Healthcare Costs and the Human Genome

Nicholas Wade had a very interesting article Tuesday morning in the New York Times (here). He reports that Dr. Stephen R. Quake, a Stanford University engineer, has invented, designed and built a machine, called the Heliscope Single Molecule Sequencer. The machine is about the size of a refrigerator and costs $1,000,000 to buy from Helicos Biosciences, a company founded by Dr. Quake.

The machine works by splitting the double helix of DNA into single strands and breaking the strands into small fragments that, on average, are 32 DNA units in length. Light emitted when a new helix is formed on these fragments characterizes the sequence in the original DNA. The light is computer analyzed and compared to human genome sequences already on file. This is done for billions of these small fragments and the differences in the test sequences from those most common in the general population identify what makes that individual unique.

Only Seven Individual Genomes Recorded to Date

Dr. Quake has decoded his own genome. He says the cost was $50,000. It is not clear if this was the operational cost only, or includes some depreciation allowance for the machine. I infer that some machine capital cost was included based on the time of analysis (four weeks) and the human involvement (three people). I was amazed to learn that Dr. Quake is only the seventh individual to be decoded in history. The first individual human genome decoding was in 2003. Only two of these individuals are named: J. Craig Venter, a pioneer of DNA decoding; and James D. Watson, the co-discoverer of the DNA double helix. Dr. Quake has compared his genome with those two and found the DNA overlaps shown in the following graphic from the NYT.

In addition to the seven individuals, the Human Genome Project has mapped the DNA sequences of a large group collectively to represent a human population “mosaic”. The project website is here. In spite of the Human Genome Project being declared complete in 2003, approximately 8% of human sequences remain to be characterized according to Wikipedia.

Technology Roadmap

Wade gives a brief history and projection of the future:

For many years DNA was sequenced by a method that was developed by Frederick Sanger in 1975 and used to sequence the first human genome in 2003, at a probable cost of at least $500 million. A handful of next-generation sequencing technologies are now being developed and constantly improved each year. Dr. Quake’s technology is a new entry in that horse race.

Dr. Quake calculates that the most recently sequenced human genome cost $250,000 to decode, and that his machine brings the cost to less than a fifth of that.

“There are four commercial technologies, nothing is static and all the platforms are improving by a factor of two each year,” he said. “We are about to see the floodgates opened and many human genomes sequenced.”
He said the much-discussed goal of the $1,000 genome could be attained in two or three years. That is the cost, experts have long predicted, at which genome sequencing could start to become a routine part of medical practice.

(My underlining added for emphasis.)

The dramatic progression of cost per individual for genome characterization is shown in the following graph:

Dr. Quake’s process is accurate to within 5 errors per 100,000 sequences. George Church, a leading biotechnologist at the Harvard Medical School, has said that the next real breakthrough in this technology should see a cost of $5,000 per individual with only one error per 100,000 sequences. I have put my SWAG (Stupid Wild Ass Guess) regarding a possible timing for achieving Church’s breakthrough requirement on the graph in red. The reader should recognize the great uncertainty in putting a breakthrough on a future timeline.

What is the Potential Market for Decoding Machines?

That is a difficult question to answer because the competing technologies are still rapidly evolving. With Dr. Quake’s machine, the time for analysis was four weeks with three persons involved. To do market estimates, one has to make assumptions about how analysis time per person will decline and how many individuals will be decoded per unit of time.

One thing is evident. At whatever level of usage, the cost of the machines is a small part of the total cost. Look at 100,000 people per year with the Quake machine (red region). Assume a 10 year machine life. Let’s use $20,000 per test (40% of Dr. Quake’s processing cost of $50,000, assuming economies of scale with repetition of procedures). With these assumptions, the machine cost (at $1 million per machine) of $7.69 billion is about 1/3 of the the cost of 10 years of processing ($20 billion). If the machine life is 20 years the machine cost is 1/6 of processing cost.

If we assume that the future high throughput technology (blue region) can be accomplished withmachines costing $1 million, the capital costs per 100,000 tests per year is lowered to $962 million. Processing costs (at $5,000 per test) for 10 years is $5 billion. Capital cost is now less than 20% of the processing cost. Only at $1,000 per individual do the two costs become comparable over 10 years.

The potential market for these machines appears to be of the order of 1,000 to a few thousand machines per year, or an annual cap ex of $1 to $5 billion. This will depend on how much benefit can be obtained from application of individual genomes to curing and preventing disease. That question will be discussed next.

What Benefits Might Result?

Two outcome benefits are expected. One of these is related to the identification of genetic predisposition to specific diseases, enabling early intervention to prevent disease development or slowing progression from early stages. The second outcome expected is the more accurate diagnosis of presented symptoms, which will enable focused treatments and reduced use of other diagnostic testing.

Are the benefits worth the cost? That is difficult to answer because the application of the technology is in its infancy. It has been found that many diseases turn out to be caused by a complex combination of a number of rare variants. Identifying the specific individual DNA sequences responsible for many diseases has not been possible. It is possible that statistical comparison of each individual to a pool of genomes for individuals with specific diseases will be the procedure first used to narrow the scope of further diagnostic tests. Eventually, of course, specific complex genetic patterns may be identified. In early applications probabilistic applications are likely.

Is There an Economic Case?

An assessment can be made with a macro analysis. The total health care bill for the U.S. annually is of the order of $2 trillion. If we take 1,000,000 DNA decoding tests annually as a benchmark, the cost would be $11 billion per year, with a 10-year cap ex amortization. The cost of this program at this level is less than 1% (0.55%) of total health care costs. Thus, the cost barrier to implementation is low. If savings of only a few percent are realized by eliminating unnecessary tests, errant therapies resulting from misdiagnosis and the early treatment and prevention of many diseases, this could have a huge cost reduction impact and an improved health result for the populace.

We should look at the cost reduction and outcome improvements not only for the year of expense for sequencing the genome of an individual. The genome is time invariant (barring a spontaneous mutation) and all future diagnosis and treatments will be improved from the one-time sequencing.

Be Careful What You Wish For

In the NYT article, Dr. Quake explains an alarming discovery in his genome: He found that he carried a marker for heart disease. From the NYT:

Dr. Quake said that analysts were annotating his genome and had found a variant associated with heart disease. Fortunately, Dr. Quake inherited the variant from only one parent; his other copy of the gene is good.
“You have to have a strong stomach when you look at your own genome,” he said.
Dr. Quake said he was making his genome sequence public, as Dr. Venter and Dr. Watson have done, to speed the advance of knowledge.
Some people may decline the opportunity to have their genome analyzed. They may prefer to follow the maxim: Ignorance is bliss. Others may worry that “defects” in their genome could be used to discriminate against them, in employment or with respect to life insurance, for example.

If it can be determined that medical outcomes are much improved for many people at a lower cost, an incentive to participate in genome screening could be lowered medical insurance premiums. If someone preferred to pay for the higher cost of ignorance and face the risk of being surprised at some time in the future by an advanced stage of disease that might have been preventable, of course they should have that choice.


We have a medical and technological revolution unfolding. The first process for decoding the human genome was developed by Frederick Sanger in 1975. It took many years for the early work to evolve to the Human Genome Project and, from there, to the new decoding procedures of today. The progress in the near future is expected to be very rapid. In the NYT article, Nicholas Wade presented the following quote from Dr. Quake:

“There are four commercial technologies, nothing is static and all the platforms are improving by a factor of two each year,” he said. “We are about to see the floodgates opened and many human genomes sequenced.”

With America looking for ways to bring a new level of cost control to spiraling health care costs, what could be better that new technology that not only reduces costs, but also produces better outcomes?

By John Lounsbury,
John Lounsbury — Seeking Alpha