Read the Full Video Transcript

Martin Gleave: Thank you for the opportunity to speak. And I was assigned this topic, which I think we are all challenged in terms of defining, even though we all kind of know what it is when we see it. I think risk stratification is a key part of optimizing outcomes in high-risk prostate cancer, and we've been trying to estimate this risk with a number of different approaches over the past decades: the Partin Tables, D'Amico criteria, various nomograms, as well as consensus definitions like NCCN. And I'll get into many of these over the next 10 minutes or so.

And our focus here is not so much on the broad spectrum of low and intermediate risk, but really how are we stretching out and better defining the high-risk spectrum of localized prostate cancer? And this matters, as in this review by Derya Tilki, there are roughly seven definitions for high-risk prostate cancer. And when you look at BCR-free survival or metastasis-free survival, these are stretched out across a pretty broad risk category. And it ranges from, at least if you look at the endpoint that matters, metastasis-free survival—it varies from 77% if you just look at PSA greater than 20, to about 56% if you just look at high-grade Gleason nine or 10.

So again, there's room for improvement in terms of better categorizing risk, which is important when you're talking about intensifying, de-intensifying treatment, or for clinical trial criteria. And I'll just highlight this one paper from Hopkins, a retrospective review of 750 patients: NCCN high-risk localized prostate cancer defined by Gleason score eight to 10, PSA greater than 20, or clinical stage T3. And they defined a very high-risk group as those with primary pattern five, greater than five cores of Gleason score eight to 10. Or again, I think most importantly here, the accumulation of multiple high-risk factors tend to then increase the risk of metastasis or cancer-specific mortality several-fold.

When you look at defining criteria for entrance into clinical trials and how we've struggled with this, CALGB is a classic example of this where 750 men with high-risk disease were intended to enroll in this study, did enroll in this study. Initially, the risk criteria were determined by the Kattan nomogram predicting a probability of freedom from biochemical recurrence of less than 60%. But they found that accrual was hampered by this. It was too stringent a definition, and the protocol was later amended to enroll men with Gleason score eight to 10 and less than a 60% probability by a Kattan nomogram. So again, defining risk also has to include the proportion of patients who may be ineligible for that trial.

Another evolving area in defining risk is estimating volume. Clearly, risk is determined by tumor volume, which is stage as well as number of cores by grade and also by PSA. And I think we're all familiar with the variability here. You can either be a T3a, Gleason score four plus four, and multiple cores, and a PSA of 40 is going to be worse than a similarly staged T3a, and a Gleason score four plus four, and only two cores with the same PSA. And so CAPRA has set out to better define risk on the basis of percent of positive cores. So if you're greater than 33%, integrating this into their definition of risk better predicted prostate cancer-specific mortality, independent of treatment modality.

Similarly, NCCN high-risk versus very high-risk was improved by recording the plus or minus less than or greater than 50% involvement by cores. And I think when you look at guidelines, nomograms for risk, NCCN unfavorable intermediate risk involves one or more of two or three intermediate risk factors, which are T2b, Gleason grade two or three, and a PSA of 10 to 20, or grade group three or greater than 50% positive cores, no very high-risk features and only one high-risk feature of either T3, grade group four or five, PSA greater than 20. And then you've got the very high-risk group, which includes T3b, grade group five, or greater than one risk factor.

With STAMPEDE, which is more geared towards a perhaps radiotherapy-targeted group, two or more risk factors of T3-4, PSA greater than 40, or Gleason eight to 10. The only issue I have with these definitions is I think it's very difficult to quantify and differentiate a T2b versus a T3a cancer. It's very dependent on interpretation and subject to bias along the way.

So I think, again, inaccuracy in distinguishing T2 versus T3, and we have to, I think, evolve its use and include more volume of positive cores. Nomograms have been developed, and I list a few here in predicting lymph node metastasis, and these I think will have to be modified in the future based upon the role of PSMA PET in this stage as well.

I think that we're all familiar with the challenges. I think this is a multifocal, multiclonal disease, which underpins intra-patient spatial heterogeneity. We have imperfect imaging and biopsy with undersampling that limits accuracy of prognostic subgrouping. These are all the things that we're struggling with moving forward. T stage is highly subjective. Grade-defined high risk can be misleading as to whether or not you're high-volume four plus four or low-volume four plus four versus a high-volume four plus three, or a high-volume three plus four with a higher PSA.

How do we weigh systematic versus MRI-targeted cores in determining risks? This is an evolving area because of patterns of practice regarding whether or not you do targeted plus or minus systematic biopsies. But I think in general, the most important factors are the presence of pattern five and volume of pattern four, and increasingly the use of PSMA PET to better estimate risk of regional versus non-regional metastasis.

There are a number of trials that are now integrating risks for inclusion criteria that are contemporary, PROTEUS and GUNS, all enriched for either high volume, high intraprostatic tumor volume and grade, and ultimately risk of metastatic recurrence. And in these trials, they define high-risk localized prostate cancer as either Gleason score greater than or equal to four plus three, and at least one of the following: any combination of four plus three and/or four plus four in greater than six systematic cores; any combination of Gleason score four plus three and/or eight in greater than three systematic cores with a high PSA; any Gleason score nine in at least one systematic core; or at least two systematic or targeted cores with Gleason score eight.

All attempting to actually incorporate high intraprostatic volume along with grade as an estimate of risk and metastasis. ASCO guidelines for tissue-based biomarkers—obviously, genomics is helping to stratify risk, and Dan is going to talk about this in the future. But I think currently, these are very useful in differentiating risk at the low and intermediate risk level, rather than further categorizing risk in the high-risk groups.

When we look at treatment intensification, I just give this one example from STAMPEDE, a phase three randomized trial in the M0 subgroup, very high-risk group randomized to receive radiation and ADT plus or minus ABI over 1,900 patients with high-risk "non-metastatic disease," although they could be N+, and at least two of either T3 or T4, Gleason eight to 10, or PSA above 40, or relapsing within less than 12 months of ADT or greater than 12 months post-treatment, and a PSA greater than four. And their intensification by adding ABI significantly reduced metastasis-free survival and overall survival. Again, identifying a subgroup that was very high risk that benefited from intensification with ARPI doublets.

So again, just ending, I think that clearly defining high risk is a spectrum, and we can extend this risk into low and intermediate risk, but in a high-risk group, it extends from either low-tier to very high risk, or even now regionally very high-risk disease with the use of PSMA PET. And this can be modified in a concluding slide where we can think about high-risk disease in a modified NCCN format where, again, this has to be pragmatic, useful in the clinic, and we can think about breaking down high risk into four tiers.

Tier one has no very high-risk features like current low-tier high risk in NCCN, and only one high-risk feature of less than three cores of grade group four, greater than five cores of grade group three, and either a T3a or a PSA greater than 20. Tier two includes greater than four.

Audience Member: That's one minute to go, right?

Martin Gleave: Okay, thank you.

There's something that's come up, thank you. Or a T2 risk group greater than four cores of grade group four, or less than three cores of grade group five, and two of the tier one risk factors and so on and so forth. So you can build these different tier groups based upon risk ending with a kind of a very high-risk tier four, which is either a high-grade cancer and extra-pelvic or M1a PSMA-detected metastasis and/or greater than two tier-one risk factors.

I think that increasingly N status and M status is going to be PSMA PET-derived, which leads to a selection or a stage migration that we have to incorporate, and we can think about incorporating things like histologic variants into this definition as well. So thank you for your attention.