Johns Hopkins throws (some) cold water on possible use of PCA3 test

It has been hoped (at least by some) that a possible use of the prostate cancer gene 3 or PCA3 test — approved in parts of Europe but still in clinical trials in the US — would be as a better tool than PSA tests and annual biopsies to monitor patients on expectant management (active surveillance).

Tosoian et al. set out to examine the relationship between PCA3 data and prostate biopsy results in men on the Johns Hopkins active surveillance program, and we should be clear up front that these patients all have to have very low-risk disease as specified by the so-called Epstein criteria:

  • Their PSA density must be < 0.1.
  • They must have a Gleason score of < 7.
  • They can have only one or two biopsy cores that are positive for prostate cancer.
  • No biopsy core can have > 50 percent of the core showing prostate cancer.

The researchers obtained urine specimens from 294 men with prostate cancer who had enrolled in the Hopkins expectant management initiative (and therefore met the above criteria). These patients were all followed with semiannual free and total PSA measurements, DREs, and an annual prostate biopsy. Disease progression was defined as the presence of any Gleason pattern 4 or 5, more than 2 positive biopsy cores, or more than 50 percent involvement of any core with cancer on a follow-up annual biopsy.

The results of the study showed the following:

  • 38/294 patients (12.9 percent) met the criteria for disease progression based on their follow-on biopsies
  • These 38 patients had a mean PCA3 score ( of 60.0) similar to the 256 patients without progression (50.8).
  • Detailed statistical analysis suggested that PCA3 alone could not be used to identify men with progression on biopsy.
  • After adjustment for age and date of diagnosis, PCA3 was not significantly associated with progression on biopsy.

It is apparent that — in men in this series — a single PCA3 score at the time of diagnosis could not be used to predict the patients most likely to be at risk for prostate cancer progression. However, what we still do not know is whether:

  • Sequential PCA3 tests (every 6 months or every year) might be able to predict risk for progression in men on active surveillance.
  • A single PCA3 might be able to predict risk for progression in men with an initial Gleason score of 3 + 4 = 7 who enroll in an active surveillance program and are subsequently upgraded to Gleaosn 4 + 3 = 7 or higher based on an annual biopsy.

In practical terms, the Johns Hopkins active surveillance program probably has the most restrictive criteria of any of the current active surveillance programs at a major medical center. While it would clearly have been helpful if a single PCA3 test at the time of diagnosis could have been used to predict which of these patients were most likely to progress over time, this was probably not going to be likely. We will still need more data to be able to assess the prognostic value (if any) of the PCA3 test in low-risk patients.

11 Responses

  1. In previous studies that demonstrated progression on biopsies, Hopkins reported that the increase in GS was related to missing the higher GS on the original biopsy. If that is the case, the PCA3 results reported now would not be discriminatory. What am I missing?

  2. Ralph:

    I am not sure that you are necessarily “missing” anything. The two different types of study data are not mutually exclusive.

    In some cases progression on biopsy is certainly a consequence of failure to “hit” the higher grade cancer on earlier biopsies, but that doesn’t negate the result of the current study (which is only applicable to the patients who met the Epstein criteria).

  3. Mike,

    Both studies used the same cohort of men. All these men met the Epstein criteria. In the first study they concluded:

    “CONCLUSIONS: Most progression after active surveillance occurs 1 to 2 years after diagnosis suggesting undersampling of more aggressive tumor rather than progression of indolent tumor. Even with progression most tumors have favorable pathology (27% potentially insignificant).”

    In the more recent one they concluded:

    “CONCLUSIONS: In men with low risk prostate cancer who were carefully selected for surveillance the PCA3 score was not significantly associated with short-term biopsy progression.”

    My point is that if most progression (increase in GS) was caused by missing the higher scores originally, why would progression and no progression show a difference in PCA3 scores? I think they are too early in their conclusions…

  4. Ralph: Now I am confused. They are saying that PCA3 levels can not be used to discriminate between patients with progression or non-progression. This would seem to agree with their prior statement that most increases in Gleason score on biopsy are because the initial biopsy simply “missed” the Gleason stage 4/5 cancer.

  5. Gentlemen: Careful reading indicates the main PURPOSE of the study was to determine the validity of this STATED Hypothesis: “… Urinary PCA3 levels have been associated with prostate cancer GRADE and EXTENT, suggesting a POSSIBLE role in monitoring patients on active surveillance …” (EMPHASIS mine).

    IN OTHER WORDS: Can an initial PCA3 result provide a valid clue to identifying those patients whose prostate cancer will progress?

    Conclusions: “In men with low risk prostate cancer who were carefully selected for surveillance the PCA3 score was not significantly associated with short-term biopsy progression.”

    IN OTHER WORDS: The individual PCA3 score did NOT show any significant INDEPENDENT ability to identify the tumor aggressiveness of individual prostate cancer patients who participated in the study. This is true, because the the scores of those who progressed, as identified by other means (mainly biopsy) had essentially the same — or insignificantly different — PCA3 results as those who did NOT progress.

    This all makes sense to me, and is why the study also concludes that: “Further analysis is necessary to assess the usefulness of PCA3 IN COMBINATION WITH OTHER BIOMARKERS OR IN SELECTED SUBSETS of patients undergoing surveillance.” Again, this all makes sense to me! (aka) az4peaks

  6. Makes sense to me too … I thought that was what the initial report had communicated! (But then I’m over 60 and therefore subject to risk for age-related delusional behavior!)

    Of course it does have to be noted that the data analyzed by the study’s authors could not possibly have provided a way to verify their actual study hypothesis because they only had one PCA3 reading to work from, based on urine samples taken at the time of initial diagnosis, as opposed to serial PCA3 data taken (say) every 6 months.


  7. Mike,

    Of course it could have! If men having consistently higher (or lower) PCA3 scores had clustered together in groups that displayed similar tendencies to either recur, or NOT to recur, conclusions could be have been drawn.

    If 95% of men with an initial score ABOVE a certain number (for example 50) had recurred and none or very few of the men BELOW that number did NOT recur, clearly the number would have had great prognostic significance. However that did NOT happen and men recurring evidently had similar score distributions to those men not recurring in THIS Study cohort.

    Therefore the conclusion that was drawn was that, IN THIS STUDY GROUP, no specific PCA3 score provided an independent prognostic benefit. This was exactly what they indicated they were trying to find out. This is how such studies work! – (aka) az4peaks

  8. Dear John:

    To quote you, the stated hypothesis was that “urinary PCA3 levels have been associated with prostate cancer grade and extent, suggesting a possible role in monitoring patients on active surveillance” (my emphasis).

    However, the researchers did not “monitor” PCA3 levels — they only assessed PCA3 levels at specific point in time — so, I cannot for the life of me see how this study was able to assess the value of PCA3 in monitoring anything. To monitor the value of PCA3 in patients on active surveillance, you would need to have taken PCA3 levels regularly over time and see if they demonstrated an association with risk for progression over time.

    We have no idea whether these men had “consistently higher (or lower) PCA3 scores” because these weren’t measured! All this study did was to assess the potential of a single PCA3 level to provide information about association with risk for progression at that point in time. There is no data provided about comparative risk groups.

    The conclusion drawn by the authors was that, “In men with low risk prostate cancer who were carefully selected for surveillance the PCA3 score” [as measured at a specific but unspecified point in time] “was not significantly associated with short-term biopsy progression.”

  9. The following is taken from the “Materials and Methods” section of the full paper:

    “Of the 294 subjects included in this study 140 (47.6%) submitted a single urine sample for PCA3 measurement, 120 (40.8%) submitted 2 samples and 34 (11.6%) submitted 3 samples. Additional samples were collected at subsequent semiannual examinations. A total of 482 urine samples were obtained from the 294 study participants. Of the 482 samples 471 yielded sufficient mRNA to measure PCA3, resulting in an informative specimen rate of 97.7%. Because PCA3 measurements did not show significant variation over time in patients with multiple measurements (data not shown), our analysis was conducted using the first PCA3 measurement for each individual.” (My emphasis added.)

    And this statement is taken from the discussion:

    “There are several limitations of this study that deserve mention. PCA3 measurements before prostate cancer diagnosis were not available in our study. Furthermore, urine samples were collected beginning in 2007, resulting in a relatively short followup to survey for evidence of progressive disease. Some of the cases classified as nonprogressors in our analysis may convert to having unfavorable biopsy features with longer followup. Nevertheless, most instances of progression have occurred in the first 2 years. Moreover it is unknown whether patients with unfavorable biopsy findings truly had progression or if this instead represents previously under sampled disease. Finally participants in our expectant management program were highly motivated to adhere to semiannual followup examinations and may not be representative of other contemporary patient populations.”

    This second statement also addresses Ralph’s prior comment about the relevance of understaging.

  10. Mike,

    Now we are into semantics, but your point about the use of the term “monitor” is well taken! It was probably a poor word choice for the description of what actually occurred, because it suggests that there was a “continuing” process involved. The fact is, however, that there WAS “monitoring” involved, but not of the test results as you choose to relate it and properly point out, but of the involved patient population, as the study relates, to determine possible recurrence.

    As far as the rest of the items, they are present in practically every study that is ever done. Each study has a set of such similar limitations. Identifying certain common factors is easy, but establishing a clear cause and effect factor related to them is often extremely difficult and, many times, impossible.

    All that you say has some validity, as do the points that I have made, and one does not necessarily invalidate the other, nor even address the same aspect of the discussion.

    All of this simply demonstrates how important INTERPRETATION of such study results are and demonstrates how, even knowledgeable people, can view the same data and draw varying conclusions.

    Effective communication between humans remains our most vexing problem, in my opinion, and we all speak the same basic language. Is it any wonder that the world, with all its varied cultures, languages, and traditional biases, has so many seemingly unresolvable issues? (aka) az4peaks

  11. Beginning to question the relevancy of my PCA3 score of 26. When I’m seeing mention of scores < 35 maybe being normal. PSA 7 and a negative DRE.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: