Chat GPT Recommendations

[Skippy3]

I recently joined the Chat GPT crowd. I pasted in my profile into Chat GPT and asked a question. My profile is copied below for convenience.

My Profile

I was diagnosed with stage 4 adenocarcinoma, prostate cancer March, 2021. Bone scan and follow up CT scan reported that it’s in my bones but in no soft tissues besides the prostate gland. Prostate biopsy scored three Gleason 8's and two Gleason 9's. Family friend with prostate cancer for 17 years recommended Dr Sartor at Tulane University. I was able to get an appointment and he prescribed ADT therapy, Lupron for 3 months, then Eligard every 6 months thereafter plus 1000 mg Abiraterone daily plus 0.5 mg prednisone daily. So far so good. PSA 0.05 after initial treatment. November 2021, PSMA PET scan shows uptake in clivis bone with SUV value of 5.2 (uh-oh), ribs with SUV value of 1.4, spine with max SUV of 2.4 (not good) and prostate with SUV value of 5.4 (maybe expected). I contacted 4 doctors, two medical oncologists and two radiation oncologists about going after the active spots. They all said I should continue on current drug regimen. Over time, April 2023, PSA moved down to undetectable, then over 10 months, crept up to 0.2. A second PSMA PET scan, July 2024, basically confirmed the places noted on November 2021, PSMA PET scan were still there and worse. Additionally, PC moved to some lymph nodes. Received several courses of radiation, 20 days worth, at Mayo Clinic, Rochester MN, September 2024. A follow up PSMA Pet scan, currently scheduled April of 2025 will illustrate how well the radiation worked.

My Question

Assuming the prostate cancer continues to grow what would be my next best treatment option?

Chat GPT Response

I’m sorry you’re going through this. Given your history and the recent spread of prostate cancer despite initial treatment, your next steps will likely depend on how well the radiation worked, as seen in your April 2025 PSMA PET scan. If the cancer continues to grow, here are some potential next treatment options to discuss with your oncologist:

1. Intensifying Systemic Therapy

Since you’re already on Abiraterone + Prednisone, one next step could be switching or adding another agent:

• Enzalutamide (Xtandi) – Another potent androgen receptor inhibitor that some patients switch to if Abiraterone is no longer keeping cancer controlled.

• Darolutamide (Nubeqa) + Docetaxel – Shown to improve survival in metastatic castration-sensitive prostate cancer.

2. Chemotherapy (If Needed)

If progression continues, chemotherapy may be an option:

• Docetaxel (First-line chemo for metastatic prostate cancer)

• Cabazitaxel (Second-line chemo if Docetaxel stops working)

3. Targeted Radioligand Therapy (If PSMA-Positive)

• Pluvicto (Lutetium-177 PSMA-617) – A PSMA-targeted radiotherapy that delivers radiation directly to cancer cells. If your April PSMA PET scan still shows active disease, this could be a strong next step.

4. Immunotherapy & Clinical Trials

• Pembrolizumab (Keytruda) – If genetic testing shows MSI-high or mismatch repair deficiency, immunotherapy could be an option.

• Clinical Trials – Many trials are testing combination therapies (e.g., adding PARP inhibitors, new androgen receptor blockers, or novel PSMA therapies).

5. PARP Inhibitors (If BRCA or DNA Repair Mutation Present)

• Olaparib (Lynparza) or Rucaparib (Rubraca) – Used if genetic testing shows BRCA1/2 or other DNA repair mutations.

6. Bone-Targeted Therapies (If Bone Metastases Worsen)

• Radium-223 (Xofigo) – A radiopharmaceutical that targets bone metastases to reduce pain and slow progression.

• Denosumab (Xgeva) or Zoledronic Acid (Zometa) – Strengthens bones and reduces fracture risk.

Next Steps:

• Wait for April 2025 PSMA PET scan results – If it shows improvement, your current treatment may still be effective.

• If progression is evident, consider Pluvicto or switching systemic therapy.

• Discuss genetic testing for BRCA, MSI, and other markers to guide targeted therapies.

• Consider a second opinion at a top cancer center (Mayo, MD Anderson, or another high-volume prostate cancer center).

Would you like help finding clinical trials or additional insights into any of these treatments?

My Comments

Frankly I was impressed. I liked the way it listed my options going forward in order of priority. I like that it’s personable. “I’m sorry you are going through this” What am I going to do now? I know of several trials that are looking positive. I’m going to download some data about these that Chat GPT didn’t mention, submit to Chat GPT and let it do the work of summarizing the results and comparing them.

And I’m going to investigate “Cancer Copilot” and see if it provides similar or better info.

Kind of cool. Thought you might be interested.

[StayingOptimistic]

This is a great reply from ChatGPT.

[Skippy3 in reply to StayingOptimistic]

I thought so but what do I know. I went to my PC group meeting last night and one of our long term attendees said it was the best document he’s ever received from that group. We’re all just struggling for the best answer. If Chat GPT can help us get there, so be it

I need to respond to Tall_Allen’s post below. The reference he provided is a long one with Chat GPT highlighted.

[petabyte in reply to Skippy3]

Did you try it with "Reason" selected? This activates the new o3-mini model.

Would be interesting to compare the output.

Also if you turn "memory" on it has more context. You can also edit this context if it picks things to "remember" that are not relevant.

The question (prompt) is very important, optimising the prompt is called "prompt engineering" 😉

It's a great tool but it's not magic, we all need to be cautious using the results.

[ulfhbg in reply to petabyte]

Actually there are studies done as well by science teams doing actual evaluation in different AI systems possibilities to educate, analyze and etc based on studies, trials and litterature and they actually score very well in different areas.

So to conclude they are basically ’useless, idiotic or whatever’ seems somewhat narrow from my perspective

I think you summarize it well Petabyte, it’s a great tool but it’s not magic and you should always look at the results it presents with critical eyes not believing everything. 🤗🙏

And when in doubt, the are actually several independent AI tools free at your disposal to compare with like ChatGPT, Copilot, Perplexity and probably some other that I forgot to mention or don’t know about 👍😉😁

[petabyte in reply to ulfhbg]

These are my "memories" 😉. However it seems like they are not used by o3-mini (yet).

Open AI rushed this model out to respond to the R1 model from deepseek so it is probably lacking some refinements. Even so, I've been impressed with it so far.

Memories are made of this 🙃

[Skippy3 in reply to petabyte]

Trying to understand what you mean. I asked Chat GPT for more info. See below. Not sure about the reason option

Me—Does Chat GPT remember all questions and answers and uses that information in future conversations?

Answer—ChatGPT does not retain memory between conversations. Each time you start a new chat, it does not remember past interactions unless you provide context again. However, in this conversation, I do have access to details you’ve shared so far (such as your medical history and treatment plans) to make discussions more relevant.

Would you like me to remember anything specific for future chats?

Me—I would like you to remember all conversations related to prostate cancer.

Answer—Got it! I’ll keep track of all prostate cancer-related discussions so we can build on past conversations and explore the best options moving forward. Let me know how I can help.

Me—How does one select the “Reason” option and what does it do?

Answer—Are you referring to a specific “Reason” option in a particular software, website, or application? If so, let me know where you’re seeing it, and I can help explain how to select it and what it does.

[petabyte in reply to Skippy3]

You need to turn memory on for it to remember things between sessions. It is under "Personalization", this screenshot is from the web interface. It is called the same thing in the App.

Settings

[petabyte in reply to petabyte]

This is with "Reason" selected. You can also use Search if you want it to do a websearch.

Reason

[Tall_Allen]

It's a mix of some things that are right and some things that are wrong. As I said, it's just a way of climbing Mt.Stupid faster:

prostatecancer.news/2024/12...

[ron_bucher in reply to Tall_Allen]

Great definition of AI. 🙂

[Skippy3 in reply to Tall_Allen]

My first question would be, what exactly is wrong with Chat GPT’s response? And, I take umbrage at the “mountain of stupidity” reference. If I’m climbing any mountain it’s a one way trail that gets narrower and narrower with a cliff at the end.

I consider Chat GPT a tool and not the definitive answer on anything, just like I consider Google and the internet for that matter. I also challenge the statement at the end of the document referenced, that says “The only way to avoid "Mt. Stupid" is to have humility.” Agree. “Most attending doctors at teaching hospitals are experts in their fields.” We’d like to think that. My daughter works as a cardiac ICU nurse in a teaching hospital. Only one of 5 cardiac surgeons is so well respected by the nurses that he doesn’t have a derogatory nickname. “The empowered patient will tap into their expertise.” I’d say “tap into” is the operative word. Doesn’t sound like a full throated endorsement to me. What I tell everyone I meet that has prostate cancer or any other cancer; you have to be your own patient advocate and verify as much as practical what your doctor is saying is best for you.

