Prostate NCCN Guideline Changes (Risk Stratification) (2024)

fiji128 said:

I was going through the NCCN guidelines and noticed some pretty substantial changes in version 4.2024. Mainly there is now the whole new section on risk stratification and how this now affects treatment decisions regarding AS, Tx +/- ADT and duration of ADT instead of just the old NCCN risk categories. This seems to be pushing the Decipher test pretty hard to make these decisions. What is everyone's thoughts on this?

Click to expand...

I had been using decipher quite a bit, a lot of payors cover it and sometimes I just need a tie breaker, nice to see an oncotype DX type thing finally make it into the guidelines for PCa

I'll have to check out the update, thanks

fiji128 said:

I was going through the NCCN guidelines and noticed some pretty substantial changes in version 4.2024. Mainly there is now the whole new section on risk stratification and how this now affects treatment decisions regarding AS, Tx +/- ADT and duration of ADT instead of just the old NCCN risk categories. This seems to be pushing the Decipher test pretty hard to make these decisions. What is everyone's thoughts on this?

Click to expand...

See Also

Alypz Sdn Bhd on LinkedIn: Thank you to all the organizations and individuals who joined our…Nebenwirkungen von Chemo- & Strahlentherapie auf Zähne | CURAPROXPervasive Computing in IoT, 2nd EditionMeenakshi M on LinkedIn: RFID, BLE, IoT, and drone Dental equipment and supplies manufacturing…

I was excited that they included ArteraAI. Data is stronger for unfav intermediate risk than with Decipher. More data will come for the other risk groups. It’s easier to run and cheaper than Decipher.

grenz said:

I was excited that they included ArteraAI. Data is stronger for unfav intermediate risk than with Decipher. More data will come for the other risk groups. It’s easier to run and cheaper than Decipher.

Click to expand...

Artera is PREDICTIVE and prognostic which is what you need for decision-making. Very easy to use the portal. Results in 2 weeks.

COI: I am not receiving $$ from Artera

Chartreuse Wombat said:

Artera is PREDICTIVE and prognostic which is what you need for decision-making. Very easy to use the portal. Results in 2 weeks.

COI: I am not receiving $$ from Artera

Click to expand...

Link on how to actually start using it? I have not seen any widespread roll out. Would love to start using.

I was more impressed with ArteraAI than I was Decipher based on that ability to be PREDICTIVE about benefit.

evilbooyaa said:

Link on how to actually start using it? I have not seen any widespread roll out. Would love to start using.

I was more impressed with ArteraAI than I was Decipher based on that ability to be PREDICTIVE about benefit.

Click to expand...

If you contact them rest assured they will reach out to you.

Good luck.

FYI Medicare covers

Chartreuse Wombat said:

Home

Discover the power of Personalized Precision Medicine & AI Cancer Therapy with ArteraAI. Transforming cancer therapy through innovation.

artera.ai

If you contact them rest assured they will reach out to you.

Good luck.

FYI Medicare covers

Click to expand...

In awe of the co-founder. Talk about synergies.

fiji128 said:

I was going through the NCCN guidelines and noticed some pretty substantial changes in version 4.2024. Mainly there is now the whole new section on risk stratification and how this now affects treatment decisions regarding AS, Tx +/- ADT and duration of ADT instead of just the old NCCN risk categories. This seems to be pushing the Decipher test pretty hard to make these decisions. What is everyone's thoughts on this?

Click to expand...

Man. I sense under the table stuff on this Decipher and NCCN thing. But hate the game and not the player I guess.

TheWallnerus said:

Man. I sense under the table stuff on this Decipher and NCCN thing. But hate the game and not the player I guess.

Click to expand...

Decipher makers have done a good job accumulating prognostic information across a variety of data sets. I think the reason they haven’t been able to run a predictive study post-hoc like ArteraAI is the RNA in many RTOG archive samples is degraded. All ArteraAI needs is the slides, which are all pristine condition even for old trials. Boosts their sample size and makes the results more valid

grenz said:

All ArteraAI needs is the slides, which are all pristine condition even for old trials. Boosts their sample size and makes the results more valid

Click to expand...

