The turning of the calendar from 2011 to 2012 has been a good time for me to reflect on the lessons of the year and make resolutions for the new one. It is also the opportunity to step back and look at some of the larger systemic trends in our field and think about whether we are doing as much as we can with genetics to accelerate the delivery of healthcare solutions to the patient. I’d like to tie together several threads in this article. The first is to pay homage to a man who has done more for me than any other in understanding the nature of reality, Eli Goldratt. Drawing from his lessons on systems and supply chains, I’ll then make some observations about our field’s own genetics supply chain and the consequences of not producing enough value for the end patient. Further, as 2011 marks the 40th year of the war on cancer, Goldratt’s death by the same can lead us to pause and reflect on our own vulnerability to disease and spur us to not just redouble our efforts, but to ask ourselves what we might do systemically to accelerate the translation of innovations in genetics to the clinic.
Eli Goldratt and TOC
The event that had the biggest impact on me in 2011 personally was the death of Eli Goldratt, the father of the management discipline known as the Theory of Constraints (TOC). As a fledgling entrepreneur 13 years ago when I founded Golden Helix, and for the next 7 years, I voraciously read the popular management fad books of the day and tried to implement what I understood to be best practices. As one attractive idea after another failed to bear fruit in our environment, with management “flavor of the month” fatigue setting in, I came across the Theory of Constraints body of knowledge in 2005. All of the lights went on for me, and ever since it has been a dominant paradigm through which I understand organizational systems.
An Israeli PhD physicist by training, Goldratt used corporations as his laboratory setting, and strove to apply the disciplines of the hard sciences to understanding cause and effect within organizations. He has contrasted the stellar advances in the hard sciences with the painfully slow progress humanity has made in the soft sciences — in particular the social sciences and management — and advocated over his life for the rigorous application of the paradigms of the hard sciences to the improvement of human systems. His stated life’s goal was “to teach the world to think.”
His world view as a physicist holds that there is an inherent simplicity to reality in terms of cause and effect. From his definition of complexity as degrees of freedom — a measure of the number of independent variables governing system behavior — he has made the bold statement that there are no complex systems in reality. He developed logical visual thinking tools for bypassing the high detail complexity of systems to pinpoint core causes of problems and find the leverage points of change to make dramatic and continuous improvements in organizational systems. You may see some of that influence in my article last year that looks at the systemic challenges to research productivity in our field: “Dammit Jim, I’m a Doctor, not a Bioinformatician”, as well as in the analysis of a core conflict in the cytogenetics field in my blog post “Rising Above Uncertainty; Increasing Clinical Yield in Array-Based Cytogenetics”.
Unfortunately, the TOC field has been narrowly seen by many as one that applies only to manufacturing. Indeed this is where Goldratt started, with his bestselling business novel, The Goal, being set in a factory production setting, where he began investigations into organizational systems. However, TOC is really a science of systems, of strategy, of supply chains, of conflict resolution, and even of self-knowledge. Further, it has a meta layer of thinking about how we think about reality, as laid out in his more recent book, The Choice.
Goldratt estimated a few years back that approximately 5% of world industry uses TOC in some form. Its influence is silently but powerfully felt throughout the world — whether it be aspects of Walmart’s supply chain, the adoption of critical chain project management (CCPM) as a standard for all public works projects in Japan (The Choice outsold Harry Potter in that country), or the teaching of over 8 million school children the TOC thinking skills through TOC for Education training of 200,000 educators worldwide. In a generation or two, I believe Goldratt’s contributions to the transformation of society will be recognized as monumental and enduring. I also believe the seeds he has planted through his evaporating cloud method for conflict resolution will contribute to the ending of war on this planet in centuries to come.
The genomics industry supply chain
Goldratt made huge contributions towards the understanding of industry supply chains. While we are often not conscious of it within our silos, the genomics industry as a whole is a supply chain (or several), and 2011 has seen some interesting developments in that supply chain. The most disturbing trend I see in the genomics space is the apparent commoditization of links in the supply chain before fledgling industries even have a chance to get off of the ground. The hardware and consumables appear to be commoditizing, and despite claims by some that it will cost $100,000 in informatics to make sense of the $1,000 genome, there are continuous pressures to drop the price of bionformatics, despite it being the bottleneck in the supply chain.
Why might there be pressure in a supply chain towards commoditization? Joel Spolsky wrote a nice analysis of how it is in the economic interest of a company to see the commoditization of products from other companies that complement their own. If bioinformatics is required to make sense of genomics data, and you are from a company that wants to sell more hardware and consumables, the availability of low cost bioinformatics software is apparently in your best interests. One hardware vendor took a step along these lines this year to offer a free cloud-based solution for secondary analysis of sequence data, much to the chagrin of bioinformatics companies who make their living from providing paid-for solutions. Conversely, if you are from a bioinformatics company that makes money by selling data analysis software, it is in your best interests to see the cost of genotyping go down so that more and more scientists generate data and come to you for analysis solutions. However, the drop in prices of sequencing can hardly be ascribed to any pressure that bioinformatics firms might play — let’s look closer at the supply chain for additional understanding.
Commoditization occurs when there are no significant differentiating aspects besides price between products as they become interchangeable. It does not appear with the rapid innovation in our space that we can truly call the drop in prices commoditization. Rather, I hypothesize that downward price pressures in our field stem at least in part from a fundamental principal of supply chains, so well articulated by Goldratt. He said, “as long as the end consumer has not bought, nobody has sold”. What does this mean? Picture a supplier of electronics components for the various gadgets we buy. The supplier may consider his goods sold when he sells to the manufacturer of a television set. However, the component supplier’s sales are capped by what the downstream television manufacturer can sell to the end consumer plus whatever parts are held in inventory. If the television manufacturer is unable to sell his finished goods to the end consumer, the component supplier will very quickly see his sales dramatically fall. No effort to sell more to the television manufacturer will bear fruit — who wants to tie up money in excess inventory?
