Caris GPSai
Caris GPSai™, a Genomic Prevalence Score, uses whole exome (DNA) sequencing and whole transcriptome (RNA) sequencing coupled with machine learning to aid in identifying the tissue of origin.
Leverage Artificial Intelligence to identify Tumor Origin and Guide Treatment with Caris GPSai
Typically, a Cancer of Unknown Primary (CUP) diagnosis is treated empirically and has very poor outcomes, with median overall survival less than one year1. GPSai analyzes genomic and transcriptomic data to match a tumor’s molecular signature across 21 cancer types from the Caris database. GPSai is intended to provide additional insight to help oncologists better manage cancers of unknown primary (CUP) or cases with atypical clinical presentation or clinical ambiguity, as identified by the ordering physician.
GPSai Data Results
The GPSai algorithm trained on genomic data from 34,352 cases and genomic and transcriptomic data from 23,137 cases, and was validated on an independent data set of 19,555 cases2. Study results showed:
Summary of performance in the prospective validation cohort
| Category | n | Above Threshold | Call Rate (%) | Sensitivity in Top 1(%) | Sensitivity in Top 2(%) | Sensitivity in Top 3(%) | Sensitivity in Top 4(%) | Sensitivity in Top 5(%) | Rule Outs/ Case | Rule Out Accuracy (%) |
|---|---|---|---|---|---|---|---|---|---|---|
| Global | 13,661 | 12,699 | 93 | 94.7 | 97.2 | 97.9 | 98.1 | 98.2 | 17.6 | 99.9 |
| Primary Specimen | 7,521 | 7,087 | 94.2 | 96.1 | 98.2 | 98.7 | 98.8 | 98.9 | 17.8 | 100 |
| Metastatic Specimen | 5,042 | 5,426 | 91.3 | 93 | 96 | 97 | 97.2 | 97.4 | 17.4 | 99.9 |
| Percent Tumor <20, ≤50 | 4 | 3 | 75 | 100 | 100 | 100 | 100 | 100 | 18.7 | 100 |
| Percent Tumor ≥20, ≤50 | 8,227 | 7,636 | 92.8 | 94.5 | 97 | 97.8 | 97.9 | 98 | 17.4 | 99.9 |
| Percent Tumor >50, ≤80 | 5,189 | 4,835 | 93.2 | 95 | 97.7 | 98.2 | 98.4 | 98.5 | 17.9 | 100 |
| Percent Tumor >80 | 241 | 225 | 93.4 | 96 | 96.4 | 96.4 | 96.4 | 96.9 | 18 | 99.9 |
GPSai - Finding Misdiagnosed Patients
Considering that rates of inaccurate diagnosis ranges between 3% – 9%3, GPSai provides an additional integral part of quality control and could lead to improved diagnostic accuracy.
Caris molecular profiling performed with GPSai provides a comprehensive workup.
- Ability to identify misdiagnosed patients.
- Pathology proactively contacts treating oncologists to discuss discordant results to optimize patient care.
GPSai can be added to any solid tumor order by selecting the appropriate box on the Caris requisition. Results for GPSai will appear in the final Caris report, with no additional cost or added specimen requirements. These results will provide additional insight by assessing how closely tumors match the genomic and transcriptomic signatures of cancer types to help you make more informed treatment decisions.
Example Caris Report: Caris GPSai (Genomic Prevalence Score)
Caris GPSai is not available in New York
- Massard C, Loriot Y, Fizazi K. Carcinomas of an unknown primary origin–diagnosis and treatment. Nat Rev Clin Oncol. 2011 Nov 1;8(12):701-10. doi: 10.1038/nrclinonc.2011.158. PMID: 22048624
- Abraham, J., Spetzler, D., et al. (2021) Machine learning analysis using 77,044 genomic and transcriptomic profiles to accurately predict tumor type. Translational Oncology, 14(3) 101016. https://doi.org/10.1016/j.tranon.2021.101016
- Peck, M., Moffat, D., Latham, B., Badrick, T., Review of diagnostic error in anatomical pathology and the role and value of second opinions in error preventions, J. Clin. Pathol. 71 (11) (2018) 995-1000.