Cancer registries - guardians of breast cancer biomarker information: A systematic review

Background: Breast cancer is the most common cancer and the leading cause of cancer-related death in females, with a large societal and economic impact. Decisions regarding its treatment are largely affected by the categorization into different subtypes with hormone receptor status and HER2 status being the most important predictive factors. Other biological markers play an important role for prognostic and predictive reasons. The data collection and harmonization of cancer cases are performed by cancer registries whose collection of parameters largely differs, partially including results from biomarker testing. Methods: This systematic literature review consisting of a total of 729 reports determined whether information about biomarker testing in breast cancer cases is collected and published by cancer registries worldwide. Results: The number of publications using breast cancer biomarker data from registries steeply rose with the beginning of the 21st century and some hospital-based and population-based cancer registries reacted with immediate collection of biomarker data following the recommendation of clinical guidelines. For female breast cancer, biomarkers have achieved an essential clinical value and this review points to a steady increase in the collection of biomarker data by cancer registries during the last decade. Conclusions: In the future, recommendations for biomarker data collection and coding by cancer registries may be required to ensure harmonization and comparability of the data.


Supplementary Note SN1
BC is the most common disease and the leading cause of cancer-related death of women worldwide [1][2][3]. During their lives, one out of eight to nine women is diagnosed with BC. While the incidence of BC has continuously increased in almost all European countries over the last decades, BC-related mortality has decreased. This is primarily influenced by mammographic screening programs, as the prognosis of the disease is dependent on early detection [4][5][6], and by improved treatment opti ons.
Being a multifactorial disease, most BC cases involve a multitude of environmental factors. Onl y 5-10% of BC cases are related to a genetic predisposition, most of which can be associated with mutations in the tumour suppressor genes BRCA1 and BRCA2 [7]. Other predisposing genes i ncl ude p53, ATM, PTEN, CDH1, CHEK2, PALB2, or STK11. Besides these genes which can be used for the screening of hereditary breast cancer, other DNA sequences, RNAs, proteins, or cells can be considered as biomarkers.

Supplementary Note SN3
A total of 729 publications were selected through a PRISM-based selection process for systematic review (Fig. A.1). The publications varied in the number of patients included in the analysis, rangi ng from 10 to 34 million patients (Fig. A.2). For eight of the publications the number of patients was not available. For the remaining 721 publications the average number of patients was 105101. However, half of the publications used data of less than 3712 patients and only 25% of the publications exceeded a number of 21751 patients.

Supplementary Note SN4
The review of recent clinical guidelines for BC identified the recommendation for te sti ng e strogen   and progesterone receptors as well as HER2 in the primary tumour for all newly diagnosed BC patients and, where feasible, on recurrent lesions [10,11,[24][25][26]. Ki-67 may additionally be used as a prognostic marker. Additional biomarkers are recommended for testing under specific circumstances, e.g. if patients are lymph-node negative, HER2 negative, and ER positive, uPA, PAI-1, or different multigene expression assays can be used for treatment decisions and prognosis or CEA, CA15.3, and CA27.29 may be used for early detection of recurrence [27]. Genetic testing for BRCA1/2 or using multigene tests, on the other hand, is only recommended for patients with a hi gh risk of familial predispositions [28,29].

Supplementary Note SN5
Biomarker data are analysed for prognostic or predictive reasons and are used for a variety of purposes. Each of the publications was therefore associated with one to three of the following categories: cancer risk, patient characteristics/incidence, staging, diagnosis in advanced stage, risk of recurrence or metastases, subtype differentiation, survival/mortality prognosis, and tre atme nt. An association was made when the data was differentiated into subcategories or displayed in re l ati on to one of the parameters. Differentiation into subtypes plays an important role for prognosti c and predictive purposes and was used in 551 publications, 192 of which then used the categorization into different subtypes for the prediction of survival and mortality and in 207 publications the biomarker-based subtype status was compared to or utilized for treatment decisions (Fig. A.7). 170 of the publications displayed the biomarker status as a patient characteristic or to differentiate incidence data into subgroups. The risk for a second primary tumour or the risk of recurrence or metastases was predicted by biomarker status in 103 and 37 publications, respectively. Few publications used biomarkers for staging (19) or diagnosis in advanced stage (3).

Review of clinical guidelines
In a first step, national and international clinical guidelines were reviewed to select the relevant biomarkers recommended for testing in BC cases. These guidelines included recommendati ons by "biological marker*" OR "tumor marker*" OR "tumour marker*" OR receptor* OR "gene expression") AND ("breast cancer*" OR "breast tumor*" OR "breast tumour*" OR "breast neoplasm*" OR "breast carcinoma*" OR "mamma carcinoma*") AND ("cancer registr*" OR "tumor registr*" OR "tumour registr*" OR "oncological registr*" OR "oncology registr*")). And the search string for the Cochrane library in title, abstracts, keywords was: (biomarker* OR "molecular marker*" OR "signature molecule*" OR "biological marker*" OR "tumor marker*" OR "tumour marker*" OR receptor* OR "gene expression") AND ("breast cancer*" OR "breast tumor*" OR "breast tumour*" OR "breast neoplasm*" OR "breast carcinoma*" OR "mamma carcinoma*") AND ("cancer registr*" OR "tumor registr*" OR "tumour registr*" OR "oncological registr*" OR "oncology registr*"). Few additional publications found when searching OpenGrey, PubMed NLM, and PubMed MeSH were added. The literature search was performed on March 20 th , 2017. We did not limit the search with regard to publication year, however, for temporal analyses the publications were displayed for the years fully covered only (excluding 2017). While the search string was limiting the ke ywords to the English language, publications retrieved from the search were not further selected wi th re spect to language.
In total, 1047 publications were retrieved. One author filtered their abstract or full text according to systematic review guidelines [30] to ensure that the relevant information on BC biomarkers from cancer registries was included in the study (Fig. A.1). In unclear cases, a second author was consulted and a common decision was made. 318 publications were excluded from the final analysis as they either did not contain data on BC (23), biomarkers (78), or the biomarker data were not extracted from cancer registry data (149). Some publications combined several of the excluding reasons. 98 additional publications were excluded as they were reviews or other types of articles without direct data use by CRs, leaving 729 studies for final analysis. Information on the publicati on year, the cancer registries and countries involved in data collection, the biomarker data col l e cte d, and the use of the biomarker data was extracted from the selected studies.
To avoid a possible bias due to the use of the words 'receptor' and 'gene expression' the search was repeated without these terms (Fig. A.8A). A total of 91 publications were selected for anal ysi s. The search resulted in a similar pattern of the biomarker data use with ER and PgR receptors being use d in the majority of publications (81 and 59, respectively) ( Fig. A.8B). HER2 appeared in 49 publications and Ki-67, CEA, CA15.3, CA27.29, BRCA1, BRCA2, and gene expression assays were utilized in less than ten publications, while no publications were found containing data on p53, uPA, or PAI-1. Figure A.1 Fl ow diagram of the systematic literature review. A total of 1047 publications were found for the search s tri ng conta ining keywords regarding biomarkers, breast ca ncer, a nd ca ncer registries (Supplementary Methods) until Ma rch 20 th 2017, 729 of whi ch were s elected for fi nal analysis. 318 publications were excluded from the analysis due to missin g d a ta on breast ca ncer or biomarkers, data not originating from ca ncer registries or missing direct data use.      The purposes were ca tegorized into the following groups: risk of developing ca ncer, b i om a rker s ta tus as a cha ra cteristic of the ca ncer or in combination with calculating the incidence ra te, breast cancer diagnosis in the a dva n ced s ta ge, differentiation of the ca ncer i nto subtypes, survival and mortality ra tes, the risk of recurrence or m e ta sta s es , a nd thera py decisions. The biomarkers were associated with these purposes either when they underwent s ubca teg ori z a ti on ba s ed on the data or when the data were displayed in relation to one of the parameters.

Supplementary Tables
Supplementary