Morbidity Results from the NSABP B-32 Trial Comparing Sentinel Lymph Node Dissection versus Axillary Dissection

doi:10.1002/jso.21535

J Surg Oncol. Author manuscript; available in PMC 2011 August 1.

Published in final edited form as:

J Surg Oncol. 2010 August 1; 102(2): 111–118.

doi: 10.1002/jso.21535

PMCID: PMC3072246

NIHMSID: NIHMS279646

Morbidity Results from the NSABP B-32 Trial Comparing Sentinel Lymph Node Dissection versus Axillary Dissection

Takamaru Ashikaga, PhD,¹ David N Krag, MD,¹ Stephanie R Land, PhD,^2,³ Thomas B Julian, MD,^2,³ Stewart J Anderson, PhD,³ Ann M Brown, PhD,³ Joan M Skelly, MS,¹ Seth P Harlow, MD,¹ Donald L Weaver, MD,¹ Eleftherios P Mamounas, MD,⁴ Joseph P Costantino, DrPH,³ Norman Wolmark, MD,^2,⁵ and the National Surgical Adjuvant Breast and Bowel Project (NSABP)

¹University of Vermont College of Medicine, Burlington, VT, USA

²National Surgical Adjuvant Breast and Bowel Project, Pittsburgh, PA, USA

³Department of Biostatistics, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, PA, USA

⁴Aultman Health Foundation, Canton, OH, USA

⁵Allegheny General Hospital, Pittsburgh, PA, USA

Correspondence to: Takamaru Ashikaga, Department of Medical Biostatistics, University of Vermont College of Medicine, Burlington VT, 05405, USA. Telephone: 802-656-2526; Fax: 802-656-3632; Email: Takamaru.Ashikaga/at/uvm.edu

The publisher's final edited version of this article is available at J Surg Oncol

See other articles in PMC that cite the published article.

Abstract

Background and Objectives

Three year post-surgical morbidity levels were compared between patients with negative sentinel lymph node dissection alone (SLND) and those with negative sentinel node dissection and negative axillary lymph node dissection (ALND) in the NSABP B-32 trial.

Methods

A total of 1975 ALND and 2008 SLND node negative breast cancer patients had shoulder range of motion and arm volumes assessed along with self reports of arm tingling and numbness. Relative shoulder abduction deficits and relative arm volume differences between ipsilateral and contralateral arms were calculated.

Results

Shoulder abduction deficits ≥ 10% peaked at one week for the ALND (75%) and SLND (41%) groups. Arm volume differences ≥ 10% at 36 months were evident for the ALND (14%) and SLND (8%) groups. Numbness and tingling peaked at 6 months for the ALND (49%, 23%) and SLND (15%, 10%) groups. Logistic regression correlates of residual morbidity included treatment group, age, handedness, tumor size, systemic chemotherapy and radiation to the axilla.

Conclusions

Although residual morbidity for both treatment groups was evident, the results of the NSABP B-32 study indicate the superiority of the SLND compared to the ALND treatment approach relative to post-surgical morbidity outcomes over a three year follow-up period.

