How do you reconcile discrepancies in clinical prostate cancer staging with AJCC and NCCN?
In light of the recently published APCCC 2024 paper, would you classify a radiographic T3 patient (by mpMRI) but cT2 disease (DRE) as having clinical T3 disease?
Join Mednetto vote and see how they answered.
Fundamentally, I use NCCN risk categories to help steer conversations about staging and treatment options for very low vs low vs fav int vs unfav int vs high risk diseases. Therefore, I use NCCN staging in my clinical practice and notes and incorporate mpMRI into staging. I find it comforting that NRG-GU009 allows T3-T4 disease as seen on MRI independent of DRE in criteria to define high risk patients eligible to enroll, and I've happily enrolled patients onto GU009 with just T3 disease via MRI. To call a patient high risk, I like for EPE on MRI to be pretty obvious and not a radiologist being "suspicious" for EPE or just observing capsular bulge with a small amount of associated furriness around the capsule.
There's another staging system - STAR-CAP - which is what I most commonly use when helping patients understand prognosis. It was published by Dess et al. with @Dr. First Last as last author in 2020 (PMID 33090219) and has undergone external validation a couple of times. There was a website they made available to assist with staging, though it appears nonfunctional right now: https://www.star-cap.org/. This website was nice and included pictorial representations of expected prostate mortality - hope it's back up soon.
I don’t. I use the Gleason score, PSA, T stage, PSMA PET when appropriate, MSKCC and Briganti nomograms, and medical comorbidities to guide recommendations.
This is a common question that, unfortunately, there is not a super simple answer if getting to the core of it.
What is the goal of AJCC? Risk stratification with standardized risk groups. It in no way is meant to guide treatment itself. Unfortunately, the AJCC 8th edition was made solely by 'expert opinion', which is why a large group of us created STAR-CAP as the first and only validated prognostic system that meets all of the AJCC criteria of a staging system.
- AJCC tends to not include the latest molecular or imaging features as they are not always ubiquitously used (hence why it was not until recently that breast cancer added all the molecular info in to breast cancer staging despite it being around for decades).
What is the goal of NCCN? To help 'guide' treatment/management decisions. It was not even created to be a risk stratification system (was based on D'Amico which was not a trained model but was expert opinion, and then adapted to NCCN risk groups, which were further subdivided to aid in treatment decision-making).
- NCCN tends to try to incorporate well-established prognostic/imaging/molecular features if based on high level evidence. Hence why there are no clear callouts of what you do with SVI based on MRI or PSMA PET given few of our trials that form our guidelines routinely used these modalities.
What do we end up with? A need for a nuanced understanding of how to integrate newer staging methods, AJCC, and NCCN guidelines.
Lets take T3a status:
- Trials in the 1990s and early 2000s defined this by DRE
- ECE detected pathologically generally has a more favorable prognosis than if detected by MRI which generally has a more favorable prognosis than if felt on rectal exam. This is because pT3a includes even minor microscopic disease, MRI includes only visible radiographic ECE, and DRE detects fairly extensive ECE. Hence, when you look at ProtecT in mostly low risk men, those who had surgery there was a sizable number with pT3a disease, however outcomes were great.
- So if using MRI to call T3a status, these men will have a prognosis more favorable on average than historical T3a by DRE.
Lets take M1 status:
- Very few trials have routinely staged patients with PSMA PET (POPRT is one of the few exceptions to this).
- M0 by CT/bone scan is more aggressive than M0 by PSMA PET/CT. Just look how phenomenal outcomes are on POP-RT with just RT+nodal RT plus 2 years ADT <10% of men had a BCR by 5 years. This is incredible. Even if some of these men met 2 of the 3 STAMPEDE high risk criteria (i.e., high grade, T3, PSA >40), would be crazy to add Abi to these patients given in STAMPEDE just RT+ADT (no Abi they had >15% death from prostate cancer at 5 years compared to 1% on POP-RT).
So end of the day, similar to prognostic tools like Decipher or ArteraAI, imaging is also highly prognostic and must carefully interpret contemporary stage and grade and how it applies to NCCN guidelines. Treatment intensification should be done carefully if upstaging patients based on more sensitive/accurate tools.
Hope this helps. Clearly not simple.
Best,
Dan
I generally do not use MRI for clinical staging, though with exceptions for very discordant and aggressive findings. Mostly I view MRI findings like Decipher -- a useful tool in risk stratification without actually changing risk groups. For example, consider a patient w/ PSA 11, DRE negative, Gleason 6 in 2 targeted cores from MR-identified lesions on contralateral sides of the prostate… would you consider that person cT2c? And then in conjunction with PSA >10, they now have 2 intermediate risk factors and are unfavorable intermediate… will you counsel them that the standard of care with EBRT is to include short term ADT? I think most would not, hence why I use AJCC staging, but to be fair NRG GU010 would enroll this person as unfavorable intermediate as Dr. @Dr. First Last mentioned the NRG allows MRI for T staging.
