1094P - NeoRisk: Neoadjuvant immunotherapy (NeoIT) recurrence risk assessment tool

Background

NeoIT with anti-PD-1 (PD1) is now a standard of care for patients (pts) with resectable stage IIIB–D melanoma. Although pathological response is predictive of recurrence, this variable alone cannot accurately identify those pts who will recur, particularly non-responders. We sought to build a recurrence risk assessment tool based on pts demographics, disease characteristics, pathological and imaging data.

Methods

Pts with resectable stage IIIB–D melanoma treated with PD1-based neoIT were included. Pts demographics, disease characteristics, blood parameters, pathological and imaging data at baseline (BL) and post-treatment (post-Tx), and clinical outcomes were analysed. A penalised multivariable logistic regression model was built to predict recurrence and a tool (NeoRisk) predicting the likelihood of recurrence for individual patients was generated.

Results

164 pts who underwent surgery post PD1-based NeoIT and had at least 6 months of follow-up were included; training cohort (n=114; 16 recurrences, 14%) and validation cohort (n=50; 7 recurrences, 14%). After the analysis of >50 variables, the combination of 4 variables accurately predicted recurrence (training cohort, AUC=0.91; validation cohort, AUC=0.93): 1) % of viable tumour cells in the tumour bed post-Tx (AOR=2.55, p=0.0016); 2) density of tumour-infiltrating lymphocytes (TILs) post-Tx (absence, mild, moderate, marked; e.g. marked vs absence, OR=0.39, p=0.0115); 3) difference in the % of fibrosis from BL to post-Tx (OR=0.69, p=0.0531); and 4) % tumour change from BL by RECIST (OR=2.06, p=0.0043). NeoRisk can be used to predict recurrence for individual patients; for example, Patient A with 55% of viable tumour cells and mild TILs post-Tx, 0% and 2% of fibrosis at BL and post-Tx, respectively, and 13% increase of tumour size from BL, has a risk of recurrence of 82%; while Patient B, with the same % of viable tumour cells, but moderate TILs post-Tx, 0% and 35% of fibrosis at BL and post-Tx, respectively, and 12% decrease of tumour size from BL, has a lower risk of recurrence of 20%.

Conclusions

NeoRisk can accurately predict recurrence after NeoIT, which may help tailor adjuvant treatment and radiological surveillance interval based on the predicted risk of recurrence for individual patients.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

Melanoma Institute Australia.

Funding

Has not received any funding.

Disclosure

I. Pires da Silva: Financial Interests, Personal, Invited Speaker: BMS, MSD, Roche, Novartis; Financial Interests, Personal, Other, Travel Support: BMS, Roche; Financial Interests, Personal, Advisory Board: MSD. R. Rawson: Financial Interests, Personal, Invited Speaker: MSD. M.S. Carlino: Financial Interests, Personal, Advisory Board, Consultant Advisor: MSD, BMS, Novartis, Amgen, Oncosec, Merck, Sanofi, