Arm and shoulder morbidity in breast cancer patients after breast-conserving therapy versus mastectomy

Abstract

Introduction. The objective of this study was to compare the prevalence of late effects in the arm and shoulder in patients with breast cancer stage II who had radical modified mastectomy (RM) or breast-conserving therapy (BCT) followed by loco-regional adjuvant radiotherapy with or without chemotherapy/anti-oestrogen. Material and methods. All patients had axillary lymph node dissection. At a median of 47 months (range 32–87) post-surgery, 263 women (RM: n=186, BCT: n=77) were seen during an outpatient visit and had their arm and shoulder function and the presence of lymphedema assessed by a clinical examination, interview and self-rating. Volume calculation was used to measure lymphedema. Results. In the RM group 20% had developed arm lymphedema versus 8% in the BCT group (p=0.02). In multivariate analysis lymphedema was associated with a higher number of metastatic axillary lymph nodes [OR1.14, p=0.02], RM [OR 2.75, p=0.04] and increasing body mass index (BMI) [OR 1.11, p<0.01]. In the RM group 24% had a restricted range of motion in shoulder flexion compared to 7% in the BCT group (p<0.01). Shoulder pain was reported by 32% in the RM group and by 12% in the BCT group (p=0.001). Increasing observation time, RM, and increasing BMI were significantly associated with impaired arm/shoulder function. Discussion. Arm/shoulder problems including lymphedema were significantly more common after RM compared to BCT in irradiated breast cancer patients who have undergone axillary lymph node dissection. The performance of BCT should be encouraged when appropriate, to ensure a low prevalence of arm/shoulder morbidity including lymphedema.

Modified radical mastectomy (RM) has been the standard surgical treatment of breast cancer in Norway until the late 1990's, when breast-conserving therapy (BCT) gradually became the treatment of choice when suitable in patients with primary tumours. The negative consequences on arm/shoulder function and occurrence of lymphedema after RM have been thoroughly described. However, the differences in side effect profile of RM compared to BCT have not been definitely clarified [1].

Concerning lymphedema, most reports have focused on the extent of axillary lymph node dissection and axillary radiotherapy as main causative factors. The difference in side effects after sentinel node dissection and axillary node dissection has been explored recently. In these studies patients treated with RM and BCT sometimes are analyzed as one group, and thereby the side effect profiles of these two surgical approaches are masked [2–4]. A review indicates that there is a reduced prevalence of lymphedema due to more use of BCT [5].

The few studies comparing RM and BCT have reported a significantly lower prevalence of lymphedema after BCT [6], [7] and Clark et al. [8] reported RM as a risk factor for developing lymphedema. Two other studies showed no difference concerning lymphedema when comparing RM to BCT [9], [10]. Some of these studies have not considered the relation to breast cancer stage when comparing the two surgery groups.

Concerning shoulder function, two comparative studies have described significantly less impaired function after BCT compared to RM [11], [12] and moderately impaired shoulder mobility has been reported after BCT [13]. Other studies have found no significant differences in such impairment between patients treated with BCT or RM [10], [14].

A reasonable conclusion from these comparative studies is that the findings are conflicting concerning difference in side effect profiles between RM and BCT concerning lymphedema and impairment of arm/shoulder function. On this background this cross-sectional study examined impairment in breast cancer patients who had undergone either BCT or RM, all with axillary dissection and locoregional adjuvant radiotherapy. The primary aim was to compare the prevalence of lymhedema and clinically significant long-term effects in arm/shoulder in the surgery groups. The secondary aim was to examine the associations of these findings to treatment modalities, extent of the disease and patient demographic related factors.

Material and methods

Patient sampling

Women consecutively treated for stage II breast cancer, between 1998 and 2002 at The Norwegian Radium Hospital (NRH) were invited to take part in a follow-up study in 2004/2005. The survey consisted of mailed questionnaires and an outpatient clinical examination.

