ESMO Congress 2026

As context for this readout, the PI3K/AKT target class rests on long-established science: Akt/PKB is a central kinase node downstream of growth factors and other stimuli, and its aberrant activation is directly linked to the pathophysiology of diseases including cancer ^[3]; full activation requires Thr308 phosphorylation by PDK1 and Ser473 phosphorylation by the rictor-mTOR complex ^[1]. PTEN is the tumor suppressor that opposes Akt activation, and in tumors that have lost PTEN expression, the rictor-mTOR complex has been proposed as a potential drug target ^[1]. Akt executes its roles in cell survival, proliferation and metabolism by phosphorylating substrates such as BAD ^[4], FKHRL1 ^[2] and GSK3 ^{[5] [3]}, and mTORC2 in turn regulates Akt ^[6]. Whether PTEN biallelic loss can be reliably detected therefore determines whether patients of this pathway can be identified at all.

At #ESMO26, Niven Mehra presented the PROMPT study asking a different question: can plasma ctDNA replace tissue genotyping in mCRPC? With 301 matched tissue/plasma pairs, plasma alone found 82% of BRCA/HRR-altered patients and picked up 4 additional cases that tissue missed (another on-site post described the increment as adding 7% yield to tissue testing), so a plasma-first strategy with reflex to tissue at low tumor fraction is defensible for PARPi eligibility. PI3K/AKT is another story: plasma caught only 38% of alterations in this pathway, with the misses concentrated in PTEN biallelic losses.

Relative to prior knowledge, this readout advances two concrete points. First, it splits the answer to 'can plasma replace tissue' by pathway: although mechanistically PTEN loss releases the brake on Akt and makes rictor-mTOR a potential target ^[1], PROMPT shows plasma detection of PI3K/AKT alterations (38%) falls far below that of HRR (82%), so the plasma strategy cannot simply be extended to this pathway. Second, it stratifies prognostic information: PTEN loss seen in plasma carried an HR of 2.11 after adjustment for ctDNA fraction, whereas tissue-only loss was not prognostic, suggesting that PTEN loss visible in plasma itself carries a stronger prognostic signal. The operational conclusion is equally clear — a low tumor fraction plasma result should be reported as 'not informative', not 'negative'.

The limitations should also be noted: all figures above come from on-site conference posts; the threshold for low tumor fraction, the assay platform, and details beyond the 301 matched pairs and the HR of 2.11 were not disclosed, and formal results await confirmation.