乌鸦传媒

In a recent issue of , Holt et al. described the development and features of My Cancer Genome, a website launched in 2011 with the purpose of guiding treatments for cancer patients who have undergone genomic testing. Though this publicly available resource is geared mainly towards clinicians, helping them identify relevant biological pathways and match patients to appropriate targeted therapies and clinical trials, the platform can also be used for research purposes as well as for education for patients and their families.

Interestingly, the content of the platform is mainly built upon a "wiki style" concept that enables public contributions of content from international experts. However, the creators ran into the problem of scalability as the complexity and sheer volume of precision oncology data continued to skyrocket. It is particularly impressive how a creative solution to this problem was proposed and enacted, by combining "knowledge generated assertion models" and templates to automate the content rapidly and accurately, resulting in a scalable website that can keep up with the influx of genomic data amongst numerous genetic variants and cancer types.

When the website was first designed, the authors reported, there were originally 1,338 genes and alterations, with 25 diseases and 697 drugs. As we fast-forward to today, the algorithm above has truly lived up to the challenge of scalability, now boasting 18,100 genes and alteration pages, 900 disease pages, and 2,700 drug pages with 9,100 clinical trials identified.

The manually inputted and computationally quantified assertions span from therapeutic, prognostic, and diagnostic parameters all the way up to clinical trials and treatments. The main point is to prognostically, diagnostically, and therapeutically match biomarkers or alterations in genes, to diseases, drugs, and trials.

The authors provide examples in which the assertion algorithms are guided by a BRCA 1/2 alteration in breast cancer or an FLT3 ITD alteration in AML, for instance, thereby triggering the program to identify prognostic criteria and treatments for these conditions, such as a PARP inhibitor and an FLT-3 inhibitor, respectively.

As we continue to experience a rapid growth in precision oncology, it will be increasingly important to manage this "big data" in a format that is easy to understand and applicable for clinicians, researchers, and patients alike.

My Cancer Genome is an elegant approach towards this goal, and I particularly found the website to be very comprehensive and complex in its breadth of information, but yet intuitive and simple to use. The ability to generate complex pages of information in an efficient manner while allowing for dynamic updates to information as knowledge evolves is a key differentiating feature of this technology.

I also feel that it will serve as a valuable tool for molecular tumor boards as well as a platform for hypothesis generation to guide future research investigations and clinical trials. Most importantly, however, the authors of this resource demonstrate a commitment to ensuring that there are no compromises in the scientific validity of the information, by conducting validation of data against known databases and literature/guidelines using structured ontologies.

In fact, it was impressive to realize the existence of rigid quality-control processes that involve numerous crosschecks, expert review, and even a quality-control manager to continuously surveil the process.

There are clearly many more challenges to improve this resource including the significant ongoing time investment and requirement for manual curation and expert review which still seems to be a critical part of the process, aligning nuanced clinical trial eligibility criteria, and propagating data on complex cancer pathways.

Hardeep Phull, MD, is a practicing hematologist/oncologist and medical director of Oncology at Palomar Health in San Diego.

Read the study here and an interview about it here.