EMPIRE 1 results published

[Chartreuse Wombat]

18F-fluciclovine-PET/CT imaging versus conventional imaging alone to guide postprostatectomy salvage radiotherapy for prostate cancer (EMPIRE-1): a single centre, open-label, phase 2/3 randomised controlled trial

Inclusion of 18F-fluciclovine-PET into postprostatectomy radiotherapy decision making and planning significantly improved survival free from biochemical recurrence or persistence. Integration of novel PET radiotracers into radiotherapy decisions and planning for prostate cancer patients warrants...

www.thelancet.com

---

Members don't see this ad.

 

[Gfunk6]

18F-fluciclovine-PET/CT imaging versus conventional imaging alone to guide postprostatectomy salvage radiotherapy for prostate cancer (EMPIRE-1): a single centre, open-label, phase 2/3 randomised controlled trial

Inclusion of 18F-fluciclovine-PET into postprostatectomy radiotherapy decision making and planning significantly improved survival free from biochemical recurrence or persistence. Integration of novel PET radiotracers into radiotherapy decisions and planning for prostate cancer patients warrants...

www.thelancet.com

"warrants further study"

Evilcore = EXPERIMIENTAL

 

[BobbyHeenan]

"warrants further study"

Evilcore = EXPERIMIENTAL

Evilcore: nope, still gotta wait till that PSA gets to 1.0.

Me: BUt NCCN now says 0.5

Evilcore: yeah that's not in OUR guidelines

Me: *silently* the only ones that matter

 

[medgator]

Evilcore: nope, still gotta wait till that PSA gets to 1.0.

Me: BUt NCCN now says 0.5

Evilcore: yeah that's not in OUR guidelines

Me: *silently* the only ones that matter

"SABR COMET was only a phase II study"

Would love to see the evilcore talking points

 

[MegaVoltagePhoton]

I’m jealous of your access. It’s behind a paywall for me

 

[TheWallnerus]

Evilcore: yeah that's not in OUR guidelines

Me: *silently* the only ones that matter

Mystifies me why we, why ASTRO, accepts this

ASTRO needs to literally set up an anti-Evicore expert group that constantly makes press releases to major media outlets

EVICORE AT IT AGAIN:
Want to beat your prostate cancer? Evicore says "SORRY"

EVICORE's Failed Physician Corps...
Making Life-Threatening Decisions Threatening YOU

EVICORE:
You Remember That Night in College?
When You Did That Thing?
EVICORE Is Worse Than That!

EVICORE...
The Harvey Weinstein of Oncology

 

[medgator]

Mystifies me why we, why ASTRO, accepts this

ASTRO needs to literally set up an anti-Evicore expert group that constantly makes press releases to major media outlets

I've talked to a mix of docs through evilcore (and health help, AIM etc) P2Ps and a good chunk of them are in academics.... Makes you wonder

 

[RickyScott]

I've talked to a mix of docs through evilcore (and health help, AIM etc) P2Ps and a good chunk of them are in academics.... Makes you wonder

Have also found that as well.

 

[Palex80]

Here's my take on this...

The authors state:
"A radiotherapy decision attestation sheet was completed before randomisation documenting provider intent as to whether radiotherapy would be offered or not, the general treatment fields (whole pelvis vs prostate bed alone), intent to provide androgen deprivation, and duration of such therapy, so that decisions before versus after PET could be compared systematically."

Now, the authors avoided putting all these treatment decisions in a graph or text and really show how all these decisions were implemented in both groups.

A chart in the Appendix shows some decision changes in the PET-group,
Initially it seems that 45 patients were scheduled to undergo prostate bed RT and 34 patients pelvic RT.
Then, post-PET 31/45 got prostate bed RT and 14/45 got pelvic RT. So, the PET lead to a treatment intensification for about roughly a third of the patients scheduled to undergo prostate bed RT (they got the pelvis on-top, because they were cN1 on PET). At the same time, 10/34 patients scheduled to undergo pelvic RT got prostate bed RT (because the PET came back cN0) and 4/34 got nothing, since they were metastatic.

In the non-PET group 56 patients received a prostate bed RT and 25 received pelvic RT.

Unfortunately, apart from the general statement of how many patients received ADT and for how long, no further information was provided on whether or not changes occured compared to baseline. I presume, that a physician would be more willing to give ADT if he saw disease in the pelvis in the PET scan or he may ommit it if he saw nothing. Did ADT prescription intent change after the PET? We do not know.


It's striking to see the imbalance among the groups.
Already from the start, it seems that the non-PET group was scheduled to undergo considerably less treatment than the PET group.

In the non-PET group: 70% received prostate-bed only RT "at physicians choice".

In the PET-group 56% were scheduled to undergo pelvic RT from the start and then that number grew to 69% post-PET.

It is important to state that 17-19% of all patients in both groups had pN1 disease and received pelvic RT irrespective of PET or physicians choice.
So there is an unclear imbalance from the start and I do not see why the investigators did not stratify for that!

To explain the difference in event free survival between the groups the authors suggest that more precise staging with PET helped to increase treatment volumes and treat patient with pelvic disease that would otherwise be left untreated. True, 14 more patients received pelvic RT in that group than initially scheduled. At the same time however 14 patients also experienced deescalation of their treatment based on PET.
Can we really tell if omitting pelvic RT in PET-negative patients is a good idea? Can PET accurately rule out nodal disease? Do not forget the recently published paper on PET-cN0 high-risk patients from India who benefitted from pelvic RT...