Keywords: shoulder abduction, arm swelling, numbness, tingling

INTRODUCTION

Several morbidity reports of randomized trials comparing traditional axillary lymph node dissection (ALND) with sentinel lymph node dissection (SLND) followed by ALND for those with pathologically positive sentinel nodes have appeared in the literature over the last few years [1-6]. The Sentinel Node Biopsy versus Axillary Clearance (SNAC) study randomized 1088 patients and included as trial outcomes arm volume assessments estimated using a truncated cone approach, shoulder abduction using a goniometer, a variety of patient reports of arm symptoms and quality of life measures collected at baseline prior to surgery and at 3 and 6 months after surgery [1]. The Italian GIVOM trial randomized 749 patients and included trial outcomes such as arm volume assessment using a truncated cone approach, shoulder abduction using a goniometer, a variety of patient reports of arm symptoms and quality of life measures collected at baseline prior to surgery and at 3 and 6 months and annually after surgery [2]. The ALMANAC trial randomized 1,031 patients and followed patients at 1, 3, 6, and 12 months with assessment of arm swelling with the sum of conical sections and self-reported arm numbness as part of their quality of life measure [3, 4]. The Cambridge Breast Unit reported on a small randomized study of 298 patients who had assessments of sensory loss and paresthesia obtained preoperatively and at 1, 3, 6, and 12 months after surgery [5]. An early report from Italy described a randomized trial of 516 patients with follow-up interviews among a subset of 200 patients at 12 and 24 months after surgery to obtain data on pain, numbness or paresthesia, arm mobility restriction and measured arm circumferences for swelling [6]. These studies concluded that SLND resulted in less morbidity compared to ALND although the study morbidity effect sizes appeared to vary due to differing samples sizes and follow-up durations as well the inclusion in the SLND group of both sentinel node negative patients and sentinel node positive patients who received axillary dissection. In addition, most of these studies utilized the sum of truncated conical sections to approximate arm volumes in contrast to the more precise water immersion approach to quantify arm swelling used in the current report.

The primary objectives of the NSABP B-32 trial are a comparison of survival, regional control and morbidity endpoints between SLND and ALND treatment groups among clinically node negative women with operable invasive breast cancer [7]. The definitive comparison of survival and regional control endpoints remain a focus of a future report. Details of the B-32 trial technical success rates for identification of sentinel lymph nodes as well as false negative rates have been reported [8]. This report will focus on the three year longitudinal characterization and comparison of the following morbidity outcomes between the SLND and ALND treatment groups in patients with pathologically negative sentinel nodes: shoulder range of motion, arm swelling, altered sensations and the incidence of residual morbidity. The morbidity results are based upon the longest follow-up of patients to date from the largest definitive U.S. Phase III trial comparing SLND and ALND treatment groups.

MATERIALS AND METHODS

B-32 Morbidity Cohorts

Women with operable invasive primary breast cancer and clinically negative nodes were randomly assigned to receive either SLN resection followed by immediate conventional ALND (Group 1) of the remaining non-SLNs versus SLND followed by ALND only if one or more SLNs were positive on intraoperative cytology or subsequent histological examination or if no SLNs were identified (Group 2). Between May, 1999, and February, 2004, 5611 patients were entered and randomized in the study by 233 surgeons from 80 institutions in the USA and Canada (Figure 1). Randomization was accomplished centrally by the NSABP Biostatistical Center (Pittsburgh, PA, USA) with stratification by age at entry (≤ 49 years or ≥ 50 years); surgical treatment plan (lumpectomy or mastectomy); and clinical tumor size in centimeters (cm) (≤ 2.0, 2.1–4.0, or > 4.0). Since ALND was mandated in Group 2 patients with pathologically positive sentinel nodes, morbidity factors were assessed only in patients with pathologically negative sentinel nodes in Groups 1 (1975 patients) and 2 (2008 patients).

Figure 1

Flow diagram with randomization schema of the B-32 trial. Asterisk indicates that in this group, patients in whom a sentinel lymph node was not identified received an axillary lymph node dissection.

Objective Shoulder Abduction Assessments

Arm mobility in degrees was assessed at baseline and at one week, at two to three weeks and at six months post-surgery by measuring the straight lateral abduction of both the ipsilateral and the contralateral arms using a standard orthopedic goniometer to determine the angle between the lateral chest wall and the humerus.

Objective Arm Volume Assessments

Arm volume in milliliters (mls) was measured by a water displacement method using a large graduated plastic cylinder for both the ipsilateral and the contralateral arms. A mark was placed 10 cm proximal from the lateral epicondyle in a straight line to the mid-deltoid muscle. The arm was then inserted into the cylinder and the cylinder was filled up to the mark on the arm with the water displacement being recorded. The water displacement was then recorded after removal of the arm with the difference between the two displacement volumes representing the arm volume. Arm volume measurements were obtained at baseline and every six months over the three year follow-up period.

Subjective Assessments of Numbness and Tingling

Sensory nerve damage was assessed with two self-report questions regarding numbness and paresthesia by asking if the patient was currently experiencing any numbness and any tingling anywhere in the ipsilateral and the contralateral arms. These two self-reports were obtained at baseline and every six months over the three year follow-up period.

Statistical Methods

An intent-to-treat approach was used for all comparisons. Baseline comparisons between both groups were made with Fisher's Exact Test for binary items and the Kruskal-Wallis rank sum test for node count data and actual tumor sizes. Relative shoulder abduction deficits and relative arm volume differences were obtained using the formula (ipsilateral – contralateral)/(contralateral) x 100% with time specific group comparisons being made by Kruskal-Wallis tests. The deficits in range of motion and differences in arm volume was broken down by categories (< 5% deficit, 5-10% deficit; ≥ 10% deficit), and (< 5% difference; 5-10% difference; ≥ 10% difference), respectively. Group specific prevalence values of self-reported numbness and tingling for the affected arm and the associated odds ratio (OR) for the SLND group relative to the ALND group were estimated at each follow-up time point. The prevalence rates were compared using Fisher's Exact Tests supplemented with exact 95% confidence interval for the observed OR.

Residual shoulder abduction morbidity deficit was defined as a difference in shoulder abduction < 5% at baseline with a larger relative shoulder abduction deficit (≥ 5%) at the 6-month follow-up. Residual relative arm volume morbidity was defined as a relative arm volume difference < 5% at baseline and with an increased relative arm volume difference (≥ 5%) at the 36 month follow-up. Residual arm numbness and arm tingling morbidity was defined to exist among those patients who reported no pain or tingling in the affected arm at baseline and who then reported such at 36 months follow-up. Exact 95% confidence intervals using a binomial distribution were obtained for individual incidence rates and compared using Fisher's Exact Test supplemented with 95% confidence intervals. Logistic regression models were developed to explore if treatment group, age, side of affected arm, dominant hand, concordance between affected arm and dominant hand, number of sentinel nodes examined, number of nodes removed from the ipsilateral axilla, number of nodes from other locations during the sentinel node procedure, type of surgery, tumor size, systemic adjuvant and radiation therapy as well as their interactions with treatment group were associated with each of the four residual morbidity conditions using a forward and backward variable selection processes. The SAS (Version 9.1) [9], SYSTAT (Version 11) [10] and StatXact (Version 4) [11] statistical packages were used for data management and analysis. Observed p-values ≤ 5% were considered statistically significant for testing purposes.

Protection of Human Subjects from Research Risks

Each participating institution received approval by their local institutional review boards and filled assurances that were approved by the Department of Health and Human Services which were in accordance with the ethical standard of the Helsinki Declaration of 1975, as revised in 1983. Written informed consent was required for patient participation.

RESULTS

Patient Characteristics and Completeness of Follow-up Assessments

Baseline characteristics were comparable between the 1975 ALND and the 2008 SLND node negative patients (Table I). However, group differences were observed for the dominant arm being affected and receipt of radiation therapy in the supraclavicular area. A total of 1968 of the 1975 ALND patients and a total of 2006 of the 2008 SLND patients had baseline functional assessments completed. The 6-month follow-up rates for the lateral shoulder abduction assessments were 85% and 87% for these patients in the ALND and SLND groups, respectively. Baseline relative arm volume differences were obtained for 1956 ALND patients and 1996 SLND patients. Self-reported baseline arm numbness was available for 1962 and 2002 patients while baseline self-reported arm tingling was available for 1961 and 2002 patients from the ALND and SLND groups, respectively. The 36-month follow-up rates for the arm volume and the two self-reported items were similar at 73% for both groups.

