Larry H. Bernstein, MD, FCAP, Curator
http://pharmaceuticalintelligence/7/6/2014/Genomics, Proteomics and standards
This article is a look at where the biomedical research sciences are in developing standards for development in the near term.
Let’s Not Wait for the FDA: Raising the Standards of Biomarker Development – A New Series
published by Theral Timpson on Tue, 07/01/2014 – 15:03
We talk a lot on this show about the potential of personalized medicine. Never before have we learned at such breakneck speed just how our bodies function. The pace of biological research staggers the mind and hints at a time when we will “crack the code” of the system that is homo sapiens, going from picking the low hanging fruit to a more rational approach. The high tech world has put at the fingertips of biologists just the tools to do it. There is plenty of compute, plenty of storage available to untangle, or decipher the human body. Yet still, we talk of potential.
Chat with anyone heavily involved in the life science industry–be it diagnostics or pharma– and you’ll quickly hear that we must have better biomarkers.
Next week we launch a series, Let’s Not Wait for the FDA: Raising the Standards of Biomarker Development, where we will pursue the “hotspots” that are haunting those in the field.
The National Biomarker Development Alliance (NBDA) is a non profit organization based at Arizona State University and led by the formidable Anna Barker, former deputy director of the NCI. The aim of the NBDA is to identify problem areas in biomarker development–from the biospecimen and sampling issues to experiment design to bioinformatics challenges–and raise the standards in each area. This series of interviews is based on their approach. We will purse each of these topics with a special guest.
The place to start is with samples. The majority of researchers who are working on biomarker assays don’t give much thought to the “story” of their samples. Yet the quality of their research will never exceed the quality of the samples with which they start–a very scary thought according toCarolyn Compton, a former pathologist, now professor of pathology at ASU and Johns Hopkins. Carolyn worked originally as a clinical pathologist and knows first hand the the issues around sample degradation. She left the clinic when she was recruited to the NCI with the mission of bringing more awareness to the issue of bio specimens. She joins us as our first guest in the series.
That Carolyn has straddled the world of the clinic and the world of research is key to her message. And it’s key to this series. As we see an increased push to “translate” research into clinical applications, we find that these two worlds do not work enough together.
Researchers spend a lot of time analyzing data and developing causal relationships from certain biological molecules to a disease. But how often do these researchers consider how the history of a sample might be altering their data?
“Garbage in, garbage out,” says Carolyn, who links low quality samples with the abysmal non-reproducable rate of most published research.
Two of our guests in the series have worked on the adaptive iSpy breast cancer trials. These are innovative clinical trials that have been designed to “adapt” to the specific biology of those in the trial. Using the latest advances in genetics, the iSPY trials aim to match experimental drugs with the molecular makeup of tumors most likely to respond to them. And the trials are testing multiple drugs at once.
Don Berry is known for bringing statistics to clinical trials. He designed the iSpy trials and joins us to explain how these new trials work and of the promise of the adaptive design.
Laura Esserman is the director of the breast cancer center at UCSC and has been heavily involved in the implementation of the iSpy trials. Esserman is concerned that “if we keep doing conventional clinical trials, people are going to give up on doing them.” An MBA as well as an MD, Esserman brings what she learned about innovation in the high-tech industry to treatment for breast cancer.
From there we turn to the topic of “systems biology” where we will chat with George Poste, a tour de force when it comes to considering all of the various aspects of biology. Anyone who has ever been present for one of George’s presentations has no doubt come away scratching your head wondering if we’ll ever really glimpse the whole system that is a human being. If there is one brain that has seen all the rooms and hallways of our complex system, it’s George Poste.
We’ll finish the series by interviewing David Haussler from UCSC of Genome Browser fame. Recently Haussler has worked extensively on an NCI project, The Cancer Genome Atlas, to bring together data sets and connect cancer researchers around the world. What is the promise and pitfalls David sees with the latest bioinformatics tools?
George Poste says that in the literature we have identified 150,000 biomarkers that have causal linkage to disease. Yet only 100 of these have been commercialized and are used in the clinic. Why is the number so low? We hope to come up with some answers in this series.
Why Hasn’t Clinical Genetics Taken Off? (part 2)
published by Sultan Meghji on Fri, 06/20/2014 – 14:49
In my previous post, I made the broad comment that education of the patient and front line doctors was the single largest barrier to entry for clinical genetics. Here I look at the steps in the scientific process and where the biggest opportunities lie:
The Sequencing (still)
PCR is a perfectly reasonable technology for sequencing in the research lab today, but the current configuration of technologies need to change. We need to move away from an expert level skill set and a complicated chemistry process in the lab to a disposable, consumer friendly set of technologies. I’m not convinced PCR is the right technology for that and would love to see nanopore be a serious contender, but lack of funding for a broad spectrum of both physics-only as well as physical-electrical startups have slowed the progress of these technologies. And waiting in the wings, other technologies are spinning up in research labs around the world. Price is no longer a serious problem in the space – reliable, repeatable, easy to use sequencing technologies are. The complexity of the current technology (both in terms of sample preparation and machine operation) is a big hurdle.
The Analysis (compute)
Over the last few years, quite a bit of commentary and effort has been put into making the case that the compute is a significant challenge (including more than a few comments by yours truly in that vein!). Today, it can be said with total confidence that compute is NOT a problem. Compute has been commoditized. Through excellent new software to advanced platforms and new hardware, it is a trivial exercise to do the analysis and costs tiny amounts of money ($<25 per sample on a cloud provider appears to be the going rate for a clinical exome in terms of platform & infrastructure cost). Integration with the sequencer and downstream medical middleware is the biggest opportunity.
The Analysis (value)
The bigger challenge on the analysis is the specific things being analyzed as mapped to the needs of the patient. We are still in a world where the vast majority of the sequencing work is being done in support of a specific patient with a specific disease. There isn’t even broad consensus yet in the scientific community about the basics of the pipeline (see my blog posthere for an attempt at capturing what I’m seeing in the market). A movement away from the recent trend in studying specific indications (esp. cancer) is called for. Broadening the sample population will allow us to pick simpler, clearer and easier pipelines which will then make them more adoptable. It would be a massive benefit to the world if the scientific, medical and regulatory communities would get together and start creating, in a crowdsourced manner, a small number of databases that are specifically useful to healthy people. Targeting things like nutrition, athletics, metabolism, and other normal aspects of daily life. A dataset that could, when any one person’s DNA is references, would find something useful. Including the regulators is key so that we can begin to move away from the old fashioned model of clearances that still permeate the industry.
Beyond the broader issues around education I referenced in my previous post, there is a massive upgrade in the regulation infrastructure that is needed. We still live in a world of fax machines, overnight shipping of paper documents and personal relationships all being more important than the quality of the science you as an innovator are bringing to bear.
Consider the recent massive growth in wearables, fitness trackers and other instrumentation local to the human body. Why must we treat clinical genetics simply as a diagnostic and not, as it should be, as a fundamental set of quantitative data about your body that you can leverage in a myriad of ways. Direct to consumer (DTC) genetics companies, most notably 23andme, have approached this problem poorly – instead of making it valuable to the average consumer, what they’ve done is attempted to straddle the line between medical and not. The Fitbit model has shown very clearly that lifestyle activities can be directly harnessed to build commercial value in scaling health related activities without becoming a regulatory issue. It’s time for genetics to do the same thing.
Posted by @finchtalk on July 3, 2014
A few weeks back, we published a review about the development and role of the human reference genome. A key point of the reference genome is that it is not a single sequence. Instead it is an assembly of consensus sequences that are designed to deal with variation in the human population and uncertainty in the data. The reference is a map and like a geographical maps evolves though increased understanding over time.
From the Wiley On Line site:
Genome maps, like geographical maps, need to be interpreted carefully. Although maps are essential to exploration and navigation they cannot be completely accurate. Humans have been mapping the world for several millennia, but genomes have been mapped and explored for just a single century with the greatest advancements in making a sequence reference map of the human genome possible in the past 30 years. After the deoxyribonucleic acid (DNA) sequence of the human genome was completed in 2003, the reference sequence underwent several improvements and today provides the underlying comparative resource for a multitude genetic assays and biochemical measurements. However, the ability to simplify genetic analysis through a single comprehensive map remains an elusive goal.
- Maps are incomplete and contain errors.
- DNA sequence data are interpreted through biochemical experiments or comparisons to other DNA sequences.
- A reference genome sequence is a map that provides the essential coordinate system for annotating the functional regions of the genome and comparing differences between individuals’ genomes.
- The reference genome sequence is always product of understanding at a set point in time and continues to evolve.
- DNA sequences evolve through duplication and mutation and, as a result, contain many repeated sequences of different sizes, which complicates data analysis.
- DNA sequence variation happens on large and small scales with respect to the lengths of the DNA differences to include single base changes, insertions, deletions, duplications and rearrangements.
- DNA sequences within the human population undergo continual change and vary highly between individuals.
- The current reference genome sequence is a collection of sequences, an assembly, that include sequences assembled into chromosomes, sequences that are part of structurally complex regions that cannot be assembled, patches (fixes) that cannot be included in the primary sequence, and high variability sequences that are organised into alternate loci.
- Genetic analysis is error prone and the data require validation because the methods for collecting DNA sequences create artifacts and the reference sequence used for comparative analyses is incomplete.