This trial is being "sold" as a trial of PET, but actually it's certainly also a trial of "pelvic RT vs. prostate bed RT" in two very unbalanced groups in my humble view...

 

[RickyScott]

Here's my take on this...

The authors state:
"A radiotherapy decision attestation sheet was completed before randomisation documenting provider intent as to whether radiotherapy would be offered or not, the general treatment fields (whole pelvis vs prostate bed alone), intent to provide androgen deprivation, and duration of such therapy, so that decisions before versus after PET could be compared systematically."

Now, the authors avoided putting all these treatment decisions in a graph or text and really show how all these decisions were implemented in both groups.

A chart in the Appendix shows some decision changes in the PET-group,
Initially it seems that 45 patients were scheduled to undergo prostate bed RT and 34 patients pelvic RT.
Then, post-PET 31/45 got prostate bed RT and 14/45 got pelvic RT. So, the PET lead to a treatment intensification for about roughly a third of the patients scheduled to undergo prostate bed RT (they got the pelvis on-top, because they were cN1 on PET). At the same time, 10/34 patients scheduled to undergo pelvic RT got prostate bed RT (because the PET came back cN0) and 4/34 got nothing, since they were metastatic.

In the non-PET group 56 patients received a prostate bed RT and 25 received pelvic RT.

Unfortunately, apart from the general statement of how many patients received ADT and for how long, no further information was provided on whether or not changes occured compared to baseline. I presume, that a physician would be more willing to give ADT if he saw disease in the pelvis in the PET scan or he may ommit it if he saw nothing. Did ADT prescription intent change after the PET? We do not know.


It's striking to see the imbalance among the groups.
Already from the start, it seems that the non-PET group was scheduled to undergo considerably less treatment than the PET group.

In the non-PET group: 70% received prostate-bed only RT "at physicians choice".

In the PET-group 56% were scheduled to undergo pelvic RT from the start and then that number grew to 69% post-PET.

It is important to state that 17-19% of all patients in both groups had pN1 disease and received pelvic RT irrespective of PET or physicians choice.
So there is an unclear imbalance from the start and I do not see why the investigators did not stratify for that!

To explain the difference in event free survival between the groups the authors suggest that more precise staging with PET helped to increase treatment volumes and treat patient with pelvic disease that would otherwise be left untreated. True, 14 more patients received pelvic RT in that group than initially scheduled. At the same time however 14 patients also experienced deescalation of their treatment based on PET.
Can we really tell if omitting pelvic RT in PET-negative patients is a good idea? Can PET accurately rule out nodal disease? Do not forget the recently published paper on PET-cN0 high-risk patients from India who benefitted from pelvic RT...


This trial is being "sold" as a trial of PET, but actually it's certainly also a trial of "pelvic RT vs. prostate bed RT" in two very unbalanced groups in my humble view...

exactly- if all pts recieved adt and pelvic xrt per pollack rtog 0534, there would be no benefit,m and certainly negative pet can not eliminate pelvic xrt. now that i have some experience with axumin, much less impressed with it- seems to always light up in the inguinal area.

 

[Mandelin Rain]

exactly- if all pts recieved adt and pelvic xrt per pollock rtog 0534, there would be no benefit,m and certainly negative pet can not eliminate pelvic xrt. now that i have some experience with axumin, much less impressed with it- seems to always light up in the inguinal area.

I agree. Axumin seems unimpressive to me. Hopefully PSMA becomes widely available.

 

[madchemist89]

Can this trial potentially justify using fluciclovine pet more routinely in the salvage setting rather than using a psa threshold?

 

[BobbyHeenan]

exactly- if all pts recieved adt and pelvic xrt per pollock rtog 0534, there would be no benefit,m and certainly negative pet can not eliminate pelvic xrt. now that i have some experience with axumin, much less impressed with it- seems to always light up in the inguinal area.

YES! on inguinal.

I've sent two men for inguinal biopsy (negative) whent heir axumin was + there.

Also, had a meningioma (stable for many years) that was SUPER hot on axumin. weird.

 

[Ray D. Ayshun]

YES! on inguinal.

I've sent two men for inguinal biopsy (negative) whent heir axumin was + there.

Also, had a meningioma (stable for many years) that was SUPER hot on axumin. weird.

My impression from quick search is that arachnoid cells, meningioma cell of origin, seem to vacuum up a lot of solutes, and axumin is just tagged leucine

 

[Palex80]

Can this trial potentially justify using fluciclovine pet more routinely in the salvage setting rather than using a psa threshold?

No.

This trial tells you that fluciclovine PET will detect sites of prostate cancer in patients with biochemical recurrence after radical prostatectomy better than standard imaging. You still need to offer salvage treatment to all patients with a biochemical failure according to your threshold, in Europe we mostly use >0.2 ng/ml.
However, fluciclovine pet can detect disease beyond the pelvis and in these patients you can skip radiotherapy to the pelvis which will not cure them anyway. In Empire 1, it was merely 4 patients out of 79, that's 5%. Not awfully many, if you ask me and potentially not all of these positive findings of fluciclovine pet were true positives, perhaps?

