Predicting Cancer Recurrence – Genomic Health

Genomic Health is a genomic-based diagnostics company that helps predict cancer recurrence.

Genomic Health is a genomic-based diagnostics company based in Redwood City, CA. Their core product is called Oncotype DX, a genomic assay that helps quantify the likelihood of breast cancer recurrence.

Value Creation

Every year, millions of people are diagnosed with cancer.  With little information available to help support treatment decisions, people usually opt for whatever tools are possible and often pursue chemotherapy. Studies have shown that chemotherapy only helps improve patient survival rates in a low percentage of patients. Despite this low percentage of success, patients often still take the treatment because they don’t have information on how they will personally respond to it. For many, this treatment is either not needed or won’t be helpful, but they still incur the cost of the treatment and must manage both the short and longer term side-effects.

Diagnostic tests have emerged to help doctors understand more about the possible recurrence of cancer after tumors have been surgically removed. This information helps doctors and patients make more educated decisions on whether they need to or should pursue chemotherapy. This type of information is valuable because it helps patients understand whether they will benefit from chemotherapy and could potentially save them from the costs and side effects of unnecessary treatment.

Standard tests look at only a few measures such as patient age, tumor size, and tumor grade to provide an assessment of cancer recurrence. Genomic Health has innovated through the use of genetic information and analysis to provide a much more accurate assessment of cancer recurrence.  Using a 21-gene panel, Genomic Health looks at far more variables in their diagnostic test and are able to come up with a more accurate “recurrence score” for their patients.

Using Data and Analysis: Regression

By analyzing 21 genes per patient, my assumption is that they are able to use this data and analysis through regression to build a predictive model that helps generate the recurrence score they provide to patients. The equation probably looks like:

Recurrence Score = β0 + β1Gene1 + β2Gene2… β21Gene21 + ε

By collecting historic data and outcomes, they had a training set from which they could build this formula. They can then use this formula on new tumor genetic information from patients to predict the likelihood of recurrence.

Value Capture

Unlike many other diagnostic tests which come in the forms of “kits” that are sent to doctors and patients, Genomic Health aims to provide the test as a service. Tumor samples are stored in paraffin blocks and sent to Genomic Health which does all the processing, testing, and generation of results.

This is also probably advantageous because they can continue to aggregate more sample data with which they can continue to train and improve their predictive model.

Genomic Health has faced many challenges, some of which have included issues with reimbursement and general awareness. But their overall approach to using genetic data to build predictive models around personal health seems like a highly strategic space to be building out.

Previous:

Starbucks: Brewing up a data storm!

Next:

Hudson’s Bay: Big Data for the Perfect Fit

Student comments on Predicting Cancer Recurrence – Genomic Health

  1. Interesting stuff… I guess this all makes sense, but I wonder whether we see here the limitations of Data analytics. As a patient being diagnosed with cancer, do I really follow what the data tells me? If I have an output sheet telling me, you have only a 5% chance that the chemotherapy will help you (but the 5% chance could really change the game), how will that change my decision? I am just not sure how in these extreme situations the decision making process could just be completely different and maybe less rational then we expect it in other circumstances…

    Would be interesting to read something about how the adoption works and how patients react to seeing the data…

  2. Great post!!
    I wonder what type of regulations they have to face when offering these solution to doctors. and whether regulations can stop them from growing.

Leave a comment