Necessity is the Mother of Stratification: Personalized Medicine is Getting Real


Joe Stanta of eyeforpharma organized a Personalized and Translational Medicine conference last week for people who are directly involved with making personalized medicine a reality. What was especially refreshing about the conference was a healthy balance of prognosticators and practitioners—with the majority of speakers representing the latter. I came away from this conference with the feeling that the hype of personalized medicine is falling away and we are about to see a number of significant products from a variety of credible vendors.

Personalizing Therapeutics

Matthew Rosamond of PricewaterhouseCoopers opened the conference repeating the assertion that drug companies can effectively market “nichebuster” drugs (that are best in class for a subpopulation) that are safe, effective, and profitable. Even he admitted that this strategy is a tough sale to traditional pharmaceutical companies, but I did see abundant evidence for a strategy that I will call “necessity is the mother of stratification.” Companies are launching into the discovery and development process using biomarkers that help to explain the mechanism of action for the drug. If they see significant differential in patient response they can use these biomarkers to identify sub-populations for which a drug is more effective and develop a commercial diagnostic test based upon the most useful biomarkers. The therapy and diagnostic (theranostic) are then marketed in tandem.

This strategy implies that in some cases the drug company will not recover the sunk costs of development but they can use patient stratification to rescue a drug that pays the marginal cost of producing it. These are not as profitable as traditional blockbusters but are preferable to rejecting a useful compound altogether. There was an assertion that having both a rational basis for the drug’s benefit and empirical evidence of efficacy is more compelling evidence to justify reimbursement and may even raise patient compliance. It was difficult for me to determine whether this is actual experience or wishful thinking at this point.


Paired development of diagnostic and therapeutic is still an evolving business model for three reasons:

1. The time that is required to develop drugs and diagnostics are very different. Diagnostic developers want to be involved early in the process but the attrition rate of drugs is unacceptably high for them and it is difficult to know early on what biomarkers are going to be appropriate to commercialize.

2. The cost basis of the industry is quite different. Many diagnostics—particularly complex multivariate tests—will necessitate clinical trials, and the traditional cost model for tests and relatively low volume for companion diagnostics make it difficult for diagnostic companies to make a profit at traditional price points for diagnostics.

3. Reimbursement hurdles for companion diagnostics are higher than traditional diagnostics.

I had heard before that theranostics would shift the center of financial gravity from drug makers toward diagnostics companies, but I did not hear this stated last week in Boston (perhaps because of a selection bias toward pharmaceutical companies). In general, people spoke about an acceptable price of $100 per test and different models for pharmaceutical subsidizing diagnostics. In general the best targets for new diagnostic tests are expensive therapies with low response rates. In this new world, pharmaceutical companies feel that they have a lot to benefit diagnostic companies in that they have established programs for dealing with both regulators and payers.

Much of the movement is coming from small biotechs, not traditional in vitro diagnostics companies. With the cost of full sequencing approaching that of a complex test (e.g. Genomic Health’s Oncotype DX) many reason that the full sequence will become more common and reusable for clinical purposes. Healthcare providers are beginning to look at sequencing moving from research to clinical. Planning for clinical rollout is real and happening now in some institutions.

There is no consensus on what the level of integration should be between the pharmaceutical and diagnostic company. Models range from Roche and Novartis that market diagnostics to Merck which will influence the development of the diagnostic but do not see themselves in the diagnostic business. AstraZeneca seems to have deals that run the gamut with several partnership models. Even those with the most indirect ties to in vitro diagnostic companies want to leverage their existing programs to help bring companion diagnostics to market. They will often subsidize the cost of diagnostics immediately after approval to ensure that there is a whole product available to the market.


There is a clear consensus that regulation of tests will increase, particularly for complex multivariate testing (e.g. multiple genes used to determine a calculated score for likely efficacy of a drug). This raises the costs of test development because the test must go through clinical trials. The process of defining the requirements is in progress but regulation of molecular diagnostics will evolve over several years. The U.S. FDA is generally considered to have the toughest standards and there were varying strategies for addressing the biggest market (U.S.) or easiest approval (Europe) first.


The dearth of suitable biological samples is considered a major impediment to progress in research and those countries and companies that have been intentional about developing biobanks have a huge strategic asset. Unfortunately, it is very difficult to repurpose samples because of informed consent policies or technical reasons. It is often inappropriate to compare samples from different locations because proteomic or RNA samples are strongly affected by the conditions that they are collected under. A simple difference in the distance between the operating room to the repository may make samples from different locations incompatible. Access to biobanks is much more restrictive in U.S. and this gives European researchers a market advantage. Several European companies have made forward-looking decisions to encourage both biobanks and clinical record systems specifically to accelerate science.


Interestingly, there were no payers in attendance even though the meeting was only a few miles from Hartford, CT. One speaker mentioned that payers feel that the evidence for personalized medicine is not yet compelling, but pharmacy benefit managers such as Medco, Generation Health, and CVS Caremark have been very active. In general, European health ministries were considered less likely to reimburse, or more likely to force prices below worthwhile levels. It was agreed that general practitioners are not prepared to advocate for, or make use of personalized medicine.


I came away from this conference with the sense that accumulated innovations of the last half century are beginning to enter clinical practice. And the most compelling reason is that drug companies are faced with the fact that some complex diseases are highly heterogeneous and people do not all respond to medications in the same way. Personalized medicine will not disrupt the pharmaceutical industry because it is possible, but it will because it is necessary.

Don Rule is the founder of Translational Software, a company that aims to accelerate the process of bringing molecular diagnostics from the bench to the bedside. He previously worked at Microsoft for 14 years. Follow @

Trending on Xconomy

By posting a comment, you agree to our terms and conditions.

3 responses to “Necessity is the Mother of Stratification: Personalized Medicine is Getting Real”

  1. Ed Berger says:

    The reimbursement hurdle for a true companion diagnostic should be lower, not higher, than that for a novel stand-alone diagnostic. Why? Because if the diagnostic were used to stratify the treatment arm of a therapy clinical trial (attractive because it the trial can be smaller with a higher probability of success) it would be recognized in the FDA-approved indications for use. Insurers won’t have any trouble covering the test under that circumstance.

  2. Don RuleDon Rule says:

    Absolutely true that a companion diagnostic would be reimbursed more readily than a novel standalone diagnostic. But in comparison to most IVD companies bread-and-butter products it is more difficult. In fact the companion diagnostic isn’t useful at all if you are not prescribing the drug and it isn’t unusual that reimbursment for the drugs themselves lags FDA approval by a year or more. So from an IVD company’s current practices it is a long time. Which explains why many of the companion diagnostics and novel stand-alone tests are being developed by small biotechs and not the IVD companies.