To Screen or Not to Screen

As Chief of the NCI’s Diagnostic Biomarker and Technology Branch, James W. Jacobson focuses on moving discoveries in cancer biology toward clinical application, and this includes biomarker discovery. As he explains, “A biomarker is anything that you can measure that in some way correlates with a disease state of interest, whether it’s the presence of disease or something that gives you information about the future course of disease. It can be a molecular entity that you can measure at the level of the genome, at the level of RNA expression or protein expression. It can be a lipid or a glycoprotein, and the field of metabolonics is also becoming important.”

The branch’s focus is to move the discoveries in cancer biology toward clinical application, covering everything from biomarker discovery activities to the confirmation, validation, and ultimately the development of clinical tests. Another portion of its program deals with making appropriate tissue resources available to the community.

Not all biomarkers are created equal, however. What, for example, is the difference between genetic biomarkers and serum biomarkers? “That’s a distinction we don’t normally make,” says Jacobson. “A genetic biomarker is something that is an alteration in the normal DNA pattern of the individual, where there is a mutation or rearrangement of the genetic information. You can measure some of those things in the serum if the cells are lysing and DNA is being released.

“A serum biomarker refers to the presence of a protein entity within the serum that you can measure without having to take a biopsy of the tumor – so a serum biomarker would be a subset of other biomarkers that you might consider. There could also be other types of molecules in the serum – anything that exists in the serum that’s correlated with the presence of disease.”

Jacobson points out that if you’re screening for the presence of disease in asymptomatic individuals, you’ll be looking for serum biomarkers, because this involves taking blood rather than doing a biopsy. Even in high-risk individuals, a biopsy won’t be done unless there is some indication that disease is present. Serum biomarkers are also useful for other clinical purposes, such as monitoring response to therapy or monitoring for recurrence in a patient who has had cancer and has been effectively treated.

“Screening is a very complicated and difficult issue because, if you are screening an asymptomatic population, the performance of that biomarker has to be extremely robust. It has to have very high sensitivity and specificity so that you’re only detecting patients who potentially have cancer and not patients who may have other diseases that would lead to the presence of that biomarker. If it’s not highly specific, then you run the risk of triggering interventions that are unnecessary, which is not to the benefit of the patient. It’s a very difficult issue.

“Our particular program does not deal with early detection and screening – there is another group within NCI that does that. We’re more involved with patients who are being seen for some symptomatology where biomarkers are used for making a differential diagnosis, for prognosis, for determining therapy or for monitoring response to therapy.”

Screening with PSA

One serum biomarker used to detect the potential presence of cancer is prostate specific antigen (PSA), though as Jacobson specifies, it’s not actually approved for that purpose, because it is prostate specific, not cancer specific.

There are a number of other factors that can lead to an increased level of PSA, such as infection, prostatitis and other benign diseases of the prostate. Conversely, not all prostate cancers increase PSA levels. PSA was initially approved by the FDA for monitoring prostate cancer patients for recurrence, but it has become a generalized screening assay, and has led to a significant increase in the detection of prostate cancers.

The big question, says Jacobson, is whether the detection of those cancers has benefited the patients involved. “The mortality from prostate cancer has been fairly constant despite this increased detection rate. It’s still an open question as to whether or not the intervention in those patients has had a major impact on prostate cancer mortality.

“There is a decline in mortality from prostate cancer, but at what cost is really the open question. The decline has been modest, but the detection of prostate cancer by PSA has led to intervention for many patients who may not have needed it. There has been a huge increase in detection but there hasn’t been a huge decline in mortality yet.”

Jacobson explains that the use of PSA screening may have led to the detection of a lot of prostate cancers that would never have become clinically important. Prostate cancer is a very slow-developing disease, and many men who develop it, particularly the more elderly among them, would have died from other causes prior to ever having a clinical problem with prostate cancer. In the case of patients with more aggressive cancers, they may not have benefited by having detected those any earlier than the development of clinical symptoms.

“It’s a very complex issue. We definitely have seen more cancers. It’s not clear exactly what the benefit of that has been compared to the morbidity that has been associated with intervention. Many men have had radical prostatectomies – removing the prostate with the associated morbidities – and they may or may not have needed that intervention.”

CA-125 and CEA

Other antigens, such as CA-125 and CEA, are not specific enough to be used as screening for early detection. There are indications in which they can be used as diagnostic tools; CA-125 in particular is being used to monitor ovarian cancer patients for response to therapy and for recurrence.

Jacobson points out that the American Association of Clinical Oncology guidelines recommend that CEA be measured at the time of diagnosis in colorectal patients, and that this be used for monitoring response to therapy and for detection of recurrence. “The understanding is that CEA, in a colorectal cancer patient, is only elevated if there’s metastatic disease present, either spread to the liver or to the lungs. Knowing that in the beginning may change the way in which you manage those patients.

“One of the challenges and one of the things that’s being looked at very carefully now is that using any of these biomarkers as an individual entity is probably not adequate. We need to use them in combination with other information and particularly other biomarkers. I know there’s a significant amount of ongoing effort to look at CA-125 in ovarian cancer as part of a profile of biomarkers that may give you more specific information about the course of disease.

