15-year Follow-up of Comparing Mastoscopic and Conventional Axillary Dissection in Breast Cancer: A Multicentres, Randomized Controlled Trial

Backgroud: Longer follow-up was necessary to testify the exact value of mastoscopic axillary lymph node dissection (MALND). Methods (cid:0) From January 1, 2003 to December 31, 2005,1027 patients with operable breast cancer were randomly assigned to two groups: MALND and CALND. 996 eligible patients were enrolled. The end points are disease free survival and overall survival. Results (cid:0) The nal cohort of 996 patients was followed for an average of 184 months. The distribution of all events was fairly similar between two groups of patients. The incidence of local in-breast events did not differ in a signicant manner between two cohorts. Similarly, the rate of distant metastases was not signicantly different with 30.0% in MLND and 32.6% in CALND. And no signicant difference was observed in other primary tumor between two groups (p=0.46). Patients who remain alive with no event comprise a total of 37.2% in MALND and 35.4% in CALND. Other primary cancers and deaths from other causes were distributed equally between two groups. The 15-year disease-free survival rates were41.1 percent for the MALND group and 39.6 percent for the CALND group (p=0.79). MALND was found to be not inferior for overall survival (P =0.54). The 15-year overall survival rates were 49.5 percentafter MALND and 51.2 percentafter CALND (p=0.86). Probability of overall survival was not signicantly different between two groups. Conclusions (cid:0) MALND does not increase unfavorable events, and also does not affect the long-term survival of patients. Therefore, MALND should be one of the preferred approaches for breast cancer surgery.

One of signi cant advances on clinical surgery in late 20th century is a gradual maturation of theory on minimally invasive surgery and rapid development of endoscopic operation. The prevailing applications of endoscopic techniques in various operations alleviate therapy outcome obviously. Their successful practice and maturation provide a referential theory and technology drawn lessons for breast surgery development and new thought in solving therapeutic dilemma of breast surgery. Endoscopy breast surgery has been developed in nearly all breast operations for about 30 years, which has been producing good clinical results. 3,4 It is a typical operation model in combined minimally invasive technique with oncoplastic surgery. 5,6 In 2003 we initiated a multicenter randomized trial in breast cancer patients comparing mastoscopic axillary lymph node dissection (MALND) to CALND. The 5-year ndings demonstrated patients underwent MALND had less axillary pain, numbness or paresthesia and arm swelling. Aesthetic appearance of axilla was much better. There was a signi cant difference between two groups in distant metastasis. It also showed no signi cant differences in disease free survival and overall survival between two groups. 7 We now present information through an average of 15 years of post-randomization follow-up.

Methods
Between January 1, 2003 through December 31, 2005, 1027 patients with operable breast cancer at 15 institutions were randomly assigned to 1 of 2 study groups: MALND and CALND. Patients younger than 80 years with operable breast cancer were eligible for enrollment. Randomization was allocated and controlled centrally using sealed opaque envelopes containing computer-generated random numbers.
The randomized data set was recorded and maintained until all study data collection was complete. Informed consent was obtained from participating patients. The protocol was approved by institutional review board committees of all institutions in compliance with the Declaration of Helsinki and guidelines for good clinical practice. 7 To be included in study, patients had to have no axillary lymph node disease on preoperative clinical examination or lymph nodes less than 1 cm by color ultrasonic inspection.
Exclusion criteria included a previous history of another cancer, preexisting limb disease causing swelling, or previous surgery in the ipsilateral axilla. Of other 240 patients, 161 were deemed ineligible, 79 were not randomly assigned to a study group. Of 1027 randomized patients, 31 were not able to be evaluated.
The ndings below are from the 996 eligible patients, who were enrolled in study for an average of 15years. Detailed descriptions of patient-entry information, eligibility and ineligibility criteria, investigation plan, operative and irradiation procedures, and other aspects of the study have been presented previously. 10 Each surgeon had performed at least 20 MALNDs. MALND operation was performed according to the method reported previously. [1][2][3]5,7 All surgeons used the same standardized technique.
Surgeons submitted a MALND video to an external monitoring committee for assessing their surgical technique, including main operative stages and the identi cation and dissection of critical adjacent structures.
Either type of ALND consisted of a level I to II and/or III (if the swollen node with suspected metastasis was found in level II) axillary dissection. Contemporary surgical treatment includes a modi ed radical mastectomy or breast-conserving treatment. In the CALND group, breast surgery and ALND were performed by traditional open method. In MALND group, the patients rst underwent MALND and then breast surgery. Patients were treated with systemic therapy and radiation therapy according to the same standard institutional protocols (National Comprehensive Cancer Network Breast Cancer Practice Guideline).

Outcome Evaluation
The end points for overall treatment comparisons were disease-free survival and overall survival. Times to those end points were calculated from the date of operations. Recurrences of tumor in the chest wall and surgical scar were classi ed as local treatment failures. Recurrences in supraclavicular, subclavicular, in the ipsilateral axilla of patients treated with either CALND or MALND in all patients were considered to be regional treatment failures. Events considered in the determination of disease-free survival were rst local, regional, or distant recurrence of tumor; contralateral breast cancer or a second primary tumor other than a tumor in breast; and death who had no evidence of cancer. Events considered in the estimation of relapse-free survival were the rst local, regional, or distant recurrence or an event in contralateral breast that was judged to be a recurrence. Distant recurrences that occurred either as the rst recurrence or after a local or regional recurrence, contralateral breast cancers, and other second primary cancers were considered in estimation of distant-disease-free survival. Overall survival refers to survival with or without recurrence of disease. The cumulative incidence of death was estimated after a recurrence or a diagnosis of contralateral disease and the cumulative incidence of death without a recurrence or a diagnosis of contralateral disease.

Statistical Analysis
Statistical analysis was performed for the parameters in end points based on an intent-to-treat population. All tests were conducted at the two-sided 5% level of significance for treatment effects. Analyses of end results were performed separately for patients in two groups.
For time to event endpoints, Kaplan-Meier plots were presented and log-rank tests were used to test the null hypothesis of no MALND effect. End points were compared between groups using two-sample t test for continuous outcome variables and chi-square test for categorical variables. The analysis was performed with missing data left missing. Estimates are provided with their standard errors. Continuous variables were summarized using mean ± SD and categorical variables were summarized using n (%).All calculations were performed using SPSS software package (version 19.0).

Results
The distribution of patients for 996 patients nally evaluable between MALND group and CALND group according to age, histologic, tumor grade, estrogen-receptor, proliferation rate and peritumoral vascular invasion was similar, reported previously. 7 The nal cohort of 996 patients was followed for an average of 184 months (range,164 to 191 months) during which time. The patients' outcomes at 15 years are summarized in Table 1. The distribution of all events was fairly similar between two groups of patients. The incidence of local in-breast events, including axillary relapse, supraclavicular metastasis, recurrence in ipsilateral breast and cancer in contralateral breast did not differ in a signi cant manner between two cohorts and was 11.1% (55/496) in MALND and 11.6% (58/500) in CALND cohorts, respectively. Similarly, the rate of distant metastases was not signi cantly different with 30.0% (149/496) in MALND cohort and 32.6% (163/500) in CALND cohort, respectively. And no signi cant difference was observed in other primary tumor between MALND group (0.4%, 20/496), compared with CALND (0.5%, 25/500).
All patients who recurred were treated with local therapy (surgery, radiotherapy, or combination of the two) and were given systemic therapy (hormonal therapy, chemotherapy). Patients who remain alive with no event comprise a total of 185/496 (37.2%) in MALND and 177/500 (35.4%) in CALND, respectively. On the other hand, most of the treatment failures were distant metastasis (31.3%, 312/996). Other primary cancers and deaths from other causes were distributed equally between two groups (55/496, 11.1% in MALND;55/496, 11.0% in CALND).
Overall Survival At a median follow-up of 15 years, there were 634 deaths (311 in MALND group and 323 in CALND group). Compared with CALND, MLND was found to be not inferior for overall survival (P = .54, by the log-rank test). The 15-year overall survival rates were 49.5 percent (95% CI, 83.4%-90.2%) after MALND and 51.2 percent (95% CI, 81.2%-89.7%) after CALND (HR 1.00, p= .86, by log rank test) ( Figure  2). The hazard ratio for death among the patients in MALND, as compared with those in CALND, was 1.04 (95 percent con dence interval, 0.92 to 1.31; P= .58). Probability of overall survival was not signi cantly different between two groups.

Discussion
With the defects of Halsted's radical mastectomy and further extensive radical surgery, we cannot but turn back to modi ed radical surgery and breast conserving surgery and SLNB later. 3,6,8,[10][11][12] The development of this road has burned a series of painful imprint step by step in progress to minimally invasive and functional treatment.
MALND ameliorates four key techniques below. 9-12 altering operation steps. No matter whether breast conserving surgery or radical mastectomy or modi ed radical mastectomy, axillary lymph node is rst dissected and breast operation is then done. This blocks the routine of metastases via blood and lymph while breast operation. It further conforms to the operation principle of tumor surgery. altering operation method. In MALND the axilla is aired after lipolysis, the parenchymatous axilla becomes the reticular structure just like the spider web. The dissection is easily carried out by endoscopy. This reduces operation complexity and simpli es operation process. altering the view angle of operation. When patients are placed supine in operation, surgeons' view is horizontal from lateral to medial. The endoscopy in MALND can reach into the narrow and irregular axillary space in any angle, which makes the dissection easier than CALND. While in CALND, in order to minimize the size of the axillary incision as possible, the view is small and narrow. Operators should bend waist, hunch back and skew neck to try to see the inside of axillary, and dig into step by step. Operative di culty and risk increase. magnifying the local view of operation. The structures in axilla can be magni ed to 4-7 times by the help of magnifying function of endoscopic system, which are identi ed clearly and preserved further. It achieves minimal invasion of operation and reduces complication and functional damage.
In China most patients with breast cancer are belonged to middle-late stages. So ALND isn ' t abandoned now. Since Suzanne et al rst reported MALND in 1993. Several medical centers have veri ed its feasibility and safety using same method. [11][12][13][14][15][16] The lymph nodes number harvested, postoperative symptoms, drainage time and drainage liquid in MALND compared to those in CALND are no signi cantly differences. But long-term complications such as upper limb dysfunction, serious pain, edema and activity complications reduced obviously in MALND. It reaches three-fold advantages of minimal invasion, and function and cosmetic. It also gains two aims in physiologically and psychologically minimal invasion. Moreover, some authors also have done a series of work on mastoscopic SLNB, singleport, 3D and robotic MALND. 13,[14][15][16][17][18][19][20] They all stated that this technique could be a promising alternative to conventional methods.
Keshtgar et al 22 (2009) made an editorial that endoscopic breast surgery was associated with minimal scarring and postoperative pain, and it appeared that wound complications were rare events. They suggested there was a need for randomized, controlled trials to provide evidence for its safety and efficacy. Leff DR et al 4 published a long paper in Breast Cancer Res Treat in 2011. Initial results had demonstrated that endoscopic breast surgery was safe and technically feasible. Twelve ran domized controlled studies were nally included involving 1983 patients. It showed that intraoperative blood loss postoperative drainage hospital stay and postoperative complications were less in MALND. There was no signi cant difference in operation time lymph node numbers dissected hospitalization costs postoperative recurrence and metastasis rates between two groups. This study concluded that MALND can reduce trauma and postoperative complications without affecting the prognosis of patients.
Chi WM et al 5 reviewed current literatures in 2019 on outcomes, techniques and trend of endoscopicassisted breast surgery over a 20 years period. It was comparable in terms of oncological, surgical as well as aesthetic outcomes compared to conventional techniques. They considered that standardization of techniques, practice guidelines and objective outcome assessment methods might pave the way for better conduct and place it as one of the standards of care for breast cancer. Unfortunately, the cases enrolled in above studies were relative small, single center, and therefore lack a su cient testimony of application value of MALND. So, a randomized multicenter trial on a larger series of patients is a burning issue.
