8 Concordance of Tumor Response with Eribulin Use in Real-World Clinical Practice

Miami Breast Cancer Conference® Abstracts Supplement, 38th Annual Miami Breast Cancer Conference® - Abstracts, Volume 35, Issue suppl 1
Pages: 14

Bruce A. Feinberg, DO1; Jingchuan Zhang, PhD2; Jonathan K. Kish, PhD, MPH1; Shrividya Iyer, PhD2; Sarah S. Mougalian, MD3

1Cardinal Health Specialty Solutions, Dublin, OH

2Eisai Inc., Woodcliff Lake, NJ

3Yale Cancer Center, Yale School of Medicine, New Haven, CT


Although clinical trials in advanced cancer routinely adopt criteria like the Response Evaluation Criteria in Solid Tumors (RECIST) v1.1, response assessment in clinical practice may be subjective; criteria utilized to assess response are not always reported in patient records. We report on the concordance of treatment response with eribulin use in clinical practice, as evaluated through 2 assessment approaches in patients with metastatic breast cancer (mBC).


Adult female patients with mBC, initiating eribulin between 2011 and 2017 consistent with the approved United States indication, underwent chart abstraction by medical oncologists invited to participate. The best response to eribulin (complete response [CR], partial response [PR], stable disease [SD], or progressive disease [PD]) was assessed via 2 methods: 1) provider-reported best response abstracted from the medical record; and 2) response retrospectively calculated per RECIST v1.1 using measurements of target lesions provided by the treating oncologist. The proportion of concordant/discordant responses was compared between the 2 methods.


Data from 513 patients were provided by 46 medical oncologists from across the United States in practices of various sizes (from small community to hospital owned); provider-reported best responses were available for all patients, and lesion measurements were obtained for
499 (97%). Best response rates reported by providers were 8% CR, 46% PR, 17% SD, and 29% PD. Best response rates were 2% CR, 44% PR, 29% SD, and 22% PD by calculation based on lesion measurements and RECIST v1.1 criteria. Concordance of best response classification was 77% overall. Discordant cases observed were mainly due to the classification of provider-reported CR and PR/PD as PR and SD, respectively, using lesion assessments (Table).


Concordance of 77% was observed between the
2 methods used to classify best response in eribulin-treated mBC. Response calculated per established criteria like RECIST using available lesion measurements in addition to provider-reported assessments in real-world studies of mBC could help to optimize the validity of outcomes assessment in clinical practice.