A deep learning system accurately classifies primary and metastatic cancers using passenger mutation patterns

Abstract

In cancer, the primary tumour's organ of origin and histopathology are the strongest determinants of its clinical behaviour, but in 3% of cases a patient presents with a metastatic tumour and no obvious primary. Here, as part of the ICGC/TCGA Pan-Cancer Analysis of Whole Genomes (PCAWG) Consortium, we train a deep learning classifier to predict cancer type based on patterns of somatic passenger mutations detected in whole genome sequencing (WGS) of 2606 tumours representing 24 common cancer types produced by the PCAWG Consortium. Our classifier achieves an accuracy of 91% on held-out tumor samples and 88% and 83% respectively on independent primary and metastatic samples, roughly double the accuracy of trained pathologists when presented with a metastatic tumour without knowledge of the primary. Surprisingly, adding information on driver mutations reduced accuracy. Our results have clinical applicability, underscore how patterns of somatic passenger mutations encode the state of the cell of origin, and can inform future strategies to detect the source of circulating tumour DNA.

Cancer mutational signatures representation by large-scale context embedding

“…There are other methods that can also extract mutational features for each patient. Here we compared the performance of MutSpace with several other methods, including calculating the frequencies of occurrence for mutational patterns in each patient (Mut-Freq), decomposing the frequency matrix of mutational patterns using the NMF algorithm, and regional mutational density (RMD) ( Jiao et al , 2020 ; Salvadores et al , 2019 ).…”

“…Importantly, the magnitude and patterns of somatic mutations are strongly affected by exposures to exogenous and endogenous mutagens, which also serve important features to classify patients ( Alexandrov et al , 2013 ; Martincorena and Campbell, 2015 ). Previous works have shown that the tissue-of-origin of cancer can be accurately identified using the frequency of non-coding mutations ( Jiao et al , 2020 ; Temiz et al , 2015 ). However, using non-coding somatic mutations for subtype identification remains challenging due to the sparsity of mutation occurrence in the genome, the complexity of mutation patterns and the difficulty to prioritize the mutations ( Fu et al , 2014 ; Watson et al , 2013 ).…”

CUP-AI-Dx: A tool for inferring cancer tissue of origin and molecular subtype using RNA gene-expression data and artificial intelligence

“…The path to improving the tumour type classification accuracy may be to consider including other potential features such as somatic point mutations [79] and histopathology images [80] in the model. Mutational profiling of tumours is steadily being incorporated into mainstream work-up of cancer patients and recently several tissue of origin classification methods have been developed based on DNA features alone either from panel [81] whole-exome, and whole-genome sequencing (WGS) [82] . Interestingly, the reported accuracy of these methods especially when using WGS passenger mutational profiles for the tissue of origin classification is similar to using gene-expression profiling and DNA methylation classification.…”

Hunting down the dominating subclone of cancer stem cells as a potential new therapeutic target in multiple myeloma: An artificial intelligence perspective

“…Together, these results determine the evolutionary trajectories of cancer and highlight opportunities for early cancer detection"[ 40 ]. All of these must rely on "A deep learning system accurately classifies primary and metastatic cancers using passenger mutation patterns"[ 41 ] for integration of dynamic space-time changes. One such integrated platform was to scaffold the diverse datasets together, allowing them to interface not only across single-cell transcriptomics (scRNA-seq), but also across distinct cellular modalities – e.g ., a bone marrow atlas to characterize lymphocyte populations[ 42 ] – to better understand cellular identity and function beyond the taxonomic listing of clusters of cellular heterogeneity[ 43 ].…”