Dataset Information

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

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.

SUBMITTER: Narayanan PL

PROVIDER: S-EPMC7921670 | biostudies-literature | 2021 Mar

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

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

Narayanan Priya Lakshmi PL Raza Shan E Ahmed SEA Hall Allison H AH Marks Jeffrey R JR King Lorraine L West Robert B RB Hernandez Lucia L Guppy Naomi N Dowsett Mitch M Gusterson Barry B Maley Carlo C Hwang E Shelley ES Yuan Yinyin Y

NPJ breast cancer 20210301 1

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 s ...[more]

PMID: 33649333

Similar Datasets

Project description:Ductal carcinoma in situ (DCIS) is a non-invasive breast cancer that can progress into invasive ductal carcinoma (IDC). Studies suggest DCIS is often overtreated since a considerable part of DCIS lesions may never progress into IDC. Lower grade lesions have a lower progression speed and risk, possibly allowing treatment de-escalation. However, studies show significant inter-observer variation in DCIS grading. Automated image analysis may provide an objective solution to address high subjectivity of DCIS grading by pathologists. In this study, we developed and evaluated a deep learning-based DCIS grading system. The system was developed using the consensus DCIS grade of three expert observers on a dataset of 1186 DCIS lesions from 59 patients. The inter-observer agreement, measured by quadratic weighted Cohen's kappa, was used to evaluate the system and compare its performance to that of expert observers. We present an analysis of the lesion-level and patient-level inter-observer agreement on an independent test set of 1001 lesions from 50 patients. The deep learning system (dl) achieved on average slightly higher inter-observer agreement to the three observers (o1, o2 and o3) (κo1,dl = 0.81, κo2,dl = 0.53 and κo3,dl = 0.40) than the observers amongst each other (κo1,o2 = 0.58, κo1,o3 = 0.50 and κo2,o3 = 0.42) at the lesion-level. At the patient-level, the deep learning system achieved similar agreement to the observers (κo1,dl = 0.77, κo2,dl = 0.75 and κo3,dl = 0.70) as the observers amongst each other (κo1,o2 = 0.77, κo1,o3 = 0.75 and κo2,o3 = 0.72). The deep learning system better reflected the grading spectrum of DCIS than two of the observers. In conclusion, we developed a deep learning-based DCIS grading system that achieved a performance similar to expert observers. To the best of our knowledge, this is the first automated system for the grading of DCIS that could assist pathologists by providing robust and reproducible second opinions on DCIS grade.

Project description:BackgroundThe treatment and prognosis of breast ductal carcinoma in situ (DCIS) with and without microinvasion (MIC) are different. Ultrasound imaging shows that DCIS is a heterogeneous breast tumor with diverse manifestations. DCIS means that the cancer cells are confined in the duct without penetrating the basement membrane, MIC means that the cancer cells penetrate the basement membrane and the maximum diameter of any largest invasive lesion is less than or equal to 1 mm. This study was designed to evaluate how deep learning can be used to identify DCIS with MIC on ultrasound images.MethodsThe clinical and ultrasound data of 467 consecutive inpatients diagnosed with DCIS (213 with MIC) in West China Hospital of Sichuan University were collected from January 2013 to April 2019 and randomly apportioned to training and internal validation sets. An external validation set comprised data from Sichuan Provincial People's Hospital with 101 patients (33 with MIC) collected between January 2017 and December 2019. There were 2,492 original images; 66% of these were used to establish a model, and the remaining 34% were used to evaluate the model. Three experienced breast ultrasound clinicians analyzed the ultrasound images to establish a logistic regression model. Finally, the logistic regression model and five deep learning models (ResNet-50, ResNet-101, DenseNet-161, DenseNet-169, and Inception-v3) were compared and evaluated to assess their diagnostic efficiency when identifying MIC based on ultrasound image data.ResultsThe characteristics of high nuclear grade (P<0.001), necrosis (P=0.006), estrogen receptor negative (ER-; P=0.003), progesterone receptor negative (PR-; P=0.001), human epidermal growth factor receptor 2 positive (HER2+; P=0.034), lymphatic metastasis (P=0.008), and calcification (P<0.001) all showed significant correlations with MIC. The Inception-v3 model achieved the best performance (P<0.05) in MIC identification. The area under the receiver operating curve (AUC) of the Inception-v3 model was 0.803 [95% confidence interval (CI): 0.709 to 0.878], with a classification accuracy of 0.766, a sensitivity of 0.767, and a specificity of 0.765.ConclusionsDeep learning can be used to identify MIC of breast DCIS from ultrasound images. Models based on Inception-v3 can provide automated detection of DCIS with MIC from ultrasound images.

Project description:BackgroundThe role of different subtypes of tumour infiltrating lymphocytes (TILs) in breast ductal carcinoma in situ (DCIS) is still poorly defined. This study aimed to assess the prognostic significance of B and T lymphocytes and immune checkpoint proteins expression in DCIS.MethodsA well characterised DCIS cohort (n = 700) with long-term follow-up comprising pure DCIS (n = 508) and DCIS mixed with invasive carcinoma (IBC; n = 192) were stained immunohistochemically for CD20, CD3, CD4, CD8, FOXP3, PD1 and PDL1. Copy number variation and TP53 mutation status were assessed in a subset of cases (n = 58).ResultsCD3+ lymphocytes were the predominant cell subtype in the pure DCIS cohort, while FOXP3 showed the lowest levels. PDL1 expression was mainly seen in the stromal TILs. Higher abundance of TILs subtypes was associated with higher tumour grade, hormone receptor negativity and HER2 positivity. Mutant TP53 variants were associated with higher levels of stromal CD3+, CD4+ and FOXP3+ cells. DCIS coexisting with invasive carcinoma harboured denser stromal infiltrates of all immune cells and checkpoint proteins apart from CD4+ cells. Stromal PD1 was the most differentially expressed protein between DCIS and invasive carcinoma (Z = 5.8, p < 0.0001). Dense TILs, stromal FOXP3 and PDL1 were poor prognostic factors for DCIS recurrence, while dense TILs were independently associated with poor outcome for all recurrences (HR = 7.0; p = 0.024), and invasive recurrence (HR = 2.1; p = 0.029).ConclusionsImmunosuppressive proteins are potential markers for high risk DCIS and disease progression. Different stromal and intratumoural lymphocyte composition between pure DCIS, DCIS associated with IBC and invasive carcinoma play a potential role in their prognostic significance and related to the underlying genomic instability. Assessment of overall TILs provides a promising tool for evaluation of the DCIS immune microenvironment.

Project description:BackgroundThe immune microenvironment in ductal carcinoma in situ (DCIS) and its significance are not well established. This study was conducted to evaluate the immune microenvironment of DCIS including the composition of tumor-infiltrating lymphocyte (TIL) subsets and PD-L1+ immune cells and to compare it with that of invasive breast cancer.Materials and methodsA total of 671 cases including three different disease groups of pure DCIS, DCIS with microinvasion (DCIS-M), and invasive carcinoma were included in this study. CD4+, CD8+, and FOXP3+ TIL subsets and PD-L1+ immune cells were detected with immunohistochemistry using tissue microarrays and were analyzed in relation to clinicopathologic characteristics and different disease groups.ResultsIn pure DCIS, high infiltrations of CD4+, CD8+, and FOXP3+ T cells and the presence of PD-L1+ immune cells were associated with high nuclear grade, comedo-type necrosis, hormone receptor (HR) negativity, and high Ki-67 proliferation index. All immune cell infiltrations were higher in invasive carcinoma than in pure DCIS regardless of the HR status. While CD4+ T cells were more abundant than CD8+ T cells in pure DCIS, CD8+ T cells were dominant in invasive carcinoma, especially in HR-negative tumors. Within individual cases of invasive carcinoma with DCIS component, all immune cell subset infiltration was higher in the invasive component than in the DCIS component; however, CD4+ TIL infiltration did not differ between the two components in HR-negative tumors. Comparing pure DCIS, DCIS-M, and DCIS associated with invasive carcinoma (DCIS-INV), CD4+ TIL infiltration revealed a gradual increase from pure DCIS to DCIS-M and DCIS-INV in the HR-negative group, whereas FOXP3+ TIL infiltration was significantly increased in DCIS-INV than in pure DCIS in the HR-positive group. The high infiltration of FOXP3+ TIL and the presence of PD-L1+ immune cells were associated with tumor recurrence in patients with pure DCIS.ConclusionsOur study showed that the immune microenvironment differs significantly not only between DCIS and invasive carcinoma but also between pure DCIS, DCIS-M, and DCIS-INV depending on the HR status.

Project description:Endocrine therapy initiation after ductal carcinoma in situ (DCIS) is highly variable and largely unexplained. National guidelines recommend considering tamoxifen for women with estrogen receptor-positive (ER+) DCIS or who undergo excision alone. We evaluated endocrine therapy use after DCIS over a 15-year period in an integrated health care setting to identify factors related to initiation.Female Group Health Cooperative enrollees ages 18-89 years with a DCIS diagnosis during 1996-2011 were eligible for inclusion. Endocrine therapy was identified through pharmacy records. Tumor and treatment information were from tumor registry reports; demographics and other risk factors were from questionnaires and electronic medical records. Relative risks (RRs) and 95% confidence intervals (CIs) for endocrine therapy initiation were calculated using multivariable generalized linear models.We identified 727 women with a DCIS diagnosis, including 163 (22%) who initiated endocrine therapy (149 tamoxifen, 14 aromatase inhibitor). Younger women were more likely to initiate endocrine therapy (RR 1.69; 95% CI 1.16-2.46 for ages 45-54 vs. 65-74 years). Compared with breast-conserving surgery (BCS) with radiation, women who had BCS alone (RR 0.46; 95% CI 0.25-0.84) or mastectomy (RR 0.54; 95% CI 0.39-0.75) were less likely to use endocrine therapy. ER testing increased from 4% of DCIS cases in 2001 to 71% in 2011; however, endocrine therapy initiation decreased from 58% of ER+ DCIS in 2001-2005 to 37% in 2009-2011.Increasing ER testing since 2001 has not corresponded to parallel increases in endocrine therapy initiation. Age, surgery, and radiation were the primary factors associated with initiation.National guidelines recommend considering tamoxifen for women with ductal carcinoma in situ (DCIS) who are estrogen receptor-positive (ER+) or who undergo excision alone. In this study, the rapid increase in ER testing caused by tamoxifen's approval in 2000 did not lead to increases in endocrine therapy initiation, despite recognition of an increasing number of DCIS tumors as ER+ each year. Contrary to the suggested guidelines, women who had breast-conserving surgery without radiation were less likely to use tamoxifen than those who had radiation. Future Food and Drug Administration approval of new endocrine agents for DCIS (such as aromatase inhibitors) may provide an opportunity to reemphasize benefits by ER and surgery status.

Dataset Information

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

Publications

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

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets