Biomarkers travel a long distance between their conception in a basic research lab and their eventual use as a clinical diagnostic. As chair of the clinical labs department at Memorial Sloan Kettering Cancer Center, Martin Fleisher spends a good amount of time thinking about biomarkers and the route they take to get to the clinic. GT’s Jeanene Swanson caught up with Fleisher to ask about how to spot a good biomarker, the rigors of validation, and more.
Genome Technology: What types of biomarkers are currently being used in clinical diagnostic labs?
Martin Fleisher: There’s basically three types of biomarkers. There’s serum-based biomarkers, there’s urine-based biomarkers, and there’s whole blood-based biomarkers.
In urine-based biomarkers, you’re looking for a piece of DNA or some genomic material that is unique to a cancer cell, that has been broken down in the serum and is now being excreted in the urine. The presence of that biomarker in the urine is an indication that there may be a cancer present.
A serum biomarker is designed to be a little bit more specific. You can make a serum biomarker specific based on the type of assay that you use for various types of cancer. [What] we’re working on the hardest here are biomarkers for ovarian cancer and prostate cancer. The biomarker is being developed not entirely to pick up on early cancer, but it’s being developed to help the clinician decide if the therapy that the patient is on is working properly. So, the biomarker could be used for two purposes: one, to pick up a cancer and two, to tell a doctor if the therapy the patient is on is working.
A third type of biomarker is the whole blood-type biomarker. Whole blood contains all kinds of cells — red cells, white cells, and all kinds of varieties of white cells. But in patients with metastatic cancer, or cancer that is to become metastatic, there are cells present in whole blood that should not be present, and those are cancer cells. Those cells are referred to as epithelial cells; they come from a carcinoma. And if there’s an epithelial cell in the blood of a person that’s being tested, there’s no question that there’s a cancer somewhere.
GT: What are the technical challenges to validating a biomarker for clinical diagnostic use?
Fleisher: Let me walk you through what we would go through to validate a biomarker. The first thing that we have to do to validate any biomarker is to know that it’s a biomarker. You’ve got to prove that it’s a biomarker, and the way to prove it — and let me use the circulating tumor cell as an example — is to say, OK, you’ve isolated this tumor cell, this epithelial cell from peripheral blood. How do you know it’s cancer? How do you know it’s not just a malformed white cell? So, you’ve got to do some tests on that cell to prove that it is a cancer cell, and there are different types of tests that you can do. There are chemical tests and there are tests that you can do pathologically. So that has to be validated — that, in fact, what you are measuring is a cancer cell.
Once you know it’s a cancer cell, you have to know whether or not you can measure this cancer cell in patients who have a similar disease. If you’re looking at prostate cancer, you have to take a look at a hundred patients [and] see whether or not a hundred patients have the same level of cancer. You have to see whether or not you can pick up that cell in those patients — or is it very select for a certain number of those patients? And if it is select, why is it select? So you have to do a big study to find out whether or not the cells are found in all prostate cancer patients who have a particular type of cancer.
And then you have to be able to reproducibly get a result on that particular patient day after day after day. It’s not good enough to get it on Monday but miss it on Thursday or Friday. You have to set up the assay so that you can show that the assay is reproducible within the assay itself. Then you have to show that if the patient is treated, does it affect the level of that cell? Because maybe if the patient gets a certain type of therapy, you’re making a tumor shed, but it’s not a metastatic cell, it’s a cell that broke off the tumor because the therapy is working. You have to know that.
Then you really have to be able to transfer your technology from one laboratory to another. You have to make sure that this assay can work in different laboratories, not just your own. And that process is called validation. Everything that has to do with making sure that it’s a cancer cell, reproducibility of the assay, day-to-day performance of measuring that particular endpoint in the patient on different days, transferring the technology to a different laboratory — that’s all part of the validation process. Every time we develop a biomarker, that’s what we go through. It’s a very elaborate, time-consuming, and very difficult process. But when you’re finished, you can trust the results.
GT: What are the regulatory hurdles involved in getting a biomarker used in a clinic?
Fleisher: The FDA must approve all testing that goes on for diagnostic purposes. And if the test does not have FDA approval, then you have to take it through a different process. You have to get an IRB approval within a medical center — IRB is an investigational review board — to let you do the testing on patients, to get permission from the patients to do that testing. The IRB simply reviews the project to see whether or not it is scientifically sound, if there’s a possibility that it will be efficacious, and that it is ethical to do the work.
GT: You’ve spoken recently about judging biomarkers based on levels of evidence. What are these, and how can they affect whether clinical labs will use a particular biomarker?
Fleisher: There are levels of evidence that make biomarkers much more reliable in terms of whether or not you should even consider doing this test on a patient. Levels of evidence go from one to five. Those levels of evidence go from, [one], a study that was done by a group that had a large patient population and had two arms within the study, patients who had cancer or patients who didn’t have cancer. And they looked at the difference of this biomarker, and the biomarker was found to be useful because the cancer patients had higher or lower levels compared to the non-cancer patients. The study was repeated by other laboratories, other investigators. That’s the level 1 form of evidence. As you go down to level 2, 3, 4, and 5, it becomes looser and looser. For example, a level 4 is a small study that was done in a laboratory that was never reproduced by anybody else. You can’t really trust that data. So, levels of evidence in using a tumor marker are quite important. You don’t want to use a biomarker that has only been used in one laboratory and nobody else can do it.
GT: Considering the increasing large-scale genomic and proteomic analyses being done today, how do you see the field of biomarker research progressing in the next five, or even 10, years?
Fleisher: It’s going to expand at an incredible rate and will go into all kinds of areas. Biomarkers will become genomic assays, they’ll become panels of gene tests. Proteomics will become important; the technology of proteomics will be worked out and made useful. It’s going to expand enormously.