International comparisons of survival after recurrent metastatic breast cancer in four countries: A population-based study.

List of abbreviations
ABCAdvanced Breast Cancer
ASCOAmerican Society of Clinical Oncology
ASNSage-standardized net survival
BCBritish Columbia
DCOdeath certificate only
EMAEuropean Medicines Agency
ESMOEuropean Society for Medical Oncology
ERoestrogen receptor
FDAUnited States food and drug administration
HER2human epidermal growth factor receptor 2
HRhormone receptor
ICD-10International Classification of Diseases, 10th edition
MBCmetastatic breast cancer
MFImetastatic-free interval
NCCNNational Comprehensive Cancer Network
NSnet survival
PBCRspopulation-based cancer registries
PRprogesterone receptor
TNMtumor, node, metastasis
WCPPworld cancer patient population
95% CI95% confidence interval

1
Introduction
Breast cancer is the most common cancer in many countries, with 2.3 million new cases diagnosed and 670,000 deaths reported in 2022 [1]. Survival has improved in recent decades, particularly in high-income countries, where 5-year net survival exceeds 85% in Northern and Western Europe, Australia, and Canada, and reaches 90% in the U.S., most likely as a result of early detection and treatment advances [2]. As survival has increased and fewer people are diagnosed at stage IV, the number of patients living with a history of breast cancer has risen but remains unknown.
In high-income countries, clinical studies indicate that approximately 20-30% of patients diagnosed with early breast cancer will experience a recurrence that is metastatic, with rates varying by stage at diagnosis and biological subtype [[3], [4], [5]]. Despite important therapeutic advances, population-based and clinical trial data showed only modest improvements in survival for metastatic breast cancer (MBC) up to 2015 [[6], [7], [8]]. More recent data suggest that 5-year survival for advanced breast cancer has improved, in the decade 2015-2025, albeit variation by subtype and region remains [9]. Treatment guidelines from leading organizations such as the Advanced Breast Cancer (ABC) Global Alliance, National Comprehensive Cancer Network (NCCN), American Society of Clinical Oncology (ASCO), and the European Society for Medical Oncology (ESMO) emphasize individualized care based on tumour biology, metastatic sites, symptoms, and prior treatments [[10], [11], [12], [13]]. Adherence to these guidelines has been linked to significantly improved survival, better quality of life, and up to a 30% lower risk of death compared to nonadherent care [14,15]. However, survival outcomes could be further influenced by the limited number of studies on MBC and the need for more comprehensive treatment guidelines.
While disparities in de novo MBC have been studied, comprehensive research specifically examining disparities in recurrent MBC across international populations is limited. Comparisons across countries can reveal international variations and potential best practices in MBC management. Furthermore, benchmarking studies can help improve standardization in MBC reporting across countries by identifying differences in data collection methods and definitions.
The aim of this study is to investigate potential disparities in recurrent MBC survival across different population-based registries, considering determinants of survival such as metastasis site, tumour subtype, initial treatment and adherence to treatment guidelines.

2
Methods
2.1
Data sources
At present, information on recurrence is not routinely collected at cancer registries or other population-level data sources and this information is extremely scarce. Five population-based cancer registries (PBCRs) in five countries in North America and Europe confirmed availability of the necessary data. Individual patient-level data were obtained from four population-based cancer registries (PBCR) from four countries, namely Canada (British Columbia), Ireland, the Netherlands and the United States (Connecticut). Data from one country (Norway) were not received due to data sharing restrictions and not included in the current analysis. Data submitted included information on histology, morphology, basis of diagnosis, initial stage at diagnosis and treatment of the initial tumour. Cases with an initial diagnosis of non-metastatic breast cancer between 2005 and 2008 were included. Methods to identify and extract recurrence information varied across registries: In the Netherlands, recurrence data were initially collected to evaluate hospital care quality and were retrospectively extracted from patient files. While in all other registries, data on metastases are not actively sought and are only recorded if they become known in the routine course of registration activities. In the current study, only cases with metastatic recurrence between 2008 and 2010 that occurred more than 120 days after initial non-metastatic diagnosis were included and followed up until December 2015. In line with expert recommendations, recurrent MBC was operationally defined as any metastatic diagnosis occurring more than four months after an initial stage I-III breast cancer diagnosis [16].
Information on tumour-node-metastasis (TNM) stage at initial diagnosis was provided by the cancer registry. Additionally, patient-level data on age at initial and metastatic diagnosis, vital status at the end of follow-up, hormone-receptor (HR)-status and human epidermal growth factor receptor 2 (HER2) of the primary breast tumour, site of first metastasis and treatment at initial diagnosis were provided. The metastatic-free interval (MFI) defined as the time between the initial diagnosis date and date of metastatic recurrence was categorized into three groups: less than 2 years, 2 to 5 years, and more than 5 years.
Primary malignant breast tumours (ICD 10: C50) were identified (n = 84,224). Cases diagnosed on the basis of death certificate only (DCO) or at autopsy, below the age of 15 or above 99 years at metastatic diagnosis, male breast cancer and second or higher sequenced cancer diagnosed at the same site were excluded. Of the identified cases, only those with a recorded metastatic recurrence (n = 2735) during the study timeframe were included in the analyses (Fig. S1). Breast cancer subtype was derived based on the oestrogen receptor (ER) and progesterone receptor (PR) status (HR + if ER and/or PR positive) and HER2 status, with HER2 status taking precedence: all tumours that were HER2+ were classified as HER2+, irrespective of HR status.

