Integrated tumor identification and automated scoring minimizes pathologist involvement and provides new insights to key biomarkers in breast cancer

Abstract

Digital image analysis (DIA) is becoming central to the quantitative evaluation of tissue biomarkers for discovery, diagnosis and therapeutic selection for the delivery of precision medicine. In this study, automated DIA using a new purpose-built software platform (QuPath) is applied to a cohort of 293 breast cancer patients to score five biomarkers in tissue microarrays (TMAs): ER, PR, HER2, Ki67 and p53. This software is able to measure IHC expression following fully automated tumor recognition in the same immunohistochemical (IHC)-stained tissue section, as part of a rapid workflow to ensure objectivity and accelerate biomarker analysis. The digital scores produced by QuPath were compared with manual scores by a pathologist and shown to have a good level of concordance in all cases (Cohen's κ40.6), and almost perfect agreement for the clinically relevant biomarkers ER, PR and HER2 (κ40.86). To assess prognostic value, cutoff thresholds could be applied to both manual and automated scores using the QuPath software, and survival analysis performed for 5-year overall survival. DIA was shown to be capable of replicating the statistically significant stratification of patients achieved using manual scoring across all biomarkers (Po0.01, log-rank test). Furthermore, the image analysis scores were shown to consistently lead to statistical significance across a wide range of potential cutoff thresholds, indicating the robustness of the method, and identify sub-populations of cases exhibiting different expression patterns within the p53 and Ki67 data sets that warrant further investigation. These findings have demonstrated QuPath's suitability for fast, reproducible, high-throughput TMA analysis across a range of important biomarkers. This was achieved using our tumor recognition algorithms for IHC-stained sections, trained interactively without the need for any additional tumor recognition markers, for example, cytokeratin, to obtain greater insight into the relationship between biomarker expression and clinical outcome applicable to a range of cancer types.

Validation of the systematic scoring of immunohistochemically stained tumour tissue microarrays using QuPath digital image analysis

“…[18][19][20] The latter approach to p53 analysis has been applied to this colon cancer cohort and reported separately, finding that aberrant extremes of p53 immunoexpression (diffuse intense or completely absent) was associated with significantly poorer unadjusted disease-specific survival when compared with 'wild-type' expression (P = 0.003). 4 We have demonstrated the feasibility and interobserver reproducibility of CD3 and p53 immunoscoring using QuPath in the TMA setting, even with limited experience of digital pathology images and minimal QuPath training. This may be extrapolated to equivalent tissue samples and to other markers showing the similar patterns of staining.…”

“…However, given recent developments in the understanding of p53 biology, and the relationship between different p53 mutations and immunoexpression patterns, it is now considered more appropriate to analyse p53 staining by comparing normal or ‘wild‐type’ staining with aberrant extremes of staining (‘mutation type’) . The latter approach to p53 analysis has been applied to this colon cancer cohort and reported separately, finding that aberrant extremes of p53 immunoexpression (diffuse intense or completely absent) was associated with significantly poorer unadjusted disease‐specific survival when compared with ‘wild‐type’ expression ( P = 0.003) …”

“…QuPath has also been utilised in recently published studies to score biomarkers demonstrating cytoplasmic (cyclooxygenase-2, 3hydroxy-3-methylglutaryl coenzyme-A reductase) and membranous (HER2) tumour cell staining. [4][5][6] However, biomarkers with more complex patterns of staining, such as mismatch repair proteins or immune checkpoint inhibitors, for example programmed cell death ligand 1 (PD-L1), are much more challenging for digital scoring, as multiple cell populations are stained (tumour and inflammatory) and distinction between tumour and inflammatory cell staining can be difficult when the pattern of staining is patchy. Recently, convolutional neural networks have shown great promise for pathology image analysis, and currently represent the state-of-the-art whenever such a rigorous cell classification is required.…”

“…Recent studies have evaluated the application of QuPath to the scoring of a variety of different biomarkers in breast and colon cancer tissue cohorts, mainly in TMA format. [3][4][5][6] In this study, we aim to demonstrate the functionality and reproducibility of biomarker scoring using QuPath image analysis involving users from different backgrounds (pathology, computer science, biochemistry) and with varying levels of experience of digital image analysis. Two well-established and biologically distinct tissue biomarkers, CD3 and p53, were evaluated in IHCstained TMAs from a large cohort of colon cancers by the three study reviewers using QuPath, and interobserver reproducibility of raw scores and downstream survival analyses was assessed.…”

Innervation of the hip joint capsular complex: A systematic review of histological and immunohistochemical studies and their clinical implications for contemporary treatment strategies in total hip arthroplasty

“…Although the majority of papers used the Freeman and Wyke [29] classification, or modifications of this, to identify mechanoreceptors, none have stated the use of a computerised identification. As research identifying immunological markers using computerised software compared to by a trained investigator produced highly similar results, this methodology may produce more reliable results [72]. Furthermore, computerised identification may be more efficient than manual assessment, and also could help prevent oversight in detecting mechanoreceptors.…”