For example, my first urologist, that was good on the DiVinci machine, in the first year of treatment, never gave me a DRE or asked for a PSA test until it was too late. At our first meeting I showed him my accelerating prostate numbers but he was unconcerned. After my GP’s bloodwork indicated high PSA the urologist eventually got on the stick and performed a biopsy. After the results came in he wanted to schedule a prostatectomy the next week, like he was going to save me. The bone scan soon showed that procedure was unnecessary. I was already metastatic. Later, a local oncologist wanted me to immediately start taking Xgeva even though it was not approved by the FDA for hormone sensitive prostate cancer. Even my world famous prostate cancer doctor, Dr Sartor, changes his mind over the years as the science and experience evolves. That’s what I love about this a guy. He is not dogmatic in his thinking. I do my own “Mt Stupid” research so I can challenge him on things. I think he enjoys the discourse. In any case, it’s my life and I’m going to research what I can so I have some idea what’s to come and if there is something better I can do about it. Stay tuned, I’ll have a good post the later part of April.

[Tall_Allen in reply to Skippy3]

Skippy, since you apparently like research, I'll let you go through your ChatGPT garbage dump to see what's wrong with it. "Mt.Stupid" is a term I picked up from the Internet. I just came across the neologism, "agnorance" to describe the combination of arrogance and ignorance that we get from Internet research as a result of the Dunning-Kruger Effect. Perhaps I will relable it.

Both the patient and the doctor have their own jobs to do under a shared decision-making scenario.. The patient's job is to verbalize what is important to him. No one can do that for him. The doctor's job is to provide expertise, so that the patient can make decisions that are right for him.

There is no way that any patient can know as much as a top oncologist at an NCI teaching hospital. The patient doesn't have time to duplicate years of classroom learning, let alone the practical experience of dealing with patients and the peer-to-peer learning that is gained at conferences and day-to-day interactions. Certainly, get second opinions, but a patient is foolish to put his cursory Internet research ahead of the doctor's expertise. There is nothing wrong with asking questions, and if there is any value to patient's research, it is to learn "Medspeak" so that they both are speaking the same language. A patient asks questions; the doctor answers to the best of his ability -- "I don't know" is a proper answer too.

All of science is provisional and subject to error. A good doctor knows that better than any patient can. That's why patient's often stay agnorant while doctors slowly gain expertise.

[janebob99]

It's a useful piece of a large puzzle. If it suggests things that you follow up with your doctor(s), and then you research and study in depth using Google Scholar, then more power to you.

[petabyte in reply to janebob99]

If you are searching for studies perplexity.ai mentioned by  ulfhbg is a good choice. You can set it in "academic mode" (only on the web UI I think).

It then provides quite concise answers with references and suggested follow up questions (these are weaknesses in chatgpt)

I use it as a better search engine.

Not all of the sources are reputable however, it has provided links to HU for example 🤣.

[dhccpa]

Thanks. Looked like good advice.

[j-o-h-n]

Next time ask Chat GPT to pick the winning Lotto numbers (before that day's drawing) for you. The response I got, was to pick the winning numbers out of a hat.

Good Luck, Good Health and Good Humor.

j-o-h-n

[PTvsPC]

ChatGPT can be an excellent starting point. The version I use provides citations after each assertion. I have tested its answers for topics I already know the answers to for several months now and its accuracy is improving from what I can tell. Still, I look at it just as a starting point, not the end.

Also, medical schools are now incorporating their own AI models into research. These AI tools search medical databases and libraries only - not the general internet.

Clearly, this is only the beginning of AI as a tool.

[PSAed]

I just don't get it, i.e. the use of AI for PCa treatment advice. The response above looks like a list of every treatment known. Its like reading the side effects list of medications, its so extensive it becomes useless for the patient but essential for the manufacturer's lawyers. I admit I am not in a position to understand all those treatment effects on PCa or when and what should be used, to me thats a matter for the experienced human professionals. In my book TA is in that category.

[Skippy3 in reply to PSAed]

Even though my PSA is low it has increased in value implying my cancer is now castrate resistant. As far as I can tell the Chat GPT response, i.e. the priority of treatment, was pretty accurate for my level of cancer. Of course my doctor and I have the option to skip around or to do something altogether different. Today there are even more choices.

[petabyte]

As you may know from my posts above I find AI tools helpful while acknowledging their limitations.

This appeared in the BBC yesterday.

bbc.com/news/articles/c0m17...

[petrig]

In Finland🇫🇮 we have very old saying about fire:"Tuli on hyvä renki mutta huono isäntä". (Fire is a good farmhand, but a bad master”) Same goes with AI.