Treatment Related Impairments in Arm and Shoulder in Patients with Breast Cancer: A Systematic Review

Background Breast cancer is the most common type of cancer in women in the developed world. As a result of breast cancer treatment, many patients suffer from serious complaints in their arm and shoulder, leading to limitations in activities of daily living and participation. In this systematic literature review we present an overview of the adverse effects of the integrated breast cancer treatment related to impairment in functions and structures in the upper extremity and upper body and limitations in daily activities. Patients at highest risk were defined. Methods and Findings We conducted a systematic literature search using the databases of PubMed, Embase, CINAHL and Cochrane from 2000 to October 2012, according to the PRISMA guidelines. Included were studies with patients with stage I–III breast cancer, treated with surgery and additional treatments (radiotherapy, chemotherapy and hormonal therapy). The following health outcomes were extracted: reduced joint mobility, reduced muscle strength, pain, lymphedema and limitations in daily activities. Outcomes were divided in within the first 12 months and >12 months post-operatively. Patients treated with ALND are at the highest risk of developing impairments of the arm and shoulder. Reduced ROM and muscle strength, pain, lymphedema and decreased degree of activities in daily living were reported most frequently in relation to ALND. Lumpectomy was related to a decline in the level of activities of daily living. Radiotherapy and hormonal therapy were the main risk factors for pain. Conclusions Patients treated with ALND require special attention to detect and consequently address impairments in the arm and shoulder. Patients with pain should be monitored carefully, because pain limits the degree of daily activities. Future research has to describe a complete overview of the medical treatment and analyze outcome in relation to the treatment. Utilization of uniform validated measurement instruments has to be encouraged.


Introduction
Breast cancer is the most common type of cancer in women in the developed world. Due to new treatment modalities, breast cancer survival has improved over time. However, as a result of breast cancer treatment, many patients suffer from adverse effects and have serious complaints in their arm and shoulder e.g. decreased joint mobility, muscle strength, pain and lymphedema, leading to limitations in activities of daily living and participation in work, sports and leisure activities. [1][2][3] In a prospective Australian study, 62% of the population still suffered from at least one impairment as a complication of breast cancer treatment and 27% suffered from two to four impairments after six years. [4] Reported variability in onset and severity of upper limb symptoms of patients with breast cancer reported in studies is large [5] and a systematic overview of risk factors related to medical treatment is lacking. This information is of direct clinical relevance, as early physical therapy intervention for these complaints as well as surveillance of patients at risk for developing impairments in daily activities reduces the need for intensive rehabilitation and the associated costs. [6] Based on the misconception that disabilities such as decreased range of motion, pain and lymphedema will resolve over time without intervention, combined with denial of the possible benefits of physical therapy interventions, this has led to the inadequate monitoring of disabilities. [7] To the best of our knowledge, this is the first systematic review with an evidence synthesis on the physical adverse effects of all components of breast cancer treatment, analyzed for each treatment modality, on impairments in the arm and shoulder, leading to limitations in activities that potentially warrant treatment. If the clinician is aware of the risk of adverse effects of the treatment, clinical reasoning regarding surveillance and the early detection of impairments in patients at risk can be applied in a systematic way.
In this article, we present a systematic literature review of the adverse effects of breast cancer treatment in terms of development of constraints in the arm and shoulder in patients with stage I-III breast cancer who underwent curative treatment. We describe the adverse effects for treatment-induced disorders of the musculoskeletal system -classified by International Classification of Functioning, Disability and Health (ICF) domains [8] -and assess the influence of pre-existing comorbidity. More specifically, the following key question is answered in this systematic review: which adverse effects related to breast cancer treatment predict persistent impairments in function and structures of the upper extremities/ thorax, e.g. reduced joint mobility, reduced muscle strength, pain, lymphedema and limitations in daily activities?

Study selection criteria
Search strategy. We conducted a systematic literature search using the databases of PubMed, Embase, CINAHL and Cochrane. Published studies in English, French and German language were eligible for inclusion. We started with the inclusion of eligible meta-analyses and systematic reviews, and then considered the inclusion of prognostic cohort studies, case-control studies and cross-sectional studies that were not included in published systematic reviews. To minimize bias, only studies with at least 100 patients were included. Studies which had already been included in systematic reviews or meta-analyses were not analyzed separately. To allow for an adequate follow-up and description of late adverse effects, only studies with a follow-up period of at least 3 months were included. When more publications of the same study were published, data were extracted from the most recent publication. As we were merely interested in adverse effects in relation to current medical practice, studies published from January 2000 to October 2012 were included. The search strings are listed in table 1.
Patients. Studies on patients with curatively treated breast cancer (Stage I-III) were included.
Outcomes. The following health outcomes were extracted: impairment in functions and structures in the upper extremity and upper body (reduced joint mobility, reduced muscle strength, pain, and lymphedema), and limitations in daily activities of the upper extremity. Outcomes had to be measured with instruments for which validation studies were published, or for which the authors described validation before initiation of the study.
Description of adverse effects of the medical treatment was divided into effects within the first 12 months and late effects (. 12 months). When outcome measures of severe cases were presented as well, these were presented between brackets in table 2.

Quality assessment
We evaluated the methodological quality of the included studies to test generalizability and possible bias. Studies were rated using the Oxford Centre for Evidence-Based Medicine, 2011 appraisal sheets and levels of evidence (see table 3) [9]. Two authors (JH + CB) independently scored each item of the appropriate scoring sheet. Disagreements were discussed together or if appropriate in the research group. If the item was well described and its quality was good, a plus (+) was assigned, plus-minus (6) was assigned if the item was incompletely described, and minus (-) was used if the item was not clearly described or not described at all. Five items were used to score systematic reviews leading to a maximum score of 100% (see table 4 and 5). Only systematic reviews including meta-analysis could achieve a full score of 100%. For cohort studies, six items were scored. Since the type of surgical treatment may influence health outcomes, articles describing radiotherapy treatment not taking into account the type of surgical treatment were given no score to the item ''Subgroups with different prognosis identified''. A full score was assigned to studies assessing the outcome ''lymphedema'' with measurements of the full arm, using tape measurements to calculate volume, water volumetry, perometry or bio-impedance spectroscopy (BIS). When other methods of multiple tape measurement were used, plus-minus was assigned to ''validated outcome'' criterion. If the Common Terminology Criteria for Adverse Events (CTCAE) was used as a measurement instrument for lymphedema no score was given, because only one location was measured. Questionnaires on lymphedema were given plus-minus, as these questionnaires led to a higher incidence percentage in relation to volumetric measurements. [10] In selecting studies with a quality score of .50% we aimed at reducing the risk of bias of the included studies resulting in more robust conclusions of our review.

Synthesis
First, we described detailed characteristics and the main findings of the included systematic reviews, RCTs, and cohort studies, as reported by the authors of the included studies. Second, we assessed adverse effects per impairment and activity limitations for each medical intervention and combination of medical interventions. Adverse effects were assessed for short-term impact (# 12 months follow-up) and long-term impact (.12 months followup). If a study did not identify which part of the treatment caused Table 1. Search string adverse effects.

Results
We identified 804 unique articles, of which 116 were eligible for full-text assessment (see figure 1 for a flow diagram). Of these, 54 studies were excluded because they did not meet the inclusion criteria. Another 23 studies were excluded because they had already been included in one or more systematic reviews (15) or had a quality rating #50% (8). Finally, 39 articles were included. In the syntheses 13 articles could not be included because adverse effects were not analyzed separately for each treatment modality.

Methodological quality of the included studies
The methodological quality of the included studies ranged from 60% to 90% for the systematic reviews (see table 4), and from 58% to 100% for prognostic studies and RCTs (see table 5). In four systematic reviews, the search strategy was limited to one database only. [1,[11][12][13] Results in four systematic reviews were not pooled due to the heterogeneity of the data. [11,12,14,15] The majority of the cohort studies presented validated outcome measures, while seven of the 32 studies described outcome by a self-generated and self-validated questionnaire [3,16,17] or performed incomplete measurements. [18][19][20][21] In six studies, a description of the outcome was incomplete. [22][23][24][25][26][27]. Table 2 presents a detailed overview of the results of the included studies. Six systematic reviews and 29 cohort studies presented analyses regarding the origin of the adverse effects. Some studies analyzed the relationship of the adverse effects in relation to comorbidity, age or BMI.