 

[DoctwoB]

Here's my take on this...

The authors state:
"A radiotherapy decision attestation sheet was completed before randomisation documenting provider intent as to whether radiotherapy would be offered or not, the general treatment fields (whole pelvis vs prostate bed alone), intent to provide androgen deprivation, and duration of such therapy, so that decisions before versus after PET could be compared systematically."

Now, the authors avoided putting all these treatment decisions in a graph or text and really show how all these decisions were implemented in both groups.

A chart in the Appendix shows some decision changes in the PET-group,
Initially it seems that 45 patients were scheduled to undergo prostate bed RT and 34 patients pelvic RT.
Then, post-PET 31/45 got prostate bed RT and 14/45 got pelvic RT. So, the PET lead to a treatment intensification for about roughly a third of the patients scheduled to undergo prostate bed RT (they got the pelvis on-top, because they were cN1 on PET). At the same time, 10/34 patients scheduled to undergo pelvic RT got prostate bed RT (because the PET came back cN0) and 4/34 got nothing, since they were metastatic.

In the non-PET group 56 patients received a prostate bed RT and 25 received pelvic RT.

Unfortunately, apart from the general statement of how many patients received ADT and for how long, no further information was provided on whether or not changes occured compared to baseline. I presume, that a physician would be more willing to give ADT if he saw disease in the pelvis in the PET scan or he may ommit it if he saw nothing. Did ADT prescription intent change after the PET? We do not know.


It's striking to see the imbalance among the groups.
Already from the start, it seems that the non-PET group was scheduled to undergo considerably less treatment than the PET group.

In the non-PET group: 70% received prostate-bed only RT "at physicians choice".

In the PET-group 56% were scheduled to undergo pelvic RT from the start and then that number grew to 69% post-PET.

It is important to state that 17-19% of all patients in both groups had pN1 disease and received pelvic RT irrespective of PET or physicians choice.
So there is an unclear imbalance from the start and I do not see why the investigators did not stratify for that!

To explain the difference in event free survival between the groups the authors suggest that more precise staging with PET helped to increase treatment volumes and treat patient with pelvic disease that would otherwise be left untreated. True, 14 more patients received pelvic RT in that group than initially scheduled. At the same time however 14 patients also experienced deescalation of their treatment based on PET.
Can we really tell if omitting pelvic RT in PET-negative patients is a good idea? Can PET accurately rule out nodal disease? Do not forget the recently published paper on PET-cN0 high-risk patients from India who benefitted from pelvic RT...


This trial is being "sold" as a trial of PET, but actually it's certainly also a trial of "pelvic RT vs. prostate bed RT" in two very unbalanced groups in my humble view...

My other main concern is the 4 patients in the PET group found to be metastatic were not included in the event analysis. This is putting a finger on the scales IMO, because presumably an equal number of subclinical (by conventional imaging) metastatic patients were randomized to the non-PET arm and were included in analysis, With 70 something patients in each arm, that 12% Biochemical difference looks a lot worse if you include those 4 patients who presumably would have failed.

Of course you can argue you spared those patient's unnecessary treatment and the PET was beneficial, but in terms of PET guided therapy improving outcomes it is wrong to exclude them IMO.

 

[madchemist89]

Sorry, I meant routinely using fluciclovine pet despite having a low PSA and low likelihood of showing anything on imaging. We have been doing salvage on patients with psa of 0.2 or lower if trending upward, but if psa below 0.7 we have not been ordering axumin scans. This study did not seem to use any psa threshold for imaging, so I wonder with a median psa of 0.34 are we justified in ordering these scans in patients with lower psa? In general, most of our patients are referred for salvage with psa well below 1.

 

[Mandelin Rain]

Sorry, I meant routinely using fluciclovine pet despite having a low PSA and low likelihood of showing anything on imaging. We have been doing salvage on patients with psa of 0.2 or lower if trending upward, but if psa below 0.7 we have not been ordering axumin scans. This study did not seem to use any psa threshold for imaging, so I wonder with a median psa of 0.34 are we justified in ordering these scans in patients with lower psa? In general, most of our patients are referred for salvage with psa well below 1.

Yes. The issue is that UHC will only approve at a level of >1, even though we need it well before then and there is pretty good data on sensitivity and specificity below that level.

 

[Palex80]

Sorry, I meant routinely using fluciclovine pet despite having a low PSA and low likelihood of showing anything on imaging. We have been doing salvage on patients with psa of 0.2 or lower if trending upward, but if psa below 0.7 we have not been ordering axumin scans. This study did not seem to use any psa threshold for imaging, so I wonder with a median psa of 0.34 are we justified in ordering these scans in patients with lower psa? In general, most of our patients are referred for salvage with psa well below 1.

The lower the PSA, the more useless the scan will be...

 

[Ray D. Ayshun]

I thought data re psa <1 was psma. Is axumin as sensitive?

 

[ramsesthenice]

exactly- if all pts recieved adt and pelvic xrt per pollack rtog 0534, there would be no benefit,m and certainly negative pet can not eliminate pelvic xrt. now that i have some experience with axumin, much less impressed with it- seems to always light up in the inguinal area.

My other main concern is the 4 patients in the PET group found to be metastatic were not included in the event analysis. This is putting a finger on the scales IMO, because presumably an equal number of subclinical (by conventional imaging) metastatic patients were randomized to the non-PET arm and were included in analysis, With 70 something patients in each arm, that 12% Biochemical difference looks a lot worse if you include those 4 patients who presumably would have failed.

Of course you can argue you spared those patient's unnecessary treatment and the PET was beneficial, but in terms of PET guided therapy improving outcomes it is wrong to exclude them IMO.

You are right and honestly I have no idea how the editorial team at Lamcet let that slide. Oh, we are just going to let them exclude 5% of the patients in one arm of a randomized trial post randomization. This is exactly the same thing as excluding patients in a neoadjuvant arm who end up not getting surgery after therapy. It’s a criteria that can only be met in one arm. It is the definition of bias.

 

[Chartreuse Wombat]

You are right and honestly I have no idea how the editorial team at Lamcet let that slide. Oh, we are just going to let them exclude 5% of the patients in one arm of a randomized trial post randomization. This is exactly the same thing as excluding patients in a neoadjuvant arm who end up not getting surgery after therapy. It’s a criteria that can only be met in one arm. It is the definition of bias.

Agree. Lancet reviewers should have called this out with at the very least a commentary

 

[Ray D. Ayshun]

I don't agree it's the same as patients not proceeding to surgery. It's fair to exclude them from pfs read out, I think, as thats the only real read out we can look at on a short enough timeline and there's no way to count them as a progression if the readout is psa based. This approach is meant to be about treating patients correctly. Presumably, those 5 patients were able to be started sooner on proper systemic therapy for patients with metastatic disease. In turn, I don't think it would be fair to exclude them from os stats, but that won't be meaningful until 8-12 years. We all know what the right answer here is, there's just no way to run a trial ethically that would satisfy everyone, which best I can tell, would entail performing an axumin scan on everyone, and ignoring the results of the scan in some.

 

[Chartreuse Wombat]

I don't agree it's the same as patients not proceeding to surgery. It's fair to exclude them from pfs read out, I think, as thats the only real read out we can look at on a short enough timeline and there's no way to count them as a progression if the readout is psa based. This approach is meant to be about treating patients correctly. Presumably, those 5 patients were able to be started sooner on proper systemic therapy for patients with metastatic disease. In turn, I don't think it would be fair to exclude them from os stats, but that won't be meaningful until 8-12 years. We all know what the right answer here is, there's just no way to run a trial ethically that would satisfy everyone, which best I can tell, would entail performing an axumin scan on everyone, and ignoring the results of the scan in some.

Once randomized always analyzed. EBM 101

 

[Krukenberg]

Once randomized always analyzed. EBM 101

With a PSA/recurrence driven primary endpoint, I’m not sure what they are supposed to do with those patients if not exclude them. The alternative is to count them as failures, which is biased against the PET arm

 

[Ray D. Ayshun]

Once randomized always analyzed. EBM 101

Was this not accounted for upfront in the protocol? 5 progressions on day 0 would not represent reality.

 

[Chartreuse Wombat]

Was this not accounted for upfront in the protocol? 5 progressions on day 0 would not represent reality.

You are drinking the koolaid. Once randomized always analyzed. It is the only way...

 

[Krukenberg]

You are drinking the koolaid. Once randomized always analyzed. It is the only way...

Modified intention to treat is very common and accepted practice. You’re being too dogmatic.

 

[Palex80]

With a PSA/recurrence driven primary endpoint, I’m not sure what they are supposed to do with those patients if not exclude them. The alternative is to count them as failures, which is biased against the PET arm

They did not have to count them as failures immediately. It all has to do with how they defined their primary endpoint.

This is how they defined it:
"The primary endpoint of the study was 3 year event-free survival. Events were defined as: PSA of 0·2 ng/mL higher than the postradiotherapy nadir, followed by another rise; persistent PSA, imaging or digital rectal examination failure; or initiation of systemic therapy."

They could have easily adapted this to:
"The primary endpoint of the study was 3 year event-free survival. Events were defined as: PSA of 0·2 ng/mL higher than the postradiotherapy nadir, followed by another rise; persistent/rising PSA, conventional imaging or digital rectal examination failure; or initiation of systemic therapy."

Then, they would have counted events in these PET-cM1 patients as soon as one of these things happened. Likely, within the first 3 months post PET.

They also specified in the protocol how to treat patients which came up as cM1 in the PET:
"In the 18F-fluciclovine-PET/CT group, radiotherapy decisions were rigidly determined by PET findings, which were categorised as: extrapelvic or skeletal uptake (no radiotherapy); pelvic nodal uptake (radiotherapy to pelvis [45·0–50·4 Gy in 1·8 Gy fractions] plus prostate bed [64·8–70·2 Gy in 1·8 Gy fractions]); uptake in prostate bed only (radiotherapy to prostate bed); or no uptake (radiotherapy to prostate bed)."

Since they defined in the protocol how to treat those patients (no radiotherapy) these patients are STILL part of the trial. They are not excluded!
And thus, they should be analyzed as part of the trial. It is certainly not ok to exclude them from the analysis and this is why:

The trial was designed to answer the question if PET-based decisions on salvage RT make sense and uses EFS as a primary endpoint. There were 4 patients with PET-cM1-disease in the PET-arm. We can assume that there were also 4 paitents with PET-cM1-disease in the standard-imaging-arm. These patients were however not removed from the analysis, since noone knows who they were because no PET was ever performed.

However, we can assume that they also failed in terms of EFS as rapid as the patients in the PET-arm, since they only received treatment in the pelvis/prostatic bed and had untreated distant metastasis left untouched. By leaving these "undetected-cM1-because-no-PET-was-done-patients" in the analysis but excluding the "PET-cM1-patients" from the analysis the results are being manipulated.

 

[Ray D. Ayshun]

I don't disagree with the sentiments regarding trial creation and interpretation. The problem I'm having here is that we also know that molecular imaging in this setting is superior to conventional imaging. The most important read-out in the long-run is OS for the ENTIRE cohort, and I would hope they don't plan on excluding those patients in that analysis. My concern is how long it will take to get to that end-point, as with every prostate trial. My thinking about this trial is that it should be answering a question regarding the role of axumin scans in getting these patients correctly treated, while others are concerned with how axumin scans would inform contours in this population. In order to answer the former question, perhaps ignoring reception of systemic therapy as an indicator of progression is warranted, as this shouldn't be done in the first place without clinical support that there has been progression, whether imaging, clinical, or biochemical, and it would be important to know the consequences of prolonging reception of systemic therapy in patients with conventionally subclinical metastatic disease. The way to answer the question re axumin scans in informing volumes is to scan everyone, remove the M1 patients, and then ignore the scans during the planning process in half the patients, which is unethical, though something akin to this was done, I believe, in one of the earlier PSMA PET papers for patients with oligometastatic disease. In any case, I don't like excluding the patients that were excluded, but simply including them as failures on day 0 or so, as would be required since they should have initiated systemic therapy, would also cloud the ultimate veracity of the trial. The way they chose to interpret things seems the least cloudy.

 

[Mandelin Rain]

I thought data re psa <1 was psma. Is axumin as sensitive?

Diagnostic accuracy of axumin is 72% <1, 83% 1-2, and 100% >2.

 

[Ray D. Ayshun]

Diagnostic accuracy of axumin is 72% <1, 83% 1-2, and 100% >2.

Not disagreeing as those numbers are pretty good. OTOH:

https://ascopubs.org/doi/abs/10.1200/JCO.2019.37.7_suppl.15

I think PSMA is better in the long-run, which is why I'm still frustrated with the snails pace that PSMA PET has been replacing axumin here in the states.

 

[ramsesthenice]

I don't disagree with the sentiments regarding trial creation and interpretation. The problem I'm having here is that we also know that molecular imaging in this setting is superior to conventional imaging. The most important read-out in the long-run is OS for the ENTIRE cohort, and I would hope they don't plan on excluding those patients in that analysis. My concern is how long it will take to get to that end-point, as with every prostate trial. My thinking about this trial is that it should be answering a question regarding the role of axumin scans in getting these patients correctly treated, while others are concerned with how axumin scans would inform contours in this population. In order to answer the former question, perhaps ignoring reception of systemic therapy as an indicator of progression is warranted, as this shouldn't be done in the first place without clinical support that there has been progression, whether imaging, clinical, or biochemical, and it would be important to know the consequences of prolonging reception of systemic therapy in patients with conventionally subclinical metastatic disease. The way to answer the question re axumin scans in informing volumes is to scan everyone, remove the M1 patients, and then ignore the scans during the planning process in half the patients, which is unethical, though something akin to this was done, I believe, in one of the earlier PSMA PET papers for patients with oligometastatic disease. In any case, I don't like excluding the patients that were excluded, but simply including them as failures on day 0 or so, as would be required since they should have initiated systemic therapy, would also cloud the ultimate veracity of the trial. The way they chose to interpret things seems the least cloudy.

The problem is that the trial never should have been done as designed. You are exactly correct in that the question we really want to know is more along the lines of does molecular imaging help us decide how to best manage patients. The problem is here is the question they asked:

We aimed to evaluate the role of 18F-fluciclovine-PET/CT in improving cancer control compared with conventional imaging (bone scan and either CT or MRI) alone for salvage postprostatectomy radiotherapy.

This is a difficult question to answer. They are literally asking do patients do better because we used PET imaging. That being the case, you would really have to consider cancer control of the entire cohort after randomization. It doesn't take a lot of statistical training to look at their numbers (including CIs) and see that if we assume (and it is a safe assumption) that these excluded patients would have been biochemical failures, they would not have seen a significant difference in DFS between the groups. Now, they would have still seen a trend with a roughly 8% difference in DFS. To answer their specific question, they should have anticipated a percentage of patients who would have a positive PET (and 5% would have been reasonable), accepted they would be considered failures on day 1, and powered their trial to account for this. How many patients would that require? Probably around 3x as many as they enrolled. As designed, the question is almost unanswerable and their IRB never should have approved the trial as it is grossly underpowered. I also take issue with the reviewers and the editor at Lancet because in reality they absolutely did not meet their primary endpoint and the paper should not have been published in in its current form in a high impact journal.

