Real world treatment sequencing patterns in secondary breast cancer (SBC): Pathway visualisation using national datasets.
Session type: Poster / e-Poster / Silent Theatre session
Treatment pathways in metastatic breast cancer are complex. The accelerated adoption of new medicines has resulted in an uncertain evidence base supporting their use. Uncertainties are related to the mismatch between trial-recruited and real-world populations and variation in the order of sequential drugs.
Published examples describing real-world practice in SBC are scarce, mainly due to the complexity of the clinical pathways that rely on a mixture of chemotherapy, endocrine therapy and biologicals, often over a long period. We demonstrate how new opportunities in routine healthcare data allow a highly granular description of real-world treatment pathways and how this varies in light of patient (pt) case-mix.
Scottish nationally available data source datasets for linkage included the National Cancer Registry, Scottish Morbidity Record, the National Cancer Quality Audit and the national Prescribing Information System. Scottish CHI number was the universal identifier for linkage. Key baseline characteristics included age, de-novo presentation, prior adjuvant treatments, co-morbidities, concomitant medications and socioeconomic status. Targeted and random sampling manual review was used to quantify missing data. R version 3.6 was used for analysis.
345 pts were identified of which 276 had ER+HER2- SBC between 2012-2017. First line therapy included 68% (235 patients) endocrine therapy, 17% (59 pts) chemotherapy, 14% (50 pts) received no treatment. Subsequent treatment decisions, including best supportive care and death, have been tracked to identify 70 unique pathways with up to 8 lines of treatment. Graphical representation of treatment pathways is made using Sankey plots. Detailed data quality reports describe missing data rates over time and a comprehensive guide for analysts has been produced as a wiki [https://blogs.ed.ac.uk/canceroutcomes/edinburgh-cancer-informatics-wiki/].
It is now possible to describe treatment sequences using routine, nationally available administrative healthcare data. Pathways are complex and do not always conform to standard guidelines. Interpretation requires modern graphical visualisation methods.