Adverse effects
In most studies, different subgroups were identified based on surgical treatment. Four studies [17,[28][29][30] focused only on patients that underwent ALND. One systematic review [1] and one cross-sectional study [27] focused on the adverse effects of radiotherapy. The adverse effects of aromatase inhibitors focused on musculoskeletal pain. [11,22,31] Zhou et al. described aromatase inhibitors in combination with zoledronic acids and pain. [11].
Synthesis per outcome measure is summarized and presented in table 6, including levels of evidence.
Regarding ALND as a medical intervention, one systematic review reported a reduction in ROM in abduction and flexion ranging from 132-175u, which was reported in 1-67% of the patients. [15] Regarding SNB, a second systematic review described a reduction in ROM. [12] Percentages of patients with ROM reduction varied from 6%-31% after 12 months, and reduced to 0%-9% after 24 months. Regarding ALND (directly or after SNB) vs. SNB, change of ROM in the third systematic review was reported in 9%-56% vs. 3%-24% of the patients, or in a mean difference of 1u-20u within 12 months and 8%-20% vs. 0%-4% over 12 months. [13] Odds Ratios (ORs) in the included studies of this systematic review ranged from 1.02-9.0 for goniometric measurements. [13] One cohort study described a reduced ROM of 21% vs. 56% at 6 months and 6% vs. 9% at 12 months, with an OR of 1.56 at 12 months. [32] Another cohort study reported reduced ROM at six months and . 12 months in a study population in which 71% underwent ALND. Reduction was present in 60% and 11% in flexion/abduction and 25% and 5% in external rotation [33]. ROM reduction was related to ALND, a greater number of lymph nodes removed, cording, seroma, mastectomy, stage II, hand dominance, BMI $ 25 and older age (.65 years).
Regarding mastectomy vs. lumpectomy, one systematic review presented an OR of 5.67 for mastectomy as a risk factor for reduced ROM. [15] In one cohort study, ROM reduction was present in 33% of the study population [34]. Mastectomy was indicated as risk factor. Regarding ALND and mastectomy vs. ALND, lumpectomy and radiotherapy reduced ROM was described at one, 12 and 24 months in overall percentages and percentages with severe reduction. Percentages reduced from 68% vs. 73% to 23% vs. 30%. [28] Regarding radiotherapy vs. no radiotherapy, one systematic review presented ORs of 2.07-12.30, a relative risk (RR) of 4.6 and reduced ROM in 34%-52% vs. 4%-20% of the study population in the included studies. [13] One large cohort study presented an OR of 2.48 for radiotherapy as a risk factor for ROM reduction. [32] Regarding axillary radiotherapy vs. no axillary radiotherapy, the risk of decreased ROM was analyzed in two systematic reviews (RR 2.6; OR 1.67). [1,15] A third systematic review reported changes in joint mobility in 14% vs. 2% of the patients in one included study; ORs in other included studies ranged from 1.70-6.83 for goniometric measurements. Regarding radiotherapy to the axilla and chest vs. radiotherapy to the chest, the same systematic review presented an RR of 1.7 in one included study and reduced ROM in 20%-49% vs. 4%-14% of the study population in other included studies. [13] Regarding chemotherapy vs. no chemotherapy, one large cohort study reported an OR of 0.73 of chemotherapy as a risk factor for ROM reduction. [32].
In synthesizing the results from the included studies, we found level 1 evidence for mastectomy and radiotherapy to the axilla as risk factors for reduced ROM in abduction, flexion and external rotation, and level 2 evidence for ALND and radiotherapy to the chest wall.
Reduction in muscle strength. Reduced muscle strength was reported in four systematic reviews [12,13,15,36] and five cohort studies. [17,18,20,33,37].  Regarding ALND, one systematic review described reduced muscle strength (OR 3.03) [15]. One cohort study described reduced muscle strength in 28% of the study population [20]. Regarding SNB, a second systematic review reported reduced muscle strength in 17%-19% of the patients after sentinel node biopsy and 11% in the long-term. [12] This systematic review identified patients with young age (,50 years) as a risk factor for muscle strength impairment based on results of one large study comparing ALND vs. SNB. Regarding ALND (directly or after SNB) vs. SNB, a third systematic review reported weakness in 48% vs. 16% of the patients, with loss of abduction strength of 12-15 Nm, loss of grip strength of 12-41 Nm in the included studies and ORs ranging from 5.14-8.82 reported in the included studies. [13].
