Monday, November 08, 2010

Is the KIF6 statin test debunked?

KIF6 codes for a kinesin, which transports cargo along intracellular microtubules.

A recent paper and accompanying editorial in JACC are sure to spark controversy. The paper was an analysis of 19 case-control studies, looking at the association of the KIF6 polymorphism with CAD risk. The findings:

Results: A total of 17,000 cases and 39,369 controls of European descent as well as a modest number of South Asians, African Americans, Hispanics, East Asians, and admixed cases and controls were successfully genotyped. None of the 19 studies demonstrated an increased risk of CAD in carriers of the 719Arg allele compared with noncarriers. 

The somewhat overhyped editorial, titled The KIF6 Collapse, declared not only that KIF6 is not a marker of increased coronary risk, but also that it cannot be relied upon to predict statin responsiveness.

There are several flaws in this analysis, not the least of which is the fact that a significant number of patients in these case-control studies were undoubtedly taking statins. Given the evidence that the KIF6 polymorphism predicts both risk and the ability of statins to reduce that same risk, a valid study would have to be confined to patients not taking statins. The editorial writers anticipated and made an unconvincing attempt to refute that objection (unconvincing to me, anyway).

A little background. KIF6 is a common genetic variation (polymorphism). There is a commercially available test. The claim that it is a marker of risk is based on analysis of sera from multiple randomized controlled trials. In these studies it proved to be a modest predictor of risk, with hazard ratios on the order of 1.2, similar to traditional risk factors. More important is the ability of the test to predict statin responsiveness. Given the current debates over the cost effectiveness of statins in primary prevention, particularly in low risk populations, the utility of such a test would be enormous.

Here are some estimates of statin effectiveness in KIF6 non-carriers versus carriers. (These findings were presented by Dr. H. Robert Superko at the 36th Annual Tutorials in the Tetons Update in Cardiovascular Disease):

In terms of numbers needed to treat (NNT) to prevent one event, in KIF6 non-carriers versus carriers, respectively, in various statin trials:

CARE 72 vs 20
WOSCOPS greater than 100 vs 18
PROSPER 83 vs 16
PROVE IT 125 vs 10

Note that WOSCOPS was a primary prevention study, but with high lipid entry criteria. Specimens from the JUPITER trial are currently being analyzed for KIF6. Those results should shed light on the utility of the KIF6 test in low risk populations.

These results are impressive. Pending the results of the JUPITER analysis, I would consider taking the KIF6 test if faced with a less than clear choice as to whether to take a statin.

1 comment:

Tom said...

I'm curious to know your thoughts now that Jupiter has been released showing no statistical significance to those with KIF-6 genotyping.