Abstract 4246: Comparison of pathology versus IHC-based ovarian carcinoma histology assignment using gene expression, DNA methylation, and clinical outcome data
Bibliographic record
Abstract
Abstract Background: Epithelial ovarian cancer (EOC) is composed of five major histologic types: 1) high-grade serous carcinoma (HGSC), accounting for most cases (∼70%); and the rarer 2) clear cell, 3) endometrioid, 4) mucinous, and 5) low- grade serous carcinoma (LGSC). As EOC risk factors are histology specific, as are the site of precursor lesions, accurate subtyping is critical to understanding EOC prognostic factors and etiology. Our aim was to compare two currently employed histology assignment strategies using gene expression, DNA methylation, and clinical outcome data Methods: Histology assignment based on: a) pathologist, and b) an integrated pathologist and IHC prediction algorithm (pathIHC) (using ARID1A, CDKN2A, DKK1, HNF1B, MDM2, PGR, TP53, TFF3, VIM, and WT1 staining patterns), was compared for tumors of all histologies at the Mayo Clinic, Rochester. The 500 most variable probes from Illumina Methylation450 BeadChips (N = 259) and Agilent 4×44K expression arrays (N = 245) were used to perform unsupervised hierarchical clustering. Fisher's Exact test was used to test the association of clusters with histology. Cox proportional hazards regression analysis was used to test the association of clusters and histology with time to progression (TTP) and time to death (TTD). Results: Eighty-two percent of tumors were concordant between histology assignment strategies. Clustering based on methylation data produced two distinct clusters. PathIHC produced more homogenous clusters (p-value = 2.8×10-16) than pathology alone (p = 1.3×10-9). Cluster M1, characterized by high levels of methylation across almost all probes, was enriched for the rarer EOC subtypes; cluster M2, characterized by moderate levels of methylation across 50%-75% of probes, was predominantly HGSC. These patterns were observed irrespective of histology assignment strategy; however, when using histology by pathologist, cluster M2 had more endometrioid and LGSC tumors interspersed with HGSC tumors. Clustering based on expression data also produced two clusters. PathIHC produced slightly more homogenous clusters (p = 9.0×10-11, pathology alone, p = 2.3×10-10). Cluster E1 was enriched for the rarer EOC subtypes; cluster E2 was primarily serous lineage tumours (HGSC and LGSC), particularly when using histology by pathIHC. Considering only clinical outcomes, histologies by pathology were more significantly different in terms of TTP (p = 1.1×10-6) and TTD (p = 1.7×10-3) than histologies by pathIHC (TTP, p = 5.4×10-6; TTD, p = 5.7×10-3). Conclusions: Histology by pathIHC produced more homogeneous clusters in both the methylation and expression data; thus, molecular data supports this strategy. Differences in clinical outcomes (TTP and TTD) between histology groups were more pronounced when using assignment by pathologist; thus, clinical data supports this strategy. Citation Format: M A. Earp, S J. Winham, S M. Armasu, B L. Fridley, M C. Larson, Z C. Fogarty, K R. Kalli, C Wang, G L. Keeney, J M. Cunningham, S Ramus, M Kobel, E L. Goode. Comparison of pathology versus IHC-based ovarian carcinoma histology assignment using gene expression, DNA methylation, and clinical outcome data. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4246. doi:10.1158/1538-7445.AM2015-4246
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 | 0.004 | 0.004 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
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".