Abstract
BACKGROUND AND OBJECTIVE:
Distant metastatic spread in breast cancer patients is a complex phenomenon involving several prognostic factors. We focused our analysis on early metastatic breast cancer (EMBC) (occurring during the first 36 months) versus late metastatic breast cancer (LMBC) (occurring beyond 3 years) in order to ascertain their possible differential predictive factors.
METHODS:
diagnostic, surgical, and follow-up data were assessed for consecutive patients with breast cancer undergoing surgery between 1997 and 2019. We analysed the predictive factors for distant metastasis using both univariate and multivariate analysis.
RESULTS:
The median follow-up for this cohort of 2708 patients was 89 months. The median metastasis-free interval (FMI) for metastasis patients was 38 months (17 months for EMBC group and 76 months for LMBC group). Distant metastases developed in 12.9% (350/2708); 48% (168/350) of them as EMBC and 52% (182/350) as LMBC. Loco-regional recurrence and nodal extracapsular extension were the only common predictors for both.
CONCLUSIONS:
EMBC and LMBC appeared as two separate conditions, with a different outcome. In the EMBC group, tumour proliferation related factors were significant (histological grade, tumour size, body mass index), whereas for LMBC, other slow-acting factors seemed to be involved (screening program, tumour burden, bilateral tumour).
Introduction
In spite of considerable improvement in diagnosis and adjuvant therapy of breast cancer (BC), about 20–30% of patients eventually present with distant metastatic spread on follow-up [1]. Once established, such an advanced stage of the disease has no cure, and patient mean survival is estimated at 2–3 years [2]. The event of distant metastatic spread seems a complex phenomenon involving several predictive factors [3,4]. It also entails important differences among patients regarding organ tropism, extent of disease [1–5], and survival. Although a clear correlation between BC phenotype and metastatic spread has been reported [6], additional factors may also weigh in, including tumour size, lymph-node disease, histologic grade, lympho-vascular invasion, and loco-regional recurrence [7–9]. Distant organs most commonly involved are bone, liver, lung, and the brain [5,10]. Metastatic-free interval (MFI) seems relevant, as patients sustaining early metastatic spread tend to show a worse prognosis [4,11,12]. This suggests an increased biologic aggressiveness and/or resistance to systemic therapy [13].
The aim of our present study was to ascertain the specific and differential predictive factors for early (<36 months) and delayed (>36 months) distant metastatic spread in a cohort of BC patients.
Material and methods
We performed a retrospective analysis that was based on a prospectively built observational database including 2708 consecutive patients with invasive breast cancer referred to our Breast Unit for surgical treatment between January 1, 1997 and December 31, 2019. The years 2018 and 2019 were included in the analysis in order to account for cases of primary or “de novo” metastases, although we are aware that such patients may not be strictly considered early metastatic cases. Written informed consent was obtained for all invasive procedures and surgery, as well as for inclusion of patient data in the database. The present study was approved by the Research Ethics Committee of our centre. All enrolled patients belonged to our regional Public Health System.
Our database included multiple conventional baseline and follow-up variables: age, diagnosis on mass-screening programme, body mass index (BMI), tumour size, histological tumour type and grade, hormone receptor status, tumour phenotype, lympho-vascular invasion, lymph-node status, surgical procedures, systemic and radiation therapy regimes, local recurrence, distant recurrence and death, among others. As such, the database was created in 1997 and its content has been updated weekly with new cases ever since. Information on active patients is updated following each outpatient visit. For the rest, a yearly update is completed by our data manager.
Owing to the fact that we started collecting patient data during January 1997, to the original variables, others have been added and registered along the years. For instance, the local population-based BC screening programme was implemented in 2002, including biennial mammograms for women aged 50–69 years.
Individual patient treatment was always based on the corresponding protocol of our Unit, which follows both national and international guidelines. Most commonly, chemotherapy regimens combining anthracyclines and taxanes, or hormone therapy regimens with tamoxifen or aromatase inhibitors were used. From 2003 on, the use of trastuzumab was instituted for Her2+ patients. Neoadjuvant therapy was based on regimes similar to the adjuvant setting.
