How accurate are gallium-68 PSMA PET/CT scans in the “real world”?

A new article in the journal Cancer Imaging has provided us with an interesting set of “real world” data on the accuracy of gallium-68 PSMA PET/CT scans in the management of prostate cancer.

This article by Keidar et al. is available as a free full-text report, and is based on data from two large, academic medical centers in Israel over a 24-month period. The same standard protocol is in place at both institutions with respect to the use of gallium-68 PSMA PET/CT scans.

Here are the core data provided in this study:

  • 445 68Ga-PSMA studies were carried out in 438 men.
  • The average (mean) age of the men was 72.4 years (range, 51 to 92 years).
  • The average (mean) PSA levels of the men were
    • 46.9 ng/ml at diagnosis (range, 0 to 4,000 ng/ml: median, 11.0 ng/ml)
    •  18.4 ng/ml at the time of the PET/CT scan (range, 0.05 to 533 ng/ml; median, 4.3 ng/ml)
  • The men had been referred for 68Ga-PSMA scanning for any one of five possible reasons:
    • Staging after initial diagnosis with high-risk disease (n = 112, 25 percent) — Group A
    • Suspected bone metastases (n = 11, 2 percent) — Group B
    • Response assessment after initial treatment for prostate cancer (n = 30, 7 percent) — Group C
    • Follow-up after treatment for prostate cancer (n = 22, 5%) — Group D
    • Biochemical failure after initial treatment for prostate cancer (n = 270, 61 percent) — Group E
  • 319/445 studies (72 percent) led to the identification of 68Ga-PSA avid disease sites (i.e., a positive scan finding) among
    • 94/112 (84 percent) men in Group A
    • 5/11 (45 percent) men in Group B
    • 22/30 (73 percent) men in Group C
    • 10/22 (45 percent) men in Group D
    • 181/270 (67 percent) men in Group E
  • 68Ga-PSMA avid lesions were most commonly detected in
    • The prostate itself (n = 193, 43 percent)
    • The immediate (loco-regional) area close to the prostate (n = 51, 11 percent)
    • The abdomino-pelvic lymph nodes (n = 129, 29 percent)
    • Distant tissues, i.e., as true metastases (n = 158, 36 percent)
  • Distant 68Ga-PSMA-avid metastases were commonly seen in men with biochemical recurrence (14/21 lesions), but not seen in men referred for staging (p < 0.013).
  • The rate of detection of active prostate cancer was clearly associated with the patients’ PSA levels at the time of their scans , as follows:
    • 31 percent among men with PSA levels of 0 to 0.99 ng/ml
    • 63 percent among men with PSA levels of  1 to 1.99 ng/ml
    • 74 percent among men with PSA levels of 2 to 3.99 ng/ml
    • 77 percent among men with PSA levels of 3 to 9.99 ng/ml
    • 90 percent among men with PSA levels of ≥ 10 ng/ml
  • 96 non-malignant 68Ga-PSMA avid foci were observed in 81 studies, of which the most common were in
    • Lymph nodes (n = 36, 38 percent)
    • Non-malignant bone lesions (n = 21, 22 percent)
    • Thyroid nodules (n = 9, 9 percent)
    • Ganglions (n = 9, 9 percent)
    • Lungs (n = 8, 8 percent)

The bottom line, based on this study, appears to be that

  • Gallium-68 PSMA PET/CT scans has findings that correlate relatively well with autopsy-based findings, indicating that 68Ga-PSMA PET/CT is an accurate detection tool in patients with metastatic prostate cancer.
  • False positive results (for risk of prostate cancer) were observed in 96/445 gallium-68 PSMA PET/CT scans (22 percent) in this “real world” cohort of patients.
  • We do not have data from this study on the rate of false negative findings.
  • For men with a PSA level of < 2.00 ng/ml at the time of their gallium-68 PSMA PET/CT scan, there appears to be the strong possibility of false negative findings.

These data suggest that gallium-68 PSMA PET/CT scanning has considerable potential value in the work-up, monitoring, and management of men with PSA levels of 2 ng/ml and higher at the time of their scan. However, the value of such scans in men with lower PSA levels appears to be open to more question. Clearly, recurrent or evident areas of prostate cancer can be observed in some men with PSA levels of < 2 ng/ml. However, what is harder to know is the risk for false negative results in such patients in the “real world” of medical practice.

It is also worth noting that, in the conduct of this study, all scans were were being reviewed retrospectively (with full knowledge of the patient’s clinical history and results of previous imaging studies) by either two specialists in nuclear medicine or a specialist in nuclear medicine and a radiologist. The final results  were reported in consensus by one or other pair of physicians. The importance of having the results of this type of scan carefully reviewed by specialists who understand and have experience in reviewing scans of this type should not be under-estimated.

14 Responses

  1. This study raises questions.

    I recently had this scan done at UCSF due to a rapidly rising PSA (from 0.3 to 0.7 to 1.6 in 6 months) after stopping ADT. A bone scan scan 6 weeks prior was negative.

    The 68Ga PSMA PT/CT scan found radiotracer uptake in the left third rib and the right scapula without corelation by CT. These seem like unusual places for bone mets. Eighteen months ago I had one bone met in my right femur imaged by an Axumin PET/CT scan. I hit it with SBRT and it’s no longer there.

    In reading this study I note false negatives and false positives. How would one go about determining this? By biopsy? By some other type of scan?


  2. What is Sitemaster’s reasoning for a strong possibility of false negatives for men with PSA < 2.00 ng/ml?

    Shouldn't the false positive percentage be 81/445, 18% not 22% comparing like to like, i.e. studies to studies, not avid foci to studies?

  3. Dear Bob:

    None of the scans we have available today are anything approaching 100% accurate in men with relatively low PSA levels like yours. They provide “guidance” not absolute truth.

    On the other hand, I am a little puzzled by your statement that the two places showing uptake of the 68Ga PMSA are “unusual”. They don’t sound any more unusual to me than others. Metastases do not necessarily appear in some sort of orderly fashion.

    With regard to what “the best” thing is to do about these, I think you are going to need to do two things. First, talk to the doctors about how they are interpreting these scan results. They will be well aware of the risk for false positives associated with this type of scan. Second, talk to the doctors about whether the potential benefits associated with repeat “zapping” of these possible mets with SBRT outweigh any risks. They might suggest that you go back onto the ADT for 6 months and combine that with the SBRT, but that’s really a conversation you have to have with the people at UCSF.

    You certainly could have bone biopsies done on the two sites identified, but bone biopsies aren’t a lot of fun, and so you should discuss those with the doctors too. I know of no scan that can be used to resolve the issue of whether the results of the 68Ga PMSA scan are truly positive or not.

  4. Dear Rick:

    I am sorry but I don’t understand your questions/comments at all.

    The comment about the problem with a high risk for false negatives can be clearly identified in the paper and in the data above. The rate of detection of active prostate cancer was < 65% in all men with a PSA level of < 2 ng/ml at the time of their scan. That's not exactly a reassuringly accurate response rate. It's only about 75% in the men with a PSA level of 2 to 9.99 ng/ml.

    With regard to the false positive rate, any occurrence of a site of gallium-68 PSMA uptake (an "avid focus") that is actually non-malignant is, by definition, a false positive result. There were 96 of these results in 445 scans of 438 men. The fact that they occurred in 81 of the scans is irrelevant.

    One could, in fact, very reasonably argue that the false positive rate was the number of incorrect results divided by the total number of positive results for gallium-68 PSMA uptake, i.e., 96/319 = 30%.

    I am not sure what your point is. What people need to understand is that while these newer scans are quite certainly better than the older ones, they are still a very long way from being accurate for men with relatively low PSA levels, many of whom seem to be under the impression that they are going to get highly accurate results from these scans when their PSA level blips up from (say) < 0.1 to 0.25 ng/ml after first-line treatment. These data clearly show that that is not the case out in the real world.

    Also, see Bob’s question/comment based on his own experience at UCSF, which is highly relevant since his PSA level at the time of his own scan was 1.6 ng/ml.

  5. I guess that as usual I was looking for a silver bullet again where none exists. I talked myself into believing that this new scan was bulletproof if I waited for PSA to climb to near 2.0. Based on my doubling rate I figured my PSA was around 2.0 when I had the scan which appeared optimal based on statistics of scan efficiency at varying PSA ranges. The question of false positives didn’t even dawn on me.

    So I’ve restarted ADT3 and Xgeva and need to think hard about applying SBRT to what could be benign tumors. And I stand corrected about my “unusual” mets. Actually I’m very fortunate to have so few bone nets and no visceral mets.

  6. I believe there is a serious flaw in this study that misled the authors to believe there were more false positives than there actually were. In this study, “Any focal 68Ga-PSMA uptake … and with a corresponding morphological abnormality on CT was considered pathological and suspicious for malignancy.”

    So what they did was they ruled out malignancy for any site of 68Ga-PSMA avidity that they could not confirm with a lesion on the CT scan. I think this was a mistake. This PET scan may pick up sites of cancer before there is a lesion large enough to be seen on CT.

    Anecdotally, I just saw a Na18F PET scan next to a 68Ga-PSMA-11 scan from the same patient at about the same time. The NaF PET only showed two spots while the Ga-68-PSMA PET lit up like a Christmas tree with multiple bone metastases, particularly showing bone marrow involvement that was not evident from the CT. In fact, the radiologist noted that the many of the bone metastases were not correlated with CT findings. CT (or MRI) is important for locating metastases, but it should not be used to confirm the PET findings. The PET is better than the CT.

    FYI: This article shows the genuine sources of false positives, which are rare.

  7. Dear Allen:

    I think I’ll let you fight that one out with the authors.

  8. Thank you Allen …. I am truly concerned that some of the editorial is scaremongering. All techniques have false positives and negatives; personally I would seek out the most current technology, gallium-68 and 18F-Pyl, as better, earlier indicators.

    We do people like Bob no favors by scaring him off scans that are likely to show metastasis earlier, and thus allow him to commence treatment sooner that we know is to his advantage.

  9. TA

    I and probably many others (including the Sitemaster) would really enjoy seeing a dialogue between you and the authors to sort out the differences of opinion you describe.

    I too was confused about the non-correlation by CT of the two bone mets found by PET; wondering why there should have been correlation in the first place.

  10. Dear Rick:

    (1) I am not sure what you are intending to imply when you write that “some of the editorial is scaremongering”. I merely reported what the authors stated in their paper. Could they be wrong? Possibly. In which case, please take that up with the authors. I am merely the messenger.

    (2) You also write that, “All techniques have false positives and negatives”. In this you are correct, but there are lot of people who seem to have got the idea that certain recent types of PET scan are near to infallible, and that is quite certainly not the case at all. We saw this all before when [11C]choline PET/CT scans first became available, and it took about 2 years before it was clear that these scans really shouldn’t be done in anyone with a PSA level of < 2 ng/ml.

    Just because something is shiny and new doesn't make it the best thing since sliced bread. Come to think of it, unsliced bread usually tastes a good deal better that sliced bread too!

  11. I think this all raises a question worth considering. Whenever there is a new diagnostic test, it is evaluated against some “gold standard” — either (1) the previous best test or (2) something with (hopefully) close to perfect positive predictive value (PPV) and negative predictive value (NPV). For any who don’t know, PPV is the percentage of all positives that are true positives (and not false positives); NPV is the percentage of all negatives that are true negatives (and not false negatives).

    For the 68Ga-PSMA-11 PET scan, some have used histological analysis of metastases as the “gold standard”; for example, in <a href= German study, 42/319 patients underwent histological analysis of pelvic lymph nodes.

    They found the PPV was 100% of lesions, and the NPV was 91.4% of lesions. In other words, all of the metastases that were positive on the PET scan were confirmed as positive on histological examination, and there were only 30 false negatives out of the 446 lymph nodes examined in those 42 patients. I think we can agree that the PET scan did pretty well in that limited test, at least for pelvic lymph nodes. However, for lymph nodes, a Combidex MRI outperformed an [11C]choline PET/CT scan. I hope that Jelle Berentz will test Combidex head-to-head against PyL.

    The NIH is now running a trial of PyL in which lesions identified in 330 patients in two categories (high risk or recurrent) will have identified lesions (including bone metastases) biopsied. High-risk patients who choose prostatectomy will have agreement or lack of agreement with whole-mount pathology identified — an ideal “gold standard”.

    The NIH is also running another clinical trial comparing PyL to NaF PET scans in the same patients. NaF only detects bone metastases (not lymph node or visceral metastases). Those who are PyL-positive will also have a biopsy of tumors.

    These clinical trials will provide much better data about the PPV of the PET scan.

  12. Allen:

    Agreed. Completely.

  13. It makes one wonder if this new radiotracer has been rushed into use, yet as I understand it, the utility of this and other radiotracers is often being judged based on the percentage of treatment changes resulting from imaging results. So is it really worthwhile in terms of improved overall survival to have these scans at PSA levels too low to to be confirmed by CT or MRI ?

  14. I think you are asking a good question — can an advanced PET scan in recurrent men improve survival? For 18%, it may result in potentially curative pelvic lymph node treatment. We know it can also rule out unnecessary treatment. Here’s a table showing the changes to treatment decisions it has made and their significance.

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 )

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: