, year, setting, NHMRC level

Table 3

J Cancer Surviv

Average age (years)±SD: 55.9±13.3

SLNB group=57.8 (range 36–86)

NHMRC level III-3

–

–

Techniques similar to those described by Kepics [15] and Fourie [18]

hold-relax techniques, manual techniques, scapular mobilisation, PNF, HEP

Interventions

AWS did not persist >2 years surgery mark

–

2 (2.7 %) AWS cases re-presented for treatment at 5 and 7 months later

7 (9.5 %) AWS cases continued to experience difficulties 6 months post-treatment

15 (20.3 %) AWS cases required treatment>3 months; three of which were complicated by previous shoulder injury, frozen shoulder and chronic seroma

58 (78.4 %) AWS cases resolved within 3 months of treatment

Time frame to resolution not specified

Outcomes, time frame to AWS resolution

AWS axillary web syndrome, ALND axillary lymph node dissection, BCT breast conserving therapy, F female, HEP home exercise program, M male, NSAIDs non-steroidal anti-inflammatory drugs, PNF proprioceptive neuromuscular facilitation, SLNB sentinel lymph node biopsy, SEM standard error of the mean, XRT radiotherapy

ALND group=56 (range 24–85)

ALND: 6 of 115 (5.2 %) developed AWS

Mean age (years): Time frame to occurrence not specified

SLNB: 1 of 111 (0.9 %) developed AWS

T1–T2 breast cancer BCT+SLNB alone or SLNB+ALND± chemotherapy±hormone therapy

N=226 F

Time frame to occurrence not specified

One (0.6 %) patient developed AWS

Wernicke [17] (2013)

Stage II–III advanced breast cancer±chemotherapy±XRT± hormone therapy

USA

NHMRC level III-3

N=196 F

Median time (weeks) to clinic presentation=4.35 (inter-quartile range 3–8.7)

Brazil—Hospital of Cancer III/National Cancer Institute

74 (30.9 %) patients treated for AWS

Breast cancer Surgery and oncology treatment not specified

- 1–11 years 17 %

- 1–3 months 33 %

- ≤4 weeks 50 %

Time to presentation for treatment:

N=239

AWS frequency, time frame to occurrence

Diagnosis, surgery, oncology treatment

Subject characteristics

Bergmann [54] (2011)

NHMRC level IV

Physiotherapy

Australia—The Frankston Hospital Lymphoedema Service and The Lymphoedema Clinic at Chelsea Longbeach

Davies [49] (2010)

Lead author, year, setting, NHMRC level

Table 3 (continued)

J Cancer Surviv

NHMRC level IV

Fourie [18] (2009)

NHMRC level IV

Canada

Tilley [2] (2009)

NHMRC level IV

1 F, 47 years old

1 F, 37 years old

L) MRM+ALND (6 LNs removed)

Breast cancer

BCT, SLNB and ALND

Breast cancer

BCT and ALND (1/12 LN+) followed by mastectomy

Breast cancer (IDC, ER/PR +)

Spain—Manacor Hospital Foundation

Villamiel Campos [34] (2008)

Breast cancer (DCIS with tumour-positive surgical margins)

L) BCT, subsequent MRM+delayed latissimus dorsi flap breast reconstruction with expanders, adjuvant chemotherapy+XRT

Breast cancer

R) BCT+level I and II ALND (0/31 LN+); 6 cycle course of CEF chemotherapy with plan for XRT after chemotherapy

Breast cancer (IDC, grade III, ER/PR+)

Skin sparing L) mastectomy, SLNB (0/2 LN+) with immediate silicone implants reconstruction

1 F, 50 years old

1 F, 43 years old

1 F, 44 years old

1 F, 41 years old

L) Mastectomy+ALND (0/22 LN+)

Breast cancer (IDC)

NHMRC level IV

Turkey

Aydogan [37] (2008)

NHMRC level IV

USA

Koehler [26] (2006)

NHMRC level IV

Canada

Reedijk [52] (2006)

NHMRC level IV

Brazil—Ferreira de Rezende S/S Physical Therapy Clinic

Ferreira Rezende l [33] (2005)

NHMRC level IV

R) BCT and ALND (all negative LNs); adjuvant XRT

Breast cancer

Six physiotherapy sessions: moist heat, assisted flexion and abduction ROM exercises, stretching, cord stretching, HEP of arm flexion and horizontal abduction

11 physiotherapy 30–45 min sessions over 3 weeks: soft tissue techniques (gentle circular mobilisation, longitudinal tissue stretch of the cords), HEP of

At 22 days

Nil discomfort, pre-morbid shoulder ROM achieved within 11 treatments

Resolution by 16 weeks

Shoulder flexion and abduction ROM of 180°, and AWS became less visible but was still palpable at 7 weeks

AWS with shoulder ROM limitations persisted at 2, 4 and 7 weeks.

