AI-Based Analysis Enhances Prediction of Best Responders for Trastuzumab Deruxtecan

Use of a novel HER2 quantitative continuous score (QCS) was better able to provide an objective and quantitative assessment of HER2 expression to identify patients who are most likely to respond to fam-trastuzumab deruxtecan-nxki (Enhertu), according to data presented during the 2020 San Antonio Breast Cancer Symposium.

The deep learning–based image analysis was used to generate a HER2 QCS for 151 patients with metastatic breast cancer enrolled in a phase 1 trial (NCT02564900) with varying levels of HER2 expression ranging from immunohistochemistry (IHC) 0 to IHC 3+. The use of HER2 QCS identified a greater number of HER2 expressors and demonstrated a broader stratification of efficacy between patients with high and low expressions of HER2 compared with manual pathologist scoring.

The overall response rate (ORR) was 56% and median progression-free survival (PFS) was 14.1 months for patients with tumors with HER2-high expression, regardless of scoring system. The HER2 QCS identified 67 of the 75 responders from the overall population, whereas stratification using the conventional IHC score missed 27 patients.

Of the 65 patients identified as having HER2-low–expressing tumors by conventional HER2 IHC scoring, 42% of patients treated with trastuzumab deruxtecan had a response and the median PFS was 11.0 months. When investigators used HER2 QCS, the same 65 responders identified as having low HER2 expression were further stratified into a subgroup of QCS-high and QCS-low patients. HER2 QCS was able to identify 21 of the 27 responders missed by the conventional IHC score and as a result the response rate increased to 53%, with a median PFS of 14.5 months.

Similarly, these results led to the identification of patients less likely to respond to trastuzumab deruxtecan. With only 6 responders missed, those patients in the QCS-low group had an ORR of 26% and a median PFS of 8.6 months.

“The ability to identify patients in the HER2-low group who will benefit from HER2-targeted

treatment is critical for a patient population that would otherwise not be treated with anti-HER2 therapy,” Mark Gustavson said in a presentation. Gustavson is director of translational research, oncology research and development for AstraZeneca Pharmaceuticals, which develops the assessment tool and jointly develops the assessment tool with Daiichi Sankyo, Inc.

Trastuzumab deruxtecan, a HER2-directed antibody and topoisomerase inhibitor conjugate was approved in December 2019 for the treatment of patients with unresectable or metastatic HER2-positive breast cancer following treatment with at least 2 prior anti-HER2–based regimens.² In data from the DESTINY-Breast 01 trial (NCT03248492), patients with low HER2 expression (n = 43) had an ORR of 44.2%, showing the agent’s promise a population that historically does not benefit from HER2-targeted therapy.³

Manual pathologist assessed HER2 protein expression classifies tumors by the percentage of tumor cells with highest intensity and completeness of staining. Low HER2 expression is defined as IHC 1+ or IHC 2+/ISH−. As limited options are available to patients in this category, investigators developed a scoring system using artificial intelligence (AI)–based image and data analysis to better identify patients with low-level expression in anticipation of trastuzumab deruxtecan’s continued efficacy in this population.¹

Deep learning models trained on commercial breast cancer samples to detect membrane, cytoplasm and nuclei of all tumor cells. The HER2 QCS was trained using pathologists’ annotations and was validated on unseen data. Optical density (OD), or level of brown stain intensity, was computed on detected membranes to derive features that could be linked to survival prediction.

In the identification of patients with high expression of HER2, investigators used a cellular ODq10 cutoff of greater than 8 in at least 90% of cells for patients to be positive. In the subgroup analysis of the 65 responders with HER2 low expression as determined by IHC, the cellular ODq10 cutoff was determined to be greater than 7.4 in at least 90% of cells to be positive.

“These data establish a clinical proof-of-concept demonstrating that use of HER2 QCS can potentially enhance prediction of patient outcome with [trastuzumab deruxtecan] by increasing the sensitivity and specificity of identification of patients with high and low HER2 expression,” Gustavson said.

Of note, investigators also determined that patients with low or negative HER2 expression response to treatment with trastuzumab deruxtecan was driven by a majority of tumor cells expressing a minimal amount of HER2 rather than a minority of cells that express higher levels of HER2.

A final factor that was assessed was role of spatial proximity for positive cells. Patients with 95% of cells having an OD value above the determined threshold were considered to have a homogenous HER2 distribution pattern (n = 36; 55.4%). Those with less than 95% of cells meeting the threshold had a heterogenous HER2 distribution pattern (n = 29; 44.6%).¹

Similar efficacy was observed with these subgroups as compared with best performing cutoffs. The ORR for the homogenous group was 53% with a median PFS of 14.8 months. The heterogenous group had an ORR of 28% with a median PFS of 8.6 months.

Further validation of the HER2 QCS is underway. Additional investigation of trastuzumab deruxtecan is also ongoing in the phase 3 DESTINY-Breast04 trial (NCT03734029) for patients with HER2-low–expressing tumors.

References

1. Gustavson M, Haneder S, Spitzmueller A, et al. Novel approach to HER2 quantification: digital pathology coupled with AI-based image and data analysis delivers objective and quantitative HER2 expression analysis for enrichment of responders to trastuzumab deruxtecan (T-DXd; DS-8201), specifically in HER2-low patients. Poster presented at: 2020 San Antonio Breast Cancer Symposia; December 8-11, 2020; virtual. Abstract PD6-01.
2. FDA approves fam-trastuzumab deruxtecan-nxki for unresectable or metastatic HER2-positive breast cancer. FDA. Updated December 20, 2019. Accessed December 10, 2020. https://bit.ly/37OdSX4
3. Modi S, Saura A, Yamashita T, at al; DESTINY-Breast 01 Investigators. Trastuzumab deruxtecan in previously treated HER2-positive breast cancer. N Engl J Med. 2020;382(7):610-621. doi:10.1056/NEJMoa1914510