New criteria for breast symmetry evaluation after breast conserving surgery for cancer

ABSTRACT Objective: to evaluate symmetry after breast-conserving surgery (BCS) for cancer. Methods: a prospective study of patients undergoing BCS. These patients were photographed using the same criteria of evaluation. The references points used were the nipple height difference (NH), the nipple-manubrium distances (NM), nipple-sternum distances (NS) and the angle between the intramammary fold and the nipple (nipple angle; NA). ImageJ software was used. Three breast symmetry models were evaluated: excellent/others (model 1), excellent-good/others (model 2) and others/poor (model 3). The ROC curve was used to select acceptable criteria for the evaluation of symmetry. Decision tree model analysis was performed. Results: a total of 274 women were evaluated. The BCCT.core result was excellent in 5.8% (16), good in 24.1% (66), fair in 46.4% (127) and poor in 23.7% (65). The difference in NH was associated with good breast area (0.837-0.846); acceptable differences were below 3.1 cm, while unacceptable values were greater than 6.4 cm. Differences in the NM were associated with average breast area (0.709-0.789); a difference in value of less than 4.5 cm was acceptable, while values greater than 6.3 cm were unacceptable. In the decision tree combined model, a good-excellent outcome for patients with differential (d) dNH = 1 (0 to 5.30 cm) and dNM ≠ 3 (<6.28 cm); and for a poor/poor result, values dNM = 3 (> 6.35). Conclusions: the results presented here are simple tools that can assist the surgeon for breast symmetry evaluation.


INTRODUCTION
B reast cancer is associated with half of all cancer cases and 38% of cancer-related deaths in developed countries 1 . It is estimated that over 1.7 million new breast cancer cases are diagnosed annually worldwide.
Breast-conserving surgery (BCS) 2,3 combined with radiotherapy 2-4 is considered safe. Overall, 57% of women diagnosed in the early stages and 13% of those diagnosed in the late stages of the disease undergo breast-conserving treatment, and most undergo radiotherapy 5 .
For patients who undergo BCS, cosmesis is considered excellent or good in 76.3% and 47% of cases, respectively 6 . The main factors associated with asymmetry are age, higher body mass index and large tumour size 7 . Many patients undergo further breast surgery due to asymmetry, and after the second procedure, 94.5% and 88.8% of patients are satisfied after 1 and 5 years, respectively. However, a second and a third operation are required in 19.1% and 6.4% of cases, respectively 8 .
Women are generally dissatisfied with their breasts, with 42.7% reporting being displeased 9 , and 30% of women who undergo BCS are not satisfied with the aesthetic results 10 . Although there are some parameters for healthy breasts 9 , in cosmetic surgical skin marking, other particular reference points and distances are considered appropriate 11 .
Cosmetic evaluation is very subjective, and inter-examiner correlation is poor. Inter-examiner variation can be minimized after a consensus is reached After providing informed consent, the selected patients were taken to a special room containing a background symmetrograph, where points were marked on the sternal manubrium and 20 cm inferiorly.
The women were photographed from a distance of 1 metre using a Cyber-Shot DSC-H300 camera with an 8-megapixel resolution. Photographs were obtained bilaterally in an anteroposterior, lateral direction, until the mid-axillary line could be seen, as this is associated with the evaluation of the areolar angle and the pencil drop angle (PDA) 9 . While analysing the photographs, patients whose images were not suitable for evaluation using BCCT.core were excluded, along with patients who underwent central BCS without areolar reconstruction.
BCCT.core was used for the cosmetic evaluation 14,19 . The BCCT.core program performs automatic calculations of different ratios/asymmetries, including the breast volume, skin colour and scarring.
The results are given with a 4-point scale (1-excellent, 2-good, 3-fair, and 4-poor) 15,19 . These parameters were used as the standard ( Figure 1). The same images were also evaluated with ImageJ software, which was used to evaluate the following distances after calibrating the equipment with known distances: the nipple-manubrium (NM) distance, the nipple angle (NA), the nipple-sternum (NS) distance and the angle of the abducting arm associated with the pencil test (PDA). These measurements were based on a previous study 9 , and for this, the differences of the distances between the nipple-manubrium (NM),  BCCT.core output and ImageJ calculations were first transferred to the IBM SPSS for Mac® program, which was used to perform differential calculations.
Subsequently, these data were exported to the MedCalc® program, where the findings related to breast symmetry were dichotomized into excellent/others (model 1), excellent-good/others (model 2) or excellent-good-fair/ poor (model 3). The receiver operating characteristic (ROC) curve was used to evaluate the sensitivity, specificity, area and difference in the cut-off point between acceptable and non-acceptable symmetry to identify simple criteria related to good breast symmetry.
From the identified criteria, we opted to analyse those that presented better results for all cut-off points and presented an increasing linearity in relation to the

RESULTS
Of the 300 patients selected for the study, 3 (1%) patient photographs had inadequate resolution, and 23 (7.7%) underwent central BCS without areolar reconstruction, which resulted in 274 patients who were eligible for inclusion in the study.
The age of the patients ranged from 25.8 to 87.5 years (mean 58.4, standard deviation (SD) 9.8), and patients had undergone breast surgery 1 to 20.2 years prior to the study (mean 6.9; SD 4.1). Overall, 50.4% of tumours were located in the right breast. Tumour  cm were considered optimal, and results worsened as the distance increased; the worst results were observed when the difference in this distance was greater than 6.4 cm. In terms of the difference in the NM distance, the results were average, in relation to models 2 and 3, and better results were observed when the difference in this distance was less than 5.0 cm. Worse results were also associated with an increase in this distance, and poorer results were observed with a differential distance greater than 6.3 cm (Table 1).
When the conformational breast data were evaluated and when the contralateral breast served as a reference, a good parameter in relation to the PDA and CNA was observed only in model 1 (excellent x others), while an average relationship was observed with respect to the CNM distance and CNA (Table 1).
In the combined decision tree model (Tables   2 and 3), the estimated decision rule suggested that a good-excellent outcome for patients with differential (d)   Oncoplastic surgery for BCS may be used to treat large tumours and can result in wide margins without concomitant increases in complication rates [22][23][24] .
It is notable that in this study, oncoplasty was performed in only 14.6% of patients, while symmetrisation was performed in only 12%. This finding suggests that the need for contralateral breast treatment to obtain symmetry should be considered. Of the 23 patients with central tumours, 16 (70%) underwent oncoplastic surgery with a plug-flap, and in the absence of the areola, the BCCT.core calculation may have been impaired, which would have resulted in a reduced incidence of oncoplasty in this study.
A second detail to consider is the long period between the initial surgical procedure and evaluation, which was 6.9 years on average. The breast shape changes over time, and weight increases accentuate differences, especially in the treated breast, because the volume increase in an irradiated breast is smaller than that in an untreated breast due to tissue fibrosis after radiotherapy.
When factors related to breast asymmetry in patients undergoing BCS are evaluated, younger age, bulky tumours 7,25 , menopausal status, tumour size, percentage of skin resected, scar orientation 25 , maximum dose of breast radiotherapy 26 , body mass and tumours located in the superomedial and inferolateral quadrant were all associated with greater asymmetry 7 . Patients with marked asymmetry are more likely to want to undergo a breast symmetrisation procedure 7 .
Patient self-evaluations tend to be better than objective findings from a cosmetic point of view 27  BCCT.core use has progressively increased 36 , universal criteria for the evaluation of symmetry are lacking. We described simple parameters that can be used. A ROC curve evaluation demonstrated that the results were acceptable when the area was greater than 0.7; these results may be excellent (area≥0.9), good (09>area≥0. 8) or average (0.8>area≥0.7). In terms of the differential values between sides, the NM and NH differences should be mentioned. In addition, an increase in distance was associated with poorer results, and both good and average area values were observed. Differential values of 3.1 cm in the NH and 4.7 cm in the NM were associated with excellent results, and these are parameters are easy to use in clinical practice. The remaining methodological differences were not satisfactory.
Regarding the breast shape, PDA values higher than 112°, an NM distance greater than 27.8 cm and an NA of less than 6.9° were associated with more ptosis of the breast, which is associated with worse outcomes.
Matthes et al. 9 sought to establish simple and easy evaluation parameters that were considered acceptable in normal women. They noted that an NM distance of less than 25 cm, a positive nipple to intramammary fold distance and a PDA less than or equal to 90° were associated with 93% patient satisfaction.
The limitations of this study include the lack of patient evaluation before treatment, the long monitoring period, the limited number of patients undergoing symmetrisation and the lack of validation calculations in another patient sample. As this was a pilot study, sample size calculations were not performed, but a convenience sample was used in an effort to identify differential values in a larger sample of patients. Additionally, potential differential values in breast asymmetry were evaluated as opposed to factors related to symmetry. In this regard, patients with excellent results tended to be younger (51.4 years of age, SD 8.5), and the results worsened with increasing age (60.5 years of age, SD 8.8 years with poor outcomes). Furthermore, a proportionately greater number of patients in this group underwent breast symmetrisation (18.8%, p=0.08).
We presented a simple method based on patient photography, which may be used to evaluate BCS results. This method is based on a combination of differences related to the NH and NM distances.
Although more studies using the same methodology are necessary, we hope that this simple method will help surgeons in clinical practice.