2.2
Data cleaning/quality control
The data was subjected to several data quality assessments and scrutinized for specific anomalies. Discrepancies such as out-of-range dates of diagnosis or dates of death, and/or invalid vital status codes and inconsistencies in terms of morphology, age, and site were flagged and manually evaluated.

2.3
Statistical analysis
Median overall survival, from the date of recurrence, stratified by the subtype using Kaplan-Meier curves for each registry were calculated. The log-rank test was used to assess whether median overall survival differences were significant between registries within each subtype at a p-value of <0.05 indicating significance.
Net-survival (NS) estimates and corresponding 95% confidence intervals (CI) were estimated and defined as the probability of surviving cancer in the absence of other causes of death, i.e., the only cause of death is cancer [17]. To account for variation in background mortality across population, country-specific life tables were used.
One-, three- and five-year NS using the Pohar-Perme estimator for patients with a metastatic recurrence during 2008-2010 were estimated, with follow-up until December 31, 2015. Patients were followed from the date of metastatic recurrence until death, loss to follow up, or a maximum of five years, whichever occurred first. NS by subtype was estimated using the cohort approach. Survival estimates were age-standardized (ASNS) using the world cancer patient population (WCPP) weights [18]. All analysis was performed using Stata 17.0.

2.4
Sensitivity analysis
To address the heterogeneity in recurrence data collection across cancer registries, we conducted a sensitivity analysis. Given the absence of standardized protocols and the varying purposes and methods of data collection we aimed to assess the robustness of our survival estimates under more consistent follow-up conditions. For this sensitivity analysis, we examined survival in patients with a primary diagnosis between 2005 and 2012 and a recorded metastatic recurrence between 2012 and 2014 with a follow-up until December 31, 2019. This allowed for a more uniform follow-up period across registries.
To account for potential bias introduced by differences in MFI—where shorter MFIs are often associated with more aggressive disease and poorer prognosis—we conducted a sensitivity analysis stratified by MFI categories.

3
Results
3.1
Patient characteristics
The characteristics of patients with recurrent MBC at the time of initial non-metastatic diagnosis are presented in Table 1. Across all registries, 2735 women with a metastatic recurrence recorded during 2008-2010 were included (Canada (British Columbia) [n = 479]; Ireland [n = 480]; the Netherlands [n = 1687]; United States (Connecticut) [n = 89]). Mean age at initial diagnosis was similar across registries (approximately 56–59 years), while the proportion initially diagnosed with early-stage disease was lowest in Canada (BC) (49.4%) and highest in the United States (Connecticut) (89.9%), with Canada (BC) also having the largest proportion of unknown stage (18.3%). Treatment patterns at initial non-metastatic diagnosis varied: use of chemotherapy was highest in Ireland (68.8%) and the United States (68.5%) and lower in the Netherlands (55.9%), radiotherapy use was highest in Canada (BC) (88.1%) and lowest in the United States (50.6%), and endocrine therapy was most frequently recorded in the Netherlands (52.2%), with subtype-specific differences in endocrine and chemotherapy use observed across registries (Table S1).
Patient characteristics at the time of metastatic diagnosis are presented in Table 2. Mean age at metastatic diagnosis was again comparable (about 58–61 years), metastatic-free interval distributions differed modestly, and first site of metastasis varied, with bone-only metastases as the most common initial site in Canada (BC) and Ireland (32.6% and 26.0% of cases, respectively), and multiple sites predominating in the Netherlands (33.1%) and particularly in the United States, Connecticut (94.4%).

3.2
Median overall survival
Fig. 1 and Table S2 presents observed survival estimates for all subtypes combined (Fig. 1A) and stratified by subtype (Fig. 1B–D). For all subtypes combined, median overall survival from metastatic recurrence ranged from 12 months in Ireland to 18 months in the United States (Connecticut), with evidence of heterogeneity between registries (log-rank p = 0.015). By subtype, median survival was longest in the Netherlands for HR+/HER2− (21 vs 13.2 months in Ireland) and HER2+ disease (17.1 vs 10 months in Canada (BC)), whereas survival for HR−/HER2− disease was similar across countries (about 8.5–8.9 months).

3.3
Net survival – all subtypes combined
ASNS estimates are shown in Fig. 2A–C and Table S2. One-year ASNS from metastatic breast cancer recurrence for all subtypes combined ranged from 51% in Ireland to 63% in the United States (Connecticut) and the Netherlands; differences at 3- and 5-years persisted but were less marked.

3.4
Net survival – by subtype
ASNS estimates stratified by the three main subtypes are presented in Fig. 2A–C and Table S2. One-year ASNS for patients with HR+/HER2-tumours was highest in the Netherlands at 67.5% (95% CI: 64.4–70.4%) and lowest in Ireland at 54.1% (95% CI: 47.1–60.5%), with Canada (BC) and the United States (Connecticut) in between. Three- and five-year ASNS estimates for patients with these tumours were similar across registries, with the highest value in the Netherlands at 29.3% (95% CI: 26.4–32.2%), and five-year net survival ranging from 15.6% (95% CI: 11.0–21.0%) in Ireland to 14.9% (95% CI: 12.6–17.3%) in the Netherlands.
Among patients with HER2+ tumours, 1-year ASNS was highest in the Netherlands at 61.3% (95% CI: 56.0–66.1%) and lowest in Canada (BC) at 44.8% (95% CI: 34.6–54.5%), with Ireland and the United States (Connecticut) showing intermediate values. Three-year ASNS survival was relatively similar between registries, with the highest estimate in Ireland at 22.9% (95% CI: 15.2–31.6%) and the lowest in Canada (BC) at 18.9% (95% CI: 11.6–27.6%), while five-year ASNS ranged from 8.5% (95% CI: 3.9–15.3%) in Canada (BC) to 15.1% (95% CI: 8.8–22.9%) in Ireland.
For patients with HR-/HER2-tumours, 1-year ASNS ranged from 37.9% (95% CI: 28.2–47.5%) in Ireland to 44.8% (95% CI: 34.7–54.5%) in Canada (BC); 3-year ASNS ranged from 8.3% (95% CI: 4.6–13.4%) in the Netherlands to 14.8% (95% CI: 8.5–22.8%) in Canada (BC), and 5-year ASNS was largely similar across countries, ranging from 3.1% (95% CI: 1.5–5.6%) in the Netherlands to 7.6% (95% CI: 3.3–14.1%) in both Canada (BC) and Ireland.

3.5
Sensitivity analysis by metastatic-free interval
ASNS estimates stratified by MFI are presented in Fig. S2. For all subtypes combined, ASNS consistently increased with longer time between initial diagnosis and metastatic recurrence, although the size of between-registry differences varied by interval and time from recurrence. For patients with recurrence within 2 years, 1-year ASNS was fairly similar across registries, ranging from 48.1% (95% CI: 40.6–55.3%) in Canada (BC) to 57.8% (95% CI: 38.0–73.2%) in the United States (Connecticut); 3-year ASNS ranged from 13.5% (95% CI: 10.6–16.8%) in the Netherlands to 27.5% (95% CI: 12.9–44.4%) in the United States, and 5-year ASNS from 6.7% (95% CI: 4.7–9.3%) in the Netherlands to 13.9% (95% CI: 9.2–19.5%) in Ireland.
In the 2–4 year metastatic-free interval group, 1-year ASNS ranged from 48.3% (95% CI: 41.9–54.5%) in Canada (BC) to 64.7% (95% CI: 61.5–67.7%) in the Netherlands; 3-year ASNS from 20.9% (95% CI: 15.9–26.3%) in Ireland to 33.6% (95% CI: 20.6–47.1%) in the United States, and 5-year ASNS from 11.0% (95% CI: 6.8–16.4%) in Canada (BC) to 19.5% (95% CI: 9.6–32.0%) in the United States.
Survival was consistently highest in the Netherlands for patients with metastatic recurrence occurring more than 4 years after their initial diagnosis: 1-year ASNS ranged from 63.7% (95% CI: 49.7–74.8%) in Ireland to 69.0% (95% CI: 62.6–74.5%) in the Netherlands,; 3-year ASNS ranged from 16.1% (95% CI: 7.9–26.9%) in Ireland to 27.5% (95% CI: 21.9–33.5%) in the Netherlands, and 5-year ASNS from 10.2% (95% CI: 4.2–19.5%) in Canada (BC) to 16.9% (95% CI: 12.3–22.2%) in the Netherlands.

