Changes in mastectomy rates at a Brazilian public hospital over 20 years (1989 to 2008)

ABSTRACT CONTEXT AND OBJECTIVE: Recently, breast-conserving surgery (BCS) has been replacing mastectomy for breast cancer treatment. The aim of this study was to evaluate the changes in mastectomy and BCS rates and the factors relating to these shifts. DESIGN AND SETTING: A retrospective study in a Brazilian public hospital. METHODS: Pathological records from female patients who underwent surgery for breast cancer at Hospital das Clínicas, Universidade Federal de Minas Gerais (HC-UFMG), between 1989 and 2008 were reviewed. The mastectomy and BCS rates were calculated. The chi-square test was used to assess factors associated with type of surgical treatment and to compare trends in treatment type over the years. Logistic regression was used for multivariate analysis. RESULTS: From 1989 to 2008, 2050 breast cancer surgical specimens were received in our service, corresponding to 1973 patients; 1324 (64.6%) of them were from mastectomy and 726 (35.4%) from BCS. A shift from mastectomy towards BCS was observed (P < 0.001). In multivariate analysis, earlier year of surgery (P < 0.001), larger tumor size (P < 0.001), having at least one positive axillary lymph node (P < 0.001) and patients’ age greater than 68 years (P = 0.007) were predictors of mastectomy. CONCLUSIONS: There was a shift from mastectomy towards BCS in our institution over the years. This may reflect consolidation of BCS (plus radiotherapy) as an equivalent treatment to mastectomy in terms of survival and a shift to earlier diagnosis for the disease.


INTRODUCTION
Breast cancer is the most common cancer in women worldwide, and also the most common cause of cancer deaths in this group. 16][7][8] In the 1970s, Fisher's group in the United States and Veronesi's group in Italy started comparing the prognoses of patients who had undergone breast-conserving surgery (BCS; wide local excision known as lumpectomy in the United States and quadrantectomy in Italy) and mastectomy for early-stage breast cancer.[11][12][13][14] This led the United States National Institutes of Health to publish a consensus in 1991 stating that BCS could be used when possible. 157][18][19][20][21][22][23][24][25] We have not found any studies in major databases (PubMed or SciELO) comparing the rates of different types of surgery for breast cancer in Brazilian public institutions.
Many factors are known to influence the decision regarding BCS versus mastectomy.7][28][29][30] Other factors are the surgeon's preference, histopathological tumor type, positivity of axillary lymph nodes, healthcare availability, findings from imaging studies and genetic abnormalities. 17,19,31,324][35] The Clinical Hospital of the Federal University of Minas Gerais (Hospital das Clínicas, Universidade Federal de Minas Gerais, HC-UFMG), Belo Horizonte, Brazil, is a public general hospital and a reference center for breast cancer treatment.

OBJECTIVE
The aim of this study was to evaluate the changes in mastectomy and BCS rates at HC-UFMG from 1989 to 2008, and the factors relating to these shifts.

METHODS
A retrospective study comparing BCS and mastectomy rates at Age cutoff points were defined by means of the CHAID (Chisquare Automatic Interaction Detector) algorithm using the type of surgery as the response variable. 37logistic regression model with the type of surgery as the response variable was used to identify factors associated with mastectomy.Starting with the univariate analysis, variables with a statistical significance of less than 0.2 (P < 0.20) were included in the multiple logistic regression model.The stepwise backward algorithm was used to select the variables with a P value under 0.05. 38Estimates of odds ratios (OR) and the corresponding 95% confidence interval (95% CI) were obtained from univariate and multivariate analyses. 39Since some patients had bilateral tumors, the analysis was repeated after excluding these cases in order to confirm the adequacy of the model.The conclusions did not change through their removal, so these cases were kept.The goodness of fit of the regression model was assessed using the Hosmer and Lemeshow test.

RESULTS
From 1989 to 2008, 1973 female patients underwent breast surgery, with records available in the Breast Pathology Laboratory of HC-UFMG.The patients' ages ranged from 19 to 97 years (mean = 55.13 years; standard deviation, SD = 13.9 years).A total of 2050 breast cancer specimens were evaluated.One thousand three hundred and twenty-four (64.6%) of them were from mastectomies, and 726 (35.4%) were BCS specimens.Table 1 shows the distribution of frequencies according to the type of surgical procedure, history of contralateral breast cancer, laterality, pathological tumor size (pT), status of regional lymph nodes (pN), histopathology and age range.
The total number of surgeries increased over the years, from A history of contralateral breast cancer (synchronous or metachronous), which occurred in 133 cases (6.5%), was not a predictor of mastectomy (P = 0.358) in univariate analysis (Tables 1 and 3).    3 and 4).Having at least one positive axillary lymph node was also a predictor of mastectomy (P < 0.001 in both of them).The larger the tumor size was, the higher the odds were of undergoing mastectomy.Pure forms of in situ tumors were more frequently diagnosed in later years.From 1989 to 1993, the proportion of patients with in situ tumors was 7% (14 cases), and from 2005 to 2008 that proportion rose to 14.5% (87 cases, P < 0.001; Figure 2).12][13]15 In 2004, the Brazilian National Cancer Institute published a consensus stating that BCS could be used for tumors smaller than three centimeters. 40[18] In a significant percentage of cases, we were unable to estimate the tumor size (20.8%) because multiple specimens were derived from the same tumor (incisional biopsies) or because some procedures had been performed in other institutions.We were also unable to estimate lymph node status in 16.4% of the patients, because they had not undergone surgical axillary staging or had undergone this procedure in another institution.Even though this could have affected our results, tumor size and lymph node status were strong independent predictors of the type of surgical treatment.
33][34][35]41 In our study, as in many other studies, older age was found to be a predictor of mastectomy.This differs from other studies, like those by Katipamula et al. 20 (younger age was found to be a risk factor for mastectomy), McGuire et al. 23 and Adkinsson et al. 25 On the other hand, Reitsamer et al. 32 showed that there was no correlation between age and type of surgery, and Zorzi et al. 18 and Chagpar et al. 42 showed higher mastectomy rates in older women.This may be explainable by different perception of breast importance in different cultural settings.It may also be attributable to our methodology, which selected only the last procedure performed on the patient.Therefore, the older woman may in fact have undergone BCS but might have had a recurrence later on and needed a mastectomy.Sixty-seven patients (5.1% of the mastectomies) first underwent breast conserving surgery and then mastectomy in a second intervention.Some authors have also reported that adjuvant therapy (such as chemotherapy) is less prescribed to older women, especially with a poor health status, 43,44 and thus a more aggressive local therapy may have been chosen.It has been noticed that younger women have higher recurrence rates after BCS. 45,46nce the source of our study was pathological records, patients' choices were not addressed although they could have influenced mastectomy rates. 22,28,35,47me authors have concluded that, even though the public health system pays for mammograms, the adherence of women to screening mammograms is still low in Brazil, especially among less privileged women.Lima-Costa and Matos 3 showed that, in 2003, only 43% of women between the ages of 50 and 69 years (the age recommended for screening mammograms according to the Brazilian National Cancer Institute 40 ) had had a mammogram within the previous two years.Marchi and Gurgel 4 showed that a higher percentage had had a first mammogram (68%), but that less than 50% of the patients were adherent to biannual screening.Moreover, women within the public system were less likely to comply with subsequent screening. 4This could be responsible for our high percentage of patients with locally advanced tumors (11.9% infiltrating the skin or chest wall).[19] It may be possible that women with more advanced tumors are more frequently referred to our hospital, and therefore we have a higher than expected percentage of large tumors.On the other hand, even though still low, we are experiencing an increase in breast cancer screening, which has now become a focus of public health policies.SISMAMA (Information System for Breast Cancer Control), implemented by the Brazilian National Cancer Institute in 2009, 48 is a tool for following up patients with abnormal breast examinations (imaging or pathology), and is also being used to direct funding for breast cancer screening.
In our study, IDC was associated with higher chances of mastectomy in univariate analyses than were DCIS, LCIS and special-type carcinomas.In multivariate analyses, the association was not significant.This differs from some studies, like those of  Katipamula et al., 20 Chagpar et al. 42 and Lee et al., 30 which showed higher mastectomy rates among patients with ILC.
0][21] However, no Brazilian data from either public or private hospitals concerning the type of surgery performed is available in the indexed literature.Through the present study, we have been able to conclude that in our hospital, which provides public care for breast cancer, this shift did not occur until recently.30][31][32][33][34][35]47 In our study, one of these issues is the increasing frequency of early breast cancer at the time of surgery, which is a necessary condition for performing BCS, and is associated with better patient outcomes. 36

CONCLUSIONS
There was a clear shift from mastectomies towards BCS for breast cancer treatment at HC-UFMG over the 20 years of this study.This may reflect not only consolidation of BCS plus radiotherapy as an equivalent treatment to mastectomy in terms of overall survival, but also a shift to less locally advanced tumors at diagnosis, due to a higher percentage of women undergoing screening mammograms.

Figure 1 .
Figure 1.Percentage of mastectomies over time.

Table 2 .
Number (%) of specimens of breast-conserving surgery (BCS) and mastectomy operated between 1989 and 2008 year of treatment, larger tumor size (pT) and positive regional lymph node status (pN).Histopathology was not significant in the model.The results from the Hosmer and Lemeshow test showed that the model fit was good (P = 0.251).
DISCUSSIONOur study evaluated a series of women who were treated for breast cancer at a Brazilian public general hospital in the state of Minas Gerais that is also a reference center for breast cancer IDC = invasive ductal carcinoma; DCIS = ductal carcinoma in situ; ILC = invasive lobular carcinoma; LCIS = lobular carcinoma in situ.

Table 3 .
Univariate analyses on factors associated with breast-conserving surgery and mastectomy

Table 4 .
Multivariate analysis on predictors of mastectomy versus breast-conserving surgery Figure 2. Proportions of pathological tumor sizes over time.