Big Data Meets DNA
Oncology is the one area where targeted treatment has been hugely successful, because tumors are driven by genetic mutations. One of the first home runs was Genentech’s breast cancer treatment Herceptin. That drug and its companion diagnostic test were approved simultaneously in 1998, and today the drug still brings in over $1 billion a year. The test measures levels of HER2, a protein that is overexpressed in some tumors. Several other top-selling targeted cancer drugs followed.4
But in other fields, the fruits of genomics were not as obvious. Since 2005 and today, about 1,500 genetic association studies have been published.5 These have found hundreds of variations linked to a wide variety of conditions, including diabetes, Crohn’s disease, and heart disease. But most of these variations contribute only slightly to the risk of developing a particular disease.6 It’s been estimated that over $100 million has spent looking for such variations, with few clearly important targets unearthed so far.
But many more sophisticated approaches to unearthing biomarkers are emerging. This is fueled not just by a greater understanding of human biology and its complexity, but by next-generation tools that enable much more precise measurements of all types of biological markers including DNA, RNA, and proteins. For example, point errors in genome sequences have dropped from 1 in 100,000 to 1 in 10 million.7
Over the last few years, gene sequencing has become orders of magnitude faster and less expensive. Hospital systems are investing millions of dollars to create genomic medicine units. Mount Sinai alone spent $100 million on such a division. Currently, those units mainly address oncology or hard-to-diagnose diseases. But these hospitals are making the investment in part because they recognize that genomics’ hour is approaching and it will be much more integrated into medicine overall.
Researchers are also looking beyond genomic data. The Center for Assessment Technology and Continuous Health (CATCH), aims to enable “a new understanding of wellness and disease through systemically identifying and annotating patient phenotypes.” They will “improve measurement of patient phenotypes with novel technologies and devices.” By phenotype, CATCH researchers mean a much wider range of data then we previously even knew existed. They will be measuring cellular, behavioral, and other common phenotypes along with things such as the patient’s microbiome, sensor readings from respiratory cilia, and immune cell genotypes.
The amount of clinical data available is also increasing and providing new insights. Archimedes, for example, is a Kaiser Permanente spinoff that uses a mathematical model to analyze healthcare questions. One of their simulations predicted that by prescribing two generic, low-cost drugs to lower cholesterol and blood pressure, Kaiser could prevent many heart attacks and strokes among diabetes and cardiac patients.
Further, pharmaceutical companies are starting to realize that they can greatly accelerate clinical trials by using biomarkers. In 2013 GlaxoSmithKline won FDA approval for Tafinlar for melanoma. That approval was based on a trial of just 250 patients. Because the drug’s mechanism—targeting BRAF V600E—had been established, they were able to enroll many more patients (three times as many) in the Tafinlar arm, versus the control arm. As a result, the company was able to demonstrate the drug’s efficacy more quickly than is usual.8
The potential for new products from this combination of new data and next-generation technologies is immense.
Tests to guide prescribing of cancer drugs alone are a major opportunity. The Pharmaceutical Research and Manufacturers of America (PhRMA) reported that more than 900 medicines and vaccines where in development against cancer by 2012.9 The overall market for cancer treatments reached nearly $36 billion in the U.S. alone in 2012.10 Eleven of the twelve cancer drugs approved by the FDA last year alone cost more than $100,000 per year in the U.S., and some cost more than $300,000.11 Many of these drugs work only in a subset of patients, but biomarkers of response are not yet available for all of them.
Tests that can be used to predict response to such drugs will become more lucrative in the new cost-conscious healthcare environment.