The survey concerned patients who fulfilled the following eligibility criteria: 1) Curatively intended surgery, either RM or BCT at seven different hospitals, followed by radiotherapy at NRH to the regional lymph node stations and the chest wall/breast; 2) No evidence of relapse since primary treatment; 3) No other malignant diagnosis; 4) Age ≤75 years at the time of the survey. The present part of the study was restricted to patients with pT1 – pT2 disease, all with metastatic axillary lymph nodes. The medical and demographic information was obtained from the medical records.

Treatment modalities

All treatment was based on the guidelines of the Norwegian Breast Cancer Group. The primary treatment consisted of RM or BCT and axillary lymph node dissection at level I–II. During RM the entire breast was removed including stripping of the pectoral fascia. BCT consisted of excision of the tumour with cancer free tissue margins of at least 3 mm.

Post-operative adjuvant chemotherapy and/or tamoxifen were administered according to the patients’ age at presentation and the hormone receptor status of the tumour.

The target volume of radiotherapy included the breast (BCT), the chest wall (RM), the ipsilateral fossa supraclavicularis, the lymph nodes along the ipsilateral arteria mammaria interna and the ipsilateral axilla. The post-mastectomy scar was covered by a 6cm wide bolus. In case of tumour involvement of the resection margins in the mastectomy specimen, the whole chest wall was covered by a bolus.

From 1998 to 2000 radiotherapy planning at NRH was based on manual routines. This technique required an electron beam field to cover the parasternal lymph nodes. From 2000 dose planning was introduced using the treatment planning system Helax-TMS. The beam arrangement consisted of 4 half-beams. All patients in both surgery groups received 25 fractions of 2 Gy to the chest wall and eventually to the breast, and 23–25 fractions of 2 Gy to the regional lymph node stations. Patients ≤50 years of age, who had undergone BCT, also received an electron boost of 5 fractions of 2 Gy to the tumour bed.

From October 1999 patients who had removed ≥10 axillary nodes without perinodal infiltration had irradiation only to the axillary apex, as opposed to those with <10 nodes removed, who had radiotherapy to the entire axilla as was the treatment modality for all patients having radiotherapy before that time.

Arm/ shoulder morbidity

During the outpatient visit two physiotherapists with long clinical experience with breast cancer patients evaluated arm/shoulder symptoms based on a questionnaire, clinical assessment, and a structured interview.

The questionnaire, developed at the British Columbia Cancer Agency in Canada (KQAP), assessed persistent problems with arm/shoulder function, pain, stiffness and swelling after the treatment [2]. KQAP consists of eight questions rated on five point Likert scales. The raw scores were dichotomised: scores of 1 and 2 as “none/mild symptoms,” and scores of 3 to 5 as “moderate/severe symptoms”. The presence of arm/shoulder symptoms were defined as “moderate/severe symptoms”. The questionnaire, translated into Norwegian, was filled in by the patients before the clinical examination.

Clinical physical assessment comprised a goniometer measurement of flexion (forward elevation of arm) and abduction (lateral elevation of arm) of both arms. Patients with changes in arm function caused by other disorders were excluded from this examination. Our clinical experience indicated that a reduced range of motion between 10–20° did not limit the daily activities of the patients. Hence we defined a reduced range of motion of ≥25° compared with the contra-lateral side as definition of impaired shoulder function. The patients were told to report the location of tightness when performing the test.

The presence of lymphedema was assessed by volumetric calculation using five circumferential measurements on both arms, as high correlation with water displacement volumetry (r = 0.96) and a high degree of reliability has been demonstrated [15]. The summed volume of the truncated cones was calculated. The definition of lymphedema comprised: 1) A difference of ≥10% in volume between the two arms; or 2) ≥2 cm increase of any circumference in the affected arm versus the contra-lateral arm and in addition used compression garment and/or had treatment for arm edema. Severe lymphedema was defined as a difference of ≥20% in volume between the two arms. In order to standardize our measurements, the women in their invitation letter were asked to refrain from wearing compression garment for 24 hours before the examination.

Lymphedema in the affected breast was evaluated by palpation and visual estimation. Additional information of subjective arm symptoms and treatment of swelling was collected by interview.

Statistical analysis

Standard descriptive statistical methods were used. Categorical variables were assessed by χ² statistics. Due to skewed distributions non-parametric tests were used to compare groups on continuous variables. We determined clinical significance in continuous and 2×2 contingency tables by calculating effect size. Clinical significance was considered as statistically significant at ES ≥ 0.40 [16]. Univariate and multivariate logistic regression analyses were used to explore associations between different treatment modalities, patient characteristics and different symptoms. Independent variables which were statistically significant in the univariate analysis or of considerable theoretical interest were included into the multivariate analysis. The strength of the associations was expressed as odds ratios (OR) with 95% confidence intervals (95% CI). To explore strength and direction of relationship between continuous variables, scatterplots, Pearson's and Spearman's correlation coefficient were applied. The analyses were performed using SPSS for Windows, version 13. P-values of <0.05 were considered statistically significant, and all tests were two-tailed.

Ethical considerations

The study was approved by the Regional Ethical Committee of Health Region South of Norway and by the National Data Inspectorate. All patients delivered written informed consent before inclusion.

Results

A total of 415 eligible women were invited. Fifty-one patients (12%) declined the outpatient visit and 46 patients (11%) did not respond to the invitation. Of the 318 attending the outpatient consultation, 55 patients with more advanced cancer (T3-T4) were excluded in order to obtain two more homogeneous disease groups. An attrition analysis of the 318 respondents and 97 non-respondents showed no significant differences between the groups concerning demographic, treatment and medical characteristics (data not shown).

Among the 263 patients who met the inclusion criteria of the study, 186 had undergone RM and 77 BCT (Table I). The two surgical modality groups did not differ significantly as to age, surgery to the axilla or BMI. However, they showed significant difference as to extent of the disease since 54% in the RM group had pT2 tumours compared to 21% in the BCS group (p < 0.001). The number of metastatic lymph nodes was almost significantly higher in the RM group than in the BCT group (p = 0.06). However, the proportion of patients with ≥10 lymph nodes removed was somewhat higher in the BCT group (p = 0.08). We found no significant correlation between the number of lymph nodes dissected and number of metastatic lymph nodes (r = 0.23). In the RM group 41% had manual dose planned radiotherapy as compared to 22% in the BCT group (p < 0.01). In the RM group 44% had radiotherapy to the entire axilla opposed to 23% in the BCT group (p < 0.01; Table I).

Table I.  Patient and treatment characteristics in the surgery groups.

	Mastectomy (n = 186)	BCT* (n = 77)
Characteristics	Median (Range)	Median (Range)	P-value
Age	55 (35–75)	55 (30–71)	0.32
BMI (before surgery)	26 (19–45)	26 (20–49)	0.46
Observation time (months)	48 (32–87)	44 (34–84)	0.01
	N (%)	N (%)
Tumour pT1	85 (46)	61 (79)
pT2	101 (54)	16 (21)	<0.001
1–9 lymph nodes removed	56 (30)	15 (20)
≥10 ” ” ”	130 (70)	62 (80)	0.08
1–3 metastatic lymph nodes	115 (62)	59 (77)
4–9 metastatic lymph nodes	63 (34)	17 (22)
>10 metastatic lymph nodes	8 (4)	1 (1)	0.057
Manual dose planning	77 (41)	17 (22)
CT dose planning	109 (59)	60 (78)	<0.01
RT † to axilla	81 (44)	18 (23)
RT to axillary apex only	105 (56)	59 (77)	<0.01
Got chemotherapy	151 (81)	60 (78)	0.5

*Breast-conserving therapy, † Radiotherapy.

