Lymphocyte density determined by computational pathology validated as a predictor of response to neoadjuvant chemotherapy in breast cancer: secondary analysis of the ARTemis trial.
ABSTRACT: Background:We have previously shown lymphocyte density, measured using computational pathology, is associated with pathological complete response (pCR) in breast cancer. The clinical validity of this finding in independent studies, among patients receiving different chemotherapy, is unknown. Patients and methods:The ARTemis trial randomly assigned 800 women with early stage breast cancer between May 2009 and January 2013 to three cycles of docetaxel, followed by three cycles of fluorouracil, epirubicin and cyclophosphamide once every 21 days with or without four cycles of bevacizumab. The primary endpoint was pCR (absence of invasive cancer in the breast and lymph nodes). We quantified lymphocyte density within haematoxylin and eosin (H&E) whole slide images using our previously described computational pathology approach: for every detected lymphocyte the average distance to the nearest 50 lymphocytes was calculated and the density derived from this statistic. We analyzed both pre-treatment biopsies and post-treatment surgical samples of the tumour bed. Results:Of the 781 patients originally included in the primary endpoint analysis of the trial, 609 (78%) were included for baseline lymphocyte density analyses and a subset of 383 (49% of 781) for analyses of change in lymphocyte density. The main reason for loss of patients was the availability of digitized whole slide images. Pre-treatment lymphocyte density modelled as a continuous variable was associated with pCR on univariate analysis (odds ratio [OR], 2.92; 95% CI, 1.78-4.85; P?
Project description:BACKGROUND:There is a need to improve prediction of response to chemotherapy in breast cancer in order to improve clinical management and this may be achieved by harnessing computational metrics of tissue pathology. We investigated the association between quantitative image metrics derived from computational analysis of digital pathology slides and response to chemotherapy in women with breast cancer who received neoadjuvant chemotherapy. METHODS:We digitised tissue sections of both diagnostic and surgical samples of breast tumours from 768 patients enrolled in the Neo-tAnGo randomized controlled trial. We subjected digital images to systematic analysis optimised for detection of single cells. Machine-learning methods were used to classify cells as cancer, stromal or lymphocyte and we computed estimates of absolute numbers, relative fractions and cell densities using these data. Pathological complete response (pCR), a histological indicator of chemotherapy response, was the primary endpoint. Fifteen image metrics were tested for their association with pCR using univariate and multivariate logistic regression. RESULTS:Median lymphocyte density proved most strongly associated with pCR on univariate analysis (OR 4.46, 95 % CI 2.34-8.50, p < 0.0001; observations = 614) and on multivariate analysis (OR 2.42, 95 % CI 1.08-5.40, p = 0.03; observations = 406) after adjustment for clinical factors. Further exploratory analyses revealed that in approximately one quarter of cases there was an increase in lymphocyte density in the tumour removed at surgery compared to diagnostic biopsies. A reduction in lymphocyte density at surgery was strongly associated with pCR (OR 0.28, 95 % CI 0.17-0.47, p < 0.0001; observations = 553). CONCLUSIONS:A data-driven analysis of computational pathology reveals lymphocyte density as an independent predictor of pCR. Paradoxically an increase in lymphocyte density, following exposure to chemotherapy, is associated with a lack of pCR. Computational pathology can provide objective, quantitative and reproducible tissue metrics and represents a viable means of outcome prediction in breast cancer. TRIAL REGISTRATION:ClinicalTrials.gov NCT00070278 ; 03/10/2003.
Project description:Although tumor-infiltrating lymphocytes (TIL) have been associated with response to neoadjuvant therapy, measurement typically is subjective, semiquantitative, and unable to differentiate among subpopulations. Here, we describe a quantitative objective method for analyzing lymphocyte subpopulations and assessing their predictive value.We developed a quantitative immunofluorescence assay to measure stromal expression of CD3, CD8, and CD20 on one slide. We validated this assay by comparison with flow cytometry on tonsil specimens and assessed predictive value in breast cancer on a neoadjuvant cohort (n = 95). Then, each marker was tested for prediction of pathologic complete response (pCR) compared with pathologist estimation of the percentage of lymphocyte infiltrate.The lymphocyte percentage and CD3, CD8, and CD20 proportions were similar between flow cytometry and quantitative immunofluorescence on tonsil specimens. Pathologist TIL count predicted pCR [P = 0.043; OR, 4.77; 95% confidence interval (CI), 1.05-21.6] despite fair interobserver reproducibility (? = 0.393). Stromal AQUA (automated quantitative analysis) scores for CD3 (P = 0.023; OR, 2.51; 95% CI, 1.13-5.57), CD8 (P = 0.029; OR, 2.00; 95% CI, 1.08-3.72), and CD20 (P = 0.005; OR, 1.80; 95% CI, 1.19-2.72) predicted pCR in univariate analysis. CD20 AQUA score predicted pCR (P = 0.019; OR, 5.37; 95% CI, 1.32-21.8) independently of age, size, nuclear grade, nodal status, ER, PR, HER2, and Ki-67, whereas CD3, CD8, and pathologist estimation did not.We have developed and validated an objective, quantitative assay measuring TILs in breast cancer. Although this work provides analytic validity, future larger studies will be required to prove clinical utility.
