06 April 2015

The Economics Of DNA Informed Medicine

In the past, I've generally assumed that customizing medical treatment with DNA testing was a prohibitively expensive exercise for only the most affluent.

Technological advances that have brought down the cost of these kind of testing, and the perspective that comes from having it done yourself, has changed my attitude.

As I've noted at the sister blog to this one, I recently had a total genome profile of myself (and the rest of my family) commercially prepared by 23 and me.  The basic price is $99 per test, with a discount for bulk orders.  It is arguably not medical grade in accuracy and completeness, but it is still quite complete and accurate.  And, the technology will only improve over time.

Some key points about DNA testing are worth noting:

1.  Absent retroviral infections, your DNA will generally remain the same from prior to your birth until long after your death.  So, you only need one medical grade DNA test per lifetime.

2.  The data set produced by a DNA test, while voluminous, is amenable to easy and compact electronic storage.  A DNA chip measures something on the order of 600,000 to 1,000,000.  Each SNP location can be coded with redundancy to allow for error checking with less than one byte of binary information.  You can code the entire data set on an uncompressed basis with about 1 MB of data.  But, since a large share of SNPs mostly have the same value for everyone, it is also possible to use a compression algorithm that would report only atypical values at those SNPs where the values are atypical.

Bottom line: a complete genome can be easily transmitted as a modest sized e-mail attachment, and is quite readable in ordinary English so long as you know the correct designation for the SNP you are looking up.

3. Your DNA contains information relevant to many ailments you may suffer during your life and to how you will react to many medicines that you may be prescribed during your life.

4. For any given ailment or medicine, it takes a computer program no more complex than a simple cell phone app to review your genome and produce all information relative to that ailment or medicine in readable terms based on published scientific studies.  SNPedia is an amateur, open access effort that comes very close to containing all of the information needed to locate the relevant articles and prepare outputs that spell out the personal DNA considerations involved.

5. In mass production, apps like this could be bundled as a no cost add on to a prescription, much as existing prescription information and warnings is today, for medicines, and a full catalog of these apps for all ailments a particular type of doctor is likely to encounter in his or her practice, ought to be possible to produce for less than the cost of a common, standard medical treatise.  The cost to buy these apps and annual updates would be tiny relative, for example, to a doctor's medical malpractice insurance bill each month.  An estimate of a cost of $1 per patient visit would probably be high.

In short, it might cost $200-$300 per patient, per lifetime, to make a review of a patient's genome, a standard part of a typical doctor's decision to prescribe a drug or devise a treatment plan.  Somebody has to fund to basic research that goes into published journal articles on genetic associations with diseases and medicines, just as they do today, but the value added secured by that research, if routine DNA checks were part of normal medical practice, would be much greater than we have today where the dissemination of that research is far less systemic.

Today, the gathering of the research that informs how we read a patient's genome and apply it to medical practice is in its infancy.  But, we have a solid database with respect to some ailments, like breast cancer and Alzheimer's disease, and some drugs.  Over time, as the value of the research in daily practice becomes clear, the list of ailments and drugs about which we have good DNA specific information will expand dramatically.

The process of diagnosing an ailment from a set of symptoms, for example, when leveraged with the power of DNA, prior medical record, and population-wide data informed Baysean priors, is likely to become much more accurate.  There may still be cases where trial and error, or additional diagnostic tests are required.  But, the odds of landing on the right diagnosis in the smallest average number of trials, will be greatly improved.  The reduced spending on diagnostic tests avoided because of the order in which diagnostic and confirming tests are performed alone would more than pay for the implementation of DNA informed medicine.

For example, suppose someone walks into a psychiatrists office at the college health clinic at age 20 presenting with schizophrenia of an undetermined subtype, and the family history is patchy, for example, because the patient was adopted.  A quick review with the schizophrenia app of that patient's genome could quickly determine which of eight subtypes of schizophrenia that patient was most likely to have, which drugs are usually most effective to treat symptoms with that subtype, and whether anything else in that patient's DNA contraindicated for that patient for this purpose.  Many hours of sessions with the patient and observation, and many months of tinkering with treatment regimes on a trial and error basis, could be avoided.

The best doctors today intuitively incorporate prior medical records and population-wide data in their estimate of the best diagnosis, but only consider the individual patient's propensities written in their DNA in exceptional case.  But, in the near future, including this powerful dataset could become routine.

It also isn't particularly important that the DNA information piece be tightly integrated with the rest of the data.  A medical practitioner could simply include a one page report with DNA information pertinent to the task at hand with the other materials in the patient's medical record at diagnostic junctures as a matter of course, and could consider it holistically using his or her "wet ware" together with the other information in the file.  Often it would just be a check box task - a medical paraprofessional would ask the treating physician what to pull DNA information on, would run those reports, and would present them to the physician a few minutes before a visit with a patient together with the rest of the patient's file.

Thus, treating a patient with DNA informed medicine would be less like ordering imaging, like an MRI, CAT Scan, or X-ray on a case by case basis, and more like reviewing the vaccination page or allergy listing or birth certificate already in the patient's file.  Once the medical quality genome test was done once in a patient's life, nothing new would have to be ordered and no additional patient time would be required to use this diagnostic tool.  It would not lead to any meaningful delay in the diagnostic process; results of a DNA check would be available instantly once the patient's genome was in put in the system when the child was an infant.

Another virtue of this advance is that it doesn't have to be introduced in one fell sweep.  For example, Kaiser Permanente could introduce use of the system and take genomes of its new patients as a matter of course, without cooperation from any other medical providers, and could introduce it for existing patients only on request or as needed.

No comments: