Assay Development For The Clinic
Blog 3 of my intro blogs for those interested in learning more about Translational Medicine.
An important part of translating preclinical work is assay development. The point at which you start the assay development may differ based on what tests are already out there and what the preclinical team has already done.
Let’s consider a situation where there is no clinical test available and there are already preclinical studies on the biomarker you are planning to test.
Often preclinical studies involve mining for a predictive biomarker using large omics studies such as RNA sequencing or exome sequencing. Maybe it was found that expression of an RNA transcript is associated with drug response, or the transcript is useful as a pharmacodynamic biomarker.
Here are the steps you should take to create a test suitable for the clinic.
Let’s take a step back here. If not already done, evaluate whether the biomarker is suitable for a clinical test. Preclinical experiments are performed in a controlled environment, which is often impossible in the clinic. Different centers will collect the samples for this test, and while you will have a protocol to provide to each local laboratory, there will be differences. For example, one lab may be much further away than the others, so samples sit in transit longer. Patients may have samples collected late on a Friday and the sample may sit in the refrigerator for a night or two. Some labs are in warmer environments than others, and so on. So assess whether the biomarker is stable enough to be accurately measured after all this.
Another aspect is whether the biomarker has a good dynamic range. If there is a significant difference in preclinical studies but the effect size is small, the differences may be lost when the above sample differences are considered.
Consider the best test for the assay. Confused about the difference between a test and an assay? See this article.
The large omics platform used to mine for the biomarker is usually not appropriate for the clinic as they are expensive and require more material (although they can be useful in other instances such as studying the mechanism of action or emerging resistance). If other tests were performed preclinically that seem more appropriate you could start there eg a PCR test or IHC test. Or there may already be a test on the market you could validate for this purpose.
You are looking for something robust, within budget, that can ideally be used on samples in the form that they are currently collected in the clinic (eg don’t go for a test that requires fresh frozen samples when they are usually fixed in formalin). An important point here is turnaround time. Define when you will need the test results (eg will it be needed for internal decisions, or be published at a specific conference) and how the samples will be batched for testing. It is useless to create a test that requires all samples to be batched and tested at the end of the study only to find that investors expect you to present an interim update.
Another frequent issue is taking a test that has been used preclinically and assuming it has been designed correctly. Preclinical assays do not have to be as rigorously developed, and it may be that a redesign is better, or a study of various options could be explored. For example, if preclinical studies were by PCR, redesign some primers and try several options during your assay development. Your future self will thank you when you have cleaner results.
Define how samples should be collected. This is important. Once you start collecting those samples, you cannot easily change it. Any changes you make may take a while to be implemented and will compromise the results. Thoroughly research collection tubes, transit, and storage conditions for each sample. Once you know how they are collected, ensure all assay development from that point on starts with samples collected similarly.
Ensure you see the same results in a human assay. If preclinical hasn’t already done this, procure some human samples and test in vitro whether you see the same result. This is usually required for pharmacodynamic biomarkers, and in the same assay you can find out what timepoints are the most appropriate for collecting the samples, and what effect sizes you expect to see. You can also assess for dose-response at the same time.
Establish limit of blank, limit of detection, and limit of quantitation and use these limits to define the protocol. See this study performed by Labcorp to do just that. Study the boundaries of the test, eg what is the smallest amount of sample that can be accepted, what maximum transit time can be accepted?
Ensure you have appropriate controls in every run. Note that these may not be the same controls as those used in other studies with different sample types.
Determine the amount of sample you need for the test. This should not be too much that it is detrimental to the patient eg don’t collect too much at once or over a short period. It should also allow other collections required for the trial such as PK.
Define when you will collect the samples. Use the information from 4, and align it with other clinical trial factors such as when other samples are being collected (and therefore the patient can have the samples all collected in one visit) and when endpoints measurements are being performed.
Determine what you would expect to see for a decision/result eg X% better response in the biomarker positive group, or 2x decrease by day Y on treatment. In some cases, once you have your result, you could stop collection eg if a pharmacodynamic marker clearly shows it has the desired effect, you can stop further collections and testing.
In many cases, the service provider performing the assay development is also the provider that will test in the clinic. Ensure they have all the credentials to do this (eg CAP/CLIA) and store this information (along with the current Lab Director's CV). Be clear in the contract about what oversight you expect (it is important to have visibility in case you need to troubleshoot), when you expect results, and what raw and processed data you expect them to send you and when.
#clinicaltrials #biotech #drugdevelopment