Nil functional limitations with AWS resolution by 2 months from first operation.

Functional limitation and AWS persisted at 1 month after first surgery.

Pain free, full shoulder ROM with AWS resolution at 10 weeks

–

AWS biopsy

Full shoulder ROM with AWS resolution at 7 week after 6 weeks of physiotherapy treatment Physiotherapy three times per week for 4 weeks then once per week for 2 weeks: MLD, stretching, soft tissue mobilisation, myofascial release, HEP of ROM, stretches, SLD, remedial exercises; compression sleeve and gauntlet worn during exercise

Nodules resolved by 15 weeks

–

At 1 week

Days post first operation

Several week history of AWS with subcutaneous nodules at approx. 8 weeks

Within 10 days

Several week history of symptoms consistent with AWS and subcutaneous nodules at approx. 7 weeks

Gradual regression of cords and improved symptoms (i.e. ↓ pain, ↑ shoulder ROM) after 31 days. Spontaneous resolution on 83rd day

–

Nil report on time frame to AWS resolution

Two patients deceased

Outcomes, time frame to AWS resolution

–

Kepics [15] (2004)

After 14 days

1 F, 52 years old

NHMRC level IV

1 F, 55 years old

Local heat or ice, cord stretching, skin traction, myofascial release techniques, muscle stretching, scar release, posture advice, deep breathing exercise, arm mobility exercises, reciprocal pulley and finger ladder, lymphoedema education

Presented for assessment at 1 month

Age (years)=51, 58, 73, 44 and 46

UK

USA

All received post-op physiotherapy; additional physiotherapy for one patient over a few weeks to treat residual AWS after open division and biopsy of AWS

1/5 (20 %) within 6 weeks, 4/5 (80 %) at 1 month

Breast cancer (4 cases IDC, 1 case florid cribriform epitheliosis)

5F

Marcus [9] (1990)

BCT or mastectomy±axillary lymphadenectomy±implant breast reconstruction

Interventions

Time frame to AWS occurrence

Diagnosis, surgery, oncology treatment

Subject characteristics

Case series and reports for AWS (in chronological order)

Lead author, year, setting, NHMRC level

Table 4

J Cancer Surviv

1 F, 39 years old

1 F, 44 years old

1 M, 57 years old

BCT+axillary LN biopsy followed by re-excision of margins

Breast cancer (DCIS)

R) excisional breast biopsy, SLNB (0/4 LN+), adjuvant XRT, Tamoxifen

At 3 days after second surgery

At 10 days

Time frame to occurrence not specified

Nil surgery Breast cancer (IDC with DCIS)

1 month history of AWS with palpable nodules along the cord

R) axilla furuncle associated with methicillin-resistant Staphylococcus Aureus

Time frame to occurrence not specified

Post-op

At 2 weeks

Time frame to AWS occurrence

Self massage, shoulder abduction exercise, Aescuven forte (300 mg)

Patient discharged from physiotherapy at 1.5 months post-op

Occupational and physiotherapy for 5 weeks: soft tissue releases, progressed to a two-person cord traction technique, self elongation techniques, skin traction, scar massage, home stretching program

Twice daily oral Doxycycline for 10 days

–

Physiotherapy: arm rotation and extension, advice re HEP and self-massage

stretching and self-mobilisation

Interventions

Pain free shoulder abduction ROM of 170° with AWS resolution at 3 weeks

Disabilities of the arm, shoulder and hand tool score of 7.5 on discharge, an improvement of 25 points from initial score of 32.5

Ongoing symptoms (sense of restriction and tightening) at 1 year managed with home stretching program after discharge