Table I

Characteristics of SN Negative Patients by Treatment Group

Morbidity Comparisons

The prevalence of shoulder abduction deficits ≥ 10% peaked at one week of follow-up for the ALND (75%) and the SLND (41%) groups (p < 0.001) and differences continued at the 2-3 week and 6 month follow-ups (both p < 0.001) (Table II). Statistically significant arm volume differences favoring the SLND group appeared at 6 months and continued to the 36-month follow-up (all p < 0.001) (Table II). Differences in the prevalence of self-reported tingling and numbness (Table III) for the ipsilateral arm also appeared at 6-months and beyond (all p < 0.001). Compared to the ALND group, patients in the SLND group had an OR for reported tingling of 0.52 or less at all follow-up times. In addition, patients in the SLND group had an OR for reporting numbness in their affected arm of 0.21 or less at all follow-up times. The 36-month prevalence rates for tingling (13.5%) and numbness (31.1%) among the ALND group patients were greater than those for the SLND group (7.5% and 8.1%, respectively). Also, the 36 month rates in the ALND group did not return to baseline levels to the same degree as that of the SLND group of patients.

Table II

Distribution of Patients within Treatment Group by Follow-up Time and Categories of Prevalence for Relative Shoulder Abduction Deficits and Relative Arm Volume Difference.

Table III

Prevalence of Self-Reported Tingling and Numbness within Treatment Group by Follow-up time.

Residual Morbidity

The group specific incidence rates of residual shoulder abduction deficits at 6 months, and residual arm volume differences, residual arm tingling and residual arm numbness at 36 months revealed statistically significant ORs favoring the SLND group for all four residual morbidity outcomes (all p < 0.001) (Table IV). The most pronounced difference was observed for residual arm numbness which had incidence rates of 30.5% (CI: 28.1% - 33.1%) and 7.5% (CI: 6.2% - 9.0%) for the ALND and SLND groups, respectively. The smallest group difference was observed for shoulder abduction with 6 month incidence rates of 19.0% (CI: 17.1% - 21.2%) and 13.2% (11.5% - 15.0%) for the ALND and SLND groups, respectively.

Table IV

The Occurrence of Residual Morbidity at the End of Follow-up for Patients with Little or No Morbidity at Baseline.

Associations of Residual Morbidity with Patient and Tumor Characteristics

Logistic regression models focused upon the treatment effect favoring the SLND group and the potential incremental impact of measures in Table I. The incidence of residual shoulder abduction deficits were inversely associated with receipt of systemic chemotherapy (OR = 0.73, p = 0.003) and positively associated with the receipt of radiation therapy to the axilla (OR = 2.48, p = 0.037) (Table V). Residual arm volume differences were positively related to age (OR = 1.41, p = 0.006), concordance of dominant arm and affected arm (OR = 1.77, p < 0.001) and the receipt of radiation therapy to the axilla (OR = 3.47, p = 0.007). Residual arm numbness was inversely related to both age (OR = 0.66, p < 0.001) and tumor size (OR = 0.75, p = 0.020). None of the measures were associated with reports of residual arm tingling. In addition, none of the measures in Table V demonstrated a statistically significant interaction with treatment group.

Table V

Multivariate Logistic Regression Model Results for Patients who Developed Residual Morbidity at the End of Follow-up and Who had Little or No Morbidity at Baseline.

DISCUSSION

Shoulder Abduction Morbidity

Any surgery to the axilla is likely to cause an immediate but temporary decrease in shoulder abduction due to postoperative pain. Once protective behavior related to postoperative pain is no longer necessary, decreased abduction is related to other issues that mechanically restrict motion or secondarily induce pain. A range of motion of 160° to 180° is considered normal [12]. Thus, a deficit ≥ 10% for an individual with a 170° contralateral arm range of motion would indicate that the ipsilateral range of motion to be below the normal range. The percentage of patients in the SLND group who had relative shoulder abduction deficits ≥ 10% peaked at 41% at week 1 and declined to 6% at 6 months of follow-up compared to the corresponding values of 75%and 9% for the ALND group. The 6 month OR value of 0.61 appears comparable to the SNAC trial [1] and the GIVOM trial for self-reported motion restrictions [2] although not as favorable as results for subjective arm mobility restrictions cited in an earlier report [6]. While the prevalence of range of motion deficits appear time dependent, a fraction of women in both treatment groups continued to report problems at the end of the 6 month follow-up.

