Abstract B027: The development of a multiscale transcriptional atlas of sarcoma
Bibliographic record
Abstract
Abstract Objectives: Sarcomas are mesodermal cancers of bone and soft tissue of which there are >60 malignant varieties, many of which can be difficult to diagnose or subtype using traditional histopathology. A universal molecular definition of sarcoma types would therefore be an invaluable tool to the diagnostic pathologist. RNA has the potential to offer a complimentary perspective to cytogenetic- and methylation-based diagnostics, as it represents the active state of the disease at sampling and better reveals its phenotype. Recognizing the potential for RNA-based classification, we set out to create a first-generation transcriptional atlas of sarcoma. Methods: To develop transcriptional definitions of cancers with the potential to further subclassify tumor types, we designed a self-optimizing and scale-adaptive unsupervised method (RACCOON), which groups samples into hierarchically organized clusters. We used this approach on the UCSC Treehouse Childhood Cancer Compendium, a set of 2,178 pediatric and 9,400 adult tumors, 1,130 of which are sarcomas, as well as 1,735 non-neoplastic samples. We then trained an ensemble of convolutional neural networks to classify tumors to these transcriptional clusters. We have now added 624 more sarcoma samples from Toronto centers and international collaborators to better represent the breadth of sarcoma. We are actively sequencing 500 additional samples in partnership with the Gabriella Miller Kids First Research Program to yield an expanded cohort of >2,200 uniformly processed and analyzed sarcomas. Results: Sarcomas organize into two clusters at the highest hierarchical level: one characterized by entities which occur primarily in adults and resemble mature tissue, the other by primarily pediatric entities which exhibit high stemness and resemble embryonic tissue. Several included entities are not bona fide sarcomas but originate from the mesoderm (e.g., Wilms Tumor) signifying a common transcriptional identity for mesodermal neoplasms. Additionally, we demonstrate the first transcriptional subtypes of central osteosarcoma reflecting its major histotypes and representing divergent clinical courses. We also determine Ewing Sarcoma (ES) to be a distinct entity which clusters separately from all other cancers, raising questions of its origin and affinity to sarcoma. When classifying ongoing patients to the atlas, we correctly classified >85% of tumors and corrected the diagnosis of 7%. We find 14% of ES in our dataset were likely misdiagnosed CIC- or BCOR-driven sarcomas. Critically, assigned subtypes are consistent between primary and relapse pairs. Conclusion: RNA-seq is a promising tool for both subtype discovery and classifying sarcoma in ongoing patients. We have already included this tool in tumor boards to help inform patient care. Our method reveals the overarching organization of sarcoma for the first time and specifies its underlying biology. This atlas is ever-growing and is open to the community to contribute. Citation Format: Joshua O. Nash, Federico Comitani, Rose Chami, Sarah Cohen-Gogo, Astra Chang-Schwertschkow, Yael Babichev, Jodi Lees, Noa Alon, Nalan Gokgoz, Stephen Man Yu, Kyoko Yuki, Miranda Lorenti, Zhanqin Liu, Alaina McGoey, Famida Spatare, Bernarld Castro, Kim Tsoi, Hagit Peretz Soroka, Jack Brzezinski, Anita Villani, Albiruni Razak, Abha Gupta, Elizabeth Demicco, Gino Somers, Brendan C. Dickson, Jay S. Wunder, Irene L. Andrulis, David Malkin, Rebecca A. Gladdy, Adam Shlien. The development of a multiscale transcriptional atlas of sarcoma [abstract]. In: Proceedings of the AACR Special Conference: Sarcomas; 2022 May 9-12; Montreal, QC, Canada. Philadelphia (PA): AACR; Clin Cancer Res 2022;28(18_Suppl):Abstract nr B027.
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 | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.005 | 0.002 |
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".