3.6
Sensitivity analysis by recurrence period
When analyses were restricted to women with metastatic recurrence between 2012 and 2014, ASNS estimates increased across most registries and subtypes, and some between-country contrasts became more pronounced (Fig. S3). For all subtypes combined, 1-year ASNS was highest in the Netherlands at 80.5% (95% CI: 74.5–85.3%) and lowest in Canada (BC) at 63.9% (95% CI: 60.4–67.2%); 3-year ASNS similarly ranged from 45.2% (95% CI: 38.3–51.9%) in the Netherlands to 29.2% (95% CI: 25.7–32.8%) in Canada (BC), while 5-year ASNS was highest in the United States (Connecticut) at 29.4% (95% CI: 22.7–36.4%) and lowest in Canada (BC) at 16.5% (95% CI: 13.7–19.4%).
For HR+/HER2-disease, 1-year ASNS reached 84.0% (95% CI: 72.4–91.0%) in the United States (Connecticut) and 65.2% (95% CI: 60.8–69.2%) in Canada (BC), with the Netherlands showing similarly high one-year survival to the United States; 3-year ASNS was highest in the Netherlands at 47.2% (95% CI: 40.9–53.1%) and lowest in Canada (BC) at 32.6% (95% CI: 28.4–36.9%), and 5-year ASNS ranged from 16.3% (95% CI: 12.9–20.0%) in Canada (BC) to 27.2% (95% CI: 21.8–32.8%) in the Netherlands.
For patients with HER2+ tumours, 1-year ASNS was 87.5% (95% CI: 73.4–94.4%) in the Netherlands versus 59.9% (95% CI: 51.7–67.1%) in Canada (BC); 3-year ASNS ranged from 26.9% (95% CI: 20.1–34.1%) in Canada (BC) to 44.1% (95% CI: 29.3–57.8%) in the Netherlands, and 5-year ASNS from 17.7% (95% CI: 11.9–24.4%) in Canada (BC) to 26.5% (95% CI: 19.2–34.4%) in Ireland.
One-, three- and five-year ASNS values in analyses of HR-/HER2-tumours (where feasible) were similar to the main analysis.

4
Discussion
This study is the first to compare survival outcomes for patients with recurrent MBC across four population-based cancer registries in Canada, Ireland, the Netherlands and the United States. In the absence of a standardized protocol for collecting recurrence data, registries currently differ in both their data collection methods and underlying objectives. Our study highlights the need for more standardised definitions of metastatic recurrence and guidelines for harmonised collection of key clinical variables across registries to enable robust comparisons of outcomes and care. The recent ABC Global Decade Report (2015-2025) which aims to further drive progress and address critical gaps in people with advanced breast cancer, outlined the collection of high-quality data as one of the key goals to optimize care and outcomes in patients with ABC [9]. Across countries, the largest between-registry differences were seen in one-year net survival, particularly for HR+/HER2− disease, where survival was highest in the Netherlands, and these patterns were broadly consistent but more pronounced in sensitivity analyses using later recurrence periods and stratification by metastatic-free interval. Within registries, survival gradients by subtype were generally uniform—with the most favourable outcomes in HR+/HER2− and the least favourable outcomes in HR−/HER2− disease—except in Canada, where women with HER2+ recurrence experienced similarly low survival to those with HR−/HER2− disease, suggesting potential differences in access to or effectiveness of subtype-specific systemic therapies.
Several population-based and hospital-based studies have reported substantial international variation in survival outcomes, in patients with both de novo and recurrent metastatic breast cancer [[19], [20], [21]], with consistently lower survival for HR-/HER2-disease and improvements over time in HR+/HER2-and HER2+ subtypes which has been attributed to the introduction of endocrine and targeted agents such as CDK4/6 inhibitors and HER2-targeted therapies in recent years [6,8,22]. Our net survival estimates fall within the lower range of values reported for mixed metastatic cohorts, which likely reflects our focus on recurrent disease and an earlier diagnostic period, but the patterns by subtype observed are broadly consistent with previous studies [6,22,23]. Although some population-based studies have examined de novo metastatic breast cancer [6,8,21,22,[24], [25], [26]] and several hospital-based cohorts have reported on recurrent disease [25,[27], [28], [29], [30], [31], [32]], prior comparative research has rarely differentiated these groups, and we found no international comparative studies evaluating recurrent MBC survival. Our findings therefore extend the existing literature by showing that between-country differences are most pronounced in one-year net survival and persist after accounting for metastatic-free interval and calendar period.
This study underscores the methodological challenges of studying recurrent MBC using routine registry data. In the absence of international guidelines, participating registries differed in the collection of information on metastatic recurrence (retrospective case-finding, routine linkage, or passive recording) and in the completeness of key variables such as receptor status and systemic treatment. These differences may influence both the case mix (for example, preferential capture of more symptomatic or advanced recurrences) and therein survival estimates, limiting comparability between countries and over time [[33], [34], [35]]. Harmonised definitions of metastatic recurrence, minimum data items (including subtype at recurrence and systemic therapies), and standardised follow-up procedures are therefore essential to enable robust international benchmarking and monitor the impact of evolving treatments in this patient group, in line with emerging international recommendations on long-term outcome and recurrence recording in population-based cancer registries [9,16,34,36].
The consistent pattern of highest survival among women with HR+/HER2-and HER2+ recurrence and lowest among those with HR-/HER2-disease aligns with current understanding of tumour biology and treatment responsiveness [37,38]. In our study, the Netherlands showed particularly favourable 1-year ASNS for HR+/HER2-and HER2+ disease, with almost 15 percentage points higher HR+/HER2-survival than the same patient group in Ireland. This may reflect earlier and wider use of endocrine therapy and HER2-targeted agents, more systematic multidisciplinary care, or more complete capture of recurrent cases [37,39,40]. In contrast, Ireland had lower survival in HR+/HER2− disease, despite broadly similar approval timelines for key endocrine agents such as tamoxifen, anastrozole and letrozole across Europe and North America [37,[41], [42], [43], [44], [45]], and consistent recommendations that endocrine therapy should be the preferred first-line treatment in HR-/HER2+ disease. These patterns coincide with approval timelines for key targeted agents: CDK4/6 inhibitors (palbociclib approved Feb 2015 FDA accelerated/2017 full; Nov 2016 EMA), and HER2-targeted therapies (trastuzumab 1998 FDA/2000 EMA; pertuzumab June 2012 FDA/March 2013 EMA; T-DM1 Feb 2013 FDA/Sept 2013 EMA) [[46], [47], [48], [49], [50], [51], [52]]. Furthermore, we observed that a higher proportion of patients with HR+/HER2− tumours in the Netherlands received endocrine therapy compared to Ireland where higher proportions of chemotherapy were observed. While this could suggest variation in treatment practices, the observed lower survival for HR+/HER2− patients in Ireland compared to the Netherlands may also be influenced by other factors, such as incomplete or inconsistently recorded data on receptor status or systemic treatments, or by differences in how metastatic recurrence is captured across registries. For HER2+ recurrence, survival was similar in Ireland and the Netherlands but lower in Canada and could be related to provincial variation in access to HER2-targeted therapies and delays in public funding for newer agents [53,54]. This aligns with reported provincial delays in Canada for HER2 agents like pertuzumab (funded ∼2014-2018 varying by province, e.g., BC ∼2015-2016; Ontario later) and T-DM1 (∼2018+, post-pCODR 2015 recommendation), compared to earlier national reimbursement in the Netherlands (pertuzumab Jul 2013, T-DM1 Jun 2014) [[55], [56], [57]]. In cases where therapies are not covered by public insurance, patients may rely on private insurance, out-of-pocket payments, or patient access programs, creating financial burdens, especially for those without adequate coverage [39]. More granular linkage of registry, pharmacy, and clinical data will be needed to clarify whether these survival differences primarily reflect care delivery or residual data artefacts.
Sensitivity analyses supported the robustness, but also highlighted nuances, of our findings. When stratified by MFI, 1-year ASNS increased with longer intervals in all registries, consistent with shorter intervals reflecting more aggressive biology [58]. For example, across registries, 1-year ASNS was nearly 20 percentage points lower for patients with MFI <2 years than for patients with MFI >4 years. The Netherlands and the United States consistently lay at the upper end of the survival range within each MFI stratum, with differences of around 5–10 percentage points compared with Ireland and Canada in the 2–4 and >4-year groups, indicating that health system-level factors likely contribute beyond tumour biology alone. Restricting analyses to more recent recurrence years (2012-2014) yielded higher absolute net survival across most subtypes and registries, accentuating the advantages of the Netherlands in the HR+/HER2-and HER2+ disease, indicating that international differences in timely uptake and delivery of modern therapies may have widened over time rather than narrowed, as confirmed in the ABC Global Decade Report 2015-2025 [8,9,22,59]. In HR+/HER2− disease, 1-year ASNS in 2012–2014 increased by around 5–8 percentage points compared with earlier recurrence years in all registries. These patterns, together with the persistent survival ranking by country across MFI strata, support the hypothesis that differences in the timing and implementation of new systemic therapies and organised multidisciplinary care contribute to international survival gaps beyond differences in tumour biology or follow-up duration.
This international study across four high-income countries holds several strengths including its population-based design, large sample size for recurrent MBC and detailed clinical information that allow for in-depth investigations of MBC survival to be investigated. However, several limitations must be acknowledged. Subtype classification was based on receptor status of the tumour at initial diagnosis and assumed to be the same at recurrence. The lack of consistency in the methods used by cancer registries to capture recurrence mean that interpretation of results should be made with caution. Treatment data were incomplete and lacked detail on lines and sequences of systemic therapy. The high proportion of multiple sites of metastasis reported in the US (94%) compared to other registries suggests that differences across registries in this data item could be due to differences in definitions rather than true variation. Some strata, particularly HER2+ and HR-/HER2-in smaller registries, contained few events, leading to imprecise estimates. These factors may have introduced selection and information bias that may impact some of the true differences between and within countries.
Taken together, these findings have important implications for research, practice, and policy. Internationally, there is a pressing need for registries to systematically record stage at diagnosis, recurrence and key biomarkers, and to link with clinical information, so that recurrent MBC can be monitored with the same rigour as incident disease. At the health-system level, the large between-country gaps in one-year net survival and the persistent disadvantage for triple-negative disease highlight the importance of timely access to evidence-based systemic therapies, multidisciplinary management, and reduction of financial and organisational barriers to care. Future work should prioritise prospective, harmonised registry modules for recurrent MBC, evaluation of care pathways and treatment uptake across health systems, and inclusion of under-represented settings, to reduce international disparities and improve outcomes for patients living with metastatic breast cancer.

CRediT authorship contribution statement
Hanna Fink: Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing – original draft. Isabelle Soerjomataram: Supervision, Writing – review & editing. Aude Bardot: Methodology, Software, Validation, Writing – review & editing. Aline Brennan: Data curation, Resources, Writing – review & editing. Ryan R. Woods: Data curation, Resources, Writing – review & editing. Lou Gonsalves: Data curation, Resources, Writing – review & editing. Jan F. Nygård: Writing – review & editing. Serban Negoita: Methodology, Writing – review & editing. Esmeralda Ramirez-Pena: Methodology, Writing – review & editing. Karen Gelmon: Methodology, Writing – review & editing. Sabine Siesling: Data curation, Methodology, Writing – review & editing. Fatima Cardoso: Investigation, Methodology, Writing – review & editing. Julie Gralow: Investigation, Methodology, Writing – review & editing. Eileen Morgan: Conceptualization, Data curation, Funding acquisition, Supervision, Writing – review & editing.

Ethical approval
Not required.

Disclaimer and acknowledgements
Where authors are identified as personnel of the International Agency for Research on Cancer/World Health Organization, the authors alone are responsible for the views expressed in this article, and they do not necessarily represent the decisions, policy, or views of the International Agency for Research on Cancer/World Health Organization. We would like to thank Dr Melina Arnold for conceptualising the initial idea for this project and for securing the funding that enabled this work to begin. The authors would like to acknowledge the cancer registries involved for maintaining and collecting information and in particular all tumour registrars for the collection and completion of any additional information for this current study.

Funding
This work was supported by the Susan G. Komen Foundation (Career Catalyst Grant CCR19608129 to EM). The funder had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Declaration of competing interest
Given Dr. Fatima Cardoso, Dr. Karen Gelmon and Dr. Sabine Siesling roles as Editor-in-Chief, Deputy Editor and Specialty Editor: Epidemiology, respectively, Dr. Cardoso, Dr. Gelmon and Dr. Siesling had no involvement in the peer-review of this article and have no access to information regarding its peer review. The authors declare that they have no other conflict of interest.