Clinical and pathological characterization of 158 consecutive and unselected oligometastatic breast cancers in a single institution

Data about incidence, biological, and clinical characteristics of oligometastatic breast cancer (OMBC) are scarce. However, these data are essential in determining optimal treatment strategy. Gaining knowledge of these elements means observing and describing large, recent, and consecutive series of OMBC in their natural history. We collected data retrospectively at our institution from 998 consecutive patients diagnosed and treated with synchronous or metachronous metastatic breast cancer (MBC) between January 2014 and December 2018. The only criterion used to define OMBC was the presence of one to five metastases at diagnosis. Of 998 MBC, 15.8% were classified OMBC. Among these, 88% had one to three metastases, and 86.7% had only one organ involved. Bone metastases were present in 52.5% of cases, 20.9% had progression to lymph nodes, 14.6% to the liver, 13.3% to the brain, 8.2% to the lungs, and 3.8% had other metastases. 55.7% had HR+/HER2− OMBC, 25.3% had HER2+OMBC, and 19% had HR−/HER2− OMBC. The HR+/HER2− subtype statistically correlated with bone metastases (p = 0.001), the HER2+subtype with brain lesions (p = 0.001), and the HR−/HER2− subtype with lymph node metastases (p = 0.008). Visceral metastases were not statistically associated with any OMBC subtypes (p = 0.186). OMBC-SBR grade III was proportionally higher than in the ESME series of 22,109 MBC (49.4% vs. 35.1%, p < 0.001). OMBC is a heterogeneous entity whose incidence is higher than has commonly been published. Not an indolent disease, each subgroup, with its biological and anatomical characteristics, merits specific management.


Introduction
According to Hellman and Weichselbaum, some metastatic cancers do not develop diffuse metastases from the start but acquire this capacity gradually. There would therefore be an intermediate stage where metastatic spread would be limited to specific organs and the number of metastases would be low. The authors hereby introduce the concept of oligometastases [1,2].
If the benefits of curative strategies have been proven for several types of oligometastatic diseases (OMD) [2][3][4][5], the question remains controversial concerning oligometastatic breast cancer (OMBC). No randomized studies have thus far answered the question. Recently, one phase IIR/ III trial was presented as an abstract at the 2022 ASCO meeting. Including OMBC of various anatomical and biological characteristics, it failed to establish the benefits of a curative strategy [6]. Published data from this study will be necessary to draw conclusions from these results.
To confirm, or refute, the benefits of intent-to-cure strategies, Hellman and Weichselbaum defined four milestones [1,2]: -Epidemiology: "The importance of the oligometastatic state will be dependent on the size of the group of patients for whom it offers curative prospects." -Biology: "This paradigm emphasizes the importance of markers specifically related to where in the spectrum of malignancy an individual cancer is located." -Imaging: "[Its] effectiveness will be critically dependent on the specificity, sensitivity, precision, and accuracy of tumor imaging." -Treatments: "New methods of surgery or radiation therapy may allow curative treatment of such oligometastases either alone or combined with systemic therapy and focal treatment." Today, several treatments are available with low toxicity and high efficiency: surgery, stereotactic body radiotherapy, intensity-modulated radiation therapy, and percutaneous image-guided treatment [7][8][9][10]. Also available are imaging techniques with high sensitivity and specificity [11][12][13]. However, OMBC incidence remains uncertain: values in the literature are based on outdated studies [14] or come from highly selected series of OMBC [15]. Furthermore, data on OMD specific biomarkers are limited [9], and a recent review confirms that data on OMBC biology are also scarce [14]. It thus remains unclear whether a limited number of metastases at diagnosis reflect specific biology, early diagnosis of metastases, or both.
In order to determine the incidence and the clinical and biological characteristics of OMBC, data must be available from series respecting the following conditions: series must be consecutive; they must describe OMBC at the moment when the first metastases are seen, thus corresponding to synchronous and metachronous OMBC according to the ESO/ESMO and ESTRO/ASTRO classifications [5,16]; criteria pertaining to the number of organs and metastases must be consensual; the calculation method for metastases must be explicitly described. Finally, the feasibility of focal treatment must not enter into inclusion criteria, thereby avoiding a selection bias linked to practices and available resources of participating teams.
In a systematic review, Van Ommen et al. [15] selected eleven publications identifying OMBC prognostic factors. Since all of these series describe cohorts of highly selected, non-consecutive patients, they cannot provide the information sought.
Since their work, other retrospective cohorts of OMBC have been published. One is a large, retrospective and consecutive cohort of 3,447 patients with synchronous MBC analyzed with inverse-sampling probability weighting collected in the Netherlands Cancer Registry (NCR) between January 2000 and December 2007 [17]. In this cohort, patients were considered to have OMBC when the number of metastases was limited to three with no other inclusion or exclusion criteria.
Two other retrospective cohorts have more restrictive inclusion criteria. The first presents a cohort of 170 patients treated at the Chennai Oncology Cancer Institute (India) for OMBC [18]. Those with ≤ 5 metastases in total and metastases ≤ 5 cm were considered OMBC. The second is a retrospective series of 50 patients treated from 2009 to 2014 at the National Cancer Center (China) for metachronous OMBC. The series included metachronous OMBC only; primary breast tumors had to be resected; metastatic disease had to be limited to a single organ; no more than three metastases, and no brain or lymph node metastases were possible for inclusion [19].
In the absence of publications strictly respecting these criteria, we undertook a retrospective study involving all consecutive patients with synchronous or metachronous MBC starting treatment at metastatic stage at our institution from January 2014 to December 2018. MBC was considered oligometastatic if there were one to five metastases at the time of diagnosis with no other inclusion or exclusion criteria. We did not retain criteria related to the feasibility of focal treatments for primary tumors, local relapses, or distant metastases. Indeed, these criteria depend on available therapeutic techniques and physicians' choices. The latter, in particular, was therefore an inadequate criterion in a study devoted to assessing OMBC at diagnosis prior to any therapeutic intervention. Our series was then compared with the Epidemiological Strategy and Medical Economics (ESME) cohort of MBC (22,109 patients treated in France for MBC between January 2008 and December 2016) [20] and the NCR cohort with its large number of OMBC cases [17]. In the end, our analysis focuses on the initial description of OMBC, providing a maximum of information on this subject. Analysis of prognostic factors for progression-free or overall survival, not pertinent in this case, is therefore not reported.

Patient selection
We conducted a non-interventional, retrospective, and monocentric study to describe the clinical and pathological characteristics of all consecutive OMBC selected from the IUCT-O advanced breast cancer database between January 2014 and December 2018. This study was approved by our Multidisciplinary Breast Committee and our Institutional Board Committee (BEC-FO-0227).
Patient selection focused on females, aged > 18 years with newly diagnosed synchronous or metachronous OMBC. Patients were excluded whose treatment at metastatic stage had been initiated in another institution, who had either an unresectable primary tumor or an unresectable local relapse with no distant metastases or who had a history of second cancer.

Definitions
In our study, OMBC was defined as MBC with up to five distant metastases at the time of diagnosis not necessarily in the same organ. This definition is consistent with those of ESO-ESMO [16] and ESTRO-ASTRO [5] regarding thresholds of the number of metastases and organs invaded. This included patients with synchronous OMBC (maximum six months interval between diagnosis of oligometastatic disease and primary cancer diagnosis) and metachronous OMBC (more than six months interval between diagnosis of oligometastatic disease and primary cancer diagnosis) according to ESTRO-ASTRO classification [5].
Organs were divided into brain, lung, liver, bone, lymph nodes, and others (skin, pancreas, and adrenal glands). Local relapse at the time of diagnosis included relapse in the breast, ipsilateral loco-regional lymph node infiltration, limited ipsilateral cutaneous breast, or chest infiltration.
Definition and number of distant metastases were derived from Kelly et al. [21]. "For lesions in the brain, bone, lung, and liver, each radiologically identifiable lesion was considered one site of disease. For lesions in the lymph nodes, radiologic involvement of each echelon of the axillary, cervical, or mediastinal lymphatics was considered a single site of disease, even if there were multiple nodes noted in a given echelon" [22,23]. In the absence of a contralateral tumor in the breast, contralateral loco-regional lymph node invasion was considered as one metastasis. Leptomeningeal disease, malignant pleural or peritoneal effusions, and extensive cutaneous involvement were considered as diffuse disease.

Data collection
All data were collected at the time of diagnosis of the first metastases, prior to any treatment for this stage. HR and HER2 status were determined by metastatic biopsy, if available, otherwise by biopsy of local relapse or of the primary tumor. Standard guidelines were applied to any tumor analysis performed. Tumors were defined as hormone receptor positive (HR+) if estrogen receptor or progesterone receptor expression was > 10% (immunohistochemistry) according to the French Groupe d'étude des facteurs pronostiques immunohistochimiques dans le cancer du sein, Unicancer (GEFPICS-FNCLCC, Unicancer study group of immunohistochemical prognostic factors in breast cancer) recommendations [24]; (HER2) immunohistochemical (IHC) score 3 or IHC score 2 with a positive in situ hybridization (ISH) classified the cancer as HER2+ . All cancers with an IHC score of 0, 1, or 2 with a negative ISH test were considered HER2-.
Imaging techniques applied for staging evaluation -Computed Tomography (CT) scan, bone scan, Positron Emission Tomography/Computed Tomography with 18 fluorodeoxyglucose ( 18 F-FDG-PET/CT), Whole-Body Resonance Magnetic Imaging (WB-MRI), liver, brain, or spinal MRI-were recorded. We then compared our cohort with the very large ESME cohort [20] and the NCR cohort, according to our literature review, the largest sample of synchronous OMBC to date with inclusion criteria similar to ours [17].

Review of the literature
The algorithms used for the literature review are presented in Online Appendix 2.

Primary objective
-Describe the clinical and pathological characteristics of synchronous and metachronous OMBC.

Secondary objectives
-Estimate the incidence of synchronous and metachronous OMBC in our database. -Describe the clinical and pathological characteristics of OMBC according to synchronous/metachronous features. -Evaluate inconsistencies in clinical and pathological characteristics between our cohort and the NCR cohort. -Determine the specific pathological characteristics of OMBC by comparing our cohort with ESME data.

Statistical analysis
This cohort is a population-based registry, with individual data on all consecutive patients treated for MBC in our institution from January 2014 to December 2018. Population characteristics were described using the usual statistics. Continuous variables were summarized by median, range, and qualitative variables by frequency and percentage. Comparison between groups was performed using either Chisquare or Fisher's exact test for qualitative variables and the Kruskall-Wallis test for continuous variables. All statistical tests were two-sided, with p-values < 0.05 considered statistically significant. Statistical analyses were conducted using STATA v16 (StataCorp, College Station, TX, USA) software.

Patient characteristics
We identified 1,229 patients in our database of advanced breast cancer, including patients with MBC or those with no distant metastases but with either an unresectable primary tumor or an unresectable local relapse from January 2014 to December 2018. Among them, 33 presented an unresectable primary tumor or a local relapse with no distant metastases, 198 had initiated treatment for MBC in another institution, and 998 had MBC treated in our institution. Among the latter, 84.2% were considered polymetastatic, and 15.8% presented one to five metastases (Fig. 1). Among these 158 OMBC patients, medium age at diagnosis was 55 years [range . Sixty-one were synchronous OMBC (38.6%) and 97 were metachronous OMBC (61.4%). For the latter, median time to relapse was 60.4 months [range 6.0-482.2], and 28.1% had local relapse at the time of distant oligometastatic relapse. Patients with metachronous OMBC were older than those with the synchronous OMBC (p = 0.003). There was no statistical difference between synchronous and metachronous OMBC for performance status (PS) (p = 0.713) or for menopausal status (p = 0.171). There was also no statistical difference in HR and HER2 status between synchronous and metachronous OMBC (p = 0.603) ( Table 1).
Sixty-six percent of patients underwent 18 F-FDG-PET/ CT or Whole-Body Magnetic Resonance Imaging (WB-MRI) for staging evaluation; others had a CT scan associated with a bone scan. Among patients with liver metastases, 52% had a liver MRI or WB-MRI, as did all patients with brain metastases, who underwent a brain MRI (data not shown).
In the case of metachronous OMBC, median time from diagnosis of primary tumor to oligometastatic relapse was 60.4 months [range 6.0; 487.2]; this time frame correlated with HR and HER2 status and SBR grade (p < 0.001) ( Table 1). Finally, there were no statistical differences for histological subtypes (p = 0.195), SBR grade (p = 0.091), and HR and HER2 status (p = 0.603) between synchronous and metachronous OMBC (Table 1). Data are also available according to SIS subtypes (Online Appendix 1).
The proportion of patients with invasive ductal carcinoma was significantly higher in our cohort (83.4% vs. 75.3%, p = 0.018). There was a larger proportion of

Comparison with the NCR synchronous OMBC cohort
The NCR series is limited to synchronous OMBC. Therefore, we compared this series with our subgroup of synchronous OMBC (Table 4). Due to lacking data, it was also not possible to compare age at diagnosis, PS, SIS, or SBR Grade between our cohort and that of the NCR. The incidence of patients with one to five metastases was 15.8%

Discussion
Our study offers original data concerning OMBC incidence, optimal thresholds for OMBC definition, and considerations regarding OMBC biology and anatomy. In this cohort, incidence is significantly higher than the 1-10% frequently proposed. The latter value is obsolete, as has been previously shown [14]. Our result is closer to more recent data: 19 [26,27], and 31% for HR+ metachronous OMBC [21]. Weichselbaum and Hellman noted that the search for a specific therapeutic strategy for MBC was justified only if a large enough body of patients were concerned. The estimated incidence of OMBC in our study confirms that such an undertaking is indeed worthwhile since OMBC represents 15-20% of all MBC. The proportion of metachronous OMBC in our series is 61.4%. This is to be compared with the results of a recent literature review [28] that found two studies devoted to synchronous and metachronous OMBC. In both of them, the proportion of metachronous OMBC was, respectively, 83% and 87%. We must keep in mind, moreover, that the very concept of OMBC is based on a specific biological profile. Limited metastatic extension is merely the phenotypic consequence. As long as OMBC-specific biomarkers remain unknown, anatomy is the sole indicator that can be used to define OMBC. Therefore, a defining threshold number of organs and metastases must necessarily lead to identifying a subgroup of patients with a homogeneous biological profile. In some publications on OMBC [29][30][31] and MBC [32][33][34], involvement of a single organ is associated with better outcome, but this has not been consistently observed [17,18]. Other authors provide no information on this point [35][36][37][38]. Given this context, should we consider that one of the biological characteristics of OMBC is, firstly, its spread to a single organ? Consequently, should we then include this property in the anatomical definition of OMBC? Such is the position taken by Weichselbaum and Hellman [1,2]. It should be noted that this characteristic also concords with the « seed and soil» theory [39][40][41] as well as with the contemporary view of metastatic diffusion physiology, where successive genomic alterations lead progressively to polymetastatic diffusion [42][43][44][45].
The proportion of patients with one to three metastases is a majority in both our cohort of synchronous OMBC and NCR cohort, respectively, 86.9 and 84.4%. A key result of Steenbrugeen et al. is that overall survival is significantly better for OMBC with one to three metastases than for MBC with more than five metastases, whereas overall survival for patients with four to five metastases does not differ from patients with more than five metastases [17]. Therefore, three metastases as an upper threshold is clinically significant. Consequently, it may be more consistent to apply this threshold in defining OMBC than the value of five applied by ESO-ESMO and ESTRO-ASTRO in defining OMD [11,16].
In our study, the proportion of aggressive BC is higher than in the ESME MBC cohort. The same observation can be made for other cohorts: Selvarajan et al. report 77% of  [19]. These results are somewhat counterintuitive, since OMBC is traditionally described as an indolent disease [11,30,46,47]. The argument that its indolent nature could explain the observed prolonged survival must therefore be rethought.
Finally, OMBC appears to be a heterogeneous entity, both clinically and biologically. As such, it seems difficult to imagine that different subgroups would benefit from the same therapeutic strategy. This consideration questions the design of clinical trials including different subgroups of OMBC.
Nevertheless, we found some discrepancies in the pathological and anatomical features of OMBC in comparing our cohort with that of the NCR. We therefore enlarged our analysis to two other retrospective, non-consecutive series cited above [18,19]. On a first level, inclusion criteria vary widely from one series to the other: number and nature of organs invaded, number and maximum diameters of metastases, feasibility of focal treatments, and inclusion of synchronous and/or metachronous disease. Other factors also influence the number and location of observed metastases and thus the distribution of pathological subtypes: metastasis screening procedures and calculation methods. All of these biases are seen to affect conclusions of OMBC studies. The OLIGOCARE project, initiated by EORTC and ESTRO, will hopefully provide robust data on OMBC [13].
The weaknesses of the present study are quite common: as a retrospective study, data are inevitably incomplete, lacking harmonization or review of imaging or biological examinations. Nevertheless, it is a large and consecutive series. Restricting inclusion criteria to exclusively anatomical data improved the objectivity of data collection. Most importantly, however, we have developed and implemented a rigorous method for calculating metastases.

Conclusion
Our large cohort consists of consecutive patients, untreated at metastatic stage with either synchronous or metachronous OMBC.
OMBC incidence is 15.8% in our cohort and varies from 15.8% to 23.1% in the most recent cohorts where data are available. From a pathological and anatomical point of view, OMBC is a heterogeneous entity; depending on the subgroup, focal ablative treatment of all disease locations may thus not result in improved prognosis. A consequential proportion of OMBC is aggressive BC: the belief that the indolent nature of OMBC could explain observed prolonged survival must therefore be carefully weighed. No significant differences were found in either anatomical or biological characteristics between synchronous and metachronous OMBC. Comparison with other cohorts of OMBC is difficult due to discrepancies in imaging techniques and variability in counting metastases; these points all need to be harmonized.
In conclusion, it must be noted that, since oligometastatic disease has been proposed as a new paradigm, no large and randomized trial has confirmed the benefits of intent-to-cure treatment for OMBC. Furthermore, if the latter is proven, what questions will remain? Do all metastases deserve to be treated, and if so, according to what timeline? What is the optimal adjuvant treatment to implement? Could active surveillance after treating localized breast cancer lead to an increased diagnosis of OMBC, and thus a gain in survival? What is the impact on intent-to-cure strategy for OMBC in terms of quality of life and medico-economic results? The short-term question remains as to whether it is preferable to continue developing randomized studies to prove the benefits of intent-to-cure OMBC treatment or, on the contrary, to "bridge the gap" [48] henceforth, looking for answers to other questions.
Author contributions All authors contributed to the study conception and design. Material preparation, data collection, and analysis were performed by Dr. J-LL, Dr. GG, MBC, and MNM. The first draft of the manuscript was written by Dr. J-LL, and all authors commented on the previous versions of the manuscript. All authors read and approved the final manuscript.
Funding The authors declare that no funds, grants, or other support was received during the preparation of this manuscript.

Data availability
The datasets generated during and/or analyzed during the current study are not publicly available. They include personal data resulting from medical care but are available from the corresponding author on reasonable request.

Declarations
Competing Interests The authors have no relevant financial or nonfinancial interests to disclose.
Ethical approval This retrospective and observational study was approved by our Multidisciplinary Breast Committee and our Institutional Board Committee (BEC-FO-0227).
Consent to participate Informed consent was obtained from all individual participants included in the study.