Invasive breast cancer: Stratification of histological grade by gene-based assays: A still relevant example from an older data set

Abstract

Aims: A Netherlands Kanker Institute data set provided the results of gene-based assays (GBAs) and histological grades of 295 patients with invasive breast cancer. Grade is the first prognostic assay available after a cancer diagnosis. Given this time-line of actual practise, the aim was to study how gene-based assays further stratify histologic grade. Methods and results: Emphasis was placed on evaluation of a simple decision tree and on study of the recurrence score (RS). The decision tree determined three risk stratifications. Tumours that were both intermediate grade (IG) and low-RS were grouped with low grade, and tumours that were IG and high-RS were coupled with high grade. IG and intermediate-RS tumours comprised the third category. Survival analysis was performed with respect to the three stratifications. Cramer's V statistic was used for concordance analysis. The mixed grade-RS classifier showed significant survival stratification (P < 0.00001). The mixed classifier was concordant with the 70-gene assay (Cramer's V = 0.57). Recurrence score alone had a 0.59 Cramer's V with the gene assay. Because two-thirds of tumours were of either low or high grade, concordance was maintained despite the majority of classifications having been determined by grade alone. Conclusion: There is no compelling reason to test low- and high-grade tumours further by GBAs.