Let’s map the same idea to the genomics supply chain. Who are the end consumers in our supply chain and what is it they are ultimately buying? It is patients who are buying health and life, and the price pressures come from the apparent reality that the ROI in terms of health and life brought by genetics, while not negligible, is also not so large as to justify the next order of magnitude of expenditures. It is no coincidence that around 2006 and onwards we saw pharmaceutical companies dramatically cut their expenditures in genetic research. The cost of genetic research was high relative to returns, and academia was obligingly overproducing an inventory of genetic signals for disease in the literature at such a rate that pharma pragmatically let the public fund the research and read the free papers. I recall in 2004 attending a core lab directors conference and asking several directors what their load to capacity ratio was for their microarray facilities. Uniformly I found their utilization was around 20%. This was indicative of the pushing of equipment inventory into the supply chain well beyond market demand. Is the same thing likely to happen with sequencing equipment, if it hasn’t already? The only out is for clinical application to hit mainstream, and yet we are likely to be in an early adopter phase for that much larger market for the next several years.
The market where sequencing is currently in use is mostly academic research, and the time lag of turning academic innovations into high-demand clinical solutions is still several years out. Despite some amazing clinical examples of rare diseases being unraveled with sequencing, the bulk of the value being produced in the supply chain is knowledge and publications. As long as academia is the dominant end consumer in the supply chain, there is a built in cap on the growth of our industry. The individual silos in the genetics supply chain are seeking to push their products downstream to meet shareholder expectations of continued revenue growth, and despite incredible innovations in sequencing throughput, we still have not seen enough translational success to the end patients to fund the growth of the industry.
When I first met Goldratt in 2005 and he gave me advice on my company, one of the first things he asked was whether our bioinformatics software could really bring bottom line results — for instance he asked whether you could really save a drug that failed for safety or efficacy reasons using a pharmacogenetics study. Promise and optimism was high for GWAS at that time, and I responded in the affirmative that there was a good enough probability of success. Time and experience has tempered that optimism. Interestingly, though, some of the largest successes in GWAS have been in pharmacogenetics, perhaps because the drugs have not been with us long enough to see selective pressures — there are some extremely high odds associations in the GWAS of drug safety and efficacy. However, the ongoing concerns about our failure to explain more than a few percent of disease variability genetically is a major roadblock to clinical translation.
High throughput sequencing holds promise for changing that — particularly through looking at the contributions of rare variants and private mutations in pedigrees. This has large implications for clinical practice — to best inform you about your health, your doctor should strive to collect phenotypic and genetic information from your extended family. However, my prediction is that after harvesting some low-hanging fruit, we will hit a wall in a few years of how much we can tell about health from the genetics you are born with and begin to adopt a longitudinal prevention model of continuous collection of a broad spectrum of time-varying biomarkers to provide early warning of disease, as described in my earlier APEC post.
Important vs. urgent
Just a couple years ago Goldratt was telling us at a TOCICO conference how he planned to author a series of a dozen books on various aspects of the science of management, projecting that it would take until his 90’s to complete. Eli died at age 64 after a brief battle with late stage lung cancer, leaving that important work undone. One of the last things he said was “never let the important become the urgent”.
Death comes to us all, and all too soon. Each of us has lost loved family members, friends, or colleagues to diseases that could have been prevented if we only knew earlier and had a means to intervene. Did you know that still about 40% of lung cancer is discovered at stage IV, where the median expected lifespan is 8 months?
If the end consumer is buying health and long life — have we as an industry delivered? In some cases, yes, such as in the field of cytogenetics where clinical yields have increased from 3% to 20% as described in an earlier post of mine. However, taking a macro view of the health output of the system as a whole by plotting below the life expectancy of Americans over the past 50 years, the lifespan has only slowly and steadily gone up by 10 years, with no apparent increase over a linear trend. Similar curves exist for other industrialized nations.
Given the disconnect in the USA between expenditures on healthcare and longevity that I outlined in an APEC talk last year, there seems to be a major gap between monetary input and health and lifespan output. It has become close to unsustainable with about 16% of US GDP spent on healthcare.
To me the big takeaway from 2011 is that the important is our health and life, and it almost always becomes urgent — cancer, heart attacks, stroke, etc. Do we have 10 years left, 20, perhaps much less? Now is the time to make systemic changes to how we do research and development and translation into clinical application. The year 2011 saw the 40th anniversary of Nixon’s declaration of war on cancer. Yes, there has been progress, but not enough for Eli Goldratt. Will there be enough progress for you, me, and our loved ones when our time comes? Better yet, will we have advanced the field of prevention to the stage where continuous longitudinal monitoring of biomarkers allows us to detect the onset and progression of diseases and no longer catch them for us and our loved ones when it is too late?
There is an industry that continues to make significant money in healthcare by selling hope and rarely delivering. It is the snake-oil salesmen, often with powerful multi-level marketing systems to sell that hope. Science can and should deliver more than hope and promise for the future. If we are not satisfied with the progress of the last 40 years, I believe we will need to apply the scientific method not just to the research, but also on improving the organizational systems that are the vehicle for our research. I hope to write more on this topic in the coming year, with an editorial in Biostatistics coming out soon: “Learning From Our GWAS Mistakes: From Experimental Design To Scientific Method”. Let’s take this seriously in 2012 and beyond, making the systemic and structural reforms that will significantly increase the slope of the longevity curve and translate what we have learned into real value for our ultimate customers — for they are us.
…And that’s my 2 SNPs.