Lymphedema and impaired arm/shoulder function by clinical examination

Lymphedema was observed in 43 patients, 37 (20%) belonging to the RM group compared to 6 (8%) in the BCT group (p = 0.02) (Table IIA). The corresponding effect size did not show clinical significance (d = 0.36). Nine patients had severe lymphedema, all of them belonging to the RM group. Six patients in the BCT group had lymphedema in the affected breast. Among those with arm lymphedema, 80% reported that the edema had developed during the first year after surgery and 17% during the second year.

Table II.  A and II B. Lymphedema, impaired shoulder function and other symptoms in the surgery groups.

	Mastectomy (n = 186)	BCT* (n = 77)
Variable	N (%)	N (%)	P-value
A. Findings from clinical examination
Lymphedema	37 (20)	6 (8)	0.02
Lymphedema
1–3 metastatic nodes	17/115 (15)	5/59 (8)	0.34
4–9 metastatic nodes	15/63 (24)	1/17 (6)	0.17
10–16 metastatic nodes	5/8 (62)	0/1 (0)	0.44
ROM† reduction ≥25°
Flexion of arm	42 (24)	5 (7)	<0.01
ROM reduction ≥25°
Abduction of arm	66 (38)	12 (18)	<0.01
B. Findings from questionnaire‡
How swollen is your arm?	53 (29)	10 (13)	<0.01
How much pain do you have in your arm?	59 (32)	10 (13)	<0.01
How much pain do you have in your shoulder?	58 (32)	9 (12)	0.001
How stiff is your arm?	35 (19)	7 (9)	0.05
How stiff is your shoulder?	48 (26)	6 (8)	0.001
How well can you use your arm compared to before your surgery?	105 (58)	25 (33)	<0.001
How numb is your arm?	42 (23)	13 (18)	0.25
How well can you move your arm compared to before your surgery?	60 (33)	9 (12)	0.001

*Breast-conserving therapy, † Range of Motion.

‡Scores indicate persistent moderate/severe symptoms.

In both univariate and the multivariate logistic regression analyses increasing number of metastatic lymph nodes, RM and increasing BMI were significantly associated with lymphedema present as the dependent variable (Table III).

Table III.  Associations of relevant variables and lymphoedema in the total sample.

	Univariate analysis			Multivariate analysis
Variable	OR	95% CI	P	OR	95% CI	P
Age at survey	1.02	0.98–1.06	0.3
Observation time	1.02	0.99–1.05	0.2
BCT* (ref)	1.00
Mastectomy	2.94	1.2–7.3	0.02	2.75	1.03–7.33	0.04
CT dose planning (ref)	1.00
Manual dose planning	1.7	0.87–3.2	0.1
RT† to apex of axilla (ref)	1.00
RT to axilla	1.12	0.56–2.22	0.75
pT1 (ref)	1.00
pT2	1.72	0.89–3.33	0.1	1.18	0.58–2.4	0.64
Metastatic lymph nodes	1.19	1.08–1.3	0.001	1.14	1.03–1.27	0.02
No of lymph nodes removed	1.08	1.005–1.15	0.04	1.05	0.98–1.14	0.17
Body mass index	1.11	1.04–1.18	<0.01	1.11	1.04–1.19	<0.01

Impaired shoulder function is defined as ROM reduction ≥ 25.

* Breast-conserving therapy, † Radiotherapy.

Impaired flexion (by ≥25° of the ipsilateral arm) was observed in 24% of the women in the RM group and 7% in the BCT group (p < 0.01). The corresponding proportions for impaired abduction were respectively 38% and 18% (p < 0.01; Table IIA). Both these variables showed clinical significance when calculating effect size. In the RM group 16% reported tightness in the chest wall when flexing the arm in contrast to 1% in the BCT group (p = 0.001).

Multivariate logistic regression analysis showed that RM, increasing BMI and observation time were significantly associated with impaired shoulder abduction (Table IV). RM and increasing BMI were significantly associated with impaired flexion in the multivariate logistic regression analysis (Table V).

Table IV.  Associations of relevant variables and impaired shoulder abduction as dependent variable.

	Univariate analysis			Multivariate analysis
Variable	OR	95% CI	P	OR	95% CI	P
Age	1.03	0.99–1.06	0.13
Observation time	1.04	1.02–1.07	0.001	1.04	1.006–1.09	0.02
BCT* (ref)	1.00
Mastectomy	2.75	1.37–5.5	0.004	2.3	1.06–4.98	0.04
CT dose planning (ref)	1.00
Manual dose planning	2.1	1.2–3.67	0.008	0.8	0.34–1.86	0.61
RT to apex of axilla (ref)	1.00
RT † to axilla	2.22	1.25–3.88	0.006	1.19	0.54–2.6	0.67
pT1 (ref)	1.00
pT2	1.5	0.87–2.58	0.14	1.4	0.75–2.6	0.29
Metastatic lymph nodes	0.98	0.9–1.09	0.64
Lymph nodes removed	0.92	0.85–0.98	0.01	0.94	0.87–1.02	0.11
Body mass index	1.1	1.03–1.17	0.004	1.12	1.05–1.2	0.001

Impaired shoulder function is defined as ROM reduction ≥ 25.

*Breast-conserving therapy, † Radiotherapy.

Table V.  Associations of relevant variables and impaired shoulder flexion as dependent variable

	Univariate analysis			Multivariate analysis
Variable	OR	95% CI	P	OR	95% CI	P
Age	1.03	0.98–1.08	0.2
Observation time	1.06	1.03–1.1	<0.001	1.03	0.99–1.08	0.13
BCT* (ref)	1.00
Mastectomy	4.14	1.6–10.9	0.004	3.3	1.2–9.14	0.02
CT dose planning (ref)	1.00
Manual dose planning	4.98	2.5–9.9	<0.001	2.5	0.93–6.1	0.06
RT to apex of axilla (ref)	1.00
RT† to axilla	2.97	1.54–5.7	0.001	1.22	0.51–2.75	0.64
pT1 (ref)	1.00
pT2	0.83	0.4–1.58	0.6
Metastatic lymph nodes	1.05	0.95–1.2	0.3
Lymph nodes removed	0.93	0.86–1.01	0.09
Body mass index	1.08	1.01–1.16	0.02	1.09	1.02–1.18	0.02

Impaired shoulder function is defined as ROM reduction ≥ 25.

*Breast conserving therapy, † Radiotherapy.

Arm and shoulder symptoms in the questionnaire

All, except four women, completed the questionnaire. Significantly less “moderate/ severe symptoms” were reported in the BCT group compared to the RM group, for all KQAP items, except arm stiffness and numbness. In the RM group 58% and 33% in the BCT group reported moderate/severe problems with arm function (p < 0.001). In the RM group 26% reported shoulder stiffness compared to 8% of the BCT group (p = 0.001). Pain in the shoulder was reported by 32% in the RM group and in 12% in the BCT group (p = 0.001) (Table IIB).

All variables in the questionnaire with a statistically significant p-value showed clinical significance when calculating effect size (d ≥ 0.40).

Discussion

Compared to RM significantly fewer patients treated with BCT suffered from arm/shoulder problems including lymphedema. The number of metastatic lymph nodes, RM and BMI were significantly associated with lymphedema in the multivariate analysis, as were RM, increasing observation time and increasing BMI with impaired arm/shoulder function as outcome.

The difference in prevalence of lymphedema between the surgical modality groups was an unexpected finding. Since most studies on lymphedema have focus on treatment modalities to the axilla, few of these studies are comparable to ours.

A study in agreement with our results by Schünemann and Willich [6] compared patients who had undergone BCT and RM. They found 7% prevalence of lymphedema in 179 patients with BCT and 19% in 2 170 patients with RM. In another study Clark et al. [8] reported a significant difference in lymphedema between the surgery groups. Neither of these reports tested other variables eventually associated with lymphedema in multivariate analysis.

Two recent studies complemented our study by reporting significantly less severe lymphedema among patients with BCT than with RM, all referred for treatment of lymphedema [17], [18]. A non-comparative study reported a prevalence of 11% lymphedema in a BCT group with a treatment program similar to ours [19]. However, these results and ours are in contrast to a comparative study by Gerber et al. [9] who reported a slight and non-significant increase in circumferential arm measurements in a BCT group compared to a RM one. Hence, we see that the results are conflicting concerning differences in prevalence of lymphedema after BCT or RM. Additional large studies from major cancer hospitals are needed.

Nine women had not removed the armsleeve prior to examination, but eight of them still complied with the definition of lymphedema, so that procedure violation hardly affected the prevalence of lymphedema.

Many studies, including the present one, report that most cases of lymphedema develop during the first 1–2 years after primary treatment [8], [6], [20]. Accordingly, a median observation time of nearly 4 years is considered sufficient for identification of the majority of lymphedema cases.

Some studies have reported an association between lymphedema and increasing number of lymph nodes removed [13], [21], whereas the present study and other studies have not found such an association [8], [14]. In our study decreasing number of lymph nodes removed was associated with impaired abduction in the univariate analysis but not in the multivariate analysis. These inverse relationships are probably explained by the extended radiation field toward the axilla in patients with <10 nodes removed.

Increasing number of removed metastatic lymph nodes was associated with lymphedema. Similar findings have been reported in other studies [2], [6], [20].

The proportion of metastatic nodes was higher and the number of pT2 patients was significantly higher in the RM group than in the BCT group. Hence the higher prevalence of lymphedema in the RM group is partly explained by the more advanced disease in this group. However, Powell et al. [19] found that lymphedema risk was more closely tied to patient- and treatment-related factors than to disease burden in patients with stage I–II breast cancer.

The observed increasing BMI as a factor significantly associated with lymphedema has been reported in numerous other studies [17], [8].

Our results concerning arm/shoulder function are in line with Sugden et al. [11] and Lauridsen et al. [12] who described significantly better arm/shoulder function after BCT compared to RM. Kuehn et al. [14] and Ernst et al. [10] found no difference in arm mobility when comparing RM and BCT groups. However, Ernst et al., who defined a discrepancy of ≥20° as impairment, found that severe impairment in abduction occurred more often in the RM group. Another study using a 15° difference as cut-off for impaired shoulder function [22], found no correlation between patients’ self-report and clinical findings. These findings and our clinical experience make us conclude that 25% difference is a clinically more valid definition of arm/shoulder impairment.

Lauridsen et al. [12] emphasized that the dissection of the pectoral fascia during RM sometimes causing adherences between the subcutaneous tissue and the underlying muscle. Such post-operative effects might explain the tightness in the chest wall when flexing the arm mainly reported by the patients in the RM group. Radiotherapy-induced fibrosis located in the same tissue probably adds to the impairment of the shoulder function.

Bentzen et al. [23] estimated that 90% of late treatment effects in the shoulder's soft tissue after radiotherapy developed during the first 3.9 years. We therefore conclude that the median observation time of 47 months in our study is sufficient for including the majority of those patients who develop long-term impairment in arm/shoulder function.

The relatively homogeneous axillary surgery in the two groups, done by different surgeons at seven different hospitals, is considered the strength of our study. A compliance of 75% is fairly good, and the lack of significant differences in the attrition analysis implies that the results in the present study are representative of the entire population. A limitation in our study was the relatively low number of patients in the BCT group which was reflected in the wide 95%CI in the regression analysis.

Our main conclusion is that lymphedema and arm/shoulder symptoms are significantly lower after BCT compared to RM in stage II breast cancer patients. Both the surgical intervention to the breast and the extent of the disease by diagnosis play a role in the development of lymphedema, whereas our study indicates that impairment in arm/shoulder function is not associated with the extent of the disease. To reduce late impairment, early detection is of importance. The performance of BCT should be encouraged when appropriate, to ensure a low prevalence of arm/shoulder morbidity and lymphedema.