Unmasking the immune microecology of ductal carcinoma in situ with deep learning

Priya Lakshmi Narayanan,Lucia Hernandez,Yinyin Yuan,Robert B West,Naomi Guppy,Allison H Hall,Shan E Ahmed Raza,Lorraine King,Carlo Maley,Barry Gusterson,E Shelley Hwang,Mitch Dowsett,Jeffrey R Marks

doi:10.1038/s41523-020-00205-5

Abstract

Despite increasing evidence supporting the clinical relevance of tumour infiltrating lymphocytes (TILs) in invasive breast cancer, TIL spatial variability within ductal carcinoma in situ (DCIS) samples and its association with progression are not well understood. To characterise tissue spatial architecture and the microenvironment of DCIS, we designed and validated a new deep learning pipeline, UNMaSk. Following automated detection of individual DCIS ducts using a new method IM-Net, we applied spatial tessellation to create virtual boundaries for each duct. To study local TIL infiltration for each duct, DRDIN was developed for mapping the distribution of TILs. In a dataset comprising grade 2–3 pure DCIS and DCIS adjacent to invasive cancer (adjacent DCIS), we found that pure DCIS cases had more TILs compared to adjacent DCIS. However, the colocalisation of TILs with DCIS ducts was significantly lower in pure DCIS compared to adjacent DCIS, which may suggest a more inflamed tissue ecology local to DCIS ducts in adjacent DCIS cases. Our study demonstrates that technological developments in deep convolutional neural networks and digital pathology can enable an automated morphological and microenvironmental analysis of DCIS, providing a new way to study differential immune ecology for individual ducts and identify new markers of progression.

Highlights

Ductal carcinoma in situ (DCIS) is a non-obligatory precursor of invasive ductal carcinoma (IDC)
Prior to DCIS detection and segmentation, UNet was first used for tissue segmentation (Fig. 1a)
To evaluate the performance of the proposed Inception MicroNet (IM-Net) in DCIS detection and segmentation, we compared it with five state-of-the-art deep learning methods using images generated across three datasets (Fig. 1a)

Summary

Introduction

Ductal carcinoma in situ (DCIS) is a non-obligatory precursor of invasive ductal carcinoma (IDC). It is the most common mammographically detected breast cancer, predicting DCIS progression to IDC remains a major clinical challenge[1,2,3]. A recent study has categorised DCIS evolution to IDC into four models, highlighting its heterogeneity. Given the complex spatial ductule structure, ecological dynamics between individual DCIS ducts and their surrounding microenvironment are difficult to measure by eye. These limits our ability to study the influence of the microenvironment on tumour evolution and progression[7]

Objectives

Methods

Results