Bidimensional CT planning was used for adjuvant radiation therapy up to 2008, when tridimensional planning was adopted. Radiation therapy was delivered after radical surgery in selected cases.
Gene expression assays were used in order to provide further prognostic data and some prediction of chemotherapy benefit. We implemented either Oncotype® or Mammaprint® in T1-T2 positive hormone receptor/Her2 negative patients (T1-2 N0, T1-2 N1).
Phenotypical type of breast cancer for each individual patient was derived on the basis of hormone receptor status, as well as on and HER2, ki67 values, and histologic grade as follows:
Luminal A: ER+ and/or PR ≥20%, HER2−, grade 1 or 2 and/or Ki67 <20% Luminal B:
Luminal B HER2−: ER+ and/or PR ≥20%, HER2−, grade 3 or Ki67 ≥20% Luminal B HER2+: ER+ and/or PR±, HER2+ Pure HER2: ER−, PR−, HER2+ Triple negative: ER−, PR−, HER2−
Definitions
Distant metastatic spread was defined as any tumour recurrence beyond the chest wall, regional lymph-nodes, or ipsilateral breast. Sites for distant recurrence included bone, pleura/lung, liver, brain, distant lymph-nodes and others (including peritoneum, spleen, ovary, eye …). Only lung, brain, and liver disease sites were considered visceral metastases.
Early metastatic breast cancer (EMBC) was defined as any distant metastatic spread occurring during the first 36 months following the initial treatment (systemic therapy or surgery). Late metastatic breast cancer (LMBC) was defined as any metastatic spread occurring beyond 3 years of the initial treatment. We selected 36 instead of 24 months as suggested by others [1,4] for two reasons. Firstly, in a previous literature report, significant differences in mortality rates between 24 months (82.1%) and 36 months (83.3%) were not found [4]. Secondly, because in our own experience the median interval to metastasis development is about 38 months. “De novo” metastases (those occurring within three months of initial treatment) were included among early metastatic spread cases.
Patient low weight was considered for BMI under 18.5, whereas normal patient weight was considered for BMI between 18.5 and 24.9, and overweight for BMI exceeding 25.
Metastasis-free interval (MFI) was defined as the time elapsed from the first BC treatment to the appearance of the first metastatic event. All other intervals considered were defined in the same way.
Variables explored as potential prognostic factors for metastasis development were: age, cigarette smoking, alcohol intake, menopausal status, body mass index, diagnosis at mass-screening, unilateral versus bilateral disease, unifocal versus multifocal disease, histologic grade, Ki67 values, tumour size, nodal status, tumour phenotype, oestrogen/progesterone receptor status, lymphovascular invasion, genomic platform result when appropriate, adjuvant systemic therapy, surgical procedure, and locoregional relapse.
Statistical analysis
For the analysis of clinico-pathologic variables, the total number of database cases was used. For the analysis of disease recurrence events, the total number of patients was considered.
Qualitative variables were expressed as their number and percentage, while quantitative variables were expressed as their median, mean value, standard deviation and range. For the univariate analysis, comparison of qualitative variables was done using the chi-square test or Fisher’s exact test, while mean values of quantitative variables were compared using the Student’s T or the ANOVA test. Statistical significance was set at a p value of <0.05, with a two-tail approach. A multivariate analysis was performed based on the binary logit regression method, including those variables that were significant at the univariate analysis.
All statistical analyses were performed using the IBM SPSS 25.0 statistical package (Armonk, NY, USA).
Results
The median follow-up for our cohort of 2708 patients was 89 months, with a range of 1–526 months. The median MFI for the overall cohort was 84 months (range 1–526). Overall, distant metastases developed in 12.9% of patients (350/2708), 48% (168/350) of them as EMBC and 52% (182/350) as LMBC. In 26.3% of patients (92/350) only one metastatic site was diagnosed, while in 73.7% (258/350), two or more sites were diagnosed. The median FMI for metastatic patients was 38 months (range 1–335). This is not an oligometastases-oriented study. Rather, all metastases occurring during our follow-up were included.
Obviously, there were patients lost to follow-up, mainly because some of them moved to other parts of the country. Overall, 69 patients did not undergo a complete follow-up (2.5%). These had a mean follow-up of 84.4 ± 58.9 months (range 6–228, and median of 67 months).
Early metastatic breast cancer versus late metastatic breast cancer
Median MFI for the EMBC group was 17 months, with a range of 1–36 months. Median MFI for the LMBC group was 76 months, with a range of 37–335 months. Specific mortality was 83.3% (140/168) for EMBC, and 68.1% (124/182) for LMBC, a statistically significant difference. For EMBC patients, mean survival was 47.1 ± 44.2 months, with a median of 35 (range 6–242 months). For LMBC patients, mean survival was 133.7 ± 68.9 months, with a median of 117 (range 40–356 months).
In the EMBC group, the rate of visceral metastases was 91.1% (153/168), and the rate of non-visceral metastases was 8.9% (15/168). In the LMBC group, the rate of visceral metastases was 86.3% (157/182), and the rate of non-visceral metastases was 13.7% (25/182). Overall, no significant differences were seen between groups. However, as seen in Table 1, EMBC patients showed a significantly higher rate of liver metastasis, whereas LMBC patients showed a significantly higher relative rate of bone and lymph-node metastases.
Organ-specific metastatic spread rates according to time of onset
Organ-specific metastatic spread rates according to time of onset
EMBC: early metastatic breast cancer; LMBC: late metastatic breast cancer.
Bone-only metastatic spread was seen in 6.5% (11/168) of patients in the EMBC group, compared with 6.6% (12/182) of patients in the LMBC group. Lung-only metastatic spread was seen in 10.1% (17/168) of EMBC patients, and in 3.3% (6/182) of LMBC patients. Liver-only metastases were seen in 8.3% (14/168) of EMBC patients, and in 2.2% (4/182) of LMBC patients. Brain-only metastases were seen in 4.2% (7/168) of EMBC patients and in 3.3% (6/182) of LMBC patients. All these were statistically non-significant differences.
No significant differences were seen between groups when the number of metastatic organs was considered. However, metastasis to three or more distant sites were more often seen with EMBC than with LMBC patients, a trend close to statistical significance (p = 0.06).
As for the specific histopathologic diagnosis, medullary breast tumours were the least EMBC-associated type - 0%, followed by mucinous tumours - 1.6% (1/64), papillary tumours - 2.6% (1/38) and mixed ducto-lobular tumours 2.7% (1/37). The rest of histological types were associated with EMBC in excess of 5%. Thus, NOS type amounted up to 6.4% (137/2153), followed by metaplastic tumours - 7.7% (2/26), neuroendocrine and lobular - 7.3% (3/41), and 7.7% (16/209), respectively, apocrine - 10% (2/20), and invasive cribriform - 18.7% (3/16) (p = 0.032).
Table 2 shows the univariate analysis results for potential factors associated with EMBC and LMBC. Specifically, such results were as follows:
Patient age - Age was not different between both study groups. However, patients under 50 showed the highest rate of metastasis in both of them.
Cigarette smoking and alcohol intake had no influence on metastasis development in our study.
Menopausal status - Premenopausal patients showed significantly higher rates of metastasis than menopausal patients, although there were no differences between study groups.
Body mass index - It is worth noting that patients with a BMI under 18.5 showed significantly higher rates of EMBC than patient with normal BMI or than obese patients. This was not seen in the LMBC group.
Mass-screening diagnosis - Patients with BC diagnosed under a mass-screening programme had significantly lower rates of both EMBC and LMBC than the rest (1.7% versus 9% for EMBC, and 1.2% versus 9.9% for LMBC). However, there were no significant differences between study groups.
Unilateral versus bilateral disease - Rates of unilateral and bilateral disease were not different in the EMBC group (6.8% versus 4%). However, in the LMBC group the rate of metastasis development was significantly higher in patients with bilateral disease (11.5%) compared with those with unilateral disease (6.9%).
Unifocal versus multifocal disease - This was not associated with significant differences in metastasis development between study groups.
Histologic grade (HG) and Ki67 values - HG seems crucially linked with the occurrence of EMBC but not to LMBC. The risk of EMBC was clearly seen to increase as HG increased. This was also noticed for Ki67 values over 20.
Tumour size - It seems a relevant factor, especially for EMBC patients. In this group, the rate of metastatic spread went up from 1.9% in T1 patients, to 10.9% in T2 patients, through 23.7% in T3 patients, and 37.7% in T4 patients. For LMBC patients, the corresponding rates were 4.8% in T1 patients, 10.6% and 10.7% in T2–T3 patients, and 19.5% in T4 patients. Except for T2 patients, the metastatic rates showed significant differences between study groups (p < 0.001).
Nodal status (axillary tumour burden) - In both study groups, rates of metastatic spread were associated with nodal involvement. For EMBC, metastatic rates increased from 2.9% in node-negative patients to 5.3% for patients with one involved node, through 6.3% for those with two positive nodes, and 23.1% for those with three or more positive nodes. For LMBC, the corresponding rates were 4.6% for node-negative patients, 5.9% for those with one positive node, 11.3% for those with two positive nodes, and 17.6% for those with three or more positive nodes. There were no differences between study groups, except for a significant increase in LMBC rates for patients without nodal involvement.
Lymphovascular invasion - This proved to be a significant risk factor both for EMBC patients (3.2% versus 13%), and for LMBC patients (4.9% versus 11.8%).
Hormone receptor and Her2neu status - Patients with negative oestrogen-receptor (ER) status showed significantly higher rates of EMBC than ER positive patients (14.5% versus 5.1% - p < 0.001). This was not seen with LMBC (9.2% versus 6.8% - p = 0.058). Patients with negative progesterone receptor (PR) status showed significantly higher rates of both EMBC (4.7% versus 13.2% - p < 0.001) and LMBC (6.7% versus 8.6% - p = 0.004). Her2neu positive patients showed significantly higher rates of EMBC than Her2neu negative patients (9.6% versus 6.2% - p = 0.022). There were no significant differences in Her2neu status associated with the occurrence of LMBC (9.9% versus 6.8% - p = 0.076).
Tumour phenotype was significantly associated with EMBC according to the univariate, but not to the multivariate analysis. In the univariate analysis, Triple negative phenotype (TN) showed the strongest correlation, followed by the pure Her2 phenotype. There was no association of LMBC and tumour phenotype (Table 2).
Univariate analysis results for potential variables associated with early metastatic breast cancer and late metastatic breast cancer
EMBC: early metastatic breast cancer; LMBC: late metastatic breast cancer.
Patients receiving adjuvant chemotherapy showed a significantly higher rate of both early and late metastases. Patients receiving adjuvant hormone therapy showed a significantly lower rate of EMBC than those without hormone therapy, although this was not the case with LMBC (Table 2).
As for implementation of tumour gene expression assays in our patients, with or without ensuing chemotherapy, for those patients in whom testing gave a low risk profile, and thus chemotherapy was avoided, there were no EMBC cases, while LMBC rate was just 1%. For patients with a high-risk profile, thus receiving chemotherapy, EMBC rate was 3.7%, with no cases of LMBC. For those patients receiving adjuvant chemotherapy without genetic testing, which comprised the bulk of our sample because genetic testing was a late development, the corresponding metastatic rates were significantly higher (8.8% and 10.4%).
The relevant risk factors for EMBC according to the multivariate analysis were Body Mass Index, Histologic Grade, Tumour size, Nodal Extracapsular Invasion, and Loco-regional Recurrence (Table 3). The relevant risk factors for LMBC according to the multivariate analysis were Bilateral Disease, Screening Programme Diagnosis, Axillary Tumour Burden, Nodal Extracapsular Invasion, and Loco-Regional Recurrence (Table 4). There were only two common predictors for both groups, namely Loco-Regional Recurrence and Nodal Extracapsular Invasion.
Logistic regression analysis for early metastasis
Logistic regression analysis for late metastasis
Based on our present results, we believe that EMBC and LMBC could be seen as separate conditions, probably obeying to different factors, also with a different course and fate. Considerable differences in MFI and specific mortality in our series, show that indeed patients with EMBC carry a worse prognosis than patients with LMBC, and therefore that we should be increasingly aware of factors leading to such differential distant spread. A worse prognosis for early distant spread has already been suggested by others [4].
Based on the univariate analysis only, we have shown an association between both EMBC and LMBC and certain specific predictive factors. At odds with other reports [4], we have shown a statistically significant correlation between tumour phenotype and EMBC, especially with TN and Her2 tumours. Such correlation does not seem to apply to LMBC. According to some reports [17], histological grade 3 has enough prognostic relevance to shift a patient from Luminal A to Luminal B type [17,18]. In our cohort, such a variable was indeed significant for both overall and EMBC. Patients with Luminal B Her2 negative type and histological grade 3 showed the highest rate of distant metastases, exceeding that of TN and pure Her2 cases. However, no specific tumour phenotype was statistically correlated with EMBC or LMBC at the multivariate analysis. The much-proffered phenotype tumour classification per se might not be as valuable as previously thought, unless compounded with other factors [19]. As previously reported by our group [20], the risk of distant metastases for a 10–15 mm TN tumour without lymph-node disease is like that of luminal tumours of the same size.
Metastatic sites also seem to vary according to the chronology of distant spread. Thus, lung and liver metastasis, either multiple or solitary, are associated with EMBC, while bone metastases are associated with LMBC.
Adjuvant chemotherapy seems to be a significant predictive factor for both EMBC and LMBC, with patients receiving adjuvant chemotherapy showing higher rates of metastases than untreated patients, probably linked to a more advanced or aggressive disease. However, such a trend was not confirmed at the multivariate analysis. As for adjuvant hormone therapy, patients not receiving such treatment showed a significantly higher rate of EMBC than treated patients. LMBC was not associated with differences in adjuvant hormone therapy.
Gene expression results were not associated with the development of either EMBC or LMBC in our series. Probably their main contribution was avoiding chemotherapy side effects in some patients, albeit without a clear-cut benefit. A fact that has already been reported by others [21,22].
If we focus on the more robust data provided by the multivariate logistic regression analysis, loco-regional relapse stands as one of the most important ones. It may occur in synchronicity with distant metastases or forebode them by a few months. Also, there is a well-established correlation between tumour size and distant relapse [14–16], which we were able to confirm for EMBC. Nodal extracapsular invasion acts as a significant risk factor, probably related to a more aggressive tumour invasion potential. Of note, there seems to be a higher rate of EMBC in underweight patients. On the other hand, although it may impact as a significant predictive factor for LMBC, diagnosis on a mass-screening programme does not behave as a predictive factor for EMBC, probably because these are small tumours with very low axillary disease burden and no extracapsular invasion.
Our investigation has limitations as it was a retrospective analysis based on a prospectively designed database. Also, inclusion of 2018 and 2019 to account for “de novo” metastases resulted in an inappropriate follow-up for LMBC.
To sum up, our multivariate regression analysis points to substantial differences in the respective drivers for EMBC and LMBC. EMBC cases are associated with low BMI, higher histologic grade, increased extracapsular invasion, and higher rates of loco-regional recurrence. On the other hand, LMBC are associated with bilateral tumours, diagnosis on a mass-screening programme, and three or more positive axillary nodes. Only loco-regional recurrence and extracapsular invasion seem common predictors for both EMBC and LMBC.
As a corollary, we believe that those clinical factors associated with the development of EMBC and LMBC need to be especially considered during patient follow-up.
Abbreviations list
EMBC: Early metastatic breast cancer
LMBC: Late metastatic breast cancer
FMI: Metastasis-free interval
BMI: Body mass index