Abstract PO-040: Comparing digital image analysis with a manual scoring approach for quantification of p63 and BRCA1 protein expression in oropharyngeal squamous cell carcinoma
Bibliographic record
Abstract
Abstract In oropharyngeal squamous cell carcinomas (OPSCC) there is overexpression of p63. We previously identified a link between p63 and BRCA1 in OPSCC, both at a gene and protein level. However, interpretation of biomarkers based on protein expression identified using immunohistochemistry can be subject to interobserver variability. We aimed to explore the option of digital image analysis as an alternative approach to immunohistochemical protein quantification. Herein, we compare manual histological interpretation of p63 and BRCA1 expression in the context of HPV status with digital image analysis. Representative samples of a clinically annotated cohort of OPSCC (n=191) were arranged in triplicate tissue microarrays (TMA) and stained immunohistochemically for p63 (p63NCL and p63DN) and BRCA1 (D-9 and Ab1). RNA in situ hybridization for a cocktail of 18 high-risk HPV genotypes (HR-18 HPV) was used to determine HPV status. Cases were scored manually using at least two independent scorers to generate consensus Q scores. Each TMA slide was digitally scanned. These images were imported into QuPath version 0.2.3, which was used to quantify the expression of each biomarker and generate digital H scores. Statistical analysis was conducted using R version 4.2.3 and GraphPad Prism5. Spearman’s Rank correlation was used to compare the digital and manual scores for each biomarker. Kaplan-Meier curves and log rank tests were used for survival analysis. Preliminary analysis showed a strong correlation between manual scoring methods and digital scores for each of the four biomarkers: p63NCL (rs=0.77, p<0.0001); p63DN (rs=0.72, p<0.0001); BRCA D9 (rs=0.75, p<0.0001) and BRCA AB1 (rs=0.84, p<0.0001). We identified a subgroup of HPV negative patients with low p63NCL expression and high BRCA D9 expression that had poor 5-year overall survival (p<0.0001). The process of digital quantification is more labour intensive than one might anticipate. It requires significant pre-processing to remove artefacts prior to obtaining a computer-generated H score. In contrast manual Q scoring allows immediate assessment. In addition, the BRCA D-9 antibody had non-specific staining, highlighting darkly staining “dendritic-like” cells which complicated digital assessment. The results show the effectiveness and reproducibility of digital pathology scoring methods for antibody quantification when compared with traditional manual scoring methods. However manual scoring methods presently used by pathologists are significantly more time efficient when compared to newer digital techniques. There is a need for improved or automated pre-processing techniques to remove some of the pitfalls associated with this approach for future studies validating use of combined low p63NCL expression and high BRCA D9 expression to predict outcome in patients with OPSCC. Citation Format: Laura Graham, Stephanie Craig, Kris McCombe, Stephen McQuaid, Simon McDade, Jacqueline James. Comparing digital image analysis with a manual scoring approach for quantification of p63 and BRCA1 protein expression in oropharyngeal squamous cell carcinoma [abstract]. In: Proceedings of the AACR-AHNS Head and Neck Cancer Conference: Innovating through Basic, Clinical, and Translational Research; 2023 Jul 7-8; Montreal, QC, Canada. Philadelphia (PA): AACR; Clin Cancer Res 2023;29(18_Suppl):Abstract nr PO-040.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.006 | 0.007 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.003 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.005 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".