Resolution after 10 days

–

Resolution by 2 months

spread over 3 weeks. Return to work after seventh physiotherapy treatment

Outcomes, time frame to AWS resolution

ALND axillary lymph node dissection; AWS axillary web syndrome; BCT breast conserving therapy; CEF cyclophosphamide, epirubicin and fluorouracil; DCIS ductal carcinoma in situ; ER estrogen receptors; F female; HEP home exercise program; IDC invasive ductal carcinoma; LNs lymph nodes; M male, MLD manual lymphatic drainage; MRM modified radical mastectomy; PR progesterone receptors; ROM range of movement; SLD self lymphatic drainage; SLNB sentinel lymph node biopsy; XRT radiotherapy

NHMRC level IV

China

Wei [7] (2013)

NHMRC level IV

USA

Lattanzi [19] (2012)

NHMRC level IV

USA

Rashtak [4] (2011)

Mean BMI (kg/m2)±SD=23.7±3.45

ALND+BCT/mastectomy

Mean age (years)±SD=53.06±10.27 (range 32–68)

Belgium

NHMRC level IV

Breast cancer

15 F

Leduc [21] (2009)

NHMRC level IV

L) Wide local excision

Microcystic adnexal carcinoma in L) axilla

1 F, 26 years old

Craythorne [6] (2009)

UK

Diagnosis, surgery, oncology treatment

Subject characteristics

Lead author, year, setting, NHMRC level

Table 4 (continued)

J Cancer Surviv

J Cancer Surviv Fig. 1 Study attrition diagram

1243 studies retrieved from search strategy: Cochrane library N = 73 Pubmed N = 199 Medline N = 145 CINAHL N = 30 Pedro N = 0 EMBASE N = 571 AMED N = 5 Google Scholar = 220

Titles & duplicates screened Removed 878 studies Retained 365 studies

Abstracts screened Excluded 105 studies Retained 260 studies

Full text screened Excluded 220 studies Retained 40 studies

Reference list screening Two studies included

Number of potentially eligible studies reviewed for AWS diagnostic criteria N = 42

Excluded 5 studies without AWS diagnostic criteria

Final number of eligible studies N = 37

be present with both palpable and visible cords [8, 18, 50]. In a level IV study, AWS was reportedly palpable but non-visible distally on the palmar surface of the wrist and proximally around the posterior axillary border spreading over the scapula and posterior chest wall [18]. In a level III-1 study, AWS was detectable by palpation only in 17 out of 36 patients [50]. Visibility and/or palpability of AWS has been suggested to be linked to amount of adipose tissue, with AWS potentially being less visible or palpable under a thick subcutaneous layer [8, 48, 50]. Four studies ranging from level II to IV documented that patients with AWS had a lower body mass index (BMI) than those without [8,

47, 48, 50], although this association was not consistently reported [21, 53]. AWS frequency Variable AWS frequencies were reported in 21 studies (Tables 2, 3 and 4), ranging from 0.6 % [54] to 85.4 % [39]. Surgical procedure may influence the reported frequency of AWS. AWS ranged from 0.9 % [17] to 25 % [51] following SLNB with higher frequencies reported following ALND, ranging from 5.2 to 36 % [17]. One level IV study reported AWS frequency by breast procedures. Up to 88.5 % patients were diagnosed with AWS following breast conserving

J Cancer Surviv

surgery and 80 % following mastectomy [39]. Severeid et al. [27] investigated AWS frequency by diagnosis and reported that 29.4 % breast cancer patients and 58.3 % melanoma patients developed AWS. Two studies [1, 55] observed that AWS did not occur in isolated breast surgery in the absence of an axillary lymphadenectomy. All studies except one level IV study [4] reported surgical-related AWS. Natural progression Time frame to occurrence AWS appears to develop in the majority of cases within three months following axillary lymphadenectomy in the diagnosis or management of cancer. Three level IV studies reported AWS occurrence within first post-operative week [1, 7, 34]. Nineteen studies (three level II, one level III-I and remaining level IV) reported AWS occurrence between one and eight post-operative weeks [1, 2, 6, 8, 9, 15, 18, 19, 26, 27, 33, 37, 38, 47, 48, 50–52, 56]. AWS occurrence between eight and 12 weeks was found in four level IV studies [26–28, 50]. Delayed occurrence at or beyond three months was reported in three level IV studies [27, 28, 47]. Another 18 studies reported AWS occurrence post-operatively but did not specify time points of AWS onset [3–5, 13, 14, 17, 21, 26, 29, 30, 35, 38, 39, 49, 53–56]. Time frame to resolution Twenty-eight studies reported time frames to resolution of AWS symptoms, ranging from three weeks post-operatively to more than three months following surgery (Tables 2, 3 and 4). Ten studies (two level II, two level III-1, six level IV studies) reported resolution of AWS within three post-operative months [1, 5, 8, 28, 38, 47, 49–51, 55]. An additional seven level IV studies specified the time frame to AWS resolution, ranging from 10 days [4] to 83rd post-operative day [6, 7, 26, 33, 34, 37]. Nine studies documented persistence of AWS beyond three post-operative months with symptoms being reported at three months [5, 8, 47, 49, 50], four months [18], six months [5, 38, 49], eight months [56] and at one year [19]. Two level IV retrospective studies [17, 39] reported AWS resolution, although no specific time frames were reported. Recurrence Three studies documented recurrences of AWS [8, 28, 49]. In a level II study, four patients with AWS experienced a recurrence of AWS after adjuvant chemotherapy and XRT [8]. In a level IV study, two cases of AWS recurrences were reported with one patient having two occurrences of AWS, 21 months apart from initial onset [28]. A level IV academic dissertation also described two cases of recurrences months later [49].

Grading of AWS There is no standardised grading system of AWS with various grading systems reported. Two level II studies used a severity rating scale in which AWS was ranked as (0) none, (1) present and asymptomatic, or (2) present and symptomatic [3, 20]. One level IV study noted documentation of subjective ratings of mild, mild to moderate, moderate, moderate to severe or severe to describe the involvement of AWS [28]. This subjective rating reportedly took into account pain intensity, loss of shoulder ROM and reactivity or symptom aggravation during functional activity and gentle stretching [28]. However, specific criteria for each grading were not reported in the study. In another level IV study, AWS was graded by the thickness of the cord on a scale from small cord (+) to thick, pronounced cord (+++) [39]. No objective measures of cord thickness were recorded. A level IV dissertation recorded the characteristics of AWS by location, number, length (measured using a non-stretch, flexible tape measure), depth (superficial, middle, or deep), width of the cord (less than, equal to or more than 1 mm) and visibility and/or palpability [47]. Pathoaetiology The pathoaetiology of AWS is not well understood. Eight studies (one level III-1 and seven level IV) [1, 4, 5, 7, 9, 34, 39, 52] investigated the composition of AWS using histopathology, immunohistochemical staining or B-type ultrasonography (Table 5). Five level IV studies observed histologic characteristics consistent with thrombosed lymphatics [4, 9, 34, 39, 52]. Positive immunostaining for D2-40 (specific marker for the lymphatic endothelium) and LYVE1 (polyclonal antibody against human lymphatic vessel) along with negative staining for CD31 (a marker for all endothelial cells) and CD34 monoclonal antibodies suggested a lymphatic origin and excluded a venous pathology [4]. The lymphatic aetiology of AWS was further supported by a level IV study which reported the location of AWS corresponding to the superficial arm lymphatics as seen in anatomical studies of foetuses [21]. Physical findings of the size, location and course of AWS in 15 females after breast cancer surgery demonstrated that AWS followed the course of lymphatic vessels of the upper arm and forearm reaching the base of the thumb [21]. However, the data were not robust enough to ascertain this definitive conclusion given the small sample size [21]. Venous aetiology was suggested by two studies [1, 5], although one reported histopathological findings of both lymphatic and venous aetiology [1]. B-type ultrasonography was used in one level IV case study to investigate AWS aetiology [7]. Blood flow signal on colour Doppler flow imaging indicated that the cord-like structure that radiated from the axilla to the medial arm was a blood vessel. Cords that radiated to the breast and thoracoabdominal wall presented the same echo of

J Cancer Surviv Table 5

Studies investigating etiology of AWS (in chronological order)

Lead author, year

Study design

Sample no. with AWS, no. of biopsies

Findings

Histopathological examination Marcus [9] (1990)

Case series

N=5, 4 biopsied

Moskovitz [1] (2001)

Retrospective chart

N=44, 4 biopsied

Johansson [5] (2001)

Prospective cohort

N=40, 7 biopsied

Reedjik et al. [52] (2006)

Case study

N=1, 1 biopsied

Josenhans [39] (2007)

Retrospective study

N=105, 1 biopsied

Villamiel Campos [34] (2008)

Case study

N=1, 1 biopsied

Occluded vessel, consistent with a thrombosed lymphatic vessel (2 cases) Dilated perineural lymphatics with inspissated material (1 case) Collagenous tissue with few capillaries (1 case) Dilated, thrombosed lymphatics or thrombosed superficial veins, or both: - Dilated lymphatics (2 cases). One of the lymphatics contained fibrin clot. - Venous thrombosis in various stages of recanalisation (3 cases) Veinus tissue (6 cases), with an inflammatory process suggesting thrombophlebitis (4 cases) Nodule composed of a non-muscular fibrotic wall infiltrated by scattered mixed inflammatory cells. Nodule’s centre filled with thrombotic material undergoing organisation and recanalisation. Lymph vessel that had undergone fibrosis which was surrounded by fat and granulation tissue. Associated lump was found to be a cystic enlargement of the lymph vessel Fibrofatty tissues, vascular structure with luminal occlusion, thickness of wall more compatible with lymphatic veins Findings confirmed findings by Moskovitz (2001) [1]

Histopathological examination and immunohistochemical staining findings Rashtak [4] (2011) Case study N=1, 1 biopsied

B-type ultrasonography Wei [7] (2013)

Case study

N=1

subcutaneous tissue as the contralateral thoracoabdominal wall without cords [7]. The authors’ hypotheses of phlebitis was consistent with the B-type ultrasonography findings but were not definitive due to the lack of a histopathology report. A dissertation obtained ultrasonographic images of patients with AWS using a high-resolution ultrasound system of 18 MHz frequency [47]. However, no differences in skin thickness, reflector thickness, disorganisation, number of reflectors or echo density were found between the affected and unaffected side. Overall, uncertainty remains regarding the pathogenesis of AWS. Lymphatic pathogenesis was more commonly reported; however, venous origin or a combination of both could not be ruled out. Immunohistochemical tests for lymphatic-specific and conventional blood vessel markers (e.g. CD31, CD34, LYVE1, D240 or von Willebrand Factor) have recently been identified as reliable techniques for potential differentiation of

Fibroblastic proliferation surrounding the dilated lymphatic vessels, presence of valve Positive staining for D2-40 (a specific marker for lymphatic endothelium) Obvious blood flow signal on colour Doppler flow imaging, indicating that AWS presented from the axilla to the medial arm was a blood vessel

lymphatics from veins [4, 57]. It is thought that tissue retraction and patient position during axillary lymphadenectomies may predispose to lymphovenous injury, stasis and hypercoagulability [1]. Lymphatic fluid normally coagulates slower than blood due to low levels of fibrinogen and absence of platelets [58]. AWS formation is speculated to be associated with acceleration of lymphatic fluid coagulation due to the presence of thrombokinase associated with surgical trauma [58, 59]. Hypercoagulation may result in vein or superficial lymphatic thrombosis, which presents as palpable or visible cords during arm abduction [21, 39]. The release of inflammatory factors was thought to cause phlebitis via intravasation in the axilla, medial arm, breast and thoracoabdominal wall through multiple mechanisms [7]. In an academic dissertation, it was hypothesised that AWS could be related to lymphangiogenesis where newly formed lymphatic vessels

J Cancer Surviv

may be adhered to underlying tissues when attempting to reestablish lymphatic flow by reconnection to existing lymph vessels [47]. Spontaneous resolution of AWS was hypothesised to be related to successful restoration of lymphovenous flow and/or gradual resorption of redundant lymphatic vessels [18, 24, 33, 36, 59], whereas non-resolution may be related to an unsuccessful or partially unsuccessful lymphatic re-connection. There was currently insufficient empirical evidence to support either of these hypotheses [18]. Risk factors Ten studies (four level II, two level III and four level IV) described risk factors including extent of surgery, BMI, age, ethnicity, axillary metastasis, oncology treatment and healing complications for developing AWS [3, 8, 27, 30, 37, 47, 48, 50, 52, 53] (Table 6). The risk of developing AWS appears to increase with the complexity or extensiveness of surgery [1, 6], which is associated with the type of surgical procedure and the number of lymph nodes removed [1, 37, 47, 50, 52]. Two level II [48, 53] studies showed an increased AWS risk following mastectomy and ALND compared to conservative surgery and SLNB procedures. In a level IV academic dissertation, the odds of developing AWS increased 1.12 times for every unit increase in number of lymph nodes removed [47]. Younger age has been reported in three level III studies to be a risk factor for developing AWS [8, 47, 48]. Lower BMI has also been suggested in four studies (two level II, one level III and one level IV) as an AWS risk factor [8, 47, 48, 50]. It is possible that age and BMI risk factors may be related as older people are more likely to have a higher BMI [8]. One level II study found ethnicity as a risk factor with higher rates of AWS (p = 0.013) and lymphoedema (p = 0.011) found in AfricanAmerican breast cancer females post-operatively [3]. Four studies (one level II, two level III and one level IV) found no association between metastatic involvement of axillary nodes and AWS occurrence [1, 50, 53, 54]. One level III-3 found no increased risk of developing AWS with neoadjuvant chemotherapy and XRT [48]. XRT may even be protective, with a level IV study finding that patients with breast cancer who had undergone XRT were less likely to develop AWS (OR=0.44; 95 % CI=0.23–0.86) [27]. Healing complications such as numbness after intercostobrachial nerve injury, haematoma, lymphoedema, subjective oedema sensation and myofascial pain syndrome (MPS) have been suggested to potentially have an association with AWS. Although one level IV study [1] reported AWS to be unrelated to intercostobrachial nerve ligation or post-operative complications, another study provided level II evidence for an increased risk of developing AWS in patients with haematoma or upper extremity numbness from intercostobrachial nerve injury [48]. The role of

lymphoedema in AWS is unclear. It has been identified that the Best Practice for Management of Lymphoedema guidelines (2006) [60] reported AWS as one of the risk factors for developing or worsening lymphoedema; however, there were no studies cited as supporting this [47]. One level II [30] and one level IV [1] study did not find association between AWS and lymphoedema. One level II study reported females with early arm oedema (>200 ml) to have 1.54 times greater risk of AWS [53]; however, this association was not found in another study [48]. A significant association between AWS and MPS was reported in a level II study, with AWS group presenting more MPS than the non-AWS group (p=0.01) [8]. Symptoms AWS was reported as being symptomatic or asymptomatic. Symptomatic AWS presented with pain, feelings of tightness and restricted shoulder range of movement (ROM) [1, 2, 4–6, 8, 9, 13, 15, 18–20, 23, 26–29, 31, 33–35, 37–39, 45, 47, 49, 50, 52, 54, 55, 61–65]. A pulling sensation was described by more than 55 % of AW S p a t i e n t s a t t w o w e e k s , f o u r w e e k s a n d three months follow-up visits in a level IV academic dissertation [47]. Patients with AWS also reported numbness in two level IV studies [7, 19], but it was unclear if this was related to the axillary surgery. Four studies (one level II and three level IV) [6, 7, 19, 38] reported progressive resolution of pain, but ongoing symptoms of tightness and/or restriction in shoulder movement remained. No explanation was provided in any study as to why AWS became symptomatic in some patients while others remain symptom free. Pain Pain associated with AWS was described in 25 studies. Of those, six were level II studies [3, 8, 13, 29, 53, 56], three were level III [5, 46, 49] and the remainder were level IV studies [1, 2, 7, 9, 15, 18, 26–28, 33, 37, 47, 49, 51, 52, 55]. The presence of pain that authors attributed to AWS in patients following breast cancer surgery ranged from 5.4 % [48] to 24.3 % [56]. Pain was commonly reported in the ipsilateral axilla, occasionally radiating down the arm with shoulder movement [1]. One level IV study described pain in the head (similar to a headache), in the neck above the clavicle and in the contralateral breast with passive distal skin stretching of AWS in the arm [18], although it was not possible to determine from the report whether there were potentially concomitant reasons for the head and neck symptoms. MPS was reported in 60 % of patients with AWS following breast cancer surgery with ALND

J Cancer Surviv Table 6

Studies reporting risk factors associated with incidence of AWS

Risk factors

Lead author, year

Authors’ conclusions

Age

Torres Lacomba [8] (2009)

Mean age±SD of patients with AWS was lower (48.9±10.2) than those without AWS (58.1±10.9) (p