Project description:Background:The ARTemis trial previously reported that addition of neoadjuvant bevacizumab (Bev) to docetaxel (D) followed by fluorouracil, epirubicin and cyclophosphamide (D-FEC) in HER2 negative breast cancer improved the pathological complete response (pCR) rate. We present disease-free survival (DFS) and overall survival (OS) with central pathology review. Patients and methods:Patients were randomized to 3 cycles of D followed by 3 cycles of FEC (D-FEC),?±4 cycles of Bev (Bev?+?D-FEC). DFS and OS were analyzed by treatment and by central pathology reviewed pCR and Residual Cancer Burden (RCB) class. Results:A total of 800 patients were randomized [median follow-up 3.5 years (IQR 3.2-4.4)]. DFS and OS were similar across treatment arms [DFS hazard ratio (HR)=1.18 (95% CI 0.89-1.57), P?=?0.25; OS HR?=?1.26 (95% CI 0.90-1.76), P?=?0.19). Both local pathology report review and central histopathology review confirmed a significant improvement in DFS and OS for patients who achieved a pCR [DFS HR?=?0.38 (95% CI 0.23-0.63), P?<?0.001; OS HR?=?0.43 (95% CI 0.24-0.75), P?=?0.003]. However, significant heterogeneity was observed (P?=?0.02); larger improvements in DFS were obtained with a pCR achieved with D-FEC than a pCR achieved with Bev?+?D-FEC. As RCB class increased, significantly worse DFS and OS was observed (P for trend?<0.0001), which effect was most marked in the ER negative group. Conclusions:The addition of short course neoadjuvant Bev to standard chemotherapy did not demonstrate a DFS or OS benefit. Achieving a pCR with D-FEC is associated with improved DFS and OS but not when pCR is achieved with Bev?+?D-FEC. At the present time therefore, Bev is not recommended in early breast cancer. ClinicalTrials.gov number:NCT01093235.
Project description:Digital analysis of pathology whole-slide images is fast becoming a game changer in cancer diagnosis and treatment. Specifically, deep learning methods have shown great potential to support pathology analysis, with recent studies identifying molecular traits that were not previously recognized in pathology H&E whole-slide images. Simultaneous to these developments, it is becoming increasingly evident that tumor heterogeneity is an important determinant of cancer prognosis and susceptibility to treatment, and should therefore play a role in the evolving practices of matching treatment protocols to patients. State of the art diagnostic procedures, however, do not provide automated methods for characterizing and/or quantifying tumor heterogeneity, certainly not in a spatial context. Further, existing methods for analyzing pathology whole-slide images from bulk measurements require many training samples and complex pipelines. Our work addresses these two challenges. First, we train deep learning models to spatially resolve bulk mRNA and miRNA expression levels on pathology whole-slide images (WSIs). Our models reach up to 0.95 AUC on held-out test sets from two cancer cohorts using a simple training pipeline and a small number of training samples. Using the inferred gene expression levels, we further develop a method to spatially characterize tumor heterogeneity. Specifically, we produce tumor molecular cartographies and heterogeneity maps of WSIs and formulate a heterogeneity index (HTI) that quantifies the level of heterogeneity within these maps. Applying our methods to breast and lung cancer slides, we show a significant statistical link between heterogeneity and survival. Our methods potentially open a new and accessible approach to investigating tumor heterogeneity and other spatial molecular properties and their link to clinical characteristics, including treatment susceptibility and survival.
Project description:The introduction of fast digital slide scanners that provide whole slide images has led to a revival of interest in image analysis applications in pathology. Segmentation of cells and nuclei is an important first step towards automatic analysis of digitized microscopy images. We therefore developed an automated nuclei segmentation method that works with hematoxylin and eosin (H&E) stained breast cancer histopathology images, which represent regions of whole digital slides. The procedure can be divided into four main steps: 1) pre-processing with color unmixing and morphological operators, 2) marker-controlled watershed segmentation at multiple scales and with different markers, 3) post-processing for rejection of false regions and 4) merging of the results from multiple scales. The procedure was developed on a set of 21 breast cancer cases (subset A) and tested on a separate validation set of 18 cases (subset B). The evaluation was done in terms of both detection accuracy (sensitivity and positive predictive value) and segmentation accuracy (Dice coefficient). The mean estimated sensitivity for subset A was 0.875 (±0.092) and for subset B 0.853 (±0.077). The mean estimated positive predictive value was 0.904 (±0.075) and 0.886 (±0.069) for subsets A and B, respectively. For both subsets, the distribution of the Dice coefficients had a high peak around 0.9, with the vast majority of segmentations having values larger than 0.8.