Regarding lumpectomy and ALND, one systematic review reported reduced muscle strength in9%-28% of the study population. [15] Regarding ALND and mastectomy vs. ALND, lumpectomy and RT reduced muscle strength was described at one, 12 and 24 months. [28] Percentages reduced from 67% vs. 72% to 39% vs. 56% reduced muscle strength. Reductions were larger in the first 12 months compared to later measurements (see table 6).
Regarding chest radiotherapy vs. no radiotherapy, the risk of reduced muscle strength was analyzed in one systematic review. [13] Extracted data from the included studies showed ORs from 1.70-6.83 for radiotherapy as a risk factor for reduced muscle strength and one included study reported reduced muscle strength in 14% vs. 2% of the patients. Regarding axillary radiotherapy vs. radiotherapy to the chest wall, the risk of reduced muscle strength was analyzed in the same systematic review. [13] One included study reported an RR of 1.7; another study showed 59% vs. 40% of the patients with reduced muscle strength. Regarding concurrent radiotherapy and chemotherapy vs. sequential radiotherapy and chemotherapy, a fourth systematic review described the risk of reduced muscle strength by concurrent treatment with an OR of 2.09. [36].
In synthesizing the results of the included studies, we found level 1 evidence for ALND, and concurrent radiotherapy and chemotherapy as risk factors for reduced muscle strength. We found level 2 evidence for SNB, radiotherapy to the chest wall and radiotherapy to the axilla and chest as risk factors for reduced muscle strength.
Regarding ALND, one systematic review [15] and one cohort study [38] described pain 12 months post-operative. This systematic review described an OR of 4.61 and percentages of shoulder pain (9%-68%) and breast pain (15%-72%) in the individual studies. [35] The cohort study described pain in 53% of the population. [38] Regarding SNB, a second systematic review reported pain in 8%-36% of the patients within 12 months and 8%-21% at 24 months, analyzing young age (,50 years) as a predictive factor, described in one included study. [12] Regarding ALND (directly or after SNB) vs. SNB, a third systematic review reported pain during motion in one included study in 12% vs. 4% at 12 months and 9% vs. 3% at 19 months and an OR of 3.54 mentioned in another study. [13].
Regarding ALND and mastectomy vs. ALND, lumpectomy and radiotherapy pain was described at 1 month post-operatively, and at 12 and at 24 months. [28] Pain reduced from 75% vs. 82% to 42% vs. 56%. Regarding chest radiotherapy vs. no radiotherapy, one individual study in a systematic review reported at least weekly pain in 26% vs. 4% of patients (OR = 7.10), 6 to 13 years postoperatively. [13] Regarding concurrent radiotherapy and chemo- Table 5. Cont. therapy vs. sequential radiotherapy and chemotherapy a fourth systematic review reported the risk of brachial neuropathy (OR 3.14). [36] Regarding chemotherapy vs. no chemotherapy, two cohort studies found chemotherapy to be a risk factor for pain, [38] with a reported OR of 3.00. [40].
Regarding the administration of zoledronic acids vs. no zoledronic acids, one systematic review reported the relative risk (RR) of arthralgia (RR 1.16) and bone pain (RR 1.26). [11] Regarding the upfront administration of zoledronic acids compared to delayed administration, the same systematic review described an increased risk of pain (RR 1.28). Regarding exemestane vs. tamoxifen, one cohort study described an increased risk of carpal tunnel syndrome (OR 9.90). [24] In this study, 43% of the patients had a musculoskeletal disorder and 2% carpal tunnel syndrome. Another cohort study described increased pain incidence by using tamoxifen at baseline and at younger age (, 55 years). [22].
In general, pre-operative pain was a risk factor for postoperative pain (OR 5.17) and prolonged pain. [24,40] Pain was correlated with decreased muscle strength and range of motion, decreased job participation, reduced use of the affected arm in leisure activities and with lifting a gallon of milk or during heavy household chores. [33] At 6 months, pain during daily activities was less than at rest. [31,41] In contrast, one study reported an exacerbation of pain by exercise. [40] Another study reported less pain during activities compared to rest at six months postoperative and more pain at 60 months. [39] Arm-shoulder pain led to sleep disturbances (OR 3.17). [35].