They asked the wrong question. There are much better trials randomizing people with positive uptake (post imaging) to SOC systemic therapy vs systemic therapy plus radiotherapy or conventional radiation vs conventional RT + boost to involved nodes. As you said above, it is ethically very difficult to get a test and then ignore diagnostic information that you can reasonably assess renders your initial plan insufficient. You have to account for these thing when you design a trial.

This trial is also not helpful for getting any kind of approval. Do the math. They did 80 PET scans and ended up aborting radiation in 4 patients. What costs the payer more, 80 PET scans, or 4 courses of IMRT?

 

[Ray D. Ayshun]

The problem is that the trial never should have been done as designed. You are exactly correct in that the question we really want to know is more along the lines of does molecular imaging help us decide how to best manage patients. The problem is here is the question they asked:

We aimed to evaluate the role of 18F-fluciclovine-PET/CT in improving cancer control compared with conventional imaging (bone scan and either CT or MRI) alone for salvage postprostatectomy radiotherapy.

This is a difficult question to answer. They are literally asking do patients do better because we used PET imaging. That being the case, you would really have to consider cancer control of the entire cohort after randomization. It doesn't take a lot of statistical training to look at their numbers (including CIs) and see that if we assume (and it is a safe assumption) that these excluded patients would have been biochemical failures, they would not have seen a significant difference in DFS between the groups. Now, they would have still seen a trend with a roughly 8% difference in DFS. To answer their specific question, they should have anticipated a percentage of patients who would have a positive PET (and 5% would have been reasonable), accepted they would be considered failures on day 1, and powered their trial to account for this. How many patients would that require? Probably around 3x as many as they enrolled. As designed, the question is almost unanswerable and their IRB never should have approved the trial as it is grossly underpowered. I also take issue with the reviewers and the editor at Lancet because in reality they absolutely did not meet their primary endpoint and the paper should not have been published in in its current form in a high impact journal.

They asked the wrong question. There are much better trials randomizing people with positive uptake (post imaging) to SOC systemic therapy vs systemic therapy plus radiotherapy or conventional radiation vs conventional RT + boost to involved nodes. As you said above, it is ethically very difficult to get a test and then ignore diagnostic information that you can reasonably assess renders your initial plan insufficient. You have to account for these thing when you design a trial.

This trial is also not helpful for getting any kind of approval. Do the math. They did 80 PET scans and ended up aborting radiation in 4 patients. What costs the payer more, 80 PET scans, or 4 courses of IMRT?

I agree with all that. Its the typical issue with prostate trials in finding a surrogate for OS until those data mature, accepting this trial is probably too small to find an OS difference, but maybe. In any case, the real math would be 80 PETS vs 4 IMRT + 80 CT AP + 80 bone scans a my hope would be that molecular imaging would replace conventional and not come after.

 

[Chartreuse Wombat]

Modified intention to treat is very common and accepted practice. You’re being too dogmatic.

Lancet is one of the most prestigious journals. In this scenario modified ITT is very biased as has been laid out in detail from posts above. I am all for being pragmatic but to exclude these patients is just wrong. As has been stated above it clearly stacks the deck.

 

[Mandelin Rain]

Not disagreeing as those numbers are pretty good. OTOH:

https://ascopubs.org/doi/abs/10.1200/JCO.2019.37.7_suppl.15

I think PSMA is better in the long-run, which is why I'm still frustrated with the snails pace that PSMA PET has been replacing axumin here in the states.

No doubt. I'm not a huge fan of Axumin and am excited for PSMA. I'm just even less of a fan of the criteria that PSA has to be >1.0 to order as per UHC.

No one is (nor should be) waiting to 1.0 for salvage just to order this test. Makes it rather low utility.

 

[evilbooyaa]

Really interesting discussion on this paper.

I was personally offering Axumin to all salvage patients (explaining the caveat that the chances it finds something are quite low and that we would proceed with treatment as originally planned in the setting of a negative Axumin scan) and recommending it to anyone above a 0.5 anyways, and basically mandating it for anyone with PSA > 1.

Not sure if this paper changes my practice regardless of the controversy about statistical modeling. I'd be interested in how others are practicing in the Axumin era for salvage prostate cancer.

IMO, it may be the wrong path (or perhaps, not the 'cleanest' path) to reach a situation where Axumin (and eventually, PSMA-PET) is considered SOC for EVERY salvage patient rather than the tired old CT + bone scan, but that is certainly the goal I would like to see.

Since we're on the topic of Axumin pet - I wish it was being studied in definitive prostate cancer as a more sensitive marker for metastatic disease than say the same old CT and bone scan. Basically I wish all bone scans went the way of the dinosaur in anyone with localized disease.

 

[Mandelin Rain]

Prostascint 4 lyfe

 

[ramsesthenice]

Really interesting discussion on this paper.

I was personally offering Axumin to all salvage patients (explaining the caveat that the chances it finds something are quite low and that we would proceed with treatment as originally planned in the setting of a negative Axumin scan) and recommending it to anyone above a 0.5 anyways, and basically mandating it for anyone with PSA > 1.

Not sure if this paper changes my practice regardless of the controversy about statistical modeling. I'd be interested in how others are practicing in the Axumin era for salvage prostate cancer.

