Interview with Kevin Hrusovsky part 3: the role of next generation sequencing in personalized medicine
Rebecca Aris interviews Kevin Hrusovsky
As part of our personalized medicine themed month we interviewed Kevin Hrusovsky. In this, the final part of a three part interview, Kevin discusses the role that next generation sequencing will play in personalized medicine.
(Continued from Interview with Kevin Hrusovsky part 2: next generation sequencing)
Previously in this three-part interview Kevin shared with us the most recent exciting advances in the understanding of molecular interactions of diseases, in addition to next generation sequencing (NGS) and the role it plays in improving patient outcomes.
In this, the third part of the interview, Kevin addresses the challenges associated with NGS and the role NGS will play in personalized medicine.
RA: What challenges are associated with next generation sequencing?
KH: Sample preparation and workflow: Sample size continues to fall as physicians move towards less invasive sampling techniques such as fine needle aspirates and circulating tumor cells, while demand for more comprehensive and granular data increases, as we move into the era of big data where “more is more” in terms of clinical data. Given that a single sample may have to provide NGS data in addition to transcriptional, methylation and proteomic data from a much smaller sample, there is a compelling need for efficient sample prep technologies and work flows that can maximize the information content of each precious patient sample in a cost-effective manner. Eliminating false positives resulting from PCR workflows is another critical requirement to moving into the clinic. And finally, importantly, this process needs to be fast in the case of detecting rare genetic diseases in newborns, where speedy diagnosis can be life-saving.
Bioinformatics / data storage and analysis: The physical pace of large-scale high throughput sequencing has profoundly exceeded our ability to store and analyze the resultant data. While we have done a stellar job in building disruptive tools to sequence the DNA, we need to do a much better job in developing innovative bioinformatics tools that interpret the variants and translates them into actionable clinical / medical reports. A lot of people talk about this bottleneck, but so far the industry has not kept pace with advances in sample prep and the powerful NGS research tool..
“…the DNA sequence in one part of a tumor may have different mutations than another part of the tumor…”
Assuming we solve the basic technical challenges of NGS, such as sample preparation, bioinformatics and getting the all-in cost down to $1000 or less per genome, then we need to tackle the hairy issues associated with clinical sequencing – and keep in mind that the stakes are much higher:
Utility: People are still grappling with the medical value of genomic sequencing. Just because you know the patient’s DNA sequence doesn’t means you know how to treat him or her. For some diseases, such as Huntington’s disease or, the treatment does not even exist.
Accuracy: While research-grade sequencing data is fine for R&,D, it is NOT fine for clinical sequencing – DNA sequence data that is used in a clinical setting to diagnose and treat patients must overcome a much higher accuracy hurdle.
Interpretation: Interpreting a diagnosis is not easy because much of the variation in the genome remains poorly understood. Nearly all of the millions of genetic polymorphisms observed to date remain of unknown clinical significance.
Importantly, no clinical-grade general database of disease-associated mutations currently exists, and interpreting the clinical significance of mutations relies on literature and databases found in the public domain. It is important to note, that most clinicians have not been schooled in these technologies so clinical adoption of new technologies is often paced by the ability to generate a clinically actionable report from the research results.
Tumor heterogeneity: The ability to pinpoint many cancers to specific DNA mutations makes it the “killer app” disease for NGS. However, in March 2012 it was discovered that tumors are much more heterogeneous than we ever envisaged – in other words, the DNA sequence in one part of a tumor may have different mutations than another part of the tumor, and respond differently to treatment. On top of that, the patient may have multiple tumors, each of which carries different mutations. Some of our collaborations are yielding that the circulating tumor cell many times has a very different DNA / protein signature than the primary tumor. And to make matters even worse, all of those tumors are likely to mutate over time. So:
How do we sample a tumor – or tumors, in the case of metastasis – to make sure that we get a good representation of all the mutations?
Since most tumors acquire mutations and become resistant to treatment, how often do we resequence so that treatment can keep pace with the cancer? Most importantly, are we sure our sampling and sample prep processes are yielding the tumor signature that the treatment should be acting against.
Regulation: No one has figured out how clinical sequencing should be regulated.
Ethics: Sequencing often reveals information that you weren’t looking for. Which information do you give to physicians and patients? Should children be told of increased risk for adult-onset disease? What do you tell their family members who make carry the same genetic risk factors?
Intellectual Property: Currently a number of genes are protected by valid patents, so does this open up sequence providers to the risk of being sued for infringement? Even if a gene patent pool were formed to provide efficient and affordable access to genes, how would one put a value on each gene? How would such a system be policed?
Education: Most physicians &, healthcare providers do not have a good understanding of what NGS offers, or how to implement the information.
Logistics: lEectronic health records (EHR) is slowly being integrated into the healthcare system, but the logistics of NGS may overwhelm the system. If you think about the bioinformatics bottleneck just at the DNA sequence level, imagine if you overlay that with individual medical records and HIPAA compliance – it’s pretty daunting.
“…no one has figured out how clinical sequencing should be regulated.”
How much to sequence – whole exome sequencing (WES) vs. whole genome sequencing (WGS)? The jury is still out. There are many anecdotal examples where each approach has helped patients, depending on the disease.
WES provides information on the ~2% of the genome that encodes proteins – i.e. the “business” part of the genome – so it is considerably cheaper than WGS, at least for now. Given that (1) most (85%) mutations known to have a large impact on disease reside in this 2% of coding DNA and (2) it is hard to interpret mutations in non-coding DNA, a cost:benefit argument can be made for WES.
However, a recent breakthrough study from a global sequencing initiative showed that >,80% of the genome is involved in “business” – in fact, most of the part of the genome that until last month we dismissed as “junk DNA” is involved in regulating the 2% of coding DNA. While this is new information that the scientific community is still digesting, it does raise that notion the WGS is a good idea after all, and worth the additional cost and complex logistics (bioinformatics).
Reimbursement: Until tests are reimbursed, their adoption is often blocked. Most ngs testing today is not being reimbursed. There is considerable energy to change this but this level of change typically does not come easily.
RA: What needs to change in order for these challenges to be overcome?
KH: Electronic health records (EHR): The system will need to expand in terms of size and sophistication in order to hold and interpret NGS data.
User-friendly physician-decision support systems to interface with medical practitioners and the electronic medical records will be needed to guide clinicians in using genetic information to select the right drugs and doses for each patient.
Education: Being able to harness NGS and other ‘omics technologies for personalized healthcare will require a genetically literate workforce, starting with the primary care physician.
Genomics training program must be designed for use in medical school curricula, residency training, and the re-education of mature physicians.
Healthcare providers (physicians, physician assistants, nurse practitioners) will need better interpretive and communication skills when it comes to genetic information.
“The convergence of high clinical utility and low costs will be key to seeing the widespread adoption of NGS in personalized medicine…”
Clinical Practice Guidelines: These are important in guiding physicians with respect to various technologies and disease indications. Such guidelines should be established for clinical NGS for both diagnosis and treatment of patients.
In addition to providing a clear pathway for physicians in helping to apply this new tool for treating and diagnosing their patients, it should help mitigate any concerns about the risk of litigation.
RA: What do you see to be the future role of NGS in personalized medicine?
#1: Clinical diagnosis, prognosis and treatment of disease:
The convergence of high clinical utility and low costs will be key to seeing the widespread adoption of NGS in personalized medicine – and we are getting very close to that point of convergence.
Applications where NGS will likely have routine utility include:
1) Determining a patient’s prognosis
2) Determining the best course of treatment
3) Determining who would benefit from preventive drugs – or conversely, who would suffer from side effects (that is, pharmacogenomics)
4) Determining the cause of a mystery disease
#2: Routine Part of Healthcare Practice for both Treatment and Prevention of Disease:
A back of the envelope calculation by Isaac Kohane and Jay Shendure (Harvard and U.Washington, respectively), showed that if we assume an all-inclusive cost of $1000 per genome, then the cost of sequencing one’s genome at birth can be amortized over the age lifetime (78 years), which comes out to $13 per year. Given that in the US we are currently spending about $9000 per person on health care, that sounds like a good deal. Keep in mind that costs are likely to drop- further and the useful information we can glean from genomes will grow, so this will only get better.
“This means that we can start to address personalized health before birth …”
In June, a study by Jay Shendure and colleagues showed that the whole genome of a human fetus could be sequenced in a safe, non-invasive manner. This means that we can start to address personalized health before birth, including the mother’s health, which plays a significant role in shaping the baby’s entire life. Also, it opens up the exciting possibility of prenatal therapy – if we can diagnose a baby before birth, then we can start to develop treatments that will help that individual before he or she is even born.
#3: Predicting Disease Risk:
However, we should set realistic expectations and keep in mind that primary DNA sequence is only part of what influences health and disease: other factors, including epigenetics, play an equally critical role.
While knowing a person’s sequence is useful for understanding disease and finding new treatments, it does not appear to be very useful for predicting a person’s medical future.
The limitation of NGS was highlighted in a large study that came out in May 2012: to address whether we could use genomic data to predict the lifelong risk for a specific disease, researchers sequenced the genomes of a large set of identical (monozygotic) twins, and looked at their susceptibility to 24 common diseases. Despite the fact that these twins have identical genomes, it turns out that they had very little concordance in terms of who got what disease. In other words, most people have an average chance of getting a common disease, based on DNA sequence alone. The factors that tip the balance towards – or away from – disease is behaviour, lifestyle and random events. In other words, epigenetics.
This implies that sequencing a person’s genome from birth may or may not be an effective use of resources – just because you can do it doesn’t mean it’s going to be useful.
About the interviewee:
Kevin Hrusovsky, former Caliper CEO, was appointed President, Life Sciences and Technology, Perkin Elmer in November 2011, following the acquisition of Caliper Life Sciences. In addition to leading the Life Science segment of PerkinElmer, he is also responsible for Services for LST products and Informatics, which was recently moved into LST.
The union of PerkinElmer and Caliper brought together many innovative technologies which, when combined, offer what Hrusovsky believes is a remarkably comprehensive and disruptive portfolio that enables translational medicine and personalized health.
A key part of the mission at LST is to revolutionize global personalized health by facilitating early detection, next generation treatment and ultimately prevention. Importantly, given PerkinElmer’s long track record as a leader in environment health, the Company’s view that helping to sustain the earth’s ecosystem is integral to our goal of preventing and eradicating disease. As part of the drive towards personalized health, earlier this year PerkinElmer announced a new initiative – the creation of Centers of Personalized Health Innovations (cPHI). The first cPHI facility is currently being built in Hopkinton, Massachusetts, and is scheduled to be complete by mid-2013. Through these Centers, PerkinElmer will be able to leverage our innovative strengths in imaging, microfluidics, next generation sequencing, biotherapeutics, epigenetics, companion diagnostics and informatics, with a focus on enabling.
What do you see to be the future role of next generation sequencing in personalized medicine?