In conclusion, we found level 1 evidence for ALND, radiotherapy before chemotherapy, and the administration of zoledronic acids (more in case of delayed administration) as risk factors for pain. We found level 2 evidence for SNB and radiotherapy as risk factors for pain.
Regarding ALND, two systematic reviews and five cohort studies described an increased risk of lymphedema. One systematic review described an RR of 3.47. [1] A second systematic review described percentages of pain in the included studies ranging from 0%-34%. [15] Percentages in the cohort studies varied from 13%-30%. [20,29,39] BMI $30 as a risk factor for lymphedema was described in one cohort study with an OR of 4.12 [44] and in another cohort study as an increase of 4.1% or HR of 2.61 for each lymph node removed. [26] Regarding SNB, a third systematic review described percentages ranging from 3%-14% in the first 12 months to 7% in the follow-up of 60 months. [12]    Regarding ALND (directly or after SNB) vs. SNB, two systematic reviews and three cohort studies described lymphedema. One systematic review reported an RR of 3.07 (when compared to no axillary dissection 3.47), [1] while another systematic review reported an OR of 11.67. [15] In the cohort studies, percentages of patients with lymphedema varied from 3%-13% vs. 0%-9% in the first 12 months to 14%-35% vs. 5%-8% in longer follow up. [19,32,43]. Regarding mastectomy, lymphedema was described in one systematic review and one cohort study. The systematic review reported an RR of 3.28, [1] while the cohort study reported an OR of 7.48. [20] Regarding ALND and mastectomy vs. ALND, lumpectomy and radiotherapy lymphedema was described at one month post-operatively, and at 12 and at 24 months. [28] Percentages of patients with lymphedema increased from 27%-41% at one month to 33%-52% at 24 months post-operatively.
Regarding breast reconstruction vs. no reconstruction, one cohort study described lymphedema in 5% vs. 18% of the study population. [42].
Regarding radiotherapy to the chest and axilla vs. radiotherapy to the chest, two systematic reviews and one cohort study described lymphedema. One systematic review described an RR of 2.97, [1] the second an OR of 2.4. [12] The cohort study reported an OR of 3.57. [17] Regarding concurrent radiotherapy and chemotherapy vs. sequential radiotherapy and chemotherapy, one systematic review reported an OR of 2.02. [36] Regarding radiotherapy before chemotherapy vs. radiotherapy after chemotherapy, the same systematic review reported an OR of 2.11.
Regarding chemotherapy vs. no chemotherapy, one cohort study reported a Hazard Ratio (HR) of 1.46. [26] The risk of lymphedema in relation to chemotherapy was investigated in this cohort study in patients with ALND, comparing multi-agent chemotherapy with chemotherapy with anthracyclines. Regarding chemotherapy with radiotherapy vs. chemotherapy without radiotherapy, HRs in this study varied from 0.30-4.09 vs. 3.78-5. 46.
The overall incidence of lymphedema increased over time, except in one study where lymphedema decreased because of decongestive lymphatic therapy. [18] One case control study described the risk of lymphedema due to infection in patients with ALND (OR 3.80). [30] BMI $30 as risk factor for lymphedema was described in one systematic review in patients with SNB as weak evidence, not providing data [12] and in two cohort studies (OR 3.59; adjusted for ALND OR = 4.1), [44] while an OR of 2.01 was found for BMI .25. [20] One study followed patients five years after ALND and provided nomograms that indicated a BMI .30 as a risk factor as well. [29] The influence of age on the development of lymphedema was described in one systematic review and four cohort studies, indicating young age (,50 years) [12,16,32] and age .65 years [30] as risk factors and increasing by age in another cohort study. [29].
One study reported that comorbidity led to a higher incidence of lymphedema. [17] We found level 1 evidence for ALND, radical mastectomy, radiotherapy to the axilla, concurrent radiotherapy and chemotherapy, and radiotherapy before chemotherapy as risk factors for lymphedema.