Project description:The tertiary lymphoid structure (TLS) is an important source of tumor-infiltrating lymphocytes (TILs), which have a strong prognostic and predictive value in triple-negative breast cancer (TNBC). A previous study reported that the levels of CXCL13 mRNA expression were associated with TLSs, but measuring the gene expression is challenging in routine practice. Therefore, this study evaluated the MECA79-positive high endothelial venule (HEV) densities and their association with the histopathologically assessed TLSs in biopsy samples. In addition, the relationship of TLSs with the CXCL13 transcript levels and clinical outcomes were examined.A total of 108 TNBC patients treated with neoadjuvant chemotherapy (NAC) were studied. The amounts of TILs and TLSs were measured histopathologically using hematoxylin and eosin-stained slides. The HEV densities and TIL subpopulations were measured by immunohistochemistry for MECA79, CD3, CD8, and CD20. CXCL13mRNA expression levels using a NanoString assay (NanoString Technologies).The mean number of HEVs in pre-NAC biopsies was 12 (range, 0 to 72). The amounts of TILs and TLSs, HEV density, and CXCL13 expression showed robust correlations with each other. A lower pre-NAC clinical T stage, higher TIL and TLS levels, a higher HEV density, CD20-positive cell density, and CXCL13 expression were significant predictors of a pathologic complete response (pCR). Higher CD8-positive cell density and levels of CXCL13 expression were significantly associated with a better disease-free survival rate.MECA79-positive HEV density in pre-NAC biopsies is an objective and quantitative surrogate marker of TLS and might be a valuable tool for predicting pCR of TNBC in routine pathology practice.
Project description:Importance:Following recent US Food and Drug Administration approval, adoption of whole slide imaging in clinical settings may be imminent, and diagnostic accuracy, particularly among challenging breast biopsy specimens, may benefit from computerized diagnostic support tools. Objective:To develop and evaluate computer vision methods to assist pathologists in diagnosing the full spectrum of breast biopsy samples, from benign to invasive cancer. Design, Setting, and Participants:In this diagnostic study, 240 breast biopsies from Breast Cancer Surveillance Consortium registries that varied by breast density, diagnosis, patient age, and biopsy type were selected, reviewed, and categorized by 3 expert pathologists as benign, atypia, ductal carcinoma in situ (DCIS), and invasive cancer. The atypia and DCIS cases were oversampled to increase statistical power. High-resolution digital slide images were obtained, and 2 automated image features (tissue distribution feature and structure feature) were developed and evaluated according to the consensus diagnosis of the expert panel. The performance of the automated image analysis methods was compared with independent interpretations from 87 practicing US pathologists. Data analysis was performed between February 2017 and February 2019. Main Outcomes and Measures:Diagnostic accuracy defined by consensus reference standard of 3 experienced breast pathologists. Results:The accuracy of machine learning tissue distribution features, structure features, and pathologists for classification of invasive cancer vs noninvasive cancer was 0.94, 0.91, and 0.98, respectively; the accuracy of classification of atypia and DCIS vs benign tissue was 0.70, 0.70, and 0.81, respectively; and the accuracy of classification of DCIS vs atypia was 0.83, 0.85, and 0.80, respectively. The sensitivity of both machine learning features was lower than that of the pathologists for the invasive vs noninvasive classification (tissue distribution feature, 0.70; structure feature, 0.49; pathologists, 0.84) but higher for the classification of atypia and DCIS vs benign cases (tissue distribution feature, 0.79; structure feature, 0.85; pathologists, 0.72) and the classification of DCIS vs atypia (tissue distribution feature, 0.88; structure feature, 0.89; pathologists, 0.70). For the DCIS vs atypia classification, the specificity of the machine learning feature classification was similar to that of the pathologists (tissue distribution feature, 0.78; structure feature, 0.80; pathologists, 0.82). Conclusion and Relevance:The computer-based automated approach to interpreting breast pathology showed promise, especially as a diagnostic aid in differentiating DCIS from atypical hyperplasia.