Actually, we rst reported the mature experiences of MALND in 86 patients in China in 2003. The technique becomes mature and standardized gradually in ongoing exploration. We summed up a "ninestep" procedure and standardized the technique too in 315 patients in 2005. 2 We also reported the experiences on a larger series of 522 patients. 11 We have accomplished over 20,000 cases in nearly 20 years. Above 150 hospitals have accumulated beyond 100 cases per hospital in our country. From 2003 to 2005, a large of randomized prospective study in 15 hospitals in China over 1200 patients was studied for comparing MALND with CALND, with an average follow-up of 63 months in 2012. 10 The results had demonstrated that bleeding was signi cantly fewer in MALND. The operating time, lymph node numbers harvested, drainage uid, drainage time and axillary seroma were similar between two groups. The axillary pain, numbness or paralysis and arm swelling in MALND were much less than those in CALND. The cosmetic status of axilla in MALND was also better than that in CALND. On the average 5-year follow-up period, the axillary recurred totally in 11 cases, similar in two groups. There was no portimplantation in MALND. The patients with MALND had similar disease-free survival and overall-survival to that with CALND. And another important nding was significantly lower rate of distant metastases in MALND. As we know, most of death in breast cancer comes from distant-metastasis. The findings of our study raise the question of whether the signi cant difference on survival rates can be found in a longerterm follow-up (i.e., 10, 15 or 20 years). This is also another aim of the present study.
In our current trial of 1027 randomized patients, 31 were not able to be evaluated. The distribution of the patients for 996 patients nally evaluable between the MALND group and the CALND group was almost equal. They were followed for an average of 184 months. The ndings at 15 years con rm and extend our earlier results (5-year follow-up).The incidence of local in-breast events, including axillary relapse, supraclavicular metastasis, recurrence in ipsilateral breast and cancer in contralateral breast has no signi cant difference between two cohorts ( 11.1% in MALND and 11.6% in CALND). And no signi cant difference was observed in other primary tumor between MALND group (0.4%) and CALND group (0.5%).
The distribution of all events was fairly similar between two groups. All patients recurred were treated with local therapy (surgery, radiotherapy, or combination of the two) and were given systemic therapy (hormonal therapy, chemotherapy). Patients alive with no event have no signi cant difference between two groups, 37.2% in MALND and 35.4% in CALND groups respectively. On the other hand, most of treatment failures were distant metastasis (31.3%). Other primary cancers and deaths from other causes were distributed equally between two groups (11.1% in MALND;11.0% in CALND). These coincide with our previous results.
However, at a median follow-up of 15 years, the rate of distant metastases was not signi cantly different between 30.0% in MALND and 32.6% in CALND, respectively. It revises the result that distant metastases in MALND were obviously low than that in CALND (P=.04) at 5-year follow-up. 7 634 patients have died, 322 and 311 in CALND and MALND respectively (p 0.05). No significant difference in 15-year diseasefree survival rates is observed between two groups. The hazard ratio for an event among patients underwent CALND, as compared with those underwent MALND, was 1.22 (P=0.67). Compared with CALND, MALND was found to be not inferior for overall survival (P = .54). The 15-year overall survival rates were 49.5 percent after MALND and 51.2 percent after CALND (p=0.86). The hazard ratio for death among patients in MALND, as compared with those in CALND, was 1.04 (P=0.58). Probability of overall survival was not signi cantly different between two groups.
This nding continues to indicate that no signi cant difference in disease-free and overall survival among patients between two groups. Namely, the application of MALND at least has not produced any disadvantage to patients. Nevertheless, the immense bene ts brought by MALND have been appearing in recent years. And the achievement of these bene ts does not sacri ce any survival chance of patients.
The ndings presented now can dispel previous misgivings on MALND from some specialists. For example, whether the liposuction affects pathologic characteristics of lymph node and interferes with pathoanatomic study of the lymph node, whether it increases the risk of exfoliation and implantation of tumor cells from the lymph node with metastasis, and whether it raises the risk of hematogenous dissemination or metastasis.

Conclusion
This report provides rm evidence from a randomized controlled study of nearly 1000 patients followed for median 15 years that the use of MALND in breast cancer surgery does not increase the axillary relapse recurrences metastases and any other unfavorable events, and also does not affect long-term survival of patients. Therefore, MALND should be one of the preferred approaches for breast cancer surgery. Abbreviations CALND = Conventional axillary lymph node dissection Page 10/16 MALND = Mastoscopic axillary lymph node dissection.

Declarations
Ethics approval and consent to participate Informed consent was obtained from participating patients. The protocol was approved by institutional review board committees compliance with the Declaration of Helsinki and guidelines for good clinical in practice.
Consent for publication All authors are consent for publication.
Availability of data and material: not applicable.
Competing interests: There is no con ict of interest that could be perceived as prejudicing the impartiality of the research reported, and no any nancial or other potential con ict of interest.