Regarding ALND, one cohort study reported decreased degree of daily activities. [17] Regarding ALND vs. SNB one systematic review and one cohort study described an increased risk of problems in performing daily activities. [3,13] ORs were calculated in two included studies in the systematic review (resp. 3.18 and 9.23). [13] Reported ORs for performing different tasks in one of the included studies in the systematic review varied from 2.13-2.34 when stratified by age, with age between 65 and 74 years at most risk and between 40 and 54 years at least risk compared to a non-breast cancer population. Decline in one or more tasks was described in another included study (34% vs. 50%, OR 0.8). One cohort study described the avoidance of normal arm use in cases of ALND compared to SNB (p ,0.001). [3] Regarding ALND (directly or after SNB) vs. SNB vs. lumpectomy, one cohort study described a decline of activities in the first year post-operatively in 39%-44% of the patients after ALND, 18%-19% in case of SNB and 12%-19% in case of lumpectomy. [45] Regarding ALND and mastectomy vs. ALND and lumpectomy, one cohort study reported more problems in arm and shoulder function, conducting social activities and work in the lumpectomy group (p,0.001). [3] Regarding ALND and mastectomy vs. ALND, lumpectomy and radiotherapy, daily activities were described at 1 month post-operatively, at 12 and at 24 months in overall percentages and percentages with severe decline in daily activities. [28] Percentages reduced over time, with more problems in the lumpectomy group. Regarding chest wall radiotherapy vs. no radiotherapy, one systematic review reported a decline in daily activities with ORs in three individual studies (resp. 1.32, 8.0 and 10.67) and percentages of 29% vs. 4% in another included study. [13] Regarding radiotherapy to the axilla and chest wall vs. radiotherapy to the chest alone, the same systematic review reported an OR of 2.64 in one included study. Regarding chemotherapy with cyclophosphamide, epirubicin and docetaxel vs. chemotherapy with cyclophosphamide, epirubicin and fluoracil, one cohort study described a higher risk in giving up daily activities (OR 1.59). [38] Overall, 34% of the population in this study showed a decline in the level of daily activities. Overall, one cross-sectional study described a decline in activities in 31% of the population. [34] One cohort study related radiotherapy to later starting remunerable work. [41] Activity level did not return to the pre-operative level within one year, [46] and at 10 months, 83% of the patients returned to work. [41] Young age as a predictive factor for a reduced number of metabolic equivalents was described in one cohort study. [46] Another cohort study described reduced use of the affected arm in leisure activities and with lifting a gallon of milk or during heavy household chores in relation to pain and feeling weak. [33].
Comorbidity was related to a decreased level of activities in daily living. [17].
We found level 2 evidence for ALND and radiotherapy, especially when the axilla was involved, as risk factors for decreasing the degree of daily activities.

Discussion
In this systematic review, we showed that breast cancer treatment results in multiple impairments in the arm and shoulder. We analyzed adverse effects for different components of breast cancer treatment and related these to the integrated treatment of breast cancer. Previous systematic reviews, as well as a part of the cohort studies included in this study, merely focused on only a part of the medical treatment and/or outcome measurements, while others only looked at a general level, without distinction between components. By distinguishing between each treatment modality and outcome measurement, we are the first to analyze the risk of each component of breast cancer treatment. We showed that patients treated with ALND are at the highest risk of developing impairments of the arm and shoulder. Reduced ROM and muscle strength, pain, lymphedema and decreased degree of activities in daily living were reported most frequently in relation to ALND. Lumpectomy was related to a decline in the level of activities of daily living. Radiotherapy and hormonal therapy were the main risk factors for pain.
An integrated approach in assessing the adverse effects of distinct breast cancer treatment modalities on impairments in arm and shoulder function is of clinical importance. Recovery from adverse effects can be addressed in multidisciplinary treatment of patients; for example, physical therapy may be suitable for the recovery of ROM, muscle strength, lymphedema and daily activities. In general, we expect that awareness and timely referral are very relevant for patients with impairments interfering with daily activities in early recovery [47]. More attention should be paid to scapular coordination and muscle strength in the early post-operative phase, as these impairments were reported even up to six years post-operatively. [12,13,15,37] We noticed that the included studies focused more on impairments in function than on activities of daily living or participation in remunerable work, hobbies and social activities. In future research, more awareness of these issues is warranted, as performing activities is an important outcome for quality of life. This will further build the body of knowledge for regaining full recovery of activities of patients with breast cancer in a multidisciplinary approach.
Unfortunately, due to the large variety in medical treatments and outcome measures, we could not perform a meta-analysis of our data. This emphasizes the importance of uniform description of treatment, analysis of outcomes, and use of uniform measurement instruments. Validated measurement instruments are important in assessing outcomes of treatments. We found a large variability of instruments, which made it difficult to compare studies and conduct a meta-analysis. This conclusion was also stated by authors of several included systematic reviews in our study [12,13,15]. International consensus regarding measurement instruments and the way of using them should be encouraged.
From our review it became clear that reduced ROM, pain and lymphedema are the most commonly described impairments. ROM decreased, especially in the first month post-operatively. As most systematic reviews presented data only for long-term followup after treatment, reductions in the first month were less noticed, but when described in cohort studies significance existed. After 12 months, percentages of patients with reduction in ROM and differences in ROM between the affected and unaffected shoulder were reduced but still existed. Wide variation of percentages shows the variability in defining ROM impairment and the way of measurement.
The incidence of lymphedema increased over time. One study reported a very high incidence of lymphedema after one month. [28] This may be due to real lymphedema or rather seroma or radiotherapy-induced breast infection. [48].
The study of Ozcinar et al. [18] showed that treatment of lymphedema decreased its severity. In general, the reported percentages of patients with lymphedema were higher when lymphedema was measured by a questionnaire. The Norman questionnaire appeared to be sensitive for detection, but not specific, [10] and may be used as an initial tool in detecting lymphedema. Volume is the most important outcome for lymphedema diagnosis and treatment evaluation; therefore, the questionnaire should be followed by tape measurement (calculated to volume) or water volumetry or perometry. Arm volume is also associated with Body Mass Index and body composition. Therefore we advocate to use percentage difference between arms (where A is the affected arm and U is the unaffected arm) A{U U À Á |100 or to use the formula for relative volume change (RVC) to determine outcome over time.

2=U2 A1=U1 {1
Activities in daily living and participation are important parameters for quality of life. Limitation in body functions and structures may be restrictive in performing activities and participating in social events. Only one systematic review [13] and six cohort studies [3,17,28,38,41,46] described limitations in activities and only three cohort studies described problems in participation. As half of the patients with breast cancer were of working age, more attention should be paid to daily activities, work capacity, hobbies and sports.
Several limitations to our study should be noted. Our cut-off point with a quality score .50% is to some extent arbitrary and may have resulted in the exclusion of valuable data in our analysis. Main reasons for the low quality scores of excluded studies were issues with subgroup analysis, lack of outcome measures, poor presentation of results and lack of sufficient follow-up. Firstly, we analyzed which articles in our search were included in the systematic reviews. Four systematic reviews were excluded: based on treatment before 2000 or with low quality score. The review with low quality score was narrative and based on retrospective data. We therefore think the exclusion of these studies has avoided bias and contribute to the robustness of our conclusions. Based on the homogeneity of the results our choice seems to be justified. Another point is that, instead of relying on the review synthesis, it would have been a possibility to use existing reviews as sources to identify primary data, which would increase the value of the paper. We choose to follow the recommendations according the Oxford Centre of Evidence-Based Medicine. In this system systematic reviews are one of the factors in evidence classification. If it would have been possible to perform a meta-analysis the original data would have been extracted from the reviews. However, as described, this was not possible. We deemed additional analysis not to be of added value for the purpose of our paper. Therefore we used quality scores to test the credibility of the conclusions of the original authors and used these in the synthesis. Adverse effects of radiotherapy that may influence limitations in arm and shoulder function, such as fibrosis of the skin and sub cutis, were not included in our study. In addition, adverse effects of chemotherapy and target therapy on general cardiopulmonary capacity were not included. Other reported symptoms such as sleep disturbances, weight gain, cardiac function and sensory disturbances have not been reported, as have anxiety and depression, while these problems may influence the capacity of performing daily activities.

Conclusions
Patients with breast cancer suffer from constraints in arm and shoulder in the first year post-operative and at long-term followup. Patients treated with ALND are most at risk for developing impairments of the arm and shoulder. Reduced ROM and muscle strength, pain, lymphedema and decreased degree of activities in daily living were reported most frequently in relation to ALND. Lumpectomy was related to a decline in the level of activities of daily living. Radiotherapy and hormonal therapy were the main risk factors for pain.
An integrated approach in addressing the adverse effects of distinct breast cancer treatment modalities on impairments in arm and shoulder function is of clinical relevance. Patients treated with ALND require special attention to detect and consequently address impairments in the arm and shoulder. Patients with pain should be monitored carefully, because pain limits the degree of daily activities.