IMO, it may be the wrong path (or perhaps, not the 'cleanest' path) to reach a situation where Axumin (and eventually, PSMA-PET) is considered SOC for EVERY salvage patient rather than the tired old CT + bone scan, but that is certainly the goal I would like to see.

Since we're on the topic of Axumin pet - I wish it was being studied in definitive prostate cancer as a more sensitive marker for metastatic disease than say the same old CT and bone scan. Basically I wish all bone scans went the way of the dinosaur in anyone with localized disease.

I typically use some kind of PET if it is likely to help. I was able to use Gallium PSMA scans on trial for a while, then had to go back to Axumin, and now have access to the Gallium scans again (everyone will soon ). I am munch happier with them than the Axumin scans. I honestly do a risk assessment to decide who gets a PET. I look at the total PSA, doubling time, time to failure, and other high risk features. Our urologists are actually pretty aggressive about salvage radiation. If someone pops up with a PSA of 0.1 a few months after surgery and 6-8 weeks later it is 0.2 they are sending over for treatment. They don't mess around. Even a gallium PET is unlikely to be very helpful at that point. The vast majority of the time that we find anything it is regional nodes. Off hand I can only think of 1 person in the past 2 years who I aborted radiation because a PET found metastatic disease (which if you look at the above referenced trial makes sense as they only found 4/80). I typically boost the gross nodes to something a little higher (60-66 if I can). The only times I end up changing my fields are if they have a high common iliac or low AC node. I will chase a little bit higher than normal in those cases. Twice I have found a bone met in the pelvis with nodes. I put them on ADT for a few months and reassessed. In both cases they had good PSA responses and I treated the pelvis with conventional fractionation and included the bone met (not conventional+ SBRT). Both are disease free but its too early to tell. We did 2 years of ADT for both. One has been off for about 6 months and the other is still on it. Time will tell if it was the right thing to do or not.

And PET scans are being looked at for the definitive setting. It has just been slower. Axumins are crappy enough I can understand why. What you really want is a tracer that is sensitive enough to pick up tumor deposits in lymph nodes that are small enough you would overlook them on a contrasted CT. They exist and will be in everyone's hands eventually. But studies with Axumins would probably not have helped our cause and might even have been harmful (if negative).

 

[seper]

People are seeing isolated nodal relapses? That would be news in prostate cancer. You should write it up.

 

[Palex80]

We have access to PSMA-PET-CT and full reimbursement.
Indications are staging prior to definitive treatment for high & very-high disease and PSA-relapse post local treatment.

Thus, we use it a lot. I have seen all kinds of findings and not all of those findings are actually prostate cancer.
A lot of spots in bones are being picked up, especially in the ribs, which we have even done biopsies for and rarely found prostate cancer.
Once, a bony lesion came back as focal mastocytosis, apparently it can also lead to PSMA-uptake.
I've had a patient who everyone called metastatic prostate cancer, but it was all just small lymphocytic lymphoma (undiagnosed until then).

We do see isolated nodal relapses and have treated quite a few of those with salvage RT (focal or whole pelvis).

Our policy has shifted during the past couple of years from believing everything we see on the PSMA-PET-CT to verify findings with a second imaging modality or biopsy, whenever possible. There are simply a lot of false positive ones.

 

[evilbooyaa]

We have access to PSMA-PET-CT and full reimbursement.
Indications are staging prior to definitive treatment for high & very-high disease and PSA-relapse post local treatment.

Thus, we use it a lot. I have seen all kinds of findings and not all of those findings are actually prostate cancer.
A lot of spots in bones are being picked up, especially in the ribs, which we have even done biopsies for and rarely found prostate cancer.
Once, a bony lesion came back as focal mastocytosis, apparently it can also lead to PSMA-uptake.
I've had a patient who everyone called metastatic prostate cancer, but it was all just small lymphocytic lymphoma (undiagnosed until then).

We do see isolated nodal relapses and have treated quite a few of those with salvage RT (focal or whole pelvis).

Our policy has shifted during the past couple of years from believing everything we see on the PSMA-PET-CT to verify findings with a second imaging modality or biopsy, whenever possible. There are simply a lot of false positive ones.

This is my big concern with all the molecular imaging data is that I am not aware of any series doing a thorough evaluation of positive predictive value and radiologic/pathologic correlation study before just treating it as gospel.

For example, the false positive rate of PET in vulvar cancer in inguinal LNs (25% of PET avid inguinal LNs will actually be negative on lymph node dissection). What percentage of F18 (or PSMA) PET positive inguinal LNs in prostate cancer are actually negative on path evaluation? What percentage of F18/PSMA PET positive in the ribs, the spine, the soft tissues, whatever, are false positives on path evaluation?

 

[BobbyHeenan]

This is my big concern with all the molecular imaging data is that I am not aware of any series doing a thorough evaluation of positive predictive value and radiologic/pathologic correlation study before just treating it as gospel.

For example, the false positive rate of PET in vulvar cancer in inguinal LNs (25% of PET avid inguinal LNs will actually be negative on lymph node dissection). What percentage of F18 (or PSMA) PET positive inguinal LNs in prostate cancer are actually negative on path evaluation? What percentage of F18/PSMA PET positive in the ribs, the spine, the soft tissues, whatever, are false positives on path evaluation?

I'm THIS close to just ignoring inguinal nodes on axumin PET.

I'm 0/3 on biopsy proven inguinal nodes.

 

[deleted969641]

I'm THIS close to just ignoring inguinal nodes on axumin PET.

I'm 0/3 on biopsy proven inguinal nodes.

My stats are about the same. Still, it's really hard not to cover a Cloquet-ish node that lit up but was biopsy negative. I have been extending traditional EI contours further down (and up into CIs as well) for a while because I see this so often and was pleased to see updating nodal contouring guidelines come out confirming my practice. So what are you going to do if you forego biopsy and have a suspicious node at your field border? I can't tell you how many times the PTV expansion of my standard contours ends in the middle of a 1.2 cm somewhat irregular high inguinal node and how deeply unsatisfying that is.

 

[BobbyHeenan]

My stats are about the same. Still, it's really hard not to cover a Cloquet-ish node that lit up but was biopsy negative. I have been extending traditional EI contours further down (and up into CIs as well) for a while because I see this so often and was pleased to see updating nodal contouring guidelines come out confirming my practice. So what are you going to do if you forego biopsy and have a suspicious node at your field border? I can't tell you how many times the PTV expansion of my standard contours ends in the middle of a 1.2 cm somewhat irregular high inguinal node and how deeply unsatisfying that is.

If biopsy (-) I've been not intentionally covering the inguinals.

I was mostly being sarcastic...since they're so amenable to biopsy I'll probably just keep sending them. It's just annoying to me and patient(s).

 

[deleted969641]

If biopsy (-) I've been not intentionally covering the inguinals.

I was mostly being sarcastic...since they're so amenable to biopsy I'll probably just keep sending them. It's just annoying to me and patient(s).

Sorry, to clarify I would not electively cover inguinals in prostate caner in basically any circumstance (even if biopsy + I'm not sure it makes sense if you're treating that do anything but cover the positive node to a definitive dose + margin). I just meant that extend my inferior EI contour to where the vessels are totally outside of the pelvis (below femoral head) and then if the volume overlaps a suspicious node I will add a few extra slices to grab it because I am neurotic and probably slightly autistic (which leaves me with the equally unsatisfying problem of having the contralateral side not be symmetric).

 

[Chartreuse Wombat]

Sorry to be pedantic but 0/3 tells you very little.

Updated: Confidence interval with zero events (created 2001-01-19, updated 2010-07-07)

www.pmean.com

So zero to 3/n is an approximate 95% confidence interval for a data set where we observed 0 events in n patients. This approximation works very well, even if we only have a few patients.

So 95% CI is 0-100

 

[deleted969641]

Sorry to be pedantic but 0/3 tells you very little.

Updated: Confidence interval with zero events (created 2001-01-19, updated 2010-07-07)

www.pmean.com

So zero to 3/n is an approximate 95% confidence interval for a data set where we observed 0 events in n patients. This approximation works very well, even if we only have a few patients.

So 95% CI is 0-100

It's always fun to run stats in situations that make no sense.
I am reminded of this resident-education paper I saw recently: Resident attitudes and benefits of mock oral board examinations in radiation oncology | BMC Medical Education | Full Text (biomedcentral.com)

 

[Chartreuse Wombat]

It's always fun to run stats in situations that make no sense.
I am reminded of this resident-education paper I saw recently: Resident attitudes and benefits of mock oral board examinations in radiation oncology | BMC Medical Education | Full Text (biomedcentral.com)

So the inference that 0/3 proves false-positive makes sense? The figure above is silly but humans are very bad at estimating when there are no events. Even with 0/10 the upper border of the CI is 30%. I have seen many interpret 0/10 as proof that the risk is <5%. Simply not true

 

[deleted969641]

So the inference that 0/3 proves false-positive makes sense? The figure above is silly but humans are very bad at estimating when there are no events. Even with 0/10 the upper border of the CI is 30%. I have seen many interpret 0/10 as proof that the risk is <5%. Simply not true

I was actually agreeing with you and joking about the absurdity of making conclusions of an intervention based on a single digit sample size. On the other hand, medicine is not blackjack. If I do a thousand controlled studies of playing perfect strategy blackjack for an hour at Casino X vs. Casino Y, I am confident with what I know now that the results will show that my relative odds of winning are exactly the same at either (49.5%). There is exactly one variable with a known outcome. If I do a thousand controlled studies of 4000 patients with medial breast tumors receiving BCS alone or BCS and RNI, I am far less confident that the relative odds of not being dead from breast cancer with RNI will be exactly 20% higher in 15 years. There's a lot more variables at play and trials help guide the "art" of what we do, along with our experience to make risk/benefit judgements in treatment. Or else I just suck at stats (probably more likely, I definitely suck at blackjack as my lifetime odds are nowhere near 49.5%).

Buuuuutttt..... I think he was joking. I-saw-this-happen-to-a-patient-one-time-therefore-anytime-these-symptoms-happen-in-the-future-to-another-patient-i-know-exactly-what-it-is-and-what-to-do is only a lesson on the syllabus at nurse practitioner school (sorry). At medical school, we are taught better "thou shalt not question 0.05 (or 0.056)."