Biomarkers and heart risk

I watched a news report last week that claimed that a relationship had been identified between heart risk and a set of three biomarkers. An abstract of the paper, titled, Aggregate Risk Score Based on Markers of Inflammation, Cell Stress, and Coagulation is an Independent Predictor of Adverse Cardiovascular Outcomes, is available. A description of the results is also available at the Emory University website as several of the coauthors are affiliated with the university.

Unfortunately the paper itself is hidden behind a pay wall at a cost of $31.50. I sent an e-mail to the contact for the paper last week requesting a copy of the paper and have not received a reply. I must say again my position is that while I understand the business model for published papers when the authors issue a press release and grant media interview to advertise their findings they then, in my opinion, assume the responsibility to provide their research to the public. Releasing their interpretation of the results while making it difficult or expensive for the public to review their work is unacceptable. Fortunately most authors who’s papers are hidden behind a pay wall will send a copy of their paper if requested to do so.

Even without the paper itself the results cited in the abstract and in the media raise several questions and concerns. Since I don’t have the paper many of these questions do not have readily available answers. They need to be addressed.

The researchers looked at 3,415 patients at who had been referred for cardiac catheterization based on suspected coronary artery disease. They then looked at three biomakers and determined a risk level based on the results of these makers.

Question 1: How were the biomakers identified? If they were selected based on a much larger set of biomarkers then was that taking into consideration in the analysis?

Question 2: How were the thresholds for each biomaarker determined. For example the used FDP ≥1.0 μg/ml. The actual threshold could have been any of several values. If the threshold used is a standard within the medial community that is one thing. But if the threshold is based on the data then that needed to be taken into consideration in the analysis.

Both of these first two questions are the equivalent of the well known multiple comparison issue in classical statistical analysis. Without consideration of the effect of making many possible comparisons then the consequence is over confidence in the results.

Question 3: Several additional biomarkers were available for each patient based on the reasons they were referred to the hospital for cardiac catheterization. Was this information used, useful or considered in the analysis. Was the results of the catherterization used in the analysis? What was the marginal additional value of the new test based on the three biomarkers after taking into consideration the reasons for the referral and the results obtained from the catheratization?

Question 4: What actions were taken based on the results of the catheratization? This test frequently leads to immediate heart bypass surgery. This alone certainly is an indicator of the risk of additional heart attacks. So this again raised the question of what is the marginal value of the new test using the three biomarkers.

These two questions raise the question of medial costs and the value of the test. Does the new test duplicate information already available? What is missing from the analysis is the more traditional treatment A vs treatment B evaluation. When proposing an medical treatment or the use of a new test this kind of analysis is critical. In the case of a new test the appropriate analysis may well be treatment A vs the combined treatment A and treatment B.

The abstract for the paper and the Emory website summary do not provide any of the customary information on possible conflict of interest on the part of the researchers. This information likely is in the paper itself. It is worth mentioning that two of the authors are affiliated with FirstMark. That information is readily avalibale when one looks at the author affiliations listed in the abstract. When I visited their website I found that FirstMark just now introducing at test that they refer to as “the first multiple biomarker test that accurately predicts near term (within 2-3 years) risk of myocardial infarction for suspected or confirmed coronary artery disease (CAD) patients.” I am not sure if this is the test referred to in the paper.