Arm Volume Morbidity

Lymphedema is perhaps the most dreaded long term complication related to axillary lymph node surgery. Once present it implies a lifelong problem. In addition to functional impairment, lymphedema may result in a daily reminder of breast cancer and be an added psychological burden. Serious upper extremity lymphedema or treatment devices are not easily concealed adding to self conscious burden. A 10% threshold for relative arm volume differences has been used in some studies to indicate the presence lymphedema [13]. The prevalence of arm volume differences ≥ 10% for the SLND group stabilized at 7% to 9% while the ALND group remained relatively stable at 13% to 14% with the corresponding OR values ranging from 0.45 to 0.69 at follow-up. These OR values are generally comparable to that seen for the SNAC trial for within subject arm volume increases ≥ 15% for the affected arm [1] although not as favorable for the SLND group as that reported for self-reported lymphedema from the GIVOM study at 24 months of follow-up [2] and the 12 month follow-up data from the ALMANAC trial [3] and the Cambridge Breast Unit [5]. The B-32 results are also substantially less favorable for the SLND group compared to the smaller randomized trial reported by Veronesi and colleagues [6]. This report and other prior reports citing subjective arm swelling or lymphedema thus appear to favor the SLND approach relative to the ALND approach although those reporting subjective data appear to favor the SLND approach to an even greater degree compared to those studies using objective measures. In addition, the current results indicate that the percentage of women within each group who develop larger relative arm volume differences appears to remain at the same levels over the follow-up times for both treatment groups.

Reported Arm Numbness and Tingling

Although neuromotor dysfunction is rare, neurosensory dysfunction following axillary node surgery is common. Sensory nerves traversing the axillary fat pad are often sacrificed during axillary node resection. Even with nerve preservation, removal of surrounding supporting tissue can lead to sensory nerve dysfunction. Self reported arm tingling peaked at 6 months for both groups. While the prevalence of tingling decreased over time, the SLND group remained consistency lower compared to the ALND group with OR values less than 0.52. Similarly, self reported arm numbness also peaked at 6 months and decreased over time. Here the SLND group had a much lower percentage of patients reporting numbness compared to the ALND group with ORs being below 0.21. The GIVOM study cited similar results for self-reported numbness in the affected arm favoring SLND group at 24 months of follow-up (OR = 0.54) [2]. Purushotham and colleagues estimated one year cumulative sensory symptom incidence rates for numbness for the SLND (48%) and ALND (74%) groups which represented an OR of 0.32 favoring the SLND group. A similar OR value of 0.36 for the one year incidence of any paresthesia resulted from cumulative incidence rates of 66% and 84% for the SLND and ALND groups, respectively [5]. The ALMANAC trial gave estimates of self reported sensory deficits for the SLND group of 11% at 12 months while the corresponding deficit for the ALND group was 31%. The OR value of 0.28 favoring SLND group also appears comparable to the current results [3, 4]. The Veronesi report gave the level of numbness or paresthesia for the affected side for the ALND group at 68% at 12 months of follow-up and 1% for the SLND group [6]. These values represent substantially more favorable results for the SLND group on the OR scale of 0.005. As was the case for objectively assessed shoulder abduction and arm volume differences, self-reports of tingling and numbness favor the SLND group although both groups include women with self-reports of symptoms at 36 months of follow-up.

Residual Morbidity

The results of the B-32 trial indicate that the SLND group show a consistent advantage over the ALND group with statistically lower incidence rates for all four residual morbidity categories although residual morbidity remains a problem. Logistic regression models for each of the four residual morbidity categories also indicated that each model had somewhat distinct covariate predictors although residual arm tingling appeared to only be related to treatment group.

Residual shoulder abduction morbidity was most highly related to the ALND treatment group followed by the receipt of systematic chemotherapy and the receipt of radiation therapy to the axilla. Thus, the OR of having residual shoulder abduction deficits increases 1.8 fold with the receipt of both systemic chemotherapy and radiation therapy to the axilla compared to women who did not have either. While age was inversely associated with tumor size and positively associated with receipt of chemotherapy, and tumor size was positively associated with receipt of chemotherapy (all p < 0.001), the inclusion of chemotherapy suppressed the impact of both age and tumor size in this model. In addition, while chemotherapy and radiation therapy were not associated directly (p = 0.227), both appear to contribute independently to the model although with opposite effects.

A recent meta-analysis identified mastectomy as compared to lumpectomy, axillary dissection compared to sentinel node dissection, receipt of radiation therapy, and identification of positive nodes as risk factors for arm lymphedema [14]. Our results clearly note the advantage of SLND compared to ALND. In addition, the logistic regression model for residual arm volume differences also indicate that older age, treatment of the dominant side, and receipt of radiation therapy to the axilla are related to the occurrence of residual arm volume differences. Thus, women in either treatment group with these three additional states would increase their OR for having residual arm volume differences by a factor of 8.7 compared to women without any of these conditions. As was the case for residual arm abduction morbidity, age was inversely associated with tumor size and positively associated with chemotherapy, and tumor size was positively associated with chemotherapy (all p < 0.001). It appears that age suppressed the impact of chemotherapy and tumor size while concordance of the dominant hand with affected arm and radiation therapy contributed independent effects in this logistic model.

The residual arm numbness model also indicated the superiority of the SLND group. In addition, both age (≥ 50) and larger tumor sizes (T2, T3) decreased the OR of having residual numbness. Younger women (≤ 49) tended to have larger tumor sizes (p < 0.001) and larger tumor sizes were associated with having a mastectomy in contrast to lumpectomy (p < 0.001). However, radiation therapy to the axilla was not associated with type of surgery (p = 0.664) nor receipt of chemotherapy (p = 0.174). It thus appears that age and tumor size subsumed any impact that chemotherapy and radiation therapy and type of surgery might have contributed for this model.

CONCLUSIONS

The B-32 is the largest randomized clinical trial comparing ALND and SLND surgical treatment approaches with the longest follow-up duration that documents the prevalence of longitudinal morbidity and the incidence of residual morbidity at the end of the planned follow-up time periods. The trial results show a consistent advantage favoring SLND over ALND at each follow-up time point relative to objectively assessed shoulder abduction deficits using a goniometer, arm volume differences using water immersion, and subjective reports of arm tingling and numbness. These results are generally consistent with earlier randomized trial results. The incidence of residual morbidity among the SLND patients is lower than that observed for the ALND patients although the SLND patients still appear to suffer significant residual morbidity. In particular, factors related to the incidence of residual shoulder abduction for both SLND and ALND patients included radiation therapy to the axilla and not receiving systemic chemotherapy. Factors related to the incidence of residual arm volume morbidity included older age, coincidence of the affected arm being the dominant arm, and receipt of radiation therapy to the axilla. Factors related to the incidence of residual arm numbness morbidity included younger age and smaller tumor size. The current observations regarding the incidence of residual morbidity and their associated risk factors have potential implications for clinician-patient interactions that may need to address issues related to the true long term surgical risk of either SLND or ALND, the treatment of chronic pain, the need to provide patient education about the importance of post surgical treatment physical therapy, and the potential for arm lymphedema with patient education efforts targeting the prevention of complications due to lymphedema.