MRI can be useful for selecting patients for or against prostatectomy; if reasonable likelihood of EPE then RP may be less desirable given higher chance of biochemical failure. But for that same patient considering definitive RT, I think you have a much higher threshold to actually call them cT3a and, assuming no other high risk factors, now subject them to long term ADT, whereas without MRI they would get short term or none. These arguments all assume you follow the NCCN guidelines for treatment of various risk groups. I have used MRI to define T staging in rare cases, e.g extensive primary tumor invading nearby organs, consider them cT4 now very high risk for which one may add abiraterone to ADT. I would consider someone with A LOT of EPE to be cT3a and treat as such. But usually patients with very bad MRI findings will have Gleason scores to match, so I don’t see a lot of aggressive tumors where MRI has large treatment implications. But I do think applying MRI to otherwise low and favorable intermediate risk patients will have considerable stage migration and you'll be giving a lot more ADT than is probably necessary.
There are two somewhat separate issues at play - staging and assessment of the cancer for the purpose of individual patient recommendation. Staging allows us to roughly organize patients into groups for a variety of purposes including making treatment recommendations but also for research and administrative needs. Staging systems are inherently slower than any emerging research that refines them. So, clinicians will always need to incorporate newer information to refine a recommendation beyond the simple use of stage. So, I formally stage using AJCC and NCCN and then add to the newer input like those highlighted above to individualize a patient recommendation.
My first thought upon reading this was, "This is an excellent example of the importance of clinical judgment!"
Similar to armed conflict, where the weapons are often initially way ahead of the tactics, the rapid and continuing evolution of advanced imaging and molecular analysis is moving beyond our contemporary staging systems.
So, what do I do?
As others have already pointed out, you have to paint the best picture that you can with the data available, while simultaneously respecting the limitations of all this technology and realizing that it is ultimately up to you as the treating MD to make the call. The radiology report on the MR may say, "suggestive of EC extension" but what do you see?
I'll start by using the available and permitted data to assign the patient as closely as possible to the appropriate AJCC stage/NCCN Risk Group and go from there.
DRE negative, but MRI and PSMA strongly suggest T3B disease? If I believe the imaging findings, I'll treat as if Stage III with elective pelvic nodal RT, and recommend long-term ADT [which half of the guys reject, but at least have the conversation].
Gleason 3+3 in 7/12 regions but MRI is negative for PR3-5 lesion and Decipher is c/w with low-risk of progression? I'd offer treatment due to the number of positive cores [with AS of course, discussed as an option] but forego ADT. And so on.
On the surface, it is nice to have a cookbook, but it is also limiting, half the fun of this job is using your brain to put all of the pieces together and come up with a reasonable patient-specific treatment plan that respects all the clinical evidence.
The understanding of some of the limitations of DRE, and its inter-rater reliability, is evolving. It may be appropriate to update this discussion in light of the recently published 2024 Advanced Prostate Cancer Consensus Conference (APCCC) paper, where 96% of panelists voted in favor of using mpMRI to reclassify patients as cT3 if there was clear evidence of extracapsular extension despite a cT2 physical exam (strong consensus). Note: this was felt to be applicable for patients with ISUP Grade Group ≥3 disease (Gleason 4+3 or higher). Allowing mpMRI to "overrule" DRE for staging is consistent with my clinical practice as well, if there is definitive evidence of extracapsular extension (not just "possible" ECE).
While NCCN may allow for T3a to be defined by MRI, traditional risk-stratification (such as NCCN risk groups) did not, and the same story with historical recurrence risk nomograms, such as the MSKCC pre-prostatectomy nomogram, which is critical in my ability to counsel unresected patients who are deciding between RALP and RT on the risk of recurrence after RALP.
If I started inputting T3 for MRI defined disease (not on DRE), the risks of recurrence after surgery would appear much worse than what the patient would be at risk for, which is not fair to the patient or my urology colleagues for me to over-estimate the patient's risk of BCR after surgery.
Similarly, calling someone T3a because they have EPE on MRI (and not on DRE) will likely lead to over-treatment compared to calling them T1c when treating with radiation +/- ADT. We have no evidence that T3a on MRI requires LT-ADT as we would otherwise recommend for HR disease receiving definitive RT.