PSMA PET Standardized Reporting "Presentation" - Matthias Eiber
February 9, 2024
At the 2024 UCSF-UCLA PSMA Conference, Matthias Eiber advocates for standardized reporting in PSMA PET imaging, underscoring its significance in improving clinical and research communication, with frameworks like PROMISE, PSMA-RADS, PPP, and RECIP enhancing treatment decision-making and lesion assessment. Dr. Eiber envisions future advancements in validating these frameworks' prognostic value and integrating electronic tools like Pylarify AI, alongside efforts like Wolfgang Fendler's PROMISE registry to evaluate the miTNM staging system's prognostic value across different disease stages.
![UCLAUCSF video logo stroke](/images/UCLAUCSF_video_logo_stroke.png)
Biographies:
Matthias Eiber, MD, PhD, Department of Nuclear Medicine, Technical University of Munich (TUM), Munich, Germany
![UCLAUCSF video logo stroke](/images/UCLAUCSF_video_logo_stroke.png)
Biographies:
Matthias Eiber, MD, PhD, Department of Nuclear Medicine, Technical University of Munich (TUM), Munich, Germany
Read the Full Video Transcript
Matthias Eiber: Thanks, Jeremie, for the introduction at the beginning of the session. And Johannes, Tom, and Jeremie, thanks for inviting me to this wonderful meeting here. And now, as we have heard when we should use PSMA PET for what reason, my task is a little bit to talk about standardized reporting, so what we do as imaging specialists to tell our clinical colleagues what we have found, in what way, and what implication that also could have using different systems.
Here are my conflicts of interest, and then we start right away. Why do we want to have standardized reporting? There are two big scenarios for what this can be used. One is for clinical reporting. I think, as imaging specialists, we have to make sure that we always speak the same language. Therefore, we facilitate the exchange of information between centers and also disciplines so that our clinical colleagues understand us and that also our imaging colleagues understand us. And therefore, we should provide concise information and especially clear answers to clinical questions. I think this was a topic of several talks here.
Also, having standardized reporting, I think, also increases repeatability and trust in imaging reporting. We have talked about pitfalls a lot during today. And also, our main goal in the end is that we support the clinicians in tailoring their therapy and assessing prognosis. However, we have an interest in research, and therefore standardized reporting is also important. The fact that we want to validate findings, we have seen a lot of prospective studies where, for example, histological validation was done and therefore we have to standardize report our findings so that histological or confirmatory imaging for validation becomes straightforward. We should also use this so that we can pool our data and finally do this for multicenter meta-analyses. And the frameworks I will introduce are also thought for using them in prospective trials and maybe adding what PSMA can do furthermore to clinical endpoints.
Basically, what I would like to do in my talk is give an overview of the different frameworks which now exist in the literature. Jeremie has also already introduced that there is the so-called PROMISE framework, the prostate cancer molecule imaging standardized reporting. This is a system that basically describes the extent of the disease on TNM level. I will come to this a little bit more in detail later. Then we have the PSMA-RADS system, which is a system similar to the RADS system for other organs that we assess lesions or the nature of a lesion on a five-point scale.
And then we have two systems where we can look at a response after systemic treatments, so response in metastatic disease. And there we have the PPP system, which follows the path of prostate cancer disease assessment, that it discriminates progressive disease from non-progression. This is what, for example, PCWG3 criteria are doing. And then we have the framework which already was introduced a little bit by Wolfgang, the RECIP framework, which also introduces the concept that we can use molecular imaging techniques to assess response. So not only progression versus non-progression but also add the category of response. Let me go through the different systems. So the PROMISE system was in fact born here in California. I think that was like 2016, 2017. And a lot of the people who have been building this up are sitting in this room. The system was updated this year or the end of last year. And the system is basically a system that follows the pathological TNM staging. And I think all our clinicians know what TNM staging means, and therefore it is constructed to follow this path.
So how is PROMISE designed? We have suspected lesions, and then we group the suspected lesions, whether they are local lesions or whether they are regional lymphatic lesions or distant metastases. We have different levels on which we assess all the lesions. We have a certainty aspect so that we make our minds on how certain we are that this is a prostate cancer lesion or whether it is something else. I refer regarding this to the very excellent pitfall talk.
