Development of an automated combined positive score prediction pipeline using artificial intelligence on multiplexed immunofluorescence images.

Immunotherapy targeting immune checkpoint proteins, such as programmed cell death ligand 1 (PD-L1), has shown impressive outcomes in many clinical trials but only 20%-40% of patients benefit from it. Utilizing Combined Positive Score (CPS) to evaluate PD-L1 expression in tumour biopsies to identify patients with the highest likelihood of responsiveness to anti-PD-1/PD-L1 therapy has been approved by the Food and Drug Administration for several solid tumour types. Current CPS workflow requires a pathologist to manually score the two-colour PD-L1 chromogenic immunohistochemistry image. Multiplex immunofluorescence (mIF) imaging reveals the expression of an increased number of immune markers in tumour biopsies and has been used extensively in immunotherapy research. Recent rapid progress of Artificial Intelligence (AI)-based imaging analysis, particularly Deep Learning, provides cost effective and high-quality solutions to healthcare. In this article, we propose an imaging pipeline that utilizes three-colour mIF images (DAPI, PD-L1, and Pan-cytokeratin) as input and predicts the CPS using AI techniques. Our novel pipeline is composed of three modules employing algorithms of image processing, machine learning, and deep learning techniques. The first module of quality check (QC) detects and removes the image regions contaminated with sectioning and staining artefacts. The QC module ensures that only image regions free of the three common artefacts are used for downstream analysis. The second module of nuclear segmentation uses deep learning to segment and count nuclei in the DAPI images wherein our specialized method can accurately separate touching nuclei. The third module of cell phenotyping calculates CPS by identifying and counting PD-L1 positive cells and tumour cells. These modules are data-efficient and require only few manual annotations for training purposes. Using tumour biopsies from a clinical trial, we found that the CPS from the AI-based models shows a high Spearman correlation (78%, p = 0.003) to the pathologist-scored CPS.

A Preliminary Study of Deep-Learning Algorithm for Analyzing Multiplex Immunofluorescence Biomarkers in Body Fluid Cytology Specimens.

INTRODUCTION: Multiplex biomarker analysis of cytological body fluid specimens is often used to assist cytologists in distiguishing metastatic cancer cells from reactive mesothelial cells. However, evaluating biomarker expression visually may be challenging, especially when the cells of interest are scant. Deep-learning algorithms (DLAs) may be able to assist cytologists in analyzing multiple biomarker expression at the single cell level in the multiplex fluorescence imaging (MFI) setting. This preliminary study was performed to test the feasibility of using DLAs to identify immunofluorescence-stained metastatic adenocarcinoma cells in body fluid cytology samples. METHODS: A DLA was developed to analyze MFI-stained cells in body fluid cytological samples. A total of 41 pleural fluid samples, comprising of 20 positives and 21 negatives, were retrospectively collected. Multiplex immunofluorescence labeling for MOC31, BerEP4, and calretinin, were performed on cell block sections, and results were analyzed by manual analysis (manual MFI) and DLA analysis (MFI-DLA) independently. RESULTS: All cases with positive original cytological diagnoses showed positive results either by manual MFI or MFI-DLA, but 2 of the 14 (14.3%) original cytologically negative cases had rare cells with positive MOC31 and/or BerEP4 staining in addition to calretinin. Manual MFI analysis and MFI-DLA showed 100% concordance. CONCLUSION: MFI combined with DLA provides a potential tool to assist in cytological diagnosis of metastatic malignancy in body fluid samples. Larger studies are warranted to test the clinical validity of the approach.

Comparison of Afirma Gene Expression Classifier with Gene Sequencing Classifier in indeterminate thyroid nodules: A single-institutional experience.

INTRODUCTION: The Afirma test has been used in the diagnosis of cytologically indeterminate thyroid nodules to reduce diagnostic uncertainty and unnecessary surgeries. Gene Sequencing Classifier (GSC) was developed to improve the positive predictive value and overall test performance of Gene Expression Classifier (GEC). Here we present our experience comparing the performance of first-generation assay of Afirma (GEC) with the new assay (GSC). METHODS: Retrospective analysis was performed on all Bethesda III and IV cytology thyroid nodules tested with GEC and GSC. Test performance was evaluated by surgical pathology outcomes. RESULTS: In total, 167 cases were tested with GEC, of which 49% were reported as benign. Fourteen cases had surgical follow-up with 11 benign, one non-invasive follicular thyroid neoplasm with papillary-like nuclear features (NIFTP) and two malignant diagnoses. Of the 167 cases, 51% had suspicious GEC result. Fifty-seven of these suspicious GEC cases had surgical follow-up with 28 benign, nine NIFTP and 20 malignant histology. There 133 cases tested with GSC, of which 61% were reported as benign. Ten cases had surgical follow-up, all of which showed benign results and 32% of the cases were tested as suspicious. Thirty-six cases with suspicious GSC had surgical follow-up. Fourteen of them had benign, five NIFTP, and 17 malignant surgical pathology. Based on molecular testing, surgical resection could have been be prevented 61% with GSC, compared to 49% with GEC test. CONCLUSION: Our experience shows that GSC has a better test performance than GEC. Also, our data support that GSC identify more cases as benign and reduces the number of unnecessary surgeries compared to GEC.

Breast Hyperplasias, Risk Signature, and Breast Cancer.

We address the dilemma faced by oncologists in administering preventative measures to "at risk" patients diagnosed with atypical and nonatypical hyperplasias due to lack of any molecular means of risk stratification and identifying high-risk subjects. Our study purpose is to investigate a four marker risk signature, MMP-1, CEACAM6, HYAL1, and HEC1, using 440 hyperplastic tissues for identifying high-risk subjects who will benefit from preventative therapies. We assayed the markers by IHC and combined their expression levels to obtain a composite value from 0-10, which we called a "Cancer Risk Score." We demonstrate that the four marker-based risk scores predict subsequent cancer development with an accuracy of 91% and 86% for atypical and nonatypical subjects, respectively. We have established a correlation between risk scores and cancer rates by stratifying the samples into low risk (score ≤ 0.5); intermediate risk (score ≤ 5.4), and high risk (score >5.4) groups using Kaplan-Meier survival analysis. We have evaluated cancer rates at 5, 10, and 15 years. Our results show that the average cancer rates in the first 5 years among low- and intermediate-risk groups were 2% and 15%, respectively. Among high-risk group, the average cancer rates at 5 years were 73% and 34% for atypical and nonatypical subjects, respectively. The molecular risk stratification described here assesses a patient's tumor biology-based risk level as low, intermediate, or high and for making informed treatment decisions. The outcomes of our study in conjunction with the available prophylactic measures could prevent approximately 20%-25% of sporadic breast cancers.