AACR Project Genomics Evidence Neoplasia Information Exchange (GENIE) recently hit another milestone: the public release of a fourth dataset, which comes on the heels of three prior public data releases, an expansion of the number of participating institutions, and a recent publication, which was among the most read articles in JCO Clinical Cancer Informatics in July, according to the publisher.
AACR Project GENIE 101
AACR Project GENIE grew rapidly from an initial concept in early 2014 to a formal consortium that publicly launched at the AACR-NCI-EORTC International Conference on Molecular Targets and Cancer Therapeutics, November 6, 2015. The initiative continues to make rapid progress, as evidenced by three public data releases and the first expansion with new participants.
AACR Project GENIE is an international cancer registry of real-world data assembled through data sharing between 19 of the leading cancer centers in the world. The founding institutions of the consortium are the Dana-Farber Cancer Institute, Boston; Gustave Roussy Cancer Campus, Paris-Villejuif, France; The Netherlands Cancer Institute, Amsterdam, on behalf of the Center for Personalized Cancer Treatment, Utrecht, The Netherlands; Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore; Memorial Sloan Kettering Cancer Center, New York; Princess Margaret Cancer Centre, University Health Network, Toronto; The University of Texas MD Anderson Cancer Center, Houston; and Vanderbilt-Ingram Cancer Center, Nashville.
Through the efforts of strategic partners Sage Bionetworks and cBioPortal, the registry aggregates and visualizes, respectively, the historical and ongoing clinical-grade sequencing data from the tumors of patients treated at participating centers with a limited clinical data set.
It is then possible to build virtual cohorts of patients about whom detailed outcomes data can be abstracted from the relevant medical records. The ability to gain insight into genotype-outcomes relationships can be used to help develop new therapies, design better clinical trials, and personalize treatment, with the ultimate goal of improving the lives of cancer patients through improved clinical decision-making.
What has AACR Project GENIE achieved to date?
A major milestone for AACR Project GENIE was the first public release of cancer genomic data aggregated through the initiative in January 2017. This first release (v. 1.0.1) consisted of 18,804 samples from 18,324 patients contributed by the eight founding participants. The second public release (v. 2.0) increased the number of sequenced tumors to 31,706 from 29,230 patients, and the third release (v. 3.0) increased it further to 39,600 sequenced tumors from 38,207 patients. The current release (v. 4.0) includes 48,447 sequenced tumors from 46,510 patients. We anticipate data from the new participants will appear in the January 2019 (v. 5.0) release and will greatly enhance the size of the registry.
The landscape analysis of the first public data release (v. 1.0.1) provided insight into the potential utility of the registry. The analysis revealed comparable mutation rates and patterns of mutational mutual exclusivity as reported in other public databases, with an enrichment of resistance mutations, likely reflective of the largely late-stage, heavily pretreated population contained within the registry. Additionally, the ability to use the data to calculate tumor mutational burden (TMB) was also demonstrated in the paper and during a presentation during the AACR Annual Meeting 2017, which will be useful in the context of immuno-oncology studies.
The clinical actionability, defined as the ability to provide an FDA-approved therapy or assign a patient to a clinical trial, was calculated as more than 30 percent for the cohort and as high as 38 percent for breast cancer alone. Importantly, this assessment was made possible through the union of three distinct knowledge bases: My Cancer Genome, OncoKB, and Personalized Cancer Therapy, which would not have been possible outside of the consortium. Since the time of the initial publication, a fourth knowledge base, Clinical Interpretations of Variants in Cancer (CIViC), has been integrated into the registry as an additional resource. Work to further harmonize the integration of these knowledge bases is ongoing and was the subject of a Precision Medicine World Conference Silicon Valley 2018 presentation. A similar positive analysis examining the utility of the registry in drug repositioning was also presented during the AACR Annual Meeting 2017.
A major question in the oncology community is whether registries like GENIE can be used to assign patients to clinical trials and/or forecast matching and enrollment in genomically driven trials. As a proof of concept, the GENIE registry was compared to the first interim analysis of the NCI-MATCH trial and the patient match rate compared favorably with the actual reported rates from the trial, with the GENIE registry matching to two arms of the trial that had yet to accrue by the interim analysis.
As a continued service to the research community, the analyses and figures from the landscape paper are rerun with every release and made publicly available. The trends in actionability and clinical trial assignment observed in the original manuscript continue, and with each subsequent release, the trend is towards larger NGS panels and increased levels of overlap at the gene level between panels.
The current genomic data in the registry was generated using 21 unique NGS panels, five more than the 1.0.1 release, ranging from 46 to 468 genes and covering a total of 1,400 unique genes. We anticipate that the trend of increasing panel size will continue with future releases as sequencing evolves at each of the participating centers and data from new participants becomes available.
With the most recent data release, the registry now contains genomic information from 7,682 non-small cell lung carcinomas, more than 5,500 breast cancers, and nearly 5,200 colorectal cancers, representing increases of 21, 22, and 20 percent, respectively, from v. 3.0. The largest increases in major cancer types occurred in embryonal tumors (339 percent), myeloproliferative neoplasias (93 percent), and multiple myeloma (92 percent).
The most commonly mutated genes in the sequenced tumors remain TP53, KRAS, and PIK3CA. CDKN2A and CDKN2B deletions, and CCND1 and MYC amplification, are the most common copy number alterations.
Among our other recent accomplishments are the completion of two natural history studies of rare pathogenic variants in breast cancer, and the respective project teams are excited to present their findings at upcoming conferences.
What’s next for AACR Project GENIE?
AACR Project GENIE started as an experiment among an international group of leading academic centers, a coalition of the willing, that has developed into an international cancer registry through data sharing and pragmatic approaches. As we enter the next phase of the project, we look forward to expanding our collection of clinical data, adding additional new participants, and more clinical and translational research from our group and others.
Our long-term vision continues to be one where every cancer patient around the world will have their tumor genomically sequenced and their physician will use a registry, such as GENIE, to make treatment decisions.
Our modest short-term goals continue to be to amass the evidence linking genetic variants to clinical outcomes, particularly in rare tumors or rare genetic variants; to improve clinical decision making; and to catalyze clinical and translational research. Among the myriad milestones that have been met, the addition of new participating institutions and the current and future public data releases are the accomplishments that the group hold dearest, as we continue to deliver on our promise of openness, transparency, and inclusion. In addition to catalyzing research, the project has catalyzed a culture shift where data sharing is the new norm, helping to change approaches even within the AACR Project GENIE Founding Institutions.
You may also like: