Detecting Prostate Cancer with Molecular Markers: uPM3

Prostate cancer is a major public health burden. In 2005, more than 232,000 new cases and 30,000 deaths from the disease are expected.¹ One of the main ways in which we screen for prostate cancer is through the use of serum prostate-specific antigen (PSA). Recently, however, the value of PSA as a screening tool has been questioned because it has become increasingly clear that many men with a “normal” PSA level can nevertheless have prostate cancer.² New diagnostic measures are clearly needed. One approach, which is now commercially available, is the uPM3™ urine test (DiagnoCure; Quebec City, Quebec, Canada). This test is the first molecular test for prostate cancer screening. The uPM3 relies on the fact that prostate cancers have increased expression of a noncoding ribonucleic acid (RNA), differential display code 3 (DD3).³ The function of DD3 remains elusive; however, the fact that it is expressed at high levels by prostate cancers and at only very low levels by benign prostate tissue has been exploited for its possible role in prostate cancer detection. Here we review 3 recently reported studies that examined the role of uPM3 as a novel diagnostic tool for early-stage prostate cancer. For clarity, “uPM3” is the name of the diagnostic test, whereas “DD3,” now known as “PCA3,” is the name of the gene for which uPM3 tests.

For brevity and because the methodology of the 3 studies was very similar, we will describe the methods here. In all 3 studies, investigators examined the first voided urine after an “aggressive” digital rectal examination. Urine samples were then centrifuged at 4°C and urinary sediments collected. The sediment was then washed with phosphate-buffered saline and snap-frozen, or cells were lysed first and then snap-frozen. Total RNA was extracted from the sediment and analyzed with quantitative realtime polymerase chain reaction (PCR) for the expression of DD3. This allowed the accurate measurement of the number of DD3 RNA transcripts that were present in the sample. To correct for the number of prostate cells in the sample, the number of DD3 transcripts was standardized to the number of PSA RNA transcripts present within the sediment, as determined with a similar PCR-based approach. The first study used the DD3/PSA ratio, whereas the latter 2 used a mathematical calculation that was more complicated but was also largely based on the number of DD3 RNA transcripts. This value was then used to determine its ability to predict the presence of prostate cancer among men undergoing prostate needle biopsy.

DD3(PCA3)-Based Molecular Urine Analysis for the Diagnosis of Prostate Cancer

Hessels D, Klein Gunnewiek JM, van Oort I, et al.

Eur Urol. 2003;44:8-15

This was the first of the 3 studies to examine uPM3 as a diagnostic marker for prostate cancer. The group was based in the Netherlands and included some of the original investigators who first discovered DD3. First, the study confirmed that in prostate cancer tissue DD3 was up-regulated, confirming their prior work.³ However, the investigators nicely extended their findings to suggest that even when the amount of cancer in the tissue sample was small (< 10%), elevated DD3 can be detected. This suggests that even a little bit of cancer could possibly be detected within a background of largely normal tissue. This provides the theoretical support to begin looking for DD3 expression in urine, where prostate cells can be shed after an aggressive digital rectal examination.

Next, the investigators performed the uPM3 assay on 108 consecutive men undergoing prostate needle biopsy for a PSA level greater than 3 ng/mL. Of these 108 men, 24 had cancer on needle biopsy. There is no mention of the number of biopsy cores that were done, though in the Discussion section it is hinted that sextant biopsies were performed. The area under the receiver operator characteristics (ROC) curve for predicting the presence of prostate cancer for the uPM3 test was 0.72 (1.0 is perfect prediction, 0.5 is equal to the flip of a coin). Based on the graphic shape of the ROC curve, Hessels and colleagues concluded that the optimal cutoff point for predicting the presence of prostate cancer was 200 × 10⁻³ DD3 transcripts per PSA transcript. Using this cutoff point, the sensitivity of the assay to predict the presence of prostate cancer was 67% and the specificity was 83%. Unfortunately, though the values of serum PSA for each of the 108 men were given, no statistical information was presented to compare the performance of the uPM3 to that of PSA for detecting cancer. In addition, it is clear that some men with a “negative” sextant biopsy will still have prostate cancer. Therefore, whether the men with a “positive” uPM3 and a “negative” biopsy had a false-negative biopsy or a false-positive uPM3 is unclear.

DD3PCA3 RNA Analysis in Urine—A New Perspective for Detecting Prostate Cancer

Tinzl M, Marberger M, Horvath S, Chypre C.

Eur Urol. 2004;46:182-186

This is the first of 2 studies from essentially the same group. In this study, the results of uPM3 assay from 201 men undergoing prostate biopsy at the investigators’ institution were reported. Interestingly, though methodology for sample collection was essentially the same as described above, only 158 of the 201 men (79%) had enough prostate cells in the spun urine for analysis. Therefore, no assay information was available on 21% of men. Of the 158 men with usable data, 62 (39%) had prostate cancer on biopsy. Like the prior study, no mention was made of the biopsy technique or number of cores obtained. In addition, 42 of the 158 men (26%) had a prior negative biopsy.

The overall area under the ROC curve for predicting the presence of prostate cancer for the uPM3 test was 0.87-better than that seen in the prior study. Use of a uPM3 cutoff point of 0.5 value of probability resulted in a sensitivity of 82% and specificity of 76% for predicting prostate cancer on biopsy. This compared favorably to PSA assay, which at cutoff points of 2.5 and 4.0 ng/mL had a sensitivity of 98% and 87% and a specificity of 5% and 16%, respectively. The sensitivity and specificity of the uPM3 test seemed to be best among men in the diagnostic “gray” zone of PSA between 4 and 10 ng/mL. Although the results of this assay seem promising, the fact that over 20% of patients had a noninformative assay is concerning. If these patients are included in the analysis, the “effective” sensitivity is reduced from 82% to 69%-a much less impressive result.

uPM3, A New Molecular Urine Test for the Detection of Prostate Cancer

Fradet Y, Saad F, Aprikian A, et al.

Urology. 2004;64:311-315

This is the third study, and the only multicenter study, to examine the value of uPM3. Fradet and colleagues studied the uPM3 assay from 517 men undergoing prostate needle biopsy at 5 medical centers. Similar to the above study, a sizable percentage of men (14%) had noninformative uPM3 assay results. Therefore, data are presented only for the 443 men with usable data. The positive biopsy rate was 34%. Like the prior studies, no mention was made of the biopsy technique.

The overall area under the ROC curve for predicting the presence of prostate cancer for the uPM3 test was 0.86-similar to the value seen in the prior study by this same group (see above). Use of a uPM3 cutoff point of 0.5 value of probability resulted in a sensitivity of 66% and specificity of 89% for predicting prostate cancer on biopsy. The uPM3 test functioned best among men with a PSA level less than 4 ng/mL, with a sensitivity of 74% and a specificity of 91%.

In summary, these 3 studies suggest that the uPM3 test might have a reasonable sensitivity and specificity for detecting prostate cancer, which in some instances might be better than the sensitivity and specificity of PSA assay, though larger studies are clearly needed. In addition, no data on free PSA, which might help improve the prognostic value of PSA, was presented in any of the studies. However, despite the fact that uPM3 is now commercially available, several limitations need to be addressed before uPM3 becomes the standard for prostate cancer screening. The first issue is the loss of nearly 20% of patients who had noninformative data. The importance of this cannot be overstated. Our collective anecdotal experience in real-world practice using the commercially available uPM3 test suggests that the “noninformative” rate might actually be higher than 20%. The second major limitation is that unlike PSA, uPM3 seems to offer no prognostic value for the stage or grade of the cancer, though this has not been directly studied. Given that the overall prevalence of prostate cancer is extremely high, Carter’s recent statement in the New England Journal of Medicine should be recalled: “It is apparent that any approach that finds more cancers without quantifying the clinical significance of the detected disease will only increase overdiagnosis and overtreatment.”⁴ Future studies will be needed to assess whether a quantitative uPM3 test (rather than positive vs negative) will help preferentially detect higher-grade, high-stage cancers. In addition, it would be useful to know whether, among men with prostate cancer, uPM3 provides any prognostic value for predicting outcome, as PSA does. Ultimately, if these issues can be addressed, uPM3 seems to be a promising new prostate cancer biomarker.