TheWallnerus said:

Man. I sense under the table stuff on this Decipher and NCCN thing. But hate the game and not the player I guess.

Click to expand...

Chartreuse Wombat said:

Artera is PREDICTIVE and prognostic which is what you need for decision-making. Very easy to use the portal. Results in 2 weeks.

COI: I am not receiving $$ from Artera

Click to expand...

It is going to be very interesting.

As @grenz states above, ArteraAI is not doing anything novel to specimens. There is no molecular assay. It's a data processing tool.

Looking at the seminal papers and the ROC curves, the tool as it stands is a bit better than the NCCN model. It is not killer. Presumably these tools will continue to get better with more data. As an aside, I am curious what the epistemic limit is for stuff like this. It is not really personalized medicine in way that testing for ALK mutation is (or better let a gene fusion related to oncogene addiction). These ROC curves are far from perfect.

But here is the thing...I am dubious that there is any proprietary "special sauce" for running deep learning/multiple neural network type AI on large volume clinical/pathologic data. While I appreciate the small scale of startups for getting stuff off the ground, I am very worried about how patents will be filed and stakes will be claimed regarding "intellectual property". It makes sense that these models are an improvement over traditional algorithms. That these models will be claimed to be "uniquely appropriate" for given clinical scenario is unlikely to be a real thing fairly quickly. I'd hate for us to pay a lot for these tools on a per patient basis.

In principal the NIH could be continuously updating deep learning models for any number of clinical scenarios and making these tools free for practitioners. I would not be surprised if this is feasible fairly quickly.

Of course this is an exciting space for academics. All the young academic prostate bucks are on the papers.

Edit: I did find a CMS schedule for AlteraAI test. Seems to be ~$770. Compared to ADT pricing, it is reasonable. Potential market is large (~40K intermediate risk patients per year). Overhead must be quite small after initial investment.

communitydoc13 said:

I am very worried about how patents will be filed and stakes will be claimed regarding "intellectual property"

Click to expand...

The landscape for neural network patents is messy, and it's changing by the year. Just because someone gets a patent issued doesn't mean that it would be honored in court. It does mean that any would-be infringers have to think carefully about toeing the line, though.

An example I sometimes go back to is Mayo vs Prometheus, a landmark case that made its way to the Supreme Court. Prometheus had a patent that went like, "Take a blood test to measure X, where we want X to be at this level Y, and titrate your drug accordingly." (A Prometheus clinical study had identified an optimal value Y, and that was the nature of the patent.) It appears that Mayo looked at that patent and said, "That's hogwash, you can't patent that" and proceeded to infringe. Courts went back and forth on appeal before the Supreme Court ultimately sided with Mayo; the final verdict proved divisive. I sympathize with the "that's hogwash" viewpoint but also understand that there is a need to provide incentives short of a 20-year monopoly and if I were a benevolent dictator I would have trouble coming up with a general rule.

Neural networks are the same category. If we end up with half a dozen models by competing companies that are all reasonably priced, then society will be better off for it. My opinion is that a fast moving startup deserves some reward, but not a monopoly. Presumably at some point, larger companies will invest in getting new complementary sources of data (like, maybe a prospective data gathering effort that also measures RNA, and the company applies its secret sauce to these RNA samples that it does not share) and that will continue to move the needle forward with even better models.

communitydoc13 said:

I did find a CMS schedule for AlteraAI test. Seems to be ~$770. Compared to ADT pricing, it is reasonable. Potential market is large (~40K intermediate risk patients per year). Overhead must be quite small after initial investment.

Click to expand...

Roughly 2/3 of men with intermediate risk disease DON'T need ADT according to the test. Pretty straightforward that makes economic sense in most cases. Additionally with competition price may fall