Selecting a collaborating laboratory

The following are general issues that a population researcher should consider when determining which laboratory should handle the biomarker assessments and what to expect when dealing with a collaborating laboratory.

Who can provide guidance about specimen protocols and run the assays?

The CPC Biomedical Services core was established to provide consultation to CPC projects that are considering inclusion of biomarkers in their research.  Email John Thorp, the core director, to obtain guidance in selection of a lab.

What are the pros and cons of using a commercial or research lab?

There are two main choices when selecting a laboratory collaborator: a commercial lab or a research lab. In some developing country settings, the choices may be few. For research projects requiring specialized assays, it may be necessary to build local laboratory capacity to meet research project needs if those services do not already exist. In the US there are often many alternatives to consider.

Advantages of working with commercial labs are that they have generally routinized the lab assay and the associated quality control procedures, they can process specimens quickly and in large volumes, and they generally can provide automated results. A drawback may be that they are more accustomed to doing clinical diagnostic assessments, and are less familiar with non-clinical techniques that are needed for collection of specimens in non-standard settings (i.e., not in a clinic). They may not be familiar with issues related to specimens that have been stored for extended periods or that have been through thaw-refreeze cycles, and they may not be accustomed to working with researchers who want remaining specimens returned for banking after initial assay.

Research labs are more likely to be working on new assays that may make use of techniques not commercially available. This can mean the reliability of the assay may not be well established, and the assay may change during the course of the research project, making it difficult to compare results that span several months or years. Because research labs generally are not in the business of high-volume processing, there can be delays in getting specimens assayed and results reported. A research lab's protocols may not be as well documented as those of a commercial lab, making it more difficult to get the description needed for incorporating assay results in data analysis and publication.

The above issues should be discussed with potential lab collaborators during the selection process.

What about sensitivity and specificity when choosing an assay or reporting the results?

Sensitivity and specificity of a laboratory assay indicate how accurate the assay is for determining whether or not the person has the disease (or condition).

Sensitivity is the proportion of the people who truly have a disease who are identified by the assay. This is the probability of correctly identifying a diseased individual.

Specificity is the proportion of the people who truly do not have the disease who are identified by the assay as not having the disease. This is the probability of correctly identifying a non-diseased individual.

Related to sensitivity and specificity are predictive values for positive or negative tests. The predictive value for a positive test assesses how many of the people who have positive test results truly have the disease and the predictive value for a negative test assesses how many of the people with negative test results truly are not diseased.

These concepts are critical to interpreting a laboratory test because high levels of false positive or false negative results have repercussions for reporting to the study participant and interpreting the factors associated with the assay result. The expected prevalence of a disease plays an important role in determining the optimal test, i.e., the optimal combination of sensitivity and specificity. Sensitivity and specificity should be available on any assay performed, and the probabilities should be considered before committing to use an assay.

How should quality control issues be resolved?

There are several approaches to laboratory quality control (QC) issues, and the lab collaborator should provide clear, written plans about how QC will be assessed. Some typical techniques include:

• running the assay in duplicate on all or some percentage of randomly selected specimens to see if the results are consistent.
• placing laboratory-created specimens with known existence or quantity of the marker of interest in a batch of specimens for assay so that laboratory staff do not know which specimens are known positive or negative samples.
• collecting enough of a particular specimen from non-study participants to allow assay of multiple aliquots; then, including one of these aliquots with each batch of research subject specimens being assayed to determine the reliability of the assay across batches. This approach assumes little delay in assaying of batches; otherwise, it is necessary to consider degradation of specimen from storage (see next item).
• re-assaying some specimens with each batch over time, e.g., at 6-month intervals, to determine whether there is a degradation of the assay results due to prolonged storage.

What if repeated assays of the same specimen lead to different results?

When duplicates are run or when the researcher decides to have another lab re-run assays and the results are significantly different, it is important to talk with the lab director to get a recommendation on how the results should be interpreted. There may be a clear explanation, or there may be problems with the lab procedures.

What results will the lab report, especially if they are doing duplicate runs (all results, the average, the outliers)?

It is common for a lab to run duplicates of assays on all or a subset of specimens, and there are often cutpoints above which a lab automatically re-runs the assay. Each of these tests yields a new result for the specimen, and it is important to discuss with the lab ahead of time which of these values you will receive. Often labs will report only what they consider to be the final determination after any re-assessments. The project researchers may want to see all values and run their own checks on how discrepant these values are.

How quickly must assays be run to avoid degradation?

Degradation of specimens is related to the conditions under which they were collected and stored and the assay to be done. Some specimens have been successfully assayed 30 years after collection, while others must be assayed immediately after collection because refrigerating for a significant period or even freezing for a short period will make the specimen unusable. If a commercial kit is being used for the assay, the kit instructions or the manufacturer's technical support should provide guidance on storage and stability. If the assay is under development by a research lab, the lab director will have to advise about these issues.

Keep in mind that if a specimen is collected for one purpose, the collection and storage procedures may not be appropriate for other assays. Anticipating possible future uses of a specimen is important so that informed consent to collect the specimen accurately reflects ultimate assays, and so that processing and storing methods provide the most flexibility for future assays.

One technique to assess degradation is to bank a large number of aliquots from a single specimen collection point (e.g., 10 aliquots of blood from one blood draw for several individuals), and assay single aliquots over time so that results can be compared across different storage periods.