Proteomics to Diagnose Human Tumors and Provide Prognostic Information

For the clinician who is faced with treating individual patients, the article by Ornstein and Petricoin might raise the famous question from the Wendy's commercial: Where's the beef? When we hear of these Star Wars technologies and complex explanations, we are often frustrated. On the one hand, we have nothing to offer our patients right now, and on the other, our patients read about these technologies and expect them to be applied right now. This editorial will focus on the right now. I am convinced that the technologies these two world-class scientists have described will make a difference to our patients within a very short period of time. The Ideal Biomarker
Let's focus on what may be the most promising immediate benefit of this technology-the development of cancer biomarkers. This area is clearly problematic in the current practice of medicine. First, for most cancers, there are no serum or urine tests that can be used for screening or early diagnosis. We just don't have markers for pancreatic cancer, lung cancer, or colorectal cancer that we can use for screening. For the cancer with the best biomarker- prostate cancer-we have a major problem. Among men with a prostate-specific antigen (PSA) level of 4.0 ng/mL or higher (about 8% of the US population at any point in time), only about 25% actually have cancer. Additionally, as the recent results of the Prostate Cancer Prevention Trial show, about 15% of men with a normal PSA (< 4.0 ng/mL) have prostate cancer.[1] Of these, 15% have high-grade cancer-a tumor suggested to have a very high risk of progression if left untreated. Certainly, we could present even worse performance characteristics for CA-125 and carcinoembryonic antigen. Let's take the problem one step further. For many of the most common cancers (certainly prostate and kidney, probably breast, possibly others), I am not certain that we want to diagnose all of the tumors. We know that many prostate cancers are relatively indolent; this could probably be said about some fraction of tumors in other organ sites as well. Thus, what is the ideal cancer biomarker? At a minimum, (1) it is present (or elevated) in patients with the disease in whom the disease poses a risk but is curable; (2) its value is related to the volume of disease; and (3) it is readily measured, usually in a body fluid (serum, urine). Technologic Promises
So, how will the field of proteomics that Ornstein and Petricoin describe change the practice of oncology (and medicine)? We have already seen the promise of these methodologies. In a group of women with ovarian cancer and 50 unaffected women, Petricoin et al demonstrated that using surface-enhanced laser-desorption/ ionization (SELDI) technology, and examining the full range of proteins in the serum (there are perhaps 100,000 proteins in serum) rather than relying on one protein, the test achieved 100% sensitivity, 95% specificity, and a 94% positive predictive value.[2] In prostate cancer, several groups have produced results substantially superior to those achieved with PSA. In 386 cases and controls, Qu et al found that sensitivity and specificity ranged from 97% to 100%.[3] Given the heterogeneity of, for example, prostate cancer, is it surprising that a panel of proteins are more predictive of the presence or absence of cancer than a single protein? Of course not. Given that polymorphisms of the PSA promotor gene can affect PSA levels or that preexisting hypogonadism can affect PSA production (due to the androgen response elements in the PSA gene), relying on this single, highly variable protein to detect a disease that affects 16% of men in their lifetime is not reasonable.[4-6] Early Detection Research Network Trial
Finally, with our ability to analyze such enormous volumes of data (hundreds of patients, thousands of proteins), we now have the opportunity to seek the Holy Grail of cancer detection tests. It is not unreasonable to hope for a biomarker that specifically detects a tumor that is both curable and requires treatment. In fact, this is the goal of a multicenter effort conducted by the Early Detection Research Network (EDRN) of the National Cancer Institute in prostate cancer.[7] In this three-phase study, the techniques of protein profiling with SELDI will first be validated at multiple institutions. The algorithm for assignment of patterns to cases and controls will be revalidated, and the test will then be validated in a completely separate study population. The intent of this effort is to ultimately validate the test using the resources of the Prostate Cancer Prevention Trial, in which about 60% of all participants had either a diagnosis of prostate cancer or an end-of-study prostate biopsy, regardless of PSA or digital rectal examination findings. Conclusions
How soon will all of this happen? When will the practicing clinician see the beef? The answer is: very soon. It is anticipated, for example, that the biomarker validation trial of the EDRN should be completed within 2 years. Almost certainly, because of the growing number of biorepositories in which samples are available in cancer cases and controls, researchers will rapidly seek to validate these methods in other organ sites. Again, this is only one application of this technology. We are limited only by our imaginations. Both the oncology professional and the oncology patient will benefit immensely and very shortly from this explosion in knowledge and technology.