“CA-125 is being considered as a potential prognostic marker in ovarian cancer, but it’s probably more informative when it’s used in combination with other biological markers as opposed to being used alone. The biology of cancer is so complex that the expectation that a single marker will give you robust information about all patients is unrealistic, and certainly the major efforts now are to go to panels of markers that give you significantly more information.”

Facing challenges

The field of biomarker research is currently facing a number of challenges. Jacobson says that there hasn’t been the organized development of clinical tests for biomarkers in the same way that there has been for drugs. Because of this, the algorithms for the testing and validation of biomarkers are not at the same level of maturity as those for drug testing. Jacobson’s team is actively addressing the need to educate the community about what needs to be done to validate biomarkers.

“There has been a significant lack of support for assay optimization and validation of biomarkers. This is a very difficult and potentially less interesting area of research than doing biology studies, so the community is more focused on the biology and less interested in doing the hard work that’s required for marker validation.

“There’s probably less academic recognition for investigators who are doing those kinds of studies. There needs to be a significant increase in the effort to make sure they get done. One thing that is critical is to have the specimens available that are appropriate for the clinical questions being asked. When you talk about biomarkers and biomarker confirmation and validation, you need to talk about a biomarker in the context of the clinical question you’re expecting that biomarker to address. This requires that you have the appropriate samples that are representative of the clinical situation. To have those samples with long-term outcome and all of the appropriate annotation is a challenge.”

Other challenges include the issue of study design and appropriate statistical support for biomarker studies. These studies often don’t have an appropriate design up front, as Jacobson explains, “If someone carries out a series of measurements in a set of clinical specimens and then goes to their statistician to say, ‘What can you make of this?’, that’s not an appropriate design. Statisticians need to be involved from the beginning; however, there are a limited number of highly qualified statisticians who understand the nuances of biomarker validation, so there needs to be more education in that area.

“There are a whole variety of issues. For example, when you’re trying to measure a biomarker and move it into clinical application, you need to spend a lot of time on the development of the assay so that you know what the performance of that assay is before you take it into a clinical validation kind of study. You also need to know about reproducibility and robustness. That’s another issue that takes a lot of time and effort to do correctly, but is not high-reward research.”

Personalization

One of the aims of biomarker research is to develop a form of personalized medicine: trying to understand what the biological characteristics of a tumor are within an individual patient, and developing the therapeutic armament to effectively address that tumor with a minimum of side effects. This requires having information about the measurement of a whole variety of biological entities or biomarkers within that tumor and then having the therapies available to address that particular set of alterations within that tumor in an effective way.

According to Jacobson, the obvious example is Gleevec. “Gleevec was developed to address alterations in the BCR-ABL gene, a fusion of the BCR and ABL genes, in chronic myelogenous leukemia, where it is a very effective therapy. It has now been found that it can also be used for alterations in other specific diseases; for example, it is effective against alterations in the KIT gene in GI stromal tumors. Another really obvious example is Herceptin therapy for breast cancer.

“These are the kinds of therapeutic interventions that we want to move to in the future, where we’re not just giving therapies that are very poisonous and kill indiscriminately, but instead we target the therapy to kill only the cells in the tumor that we’re interested in killing.”

What are the implications of this for that for the pharmaceutical industry? There has been a lot of concern expressed that personalization will reduce the size of the market for any particular therapy that a drug company develops, because you’ll only be treating a subset of the patients who will benefit from that drug. However, says Jacobson, we may be able to identify these subsets in a large number of diseases and this should keep the market effective in terms of its size. “Instead of a therapy being used to treat a single cancer, it will be used for a whole variety of cancers where there are particular alterations that that drug will target. Whether it will be enough to offset the concerns raised remains to be seen. This is still very much a developing area of therapeutic intervention, and not all the answers are in yet.”

Jacobson does not believe that the use of biomarkers will ever replace traditional biopsies in cancer diagnosis. “Many of the biomarkers require the biopsy to get you the materials that you need to have, because many of them are tissue-based evaluations and you need to have a piece of the tumor to do the analysis that’s required. Biomarkers will always be adjunct to the actual biopsy.

“If you’re talking about biomarkers that are used for screening, they may detect the potential for the presence of cancer, but it will then take other interventions to determine whether the cancer is there – for example, an imaging modality which leads to seeing a mass and doing a biopsy, which you can then evaluate. Biomarkers are never going to be standalone diagnostics; they will be used in combination with all of the clinical information you have available to you and a whole variety of other information about the tumor, including the pathologist looking at the biopsy and understanding what those cells look like. Biomarkers don’t replace information, they add information.”

About James W. Jacobson

James is Chief, Diagnostic Biomarker and Technologies Branch, at the National Cancer Institute, NIH, HHS whose focus is to move the discoveries in cancer biology toward clinical application. Prior to joining the NCI in 1991, Jacobson was an academic researcher, and spent 11 years in the biotech industry. He is a molecular biologist by training.