The article by Drs. de la Taille, Olson, and Katz is an accurate and concise review of clinical studies for the detection of circulating prostate cancer cells using reverse transcriptasepolymerase chain reaction (RT-PCR) technology. These investigators from the Department of Urology at Columbia-Presbyterian Medical Center have as much experience as any group in the use of RT-PCR for this purpose. Initially very strong proponents of the efficacy of RT-PCR as a staging tool, they have become slightly more reserved in the current article. In this well-written review, the authors allude to a number of issues affecting RT-PCR results from prostate cancer patients that deserve further comment.
Why Do RT-PCR Results Differ?
The issue of whether RT-PCR can be used as a staging tool in prostate cancer prior to treatment has been the focus of over two dozen publications. Results of studies were initially encouraging but subsequently disappointing. At present, the overwhelming consensus is that this molecular test should not be used currently for clinical management. Yet, in the hands of the Columbia group, the results continue to look promising. This begs the question of why; what is different between this group and nearly every other site engaged in similar studies? The answer, quite simply, is unknown.
An exhaustive review by two members of our group examined the num-
erous technical differences that exist among investigators who have reported RT-PCR results. To my knowledge, there is not a pair of investigators among the major groups who are using identical techniques encompassing all aspects of the method from specimen acquisition to verification of product. Because of this tremendous variability, elucidation of the reasons why results differ among groups is most difficult.
Results of a Multicenter Consortium
To address this variability in methods, a multicenter consortium was established. For a period of over 3 years, specimens from approximately 300 patients were collected prior to radical prostatectomy, sent to a single processing center, blinded, and aliquots then redistributed to four test sites for assay. This was the first attempt to standardize specimen collection and processing; five clinical sites contributed specimens.
We were astonished to find that, despite our exhaustive efforts to select and test the best methods of acquisition, handling, shipping, processing, and so forth (using LNCaP-spiked blood specimens), there was significant loss of RNA integrity from the time of specimen acquisition to the time of testing. The RT-PCR results on identical aliquots from the four test sites were highly variable. Interestingly, quality controls, including sensitivity controls, were consistently acceptable.
Many conclusions can be drawn from this experience, but the take-home message is that cancer cells in blood may be more fragile than LNCaP cells spiked into blood, and one needs to be very cautious about the molecular integrity of specimens when there are multiple steps prior to assay. Shipping of the specimen may well be the weak link in the process.
Thus, although standardization is an important aspect of assay design and user acceptance, as recommended by de la Taille et al in the accompanying article, I would argue that there are more critical issues related to specimen handling that must be resolved first.
A Difference in Assay Design
One additional note of interest deserves comment. As stated above, there have been multiple discussions among investigators speculating as to why the Columbia assay/results were better than those of others. One afternoon, after going over minute aspects of assay design with Ralph Buttyan, PhD, of the Columbia group, we stumbled upon the fact that he initially tested an assortment of primers for PSA on clinical specimens, picked the best primer set, and then did sensitivity testing with LNCaP cells spiked into blood. To my knowledge, our group and nearly all others first screened prospective primers against LNCaP cells in blood to obtain the best sensitivity and then tested clinical specimens.
Whether or not this is an important difference remains to be evaluated. Yet, this may again highlight, as in our multisite clinical trials, that in the design of molecular assays for prostate cancer, LNCaP cells spiked into blood may not be equivalent to the clinical situation.
Detecting Prostate Cancer Cells in Bone Marrow
Three investigative groups have focused on detecting prostate cancer cells in bone marrow instead of, or in addition to, peripheral blood.[3-6] Data from these RT-PCR studies have forced us to alter our concepts of the timing and frequency of prostate cancer systemic dissemination. Although perhaps not a surprise to some, these two groups plus our own found that at least 50% of patients had RT-PCRdetectable cells in their bone marrow prior to radical prostatectomy. In matched-pair comparison studies, the frequency in bone marrow represented anywhere from twice to three times that in peripheral blood.
When we compared two bone marrow sampling sites, we found strong evidence of sampling error, leading us to speculate that probably 80% or more of patients are actually RT-PCRpositive in their bone marrow. If the frequencies found in these three studies are even remotely accurate, do further attempts to utilize peripheral blood as a staging indicator make sense? Perhaps what these extensive efforts in RT-PCR have actually accomplished, in a circuitous fashion, is to prompt us to refocus on the event that results in patient deathie, bone metastasis.
One may argue that we have yet to prove that the cells detected by RT-PCR in either the peripheral blood or the bone marrow are always cancer cells. Yet, the preponderance of evidence using normal controls or patients with benign prostatic hypertrophy is that the frequency of positivity in these series is negligible (ie, a few percent) overall. Recent studies by Polascik et al have demonstrated that prostate cancer cells can be identified and isolated from the peripheral blood of many patients prior to radical prostatectomy.
Thus, the evidence is substantial that the results obtained by RT-PCR of the bone marrow are not artifactual. This leads one to conclude that most patients do have prostate cancer cells in the bone marrow upon presentation.
Many Questions Raised
Yet, the frequency of positivity in these bone marrow specimens raises many questions, because we know historically that the clinical recurrence rate is only one-half to one-third of the RT-PCR frequency rate. Going back to a previous question, does the peripheral blood frequency, although perhaps not a good staging parameter, more accurately reflect clinical outcome? Does the high rate of positivity in the bone marrow indicate a filtering and concentration effect? If prostate cancer cells are in the bone marrow, whether by passive concentration or active adherence, why is the actual recurrence rate considerably lower?
What are the phenomenology and mechanistic attributes that determine why some patients develop a recurrence while others do not, if most already have disseminated prostate cancer cells early in the disease process? Is it simply a matter of Gleason grade? We think not. Could adaptation and growth in this new environment be a multistep process, and, if so, what are the possibilities for earlier intervention to slow or stop the process? These are just a few of the many questions that have emerged from RT-PCR studies.
To answer these questions, we will again rely heavily on RT-PCR technology. We will search for expression of known genes, and more strongly emphasize the need to develop molecular profiles from cells isolated very early in the disease process throughout the time to overt metastasis and subsequent androgen independence. We will need to enhance our efforts to isolate cells from peripheral blood and bone marrow that yield suitable cells for molecular profiling, and vigorously verify single cell amplification techniques so that profiles accurately reflect low to high gene abundance. We also envision that RT-PCR results will soon be confirmed and expanded by in situ and immunohistochemical analysis of enriched/isolated cells.
Numerous Avenues for Future Exploration
Thus, in closing, I do not share some of the negative comments that have recently surrounded RT-PCR staging efforts. The overly optimistic goals set forth only a short 6 years ago by the pioneers of these endeavors may not be fully realized, but, as a result of the recruitment of many investigators into this quest, other avenues of exploration and study are opening.
The investigators in the Columbia group are among those pioneers, and their perseverance is to be commended. While some of us will continue to explore technical considerations in an attempt to perfect the RT-PCR methodology, others will begin correlating the expression of genes associated with aggressive behavior to clinical recurrence. As more is learned about the expression of genes that favor adaptation and growth of cells within the bone marrow, many of us will be dipping into our banks of stored RNA/cDNA to test for the expression of these genes in patients for whom there is now considerable clinical follow-up. I truly see great opportunities ahead, and am not regretful in the least that we embarked on this venture to detect cancer dissemination by RT-PCR.