ACKNOWLEDGEMENTS

Elaine Elliott, Barbara Branch, and Sarah Howe for assistance in performing this study. This work was supported by the National Cancer Institute, Department of Health and Human Services, Public Health Service Grants [P30 CA022435, 5RO1CA74137-9, U10-CA-12027, U10-CA-69651, U10-CA-37377, and U10-CA-69974]

ABREVIATIONS LIST


ALMANAC	Axillary Lymphatic Mapping Against Nodal Axillary Clearance

GIVOM	Gruppo Interdisciplinare Veneto di Oncologia Mam-maria

SNAC	Sentinel Node Biopsy Versus Axillary Clearance

REFERENCES

1. Gill G. Sentinel-Lymph-Node-Based Management or Routine Axillary Clearance? One-Year Outcomes of Sentinel Node Biopsy Versus Axillary Clearance (SNAC): A Randomized Controlled Surgical Trial. Ann Surg Oncol. 2009 Feb;16(2):266–75. [PubMed]

2. Del Bianco P, Zavagno G, Burelli P, et al. Morbidity comparison of sentinel lymph node biopsy versus conventional axillary lymph node dissection for breast cancer patients: results of the sentinella-GIVOM Italian randomised clinical trial. Eur J Surg Oncol. 2008 May;34(5):508–13. [PubMed]

3. Mansel RE, Fallowfield L, Kissin M, et al. Randomized multicenter trial of sentinel node biopsy versus standard axillary treatment in operable breast cancer: the ALMANAC Trial. J Natl Cancer Inst. 2006 May 3;98(9):599–609. [PubMed]

4. Fleissig A, Fallowfield LJ, Langridge CI, et al. Post-operative arm morbidity and quality of life. Results of the ALMANAC randomised trial comparing sentinel node biopsy with standard axillary treatment in the management of patients with early breast cancer. Breast Cancer Res Treat. 2006 Feb;95(3):279–93. [PubMed]

5. Purushotham AD, Upponi S, Klevesath MB, et al. Morbidity after sentinel lymph node biopsy in primary breast cancer: results from a randomized controlled trial. J Clin Oncol. 2005 Jul 1;23(19):4312–21. [PubMed]

6. Veronesi U, Paganelli G, Viale G, et al. A randomized comparison of sentinel-node biopsy with routine axillary dissection in breast cancer. N Engl J Med. 2003 Aug 7;349(6):546–53. [PubMed]

7. Krag DN, Julian TB, Harlow SP, et al. NSABP-32: Phase III, randomized trial comparing axillary resection with sentinal lymph node dissection: a description of the trial. Ann Surg Oncol. 2004 Mar;11(3 Suppl):208S–10S. [PubMed]

8. Krag DN, Anderson SJ, Julian TB, et al. Technical outcomes of sentinel-lymph-node resection and conventional axillary-lymph-node dissection in patients with clinically node-negative breast cancer: results from the NSABP B-32 randomised phase III trial. Lancet Oncol. 2007 Oct;8(10):881–8. [PubMed]

9. SAS (Ver 9 ) SAS Institute Inc., Cary North Carolina 27511

10. SYSTAT (Ver 11) SYSTAT Software Inc., Richmond CA 94804

11. StatXact (Ver 4) Cytel Software Corporation, Cambridge MA 02139

12. Luttgens K, Hamilton N. Kinesiology: Scientific Basis of Human Motion. 9th Ed. Brown & Benchmark; Madison, WI: 1997.

13. Kligman L, Wong R, Johnston M, et al. The Treatment of Lymphedema Related to Breast Cancer-Evidence Summary Report #13-1. Cancer Care Ontario. 2003 August 22;May 22; http://www.cancercare.on.ca/pdf/pebces13-1f.pdf. 2009.

14. Tsai RJ, Dennis LK, Lynch CF, et al. The Risk of Developing Arm Lymphedema Among Breast Cancer Survivors: A Meta-Analysis of Treatment Factors. Ann Surg Oncol. 2009 April; DOI 10.1245/s10434-009-0452-2. [PubMed]