We have already heard from Louise about her PRIMARY score, and the PRIMARY score is also part of this certainty assessment, if the prostate is still in place. And this is something, Louise, I have to disagree with you a little bit because we have implemented this. And I think also with your agreement that we also use this, I know there are obstacles, I know there are challenges, but also that we agreed for biochemical recurrence after radiation therapy, so where the prostate is still in place.
In addition to that, if the prostate is removed, we use the PSA expression score. This is a five-point score Wolfgang has already shown. I will show it later. Then we grade the disease on its extent. So we have the categories of local recurrence and of T0-T4. You can find all the details for this in the respective publication. And then we do something similar for N and M staging, that we use a five-point scale for assessing the certainty that this is a malignant lesion. And then we have different categories that very closely follow the pathological categories to group the disease.
For exploratory reasons, we could add the volume of disease more for scientific aspects. But in the end, with what we come out, and I think this is essential because this is something which our clinicians like because it's a very concise and a very short description of the dissemination of the disease, that is a so-called miTNM Code. And there we code our findings based on the system, here, for example, miT2, N2, M1a, M1b, and all our clinicians have an impression of what that means if they just read this single line. And then we also add the PSA expression score, both for certainty reasons that we give our clinicians the impression of what kind of uptake our lesions have, as well as in the case of radioligand therapy, that we provide the lowest and the highest score with regards to eligibility.
I don't want to go into details of all the different TNM stages which are described, so I already say that this can be taken from the publication but also the publication is... To the publication's so-called pocket card is added, which can be really used during routine reporting to enhance and orientate on this system.
I said that the PSMA expression score is part of it, it's part of it regarding the certainty of lesions, of lesion assessment in benign versus malignant disease. We completely agree, I think, together that PSMA uptake in PET is not the only discriminator for a disease or not. We have to look at anatomical imaging. We have to have pitfalls in mind, as outlined today in the morning.
Wolfgang has already told us that there are different categories for the PSMA expression score and that if we use it for radioligand therapy, we usually select patients with lesions higher than liver, which fall into the category of PSMA expression score two and higher. And this is also the same category, which usually, if we don't take into account pitfalls or other confounders, which are usually the threshold for discriminating between unspecific and cancer-related disease, if you look at these pocket cards.
What I want to show you here is how we now, since several years, structure our report in Munich. We have created this table which we add to the end of all of our reports and we really, we go by different TNM categories. It's not that we provide our clinician a very precise report in the conclusion, but we really very strongly go with this system that we assess local disease, whether it's present or not, that we assess local regional lymph node disease, whether it's present or not. And that we group for M1a, which are distant lymph node metastases, for bone metastases, and visceral metastases. And we provide the clinicians in this way. And then at the end of the conclusion, we also provide the aforementioned miTNM Code line and we get very positive feedback because, as everybody, or as our clinicians are short on time, and usually sometimes, or usually, they only read this line here and then they have an idea about the distribution of the disease in the respective patient.
I say that I want to cover different systems. I would shortly like to introduce the PSMA-RADS system, the Prostate-Specific Membrane Antigen Reporting and Data System. Also, a new version has been published last year. This system very much parallels other RADS systems like the TI-RADS, the LI-RADS, the PI-RADS, or whatever RADS. It is not so much intended to structure the extent of the disease but more to assess the nature of a lesion.
And therefore, in my opinion, it's really a very good dictionary of typical pitfalls and how to approach them. Because especially the category of PSMA-RADS-3, which is equivocal lesions, where depending on the kind of the lesion further work or follow-up is recommended. This is given very nicely in detail how this lesion should be managed. And I think for this, the PSMA-RADS system is really a good addition to the literature. Also, for clinical practice, especially if somebody is starting with PSMA PET reporting.
I can go over this system a little bit quicker because it was already introduced. I'm now going to the systems for assessing response to metastatic disease. I start with the RECIP system, the Response Evaluation Criteria in PSMA PET. I would like to inform you a little bit about the history of this classification. The idea was, and that was also the clinical impression, and I'm sure a lot of you have a similar impression of the patients, the tumor response is related to changes in tumor burden and also the occurrence of new lesions.
So really, the intent was to account for these two different aspects, and therefore the hypothesis for this system was that partial response in tumor volume without new lesions probably has a superior overall survival than partial response and new lesions. And similarly, that progression in tumor and new lesions is probably worse than progression in tumor volume, but no new lesion. And that is how basically this RECIP classification was organized. And regarding the use of molecular imaging, quantitative molecular imaging, the idea was to include also the category of partial remission or response into a prostate cancer response assessment framework. What basically was done is from three different centers, patients were pooled who underwent lutetium PSMA treatment and had a PET prior to the first treatment and after the second treatment. Different cutoffs for volumes were tested, both for volume increase as well as for volume decrease. In the end, a cutoff of 30% decrease in the volume of PSMA-expressing disease correlated best with overall survival, similar to a 20% increase of volume, as you can, for example, see here in these Kaplan-Meier curves.
And then also the hypothesis of additional new lesions was tested. And you can also see that there was a significant difference in patients with and without new lesions. And that volume changes and the appearance of new lesions are not always 100% aligned. You can see from this graph here, where you can see on the y-axis the change in the volume on PSMA PET. And you see that here, in patients with decreasing volume, you have most of the patients with no new lesions, but you also have some patients, despite an overall decrease of volume, which show new lesions.
And similarly also on the other side, where you have patients with increasing volume and you don't have new lesions in all patients. So we have not 100% consistency between those two categories, and therefore we thought it makes sense to look into them separately. In the end, and that was also already shown, these RECIP criteria were defined, as I have introduced, the decrease in volume and the increase in volume finally now defines the four different categories.
I also want to briefly introduce another system, the so-called PPP system, the PSMA PET Progression system. This is a system which is currently much more eminence-based than other systems. It's a proposal in literature which is intended to follow the rules for a bone scan. In one part, because it was defined that the appearance of two or more new lesions on the PSMA PET defines progression. It becomes a little bit more complicated if you only have one new lesion because then you can also take clinical data, laboratory data into account, that if you have an increase in known lesions, then also you can add clinical and laboratory data to define progression.
What is the next step now after we have these different standardized reporting criteria? I think the next step is that we look whether they are associated with prognosis. There are a lot of data out there already. I think most of them are of retrospective nature. I think we have to enhance the data. I just use the remaining time to introduce two different works on this.
There is one using the PPP criteria, which looked at patients before and after docetaxel chemotherapy. And it was clearly shown that patients who were non-progressing on PPP criteria did significantly better than those who did not. And also, there are first reports out that stages defined by PROMISE, miT and N staging is associated with prognosis, which is not unexpected. Similar to the association of prognosis in this scenario, where biochemical recurrence-free survival after radical prostatectomy for primary disease is associated with histological TNM staging. And in this setting, it was tested against the miT and the miN stages, and clear differences in the patients' cohorts could be found.
How can I summarize? I think we have different systems for staging and lesion assessment. We have the PROMISE, miTNM system, which is a system that facilitates a very well-known system and structures the report for clinical needs. I think it allows easy communication with clinicians as this code line gives a rough impression of disease extent in the body. We have the PSMA-RADS system, which is a system, in my opinion, very useful for outlining the certainty of findings and also making the mind on pitfalls. I think it's more useful on the lesion-based than on patient-based, as also presented in the system.
We have systems for response, which are the PPP system, which in my opinion is very eminence-based. It has not been established on a representative patient cohort. I think this needs to be done to explore the use of the system further. It follows the PCWG3 criteria, which discriminates progressive from non-progressive. And then we have the RECIP system which introduces the concept of partial response for molecular imaging. It was developed with real-world data from radioligand therapy but needs to be extended also to other systemic treatments, and this is work that is currently underway. It sounds a little bit difficult that you have to assess volumetric changes, which is difficult in clinical routine. However, there was also a recent publication which showed that visual assessment after training in plus-30% versus minus-20% volume change is also very reliable.
What comes next then? I think what comes next or what is already out there partially is electronic and automatic tools. There is one nice software, the Pylarify AI software, which automatically can estimate the miTNM stage and also quantify tumor burden. There is software which you can use for easy establishment of the miTNM Codes. Hopefully, the PCWG4 criteria will integrate PSMA PET, and I'm hearing something in that direction, which is very good for our community, I would say. There are initial data on correlation with clinical outcome available, but I think we now have to use these new systems and look at their prognostic value. And regarding this, I would like to mention Wolfgang Fendler's initiative of a PROMISE registry on a global base. I think everybody can contribute. I think Wolfgang is more than happy to inform people who are interested, really testing the value of the miTNM staging from PROMISE in a large cohort for the outcome of prostate cancer patients in different disease stages.
Thank you very much.
Matthias Eiber: Thanks, Jeremie, for the introduction at the beginning of the session. And Johannes, Tom, and Jeremie, thanks for inviting me to this wonderful meeting here. And now, as we have heard when we should use PSMA PET for what reason, my task is a little bit to talk about standardized reporting, so what we do as imaging specialists to tell our clinical colleagues what we have found, in what way, and what implication that also could have using different systems.
Here are my conflicts of interest, and then we start right away. Why do we want to have standardized reporting? There are two big scenarios for what this can be used. One is for clinical reporting. I think, as imaging specialists, we have to make sure that we always speak the same language. Therefore, we facilitate the exchange of information between centers and also disciplines so that our clinical colleagues understand us and that also our imaging colleagues understand us. And therefore, we should provide concise information and especially clear answers to clinical questions. I think this was a topic of several talks here.
Also, having standardized reporting, I think, also increases repeatability and trust in imaging reporting. We have talked about pitfalls a lot during today. And also, our main goal in the end is that we support the clinicians in tailoring their therapy and assessing prognosis. However, we have an interest in research, and therefore standardized reporting is also important. The fact that we want to validate findings, we have seen a lot of prospective studies where, for example, histological validation was done and therefore we have to standardize report our findings so that histological or confirmatory imaging for validation becomes straightforward. We should also use this so that we can pool our data and finally do this for multicenter meta-analyses. And the frameworks I will introduce are also thought for using them in prospective trials and maybe adding what PSMA can do furthermore to clinical endpoints.
Basically, what I would like to do in my talk is give an overview of the different frameworks which now exist in the literature. Jeremie has also already introduced that there is the so-called PROMISE framework, the prostate cancer molecule imaging standardized reporting. This is a system that basically describes the extent of the disease on TNM level. I will come to this a little bit more in detail later. Then we have the PSMA-RADS system, which is a system similar to the RADS system for other organs that we assess lesions or the nature of a lesion on a five-point scale.
And then we have two systems where we can look at a response after systemic treatments, so response in metastatic disease. And there we have the PPP system, which follows the path of prostate cancer disease assessment, that it discriminates progressive disease from non-progression. This is what, for example, PCWG3 criteria are doing. And then we have the framework which already was introduced a little bit by Wolfgang, the RECIP framework, which also introduces the concept that we can use molecular imaging techniques to assess response. So not only progression versus non-progression but also add the category of response. Let me go through the different systems. So the PROMISE system was in fact born here in California. I think that was like 2016, 2017. And a lot of the people who have been building this up are sitting in this room. The system was updated this year or the end of last year. And the system is basically a system that follows the pathological TNM staging. And I think all our clinicians know what TNM staging means, and therefore it is constructed to follow this path.
So how is PROMISE designed? We have suspected lesions, and then we group the suspected lesions, whether they are local lesions or whether they are regional lymphatic lesions or distant metastases. We have different levels on which we assess all the lesions. We have a certainty aspect so that we make our minds on how certain we are that this is a prostate cancer lesion or whether it is something else. I refer regarding this to the very excellent pitfall talk.
We have already heard from Louise about her PRIMARY score, and the PRIMARY score is also part of this certainty assessment, if the prostate is still in place. And this is something, Louise, I have to disagree with you a little bit because we have implemented this. And I think also with your agreement that we also use this, I know there are obstacles, I know there are challenges, but also that we agreed for biochemical recurrence after radiation therapy, so where the prostate is still in place.
In addition to that, if the prostate is removed, we use the PSA expression score. This is a five-point score Wolfgang has already shown. I will show it later. Then we grade the disease on its extent. So we have the categories of local recurrence and of T0-T4. You can find all the details for this in the respective publication. And then we do something similar for N and M staging, that we use a five-point scale for assessing the certainty that this is a malignant lesion. And then we have different categories that very closely follow the pathological categories to group the disease.
For exploratory reasons, we could add the volume of disease more for scientific aspects. But in the end, with what we come out, and I think this is essential because this is something which our clinicians like because it's a very concise and a very short description of the dissemination of the disease, that is a so-called miTNM Code. And there we code our findings based on the system, here, for example, miT2, N2, M1a, M1b, and all our clinicians have an impression of what that means if they just read this single line. And then we also add the PSA expression score, both for certainty reasons that we give our clinicians the impression of what kind of uptake our lesions have, as well as in the case of radioligand therapy, that we provide the lowest and the highest score with regards to eligibility.
I don't want to go into details of all the different TNM stages which are described, so I already say that this can be taken from the publication but also the publication is... To the publication's so-called pocket card is added, which can be really used during routine reporting to enhance and orientate on this system.
I said that the PSMA expression score is part of it, it's part of it regarding the certainty of lesions, of lesion assessment in benign versus malignant disease. We completely agree, I think, together that PSMA uptake in PET is not the only discriminator for a disease or not. We have to look at anatomical imaging. We have to have pitfalls in mind, as outlined today in the morning.
Wolfgang has already told us that there are different categories for the PSMA expression score and that if we use it for radioligand therapy, we usually select patients with lesions higher than liver, which fall into the category of PSMA expression score two and higher. And this is also the same category, which usually, if we don't take into account pitfalls or other confounders, which are usually the threshold for discriminating between unspecific and cancer-related disease, if you look at these pocket cards.
What I want to show you here is how we now, since several years, structure our report in Munich. We have created this table which we add to the end of all of our reports and we really, we go by different TNM categories. It's not that we provide our clinician a very precise report in the conclusion, but we really very strongly go with this system that we assess local disease, whether it's present or not, that we assess local regional lymph node disease, whether it's present or not. And that we group for M1a, which are distant lymph node metastases, for bone metastases, and visceral metastases. And we provide the clinicians in this way. And then at the end of the conclusion, we also provide the aforementioned miTNM Code line and we get very positive feedback because, as everybody, or as our clinicians are short on time, and usually sometimes, or usually, they only read this line here and then they have an idea about the distribution of the disease in the respective patient.
I say that I want to cover different systems. I would shortly like to introduce the PSMA-RADS system, the Prostate-Specific Membrane Antigen Reporting and Data System. Also, a new version has been published last year. This system very much parallels other RADS systems like the TI-RADS, the LI-RADS, the PI-RADS, or whatever RADS. It is not so much intended to structure the extent of the disease but more to assess the nature of a lesion.
And therefore, in my opinion, it's really a very good dictionary of typical pitfalls and how to approach them. Because especially the category of PSMA-RADS-3, which is equivocal lesions, where depending on the kind of the lesion further work or follow-up is recommended. This is given very nicely in detail how this lesion should be managed. And I think for this, the PSMA-RADS system is really a good addition to the literature. Also, for clinical practice, especially if somebody is starting with PSMA PET reporting.
I can go over this system a little bit quicker because it was already introduced. I'm now going to the systems for assessing response to metastatic disease. I start with the RECIP system, the Response Evaluation Criteria in PSMA PET. I would like to inform you a little bit about the history of this classification. The idea was, and that was also the clinical impression, and I'm sure a lot of you have a similar impression of the patients, the tumor response is related to changes in tumor burden and also the occurrence of new lesions.
So really, the intent was to account for these two different aspects, and therefore the hypothesis for this system was that partial response in tumor volume without new lesions probably has a superior overall survival than partial response and new lesions. And similarly, that progression in tumor and new lesions is probably worse than progression in tumor volume, but no new lesion. And that is how basically this RECIP classification was organized. And regarding the use of molecular imaging, quantitative molecular imaging, the idea was to include also the category of partial remission or response into a prostate cancer response assessment framework. What basically was done is from three different centers, patients were pooled who underwent lutetium PSMA treatment and had a PET prior to the first treatment and after the second treatment. Different cutoffs for volumes were tested, both for volume increase as well as for volume decrease. In the end, a cutoff of 30% decrease in the volume of PSMA-expressing disease correlated best with overall survival, similar to a 20% increase of volume, as you can, for example, see here in these Kaplan-Meier curves.
And then also the hypothesis of additional new lesions was tested. And you can also see that there was a significant difference in patients with and without new lesions. And that volume changes and the appearance of new lesions are not always 100% aligned. You can see from this graph here, where you can see on the y-axis the change in the volume on PSMA PET. And you see that here, in patients with decreasing volume, you have most of the patients with no new lesions, but you also have some patients, despite an overall decrease of volume, which show new lesions.
And similarly also on the other side, where you have patients with increasing volume and you don't have new lesions in all patients. So we have not 100% consistency between those two categories, and therefore we thought it makes sense to look into them separately. In the end, and that was also already shown, these RECIP criteria were defined, as I have introduced, the decrease in volume and the increase in volume finally now defines the four different categories.
I also want to briefly introduce another system, the so-called PPP system, the PSMA PET Progression system. This is a system which is currently much more eminence-based than other systems. It's a proposal in literature which is intended to follow the rules for a bone scan. In one part, because it was defined that the appearance of two or more new lesions on the PSMA PET defines progression. It becomes a little bit more complicated if you only have one new lesion because then you can also take clinical data, laboratory data into account, that if you have an increase in known lesions, then also you can add clinical and laboratory data to define progression.
What is the next step now after we have these different standardized reporting criteria? I think the next step is that we look whether they are associated with prognosis. There are a lot of data out there already. I think most of them are of retrospective nature. I think we have to enhance the data. I just use the remaining time to introduce two different works on this.
There is one using the PPP criteria, which looked at patients before and after docetaxel chemotherapy. And it was clearly shown that patients who were non-progressing on PPP criteria did significantly better than those who did not. And also, there are first reports out that stages defined by PROMISE, miT and N staging is associated with prognosis, which is not unexpected. Similar to the association of prognosis in this scenario, where biochemical recurrence-free survival after radical prostatectomy for primary disease is associated with histological TNM staging. And in this setting, it was tested against the miT and the miN stages, and clear differences in the patients' cohorts could be found.
How can I summarize? I think we have different systems for staging and lesion assessment. We have the PROMISE, miTNM system, which is a system that facilitates a very well-known system and structures the report for clinical needs. I think it allows easy communication with clinicians as this code line gives a rough impression of disease extent in the body. We have the PSMA-RADS system, which is a system, in my opinion, very useful for outlining the certainty of findings and also making the mind on pitfalls. I think it's more useful on the lesion-based than on patient-based, as also presented in the system.
We have systems for response, which are the PPP system, which in my opinion is very eminence-based. It has not been established on a representative patient cohort. I think this needs to be done to explore the use of the system further. It follows the PCWG3 criteria, which discriminates progressive from non-progressive. And then we have the RECIP system which introduces the concept of partial response for molecular imaging. It was developed with real-world data from radioligand therapy but needs to be extended also to other systemic treatments, and this is work that is currently underway. It sounds a little bit difficult that you have to assess volumetric changes, which is difficult in clinical routine. However, there was also a recent publication which showed that visual assessment after training in plus-30% versus minus-20% volume change is also very reliable.
What comes next then? I think what comes next or what is already out there partially is electronic and automatic tools. There is one nice software, the Pylarify AI software, which automatically can estimate the miTNM stage and also quantify tumor burden. There is software which you can use for easy establishment of the miTNM Codes. Hopefully, the PCWG4 criteria will integrate PSMA PET, and I'm hearing something in that direction, which is very good for our community, I would say. There are initial data on correlation with clinical outcome available, but I think we now have to use these new systems and look at their prognostic value. And regarding this, I would like to mention Wolfgang Fendler's initiative of a PROMISE registry on a global base. I think everybody can contribute. I think Wolfgang is more than happy to inform people who are interested, really testing the value of the miTNM staging from PROMISE in a large cohort for the outcome of prostate cancer patients in different disease stages.
Thank you very much.