Joensuu, Heikki

Abstract The discovery of molecular subtypes has corroborated the heterogeneity of breast cancers. But in clinical routine, treatment selection relies on measuring the aggressiveness of the tumor via histopathology, which is routinely based on grading, a prognostic yet laborious, poorly reproducible, and subjective procedure defined more than 30 years ago, before molecular subtypes existed. In this paper, we present an interpretable deep learning-based computer-aided diagnosis method to automate mitotic count, one of the components of grading; the method complies with current guidelines and automates hot-spot finding on full digital whole-slide images of routinely prepared pathologic slides stained with hematoxylin and eosin. We use the computer model to query the current “one-size-fits-all” mitotic count procedure by investigating subtype-specific prognostic value of mitotic count using an external independent large-scale multicentric dataset of 1,709 breast cancers. We show that for human epidermal growth factor-2 (HER2) positive tumors, mitotic count was not prognostic. In hormone-receptor positive/HER2 negative cancers, an average mitotic count density different from clinically used cut offs predicted worse recurrence free survival in multivariable analysis, adjusted for known clinical prognosticators. We provided some first experimental evidence that the prognostic role of mitotic count should be reviewed.

Abstract Purpose: Recent studies have demonstrated a benefit of adjuvant capecitabine in early breast cancer, particularly in patients with triple-negative breast cancer (TNBC). However, TNBC is heterogeneous and more precise predictive biomarkers are needed. Experimental Design: Tumor tissues collected from TNBC patients in the FinXX trial, randomized to adjuvant anthracycline–taxane–based chemotherapy with or without capecitabine, were analyzed using a 770-gene panel targeting multiple biological mechanisms and additional 30-custom genes related to capecitabine metabolism. Hypothesis-generating exploratory analyses were performed to assess biomarker expression in relation to treatment effect using the Cox regression model and interaction tests adjusted for multiplicity. Results: One hundred eleven TNBC samples were evaluable (57 without capecitabine and 54 with capecitabine). The median follow-up was 10.2 years. Multivariate analysis showed significant improvement in recurrence-free survival (RFS) favoring capecitabine in four biologically important genes and metagenes, including cytotoxic cells [hazard ratio (HR) = 0.38; 95% confidence intervals (CI), 0.16–0.86, P-interaction = 0.01], endothelial (HR = 0.67; 95% CI, 0.20–2.22, P-interaction = 0.02), mast cells (HR = 0.78; 95% CI, 0.49–1.27, P-interaction = 0.04), and PDL2 (HR = 0.31; 95% CI, 0.12–0.81, P-interaction = 0.03). Furthermore, we identified 38 single genes that were significantly associated with capecitabine benefit, and these were dominated by immune response pathway and enzymes involved in activating capecitabine to fluorouracil, including TYMP. However, these results were not significant when adjusted for multiple testing. Conclusions: Genes and metagenes related to antitumor immunity, immune response, and capecitabine activation could identify TNBC patients who are more likely to benefit from adjuvant capecitabine. Given the reduced power to observe significant findings when correcting for multiplicity, our findings provide the basis for future hypothesis-testing validation studies on larger clinical trials.

Purpose We compared efficacy of trastuzumab versus no trastuzumab in patients with small (≤ 2 cm) human epidermal growth factor receptor 2 (HER2) –positive breast cancer treated in randomized trials. Methods A meta-analysis was conducted using data from five of the six adjuvant trastuzumab trials. Efficacy end points were disease-free survival (DFS) and overall survival (OS). Separate analyses were prospectively planned for hormone receptor (HR) –positive and HR-negative cohorts. Random effect models and Yusuf-Peto fixed effects models assessed the impact of heterogeneity on baseline hazards and treatment effects across studies. Peto-Pike cumulative incidence estimates were stratified by study and nodal status. Results Median follow-up time was 8 years. For 2,263 patients with HR-positive disease, 8-year cumulative incidence rates comparing trastuzumab versus no trastuzumab were 17.3% versus 24.3% (P < .001) for DFS and 7.8% versus 11.6% (P = .005) for OS, respectively; for 1,092 HR-positive patients with zero or one positive lymph nodes, results were 12.7% versus 19.4% (P = .005) for DFS and 5.3% versus 7.4% (P = .12) for OS, respectively. For 1,957 patients with HR-negative disease, 8-year cumulative incidence rates were 24.0% versus 33.4% (P < .001) for DFS and 12.4% versus 21.2% (P < .001) for OS, respectively; for 1,040 HR-negative patients with zero or one positive lymph nodes, results were 20.4% versus 26.3% (P = .05) for DFS and 8.2% versus 12.2% (P = .084) for OS, respectively. Conclusion Women with HER2-positive tumors ≤ 2 cm in the randomized trastuzumab trials derived substantial DFS and OS benefit from adjuvant trastuzumab. Trastuzumab-treated patients with HR-positive disease and ≤ one positive lymph node may be candidates for trials assessing less aggressive treatment approaches.

BACKGROUNDLittle is known about the factors that predict for gastrointestinal stromal tumor (GIST) recurrence in patients treated with adjuvant imatinib.METHODSRisk factors for GIST recurrence were identified, and 2 risk stratification scores were developed using the database of the Scandinavian Sarcoma Group (SSG) XVIII trial, where 358 patients with high‐risk GIST with no overt metastases were randomly assigned to adjuvant imatinib 400 mg/day either for 12 or 36 months after surgery. The findings were validated in the imatinib arm of the American College of Surgeons Oncology Group Z9001 trial, where 359 patients with GIST were randomized to receive imatinib and 354 were to receive placebo for 12 months.RESULTSFive factors (high tumor mitotic count, nongastric location, large size, rupture, and adjuvant imatinib for 12 months) were independently associated with unfavorable recurrence‐free survival (RFS) in a multivariable analysis in the SSGXVIII cohort. A risk score based on these 5 factors had a concordance index with GIST recurrence of 78.9%. When a simpler score consisting of the 2 strongest predictive factors (mitotic count and tumor site) was devised, the groups with the lowest, intermediate high, and the highest risk had 5‐year RFS of 76.7%, 47.5%, and 8.4%, respectively. Both scores were strongly associated with RFS in the validation cohort (P < .001 for each comparison).CONCLUSIONSThe scores generated were effective in stratifying the risk of GIST recurrence in patient populations treated with adjuvant imatinib. Patients with nongastric GIST with a high mitotic count are at a particularly high risk for recurrence. Cancer 2014;120:2325–2333 © 2014 The Authors. Cancer published by Wiley Periodicals, Inc. on behalf of American Cancer Society.