Project description:PURPOSE:Machine Learning Package for Cancer Diagnosis (MLCD) is the result of a National Institutes of Health/National Cancer Institute (NIH/NCI)-sponsored project for developing a unified software package from state-of-the-art breast cancer biopsy diagnosis and machine learning algorithms that can improve the quality of both clinical practice and ongoing research. METHODS:Whole-slide images of 240 well-characterized breast biopsy cases, initially assembled under R01 CA140560, were used for developing the algorithms and training the machine learning models. This software package is based on the methodology developed and published under our recent NIH/NCI-sponsored research grant (R01 CA172343) for finding regions of interest (ROIs) in whole-slide breast biopsy images, for segmenting ROIs into histopathologic tissue types and for using this segmentation in classifiers that can suggest final diagnoses. RESULT:The package provides an ROI detector for whole-slide images and modules for semantic segmentation into tissue classes and diagnostic classification into 4 classes (benign, atypia, ductal carcinoma in situ, invasive cancer) of the ROIs. It is available through the GitHub repository under the Massachusetts Institute of Technology license and will later be distributed with the Pathology Image Informatics Platform system. A Web page provides instructions for use. CONCLUSION:Our tools have the potential to provide help to other cancer researchers and, ultimately, to practicing physicians and will motivate future research in this field. This article describes the methodology behind the software development and gives sample outputs to guide those interested in using this package.
Project description:BACKGROUND:For breast cancer patients undergoing neoadjuvant chemotherapy (NAC), pathologic complete response (pCR; no invasive or in situ) cannot be assessed non-invasively so all patients undergo surgery. The aim of our study was to develop and validate a radiomics classifier that classifies breast cancer pCR post-NAC on MRI prior to surgery. METHODS:This retrospective study included women treated with NAC for breast cancer from 2014 to 2016 with (1) pre- and post-NAC breast MRI and (2) post-NAC surgical pathology report assessing response. Automated radiomics analysis of pre- and post-NAC breast MRI involved image segmentation, radiomics feature extraction, feature pre-filtering, and classifier building through recursive feature elimination random forest (RFE-RF) machine learning. The RFE-RF classifier was trained with nested five-fold cross-validation using (a) radiomics only (model 1) and (b) radiomics and molecular subtype (model 2). Class imbalance was addressed using the synthetic minority oversampling technique. RESULTS:Two hundred seventy-three women with 278 invasive breast cancers were included; the training set consisted of 222 cancers (61 pCR, 161 no-pCR; mean age 51.8?years, SD 11.8), and the independent test set consisted of 56 cancers (13 pCR, 43 no-pCR; mean age 51.3?years, SD 11.8). There was no significant difference in pCR or molecular subtype between the training and test sets. Model 1 achieved a cross-validation AUROC of 0.72 (95% CI 0.64, 0.79) and a similarly accurate (P?=?0.1) AUROC of 0.83 (95% CI 0.71, 0.94) in both the training and test sets. Model 2 achieved a cross-validation AUROC of 0.80 (95% CI 0.72, 0.87) and a similar (P?=?0.9) AUROC of 0.78 (95% CI 0.62, 0.94) in both the training and test sets. CONCLUSIONS:This study validated a radiomics classifier combining radiomics with molecular subtypes that accurately classifies pCR on MRI post-NAC.
Project description:This randomized, multicenter study compared the efficacy of docetaxel with or without capecitabine following fluorouracil/epirubicin/cyclophosphamide (FEC) therapy in operable breast cancer and investigated the role of Ki67 as a predictive biomarker. Patients were randomized to 4 cycles of docetaxel/capecitabine (docetaxel: 75 mg/m2 on day 1; capecitabine: 1,650 mg/m2 on days 1–14 every 3 weeks) or docetaxel alone (75 mg/m2 on day 1 every 3 weeks) after completion of 4 cycles of FEC (5-fluorouracil 500 mg/m2, epirubicin 100 mg/m2 and cyclophosphamide 500 mg/m2 on day 1 every 3 weeks). The primary endpoint was the pathological complete response (pCR) rate. Predictive factor analysis was conducted using clinicopathological markers, including hormone receptors and Ki67 labeling index (Ki67LI). A total of 477 patients were randomized; the overall response in the docetaxel/capecitabine and docetaxel groups was 88.3 and 87.4 %, respectively. There were no significant differences in the pCR rate (docetaxel/capecitabine: 23 %; docetaxel: 24 %; p = 0.748), disease-free survival, or overall survival. However, patients with mid-range Ki67LI (10–20 %) showed a trend towards improved pCR rate with docetaxel/capecitabine compared to docetaxel alone. Furthermore, multivariate logistic regression analysis showed pre-treatment Ki67LI (odds ratio 1.031; 95 % CI 1.014–1.048; p = 0.0004) to be a significant predictor of pCR in this neoadjuvant treatment setting. Docetaxel/capecitabine (after 4 cycles of FEC) did not generate significant improvement in pCR compared to docetaxel alone. However, exploratory analyses suggested that assessment of pre-treatment Ki67LI may be a useful tool in the identification of responders to preoperative docetaxel/capecitabine in early-stage breast cancer.