Abstract
PURPOSE:
We evaluated the relation between first site of recurrence of early breast cancer and disease profile at presentation and reported survival results, suggesting a personalized diagnostic imaging guidance during follow up.
METHODS:
Among 1400 early breast cancer treated from 2000 to 2010, 324 relapses were divided into 4 groups according to first site: A-locoregional, B-bone, C-Brain and D-visceral. We analyzed redictive factors of each group compared to a control group of 100 non relapsing patients and the remaining groups matched.
RESULTS:
In group A, patients were more likely to have histological tumor size above >2 cm, grade 1–2, HR positive and 0–3 involved lymph nodes. In group B, patients had more commonly grade 2–3, 1–3 positive lymph nodes and HR positive tumors. In group C, patients were more frequently young, with large tumor size, grade3, positive lymph nodes and HER2 positive tumors. In group D, patients were more likely to have tumors>2 cm in size, with nodal involvement, grade 3, HR negative and HER2 positive tumors. Annual recurrence rate in group A, was stable ranging between 15%–18%, within the first 3 years and peaked at 19.4% in the interval [1–2]year in group B. Median survival was 46 months in group A, 43 months in group B, with no significant difference.
CONCLUSION:
Outcome of loco-regional and bone relapses was good, suggesting that both systematic mammography and bone-scan/CT scan for high risk patients (N+, gradeIII) during the first 2–3 years may represent a tailored relevant follow-up protocol for breast cancer patients.
Introduction
Breast cancer (BC) is the most common female malignancy worldwide and also in Tunisia, as in other North African and Arab countries, incidence is rapidly growing [1]. In the absence of mammography screening, BC remains diagnosed at a palpable mass, with a median tumor size at diagnosis in Tunisia around 4 cm [1]. Multimodality therapy combining surgery, radiation therapy, chemotherapy, endocrine and targeted therapy led to a 5-year survival improvement reaching 61% in 2007 versus 50% in 1997 [2]. However, over one third of women will have a recurrence of their disease [3]. Site of recurrence is of clinical and biological importance and several series reported a predilection of particular BC sub-types for specific sites [4]. Recent studies, reported that BC metastasizing ability, is gene-driven, different from one organ to another [5,6]. It is suggested that tissular-biomolecular patterns of the organs may facilitate metastatic growth, with patterns similar or not to the primary BC [5–7]. Nowadays, there is no evidence to support a particular follow-up protocol to detect those recurrences. Randomized clinical trials failed to demonstrate a survival benefit from intensive follow-up schedule in all coming breast cancer patients [8–11]. However most studies were conducted without distinguishing patients according to their anatomo-clinical features or molecular subtype.
With the aim to evaluate the relation between first site of recurrence of early BC and disease profile at presentation, we analyzed clinical aspects and predictive factors of loco-regional, bone, brain and visceral relapse, as sites of first appearance, in a case control study of cohort of Tunisian patients treated with curative intend in the adjuvant setting. We also reported outcome results and we suggested a personalized diagnostic imaging guidance during follow up.
Patients and methods
We retrospectively reviewed a cohort of 1400 patients with histologically confirmed BC treated with curative intent in the period between 2000 and 2010, in medical oncology departments at “Abderhhamen Mami” Hospital and in “Clinique Taoufik” Hospital, Tunisia. At the moment of diagnosis, patients had complete workup (chest X-ray/abdominal ultrasound or CT scan and bone scan). Only patients who had curative adjuvant therapy and available data about follow up were reviewed. We identified 324 (23%) patients relapsing during follow up. Population characteristics and annual recurrence rate of this population according to IHC profile (triple negative, hormonal receptors (HR) positive and HER2 overexpressing) have been reported in an earlier publication [12]. Diagnosis of relapse was based on either radiological signs and/or on histological confirmation. Both loco-regional and distant metastases were included. Patients with second cancers, controlateral BCs and male sex were excluded.
Definition of site-specific relapse groups
The 324 relapsing patients were divided into 4 groups according to first site of appearance.
Comparisons were performed for each site specific group (A, B, C, D) with the control group and remaining site specific groups combined (ex: group A versus control group versus B + C + D…).
Specific site relapse: Outcome parameters
Time to relapse (TTR) was defined as time between first therapeutic action and the first event of recurrence. Relapse rate for each site group was calculated for each year interval of follow-up. It was defined as the fraction of followed patients who had recurrent disease in each specific site within each group in a 1-year period interval restricted to follow-up contribution of each specified time interval. Overall survival (OS) was defined as time from date of first therapeutic action to death as a result of any cause or loss to follow-up.
Statistical analysis
The chi-square Pearson test and Fisher’s exact test were used to evaluate associations between recurrence groups and clinico-pathological parameters. The Kruskal–Wallis test was used to evaluate associations with age, tumor size and time to relapse. Kaplan–Meier was used to estimate overall survival and log-rank test was used to compare the survival differences among recurrence groups. Statistical Package for Social Sciences (SPSS), version 16.0 was used to analyze the data. For all data analyses conducted, the significance level was set at 0.05.
This study was approved by the local ethical committee.
Results
Among 324 relapsing patients, we observed 18.5% locoregional recurrence (60, group A), 28.3% bone recurrence (92, group B), 7% brain recurrence (23, group C) and 46% visceral recurrences (149, group D). In group A, 12/60 patients had local recurrence (8 after breast conserving surgery and 4 after mastectomy), 37 patients had nodal recurrence (23 axillary, 14 sub-clavicular) and 11 had both. Eight patients had salvage mastectomy, 19 had radiation therapy (5 of them not previously irradiated), all patients received chemotherapy. In group B, 35 patients had palliative radiation (10 for pain relief, 7 for spinal cord compression, 18 for high risk lesions), 50 had chemotherapy and 67 patients had bisphosphonates. In group C, we observed 4 cases of meningeal carcinomatosis. One case of unique brain metastases was operated, and all other cases had whole brain radiation and chemotherapy. In group D, metastases were: liver (43), lung (32), pleural (27), peritoneal (12), gastric (3) and other (32). Seven patients had initial endocrine therapy, 3 patients had metastasectomy (2 liver, 1 lung) and the remaining had chemotherapy.
Site specific group characteristics
Group A: BC patients with loco regional recurrence were more likely to have histological tumor size above >2 cm, grade 1–2, HR positive and 0–3 involved lymph nodes. They have been more frequently treated with breast conserving surgery. (Table 1). Group B were more likely to have grade 2–3, 1–3 positive lymph nodes and HR positive tumors (Table 2). Group C:. Patients who developed brain metastases were more likely to be young, to have large tumor size, grade3, positive lymph nodes and HER2 positive tumors (Table 3). Group D: patients with visceral metastases were more likely to have tumors > 2 cm in size, with nodal involvement, grade 3, HR negative and HER2 positive tumors (Table 4).
Site specific recurrence outcome
Median follow-up was 76 months. Median TTR was statistically significant different between different groups. The longest TTR was observed in group A (40.5 months), the shortest in group C (24 months). Annual recurrence rate in group A, was stable ranging between 15%–18%, within the first 3 years, it decreased progressively after. Relapse rate in group B was 19.4% in the interval [1–2[year, 28.3% in the [2–3[year and 13% in the [3–4[year. It deceased to <3% after. Relapse rate in group D was 14.1% in the interval [1–2[,29.5% in the interval [2–3[year and 10.7% in the interval [3–4[year to < 5% thereafter. Brain metastases (group C) occurred mostly 39.1% within 1–2 year interval. Annual recurrence rate distribution of different groups is described in Fig. 1.
Median survival was 46 months in group A, 43 months in group B, 11 months in group C and 23 months in group D with a significant log-rank p value (<0.001). Survival at 2 years was, 79%, 72%, 0% and 50% respectively (p < 0.001). When comparing survival between group A versus group B and group D versus group C, we did not find a statistically significant difference between groups A and B (0.29), group D had a significantly better survival than group C (p = 0.001).
We considered sub group survival analysis according to first site of recurrence in 3 groups according to immunohistochemistry: [HR+, HER2-] accounting for 119 patients, [HR-, HER2-] accounting for 78 patients and [HER2+] accounting for 96 patients (Fig. 2). In [HR+, HER2-] and [HER2+] populations, we observed a statistically significant difference in overall survival, between different first sites of recurrence with locoregional and bone relapses being associated to a better outcome. However there was no impact on survival of the site of recurrence in[HR-, HER2-] population. We did not find a significant difference of survival for brain metastases (group D) between HR positive and HR negative patients (11.35 versus 11 months, p = 0.8) and between HER2 positive and HER2 negative (13.4 versus 9.7, p = 0.74) patients.
In group D, 89 patients had at least 3 metastatic sites (Bone in 67 cases, lung in 53 cases, liver in 48 cases, pleural 32 cases, Brain in 17 cases, peritoneal in 16 cases, Kidney in 5 cases, gastric in 2 cases). Unique metastatic organ (with exception of brain, bone and loco-regional) was seen in 39 cases: 21 patients had metastases only in the lung, and 19 had metastases in the liver. We observed a non significant tendency to a better median survival in patients with unique visceral metastatic site, compared to patients with multiple visceral metastatic sites, 26 months versus 22 months, p = 0.09.
Discussion
Our study focused on first relapse sites of early BC and observed that detected bone and/or locoregional relapses during the first 3 years of follow-up was associated with a better prognosis. Due to the high risk population in our country (locally advanced and young age mainly), we continue in our center, to perform systematic radiological and biological follow-up in asymptomatic patients to detect early breast cancer recurrence. Current practice guidelines such as ESMO [13], NCCN and ASCO [14] recommend only annual mammography with no further systematic biological or radiological assessment. This follow-up strategy is endorsed by five randomized trials recently reviewed by Cochrane library concluding that there is no survival or quality of life benefit conducting systematic further radiological and biological surveillance [15]. Nevertheless, all trials cited in this review did not tailor surveillance regarding patient’s characteristics and patterns of relapse. We believe that BC relapses should be me more considered as a chronic than a rapidly fatal situation. In fact, oligometastaic BC has a better outcome than the entire MBC population, 10 years overall survival ranged between 35% and 73% versus 26%–52%. This group of patients represents 1%–3% of all MBC and some among them can be “curable” [16].
One of the classical breast cancer recurrences usually treated with curative intent is loco-regional recurrence [17]. Salvage mastectomy is the standard treatment after ipsilateral breast tumor recurrence. Secondary breast conserving surgery with or without re-irradiation can also be a suitable option in some selected cases [18]. Furthermore, local recurrence after mastectomy can be cured with local excision, radiation therapy; which can be associated with hyperthermia [Harms and al, 2015]; or electro chemotherapy [19]. The key to correctly manage this excellent outcome group is early detection. In our population, tumor size>2 cm and positive 0-3 involved lymphnodes were surrogate markers for isolated loco-regional recurrence. A surprising result is the higher recurrence rate in breast conserving surgery group which can be due to the probable heterogeneity of margin status in this group; not all patients had surgery in expert centers. Mastectomy and lumpectomy combined with radiation therapy are reported to be equivalent treatments in terms of overall survival and recurrence rate [20]. Our results go along with the international guidelines, recommending local surveillance in this group. It is enhanced by the excellent outcome of this population in our study.
Bone metastasis, being the most common site of distant breast cancer, has been a growing subject of debate concerning chemoprevention. Initiating adjuvant biphosphonates therapy, in particular Zolidronic acid, showed a substantial benefit in disease free survival in early breast cancer in post-menopausal women [22]. Therefore, it is needed to define a comprehensive approach regarding breast cancer phenotype underlying metastatic bone disease. Future translational studies may find biomarkers of bone relapse based on gene signatures observed in the primary tumor. Savci-Heijink et al recently identified a gene signature in early breast cancer that could predict bone metastasis in later follow-up [23]. As in our current series, HR+, high grade and positive lymph nodes(1–3)BC patients were the more likely to develop bone metastasis. We suggest systematic bone scan surveillance in women who are HR+, have positive lymph nodes (1–3) and high-grade tumors looking to the excellent outcome of this bone alone first relapse group. In our study, adjuvant zolidronic acid was not given.
Brain metastasis occurs in 10%–20% of advanced breast cancers. It is associated with a poor prognosis, median survival is of 3–6 months [24]. Prognosis can be predicted using modified Breast-GPA Index based on Karnofsky Performance Status, age, molecular subtype and number of brain metastasis. The overall survival ranges from 2,6 months for a score of 0–1 to 28,8 months for a score 3,5-4 [25]. Multimodal local treatments including surgery, external beam radiotherapy and gamma knife surgery have shown improved survival in selected patients [26]. This may suggest the utility of early brain metastasis detection in selected population mainly among patients with node-positive disease, HR negative, high tumor grade and younger age as reported in previous reports [27,28]. As shown in our study HER2 enriched and HR negative tumors are more likely to develop brain metastasis during follow-up. Possible explanation of this particular pattern of relapse is HER2 enriched metastatic cells tropism to the central nervous system [29] and showing relatively better prognosis [26]. In our study no difference in brain metastases survival was observed among HR+, HER2+ and triple negative populations. We cannot recommend systematic brain imaging. Future results from new anti-HER2 agents (Pertuzumab, T-DM1) may show particular impact on brain metastasis that may lead to changing recommendations regarding systematic screening.
Visceral metastasis was with 149/324 pts, the most frequent site of recurrence in our population, probably due to the poor prognosis and high-relapse risk of our patients. However inside this group, prognosis is variable, according to the number of metastasis at diagnosis, site of recurrence, disease-free survival time and molecular subtypes. The prognosis goes from bad to worse from soft tissue, lung and liver recurrence [30]. Our belief is that follow-up CT-scan could permit to detect early relapse in liver or lung metastasis as local treatments have shown survival benefit in selected patients. A meta-analysis pooled 19 studies including 553 patients showed a median 5-years survival rate at 40% in patients who had hepatectomy for liver metastasis from breast cancer [31]. Those encouraging results are comparable to those obtained with resection of colorectal liver metastasis [32]. Metastasectomy of selected breast cancer secondary lung metastasis showed interesting overall survival rates ranging from 31%–82% [33]. Unique or oligometastatic disease remains rare in breast cancer relapse, the question of the relevance and value of systematic CT scan during follow up is still open.
Conclusion
Since recognition of BC molecular classification, clinical trials investigating surveillance have not evolved to the actual concept of breast cancer heterogeneous behaviour. Even if we found specific patterns of relapse in our breast cancer patients, we cannot recommend changing the actual surveillance recommendations due to the retrospective nature of our study. However, considering the evolution of therapeutic options in metastatic disease, customized strategy is needed through prospective randomized clinical trials comparing tailored surveillance protocols to the actual standard consisting in annual mammography and clinical examination [34].