Academy Exchange

 View Only

  • 1.  Chat GPT second option

    Posted 07-08-2025 11:43

    I have a patient concerned about two consecutive viral load of 60, who brought me recommendations from ChatGPT.   I'm interested in what others think:

    HIV positive 6/2004, initial regimen AZT/3TC/Efav

    10/04 Abacavir added when VL when from 50-1500, VL again undetectable

    vircoType with "virtual phenotype"

    NRTI: 41L, 210W, 211K, 215C

    NNRTI: 189I

    PI: 30N, 33F, 63P, 77I, 88D, 93L

    Switched to TAF/FTC/Efav due to anemia

    GenoSure Archive 10/16

    NRTI: M41L, L210W, N348I,

    NNRTI: N348I

    INI: none

    PI: D30N, L33F, I62V, N88D

    10/20 switched to Juluca due to "cardiac issues" and concerns

    GenoSure Archive 10/20

    NRTI: M41L, L201W, N348I

    NNRTI: Q207E, N348I

    INI: none

    PI: D20N, L33F, I62V, N88D, 

    His VL on Juluca remained undetectable 10/20 onwards until 4/25 (60) and 6/25 (60)

    I have not ordered another archive genotype test.

    Patient now very worried about a possible failing regimen, and transmission to HIV negative partner.  He has never missed a dose.  He always takes it with a meal, and does not use any medications for GERD.

    He has used the paid version of Chat GPT and another AI tool (not sure which) and has input his treatment history as well as all the mutations from the one genotype and two archive genotypes, including 43 different minor mutations and polymorphisms.

    Chat GPT gave me it's treatment recommendations:

    #1: Switch to dolutegravir/TAF/FTC.

    #2: Biktarvy

    #3: DTG + boosted DRV + optional TAF/FTC

    #4: DTG + maraviroc + boosted DRV

    Salvage Therapy "given the patient's extensive treatment and resistance history it is prudent to consider potential salvage therapy now"

    #1: DTG + Boosted DRV

    #2: DTG + DRV + maraviroc

    #3: Lenacapavir + OBT

    #4: CAB/Rilpiv

    My thoughts: Chat CPT seems very comfortable with the archive genotype testing (more comfortable than I am), and not very concerned with all those TAMS from an actual genotype.  It also seems convinced that this is a failing regimen, although I'm not as convinced.

    What do others think?



    ------------------------------
    Ben Stearn
    Dupont Circle Physicians Group
    Washington DC
    ------------------------------


  • 2.  RE: Chat GPT second option

    Posted 07-08-2025 15:33
    Edited by Joel Gallant 07-08-2025 18:39

    Hi Ben!

    I do tend to trust positive results from archive testing.  There was a study from Italy showing that archive genotype results were predictive of virologic failure. However, according to his past genotype results, his virus should be fully susceptible to the DTG and RPV in Juluca. I also don't find his two recent viral loads concerning. There's no evidence that he's failing therapy, and switching regimens might not solve the "problem."

    The first two regimens ChatGPT mentions are reasonable just because they're strong, high-barrier, guidelines-recommended 3-drug regimens. With the 41L and 210W mutations, he has at least low-level tenofovir resistance, though that's probably not a big deal in combination with BIC or DTG. After that, there may be a little AI "hallucination" going on. MVC (without even knowing his tropism)? "Salvage therapy" in someone doing find on a 2-drug regimen? Interestingly, it mentions CAB/RPV as an option, which is sort of a long-acting version of what he's already on, but with a lower barrier to resistance.

    For me, the options would be to leave him alone--if he's willing--or to switch BIC/FTC/TAF, which I'd pick over DTG + FTC/TAF mainly because it's a single-tablet regimen. Either way, I'd assure him that low-level viremia like his is common and not evidence of virologic failure, no matter what ChatGPT says!



    ------------------------------
    Joel Gallant, MD, MPH
    Johns Hopkins University
    Baltimore, MD

    AXCES Research Group
    Santa Fe, NM
    ------------------------------

    Original Message


Related Content