How to Assess Simple Screening Strategies
- PDF / 354,239 Bytes
- 6 Pages / 504 x 719.759 pts Page_size
- 57 Downloads / 230 Views
Drug Injormarron Journal, Vol. 35, pp. 413-418, 2001 Printed in the USA. All rights reserved.
HOW TO ASSESS SIMPLE SCREENING STRATEGIES DHAMMIKA AMARATUNGA, PHD Research Fellow, Preclinical Biostatistics, The RW Johnson Pharmaceutical Research Institute, Raritan, New Jersey
In early stage drug discovey screens, the efficacy of a potential drug candidate in an in vivo assay is often judged, rather informally, on the proportion of animals that respond to the drug. Only if at least a prespecified proportion of animals respond is the drug considered active enough for further testing. This is a “simple screening strategy.” Researchers are often unaware of the properties of such strategies and what effects changing the parameters of the strategy, such as the sample size, could have. It is suggeAted and demonstrated here that the properties of a simple screening strategy can be easily understood by considering its true and frlse positive rates. Key Words: False positive rate; True positive rate; Receiver operating characteristic curves
INTRODUCTION THE HUNDREDS OF compounds that emerge from initial screening efforts (such as high throughput screening) as potential drug candidates are subjected to a battery of in vivo screening assays. These in vivo screens are an important first test of the compounds in whole animal models and only a few compounds that show activity (and low toxicity) in them are admitted for further testing. Yet, the decision as to whether a compound is “active” or not in a particular assay is often based on a fairly simple decision rule. For instance, a researcher might treat 10 mice with a test compound at a certain dose level and deem the compound sufficiently active and suitable for further study should 7 or more of the treated mice respond. The properties of such decision rules,
Reprint address: Dhammika Amaratunga, The RW Johnson Pharmaceutical Research Institute. 1000 Route 202, Raritan, NJ 08869-0602. E-mail: dmaratu@prius. jnj.com.
which we call simple screening strategies, are often not fully appreciated by their users, who will, therefore, be unaware of the consequences of selecting one particular strategy over another. For instance, in the above example, is it better to use a cutoff of eight instead of seven? Or, if the amount of compound is limited, what are the consequences of reducing the number of animals from 10 to 5? Or. should a different cutoff be used if the assay is one in which a small percentage of animals respond spontaneously? Or, is a two-stage strategy preferable to a singlestage strategy? In order to get a handle on issues such as these and alleviate some of the arbitrariness in defining a strategy, it is useful to have some idea of the statistical properties of a proposed strategy. It is, in fact, quite easy to assess the properties of a simple screening strategy via its false positive and false negative rates. We have found that such assessments help researchers refine their overall screening strategy. Also, such considerations are helpful to researchers having to just
Data Loading...
