Machine learning analysis using 77,044 genomic and transcriptomic profiles to accurately predict tumor type

Abraham, Jim; Heimberger, Amy B.; Marshall, John L.; Heath, Elisabeth I.; Helmstetter, Anthony; Xiu, Joanne; Magee, Daniel; Stafford, Phillip; Nabhan, Chadi; Antani, Sourabh; Johnston, Curtis; Oberley, Matthew James; Korn, W. Michael; Spetzler, David

doi:10.1016/j.tranon.2021.101016

Cited by 24 publications

(20 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Several retrospective studies have attempted to assess the prediction accuracy of these tests performed on biopsy specimens from patients with CUP. Using correlation with clinicopathological features, the IHC profile or the identification of a latent primary as prediction comparator, these molecular based tissue of origin classifiers yield prediction accuracy from 60% to 92% [15,[21][22][23][24][25][26][27][28][29][30]. This is corroborated by a prospective study demonstrating an 84% agreement of molecular profile with clinicopathological diagnosis [31].…”

Section: Tissue Of Origin Classifier Assaysmentioning

confidence: 67%

See 1 more Smart Citation

Redefining cancer of unknown primary: Is precision medicine really shifting the paradigm?

Thibault

Fernández

Labidi-Galy

et al. 2021

Cancer Treatment Reviews

View full text Add to dashboard Cite

The concept of Cancer of Unknown Primary (CUP) has evolved with the advent of medical oncology. CUP can be difficult to diagnose and represents 2 to 5% of new cancers, therefore not exceptionally rare. Within CUPs can be identified a subset of favourable prognosis tumours, however the vast majority of CUP patients belongs to a poor prognosis group.CUP features significant oncological challenges, such as unravelling biological and transversal issues, and most importantly, improving patient's outcomes. In that regard, CUP patients' outcomes regrettably showed minimal improvement for decades and CUP remains a cancer group of very poor prognosis.The biology of CUP has two main hypotheses. One is that CUP is a subgroup of a given primary cancer, where the primary is present but cannot be seen due to its small size. The other, the "true" CUP hypothesis, states that CUP share features that make them a specific entity, whatever their tissue of origin. A true biological signature has not yet been described, but chromosomal instability is a hallmark of poor prognosis CUP group.Precision oncology, despite achieving identifying the putative origin of the CUP, so far failed to globally improve outcomes of patients. Targeting molecular pathways based on molecular analysis in CUP management is under investigation. Immunotherapy has not shown ground-breaking results, to date. Accrual is also a crucial issue in CUP trials.Herein we review CUP history, biological features and remaining questions in CUP biology, the two main approaches of molecular oncology in CUP management, in order to draw perspectives in the enormous challenge of improving CUP patient outcomes.

show abstract

Section: Tissue Of Origin Classifier Assaysmentioning

confidence: 67%

“…Even with such sophisticated tools, this yields in 71.7% of prediction in CUPs. This approach also have the originality to provide, at the same time, results about potential targetable genomic alteration [30].…”

Section: T Olivier Et Almentioning

confidence: 99%

Redefining cancer of unknown primary: Is precision medicine really shifting the paradigm?

Thibault

Fernández

Labidi-Galy

et al. 2021

Cancer Treatment Reviews

View full text Add to dashboard Cite

show abstract

“…Although more research is indeed necessary to further elucidate VHL's role in the development of such tumors, our case suggests that molecular characterization of tumors provides information not obtainable by histological examination. Molecular signatures such as GPSai TM scoring contribute to a greater understanding of tumor genesis, as such tools have been found to have an accuracy of 94% in identifying tumor type when matching with a high probability [ 15 ]. Furthermore, commercially available NGS and GPS are tools available to clinicians and can provide significant utility in establishing diagnoses and assessing responses to treatments.…”

Section: Discussionmentioning

confidence: 99%

Germline VHL Mutation Discovered in Association with EGFR-Positive Lung Cancer and Metachronous Hepatocellular Carcinoma: A Case Report

et al. 2021

Self Cite

View full text Add to dashboard Cite

VHL is a tumor suppressor gene located on chromosome 3 that is classically associated with tumors of the eye and CNS, renal cell carcinoma, and pheochromocytoma. We describe what appears to be the first report of an association between a germline VHL mutation and non-small cell lung cancer and metachronous hepatocellular carcinoma (HCC). Our case involves a 63-year-old nonsmoking male who was initially diagnosed with EGFR mutation-positive metastatic nonsquamous, non-small cell lung adenocarcinoma, who subsequently developed HCC and squamous cell carcinoma of the femur despite first-line treatment with EGFR-blocking osimertinib. Caris molecular profiling unexpectedly identified a shared underlying VHL mutation in all 3 lesions. Genetic mapping through a machine learning-based tool called Genomic Prevalence Score (GPSai<sup>™</sup>) helped determine that the femur tumor was a metastatic lesion as opposed to a separate primary and that the HCC was a distinct primary malignancy. We not only highlight the association between these tumors and a VHL mutation but also emphasize the value of next-generation sequencing and a molecular disease classifier in a patient with multiple primaries, how it helps guide therapy, and its value in guiding future studies.

show abstract

“…Also, CDH1 mutation, which is frequently reported in plasmacytoid UC, was not identified [ 9 ]. Furthermore, we applied a Genomic Prevalence Score (MI GPSai™) using whole-exome sequencing and whole transcriptome (RNA) analysis coupled with machine learning to confirm tumor origin [ 14 ]. MI GPSai predicts the tumor type with an accuracy of 94% by matching tumor molecular signature across 21 cancer types.…”

Section: Molecular Studiesmentioning

confidence: 99%

Colonic Adenocarcinoma with Plasmacytoid Feature: Histopathology and Molecular Characteristics of a Rare Neoplasm with an Unusual Presentation

Marji

Dhillon

Lauwers

et al. 2022

Case Reports in Pathology

View full text Add to dashboard Cite

Colorectal carcinoma with noncohesive tumor cells has been described in tumors with signet ring cells (mucinous adenocarcinoma and signet ring cell adenocarcinoma) and rhabdoid feature (carcinoma with sarcomatoid component). Cases of carcinoma with plasmacytoid morphology are rare in the gastrointestinal tract, and a single case of plasmacytoid colorectal carcinoma has been reported. We report the case of a 37-year-old woman who presented with urinary symptoms, hematuria, and abdominal pain. Imaging studies showed segmental sigmoid wall thickening with pericolic infiltration and focal bladder wall thickening. The cystoscopy with transurethral resection of bladder tumor revealed muscle invasion, dis-cohesive carcinoma with plasmacytoid morphology, which was initially misdiagnosed as the plasmacytoid urothelial carcinoma. Immunohistochemical stains showed the tumor cells to be positive for CDX2, CK20, and SATB2 and negative for p63, GATA3, CK7, and Uroplakin II, indicating the colorectal origin of the tumor. The subsequent colonic wall biopsy showed the same tumor. Molecular studies identified BRAF V600E, SMAD4, and p53 mutations associated with aggressive colorectal adenocarcinoma with mucinous/signet ring cell features. Further whole-exome sequencing and whole transcriptome analysis confirmed the colorectal origin of the tumor. This rare colorectal adenocarcinoma with the plasmacytoid feature may represent the signet ring cell adenocarcinoma lacking extracellular mucin or intracellular vacuole. Diagnosis of this rare histological subtype of colorectal carcinoma is important, particularly in the unusual presentation of this aggressive tumor.

show abstract

Machine learning analysis using 77,044 genomic and transcriptomic profiles to accurately predict tumor type

Cited by 24 publications

References 31 publications

Redefining cancer of unknown primary: Is precision medicine really shifting the paradigm?

Redefining cancer of unknown primary: Is precision medicine really shifting the paradigm?

Germline VHL Mutation Discovered in Association with EGFR-Positive Lung Cancer and Metachronous Hepatocellular Carcinoma: A Case Report

Colonic Adenocarcinoma with Plasmacytoid Feature: Histopathology and Molecular Characteristics of a Rare Neoplasm with an Unusual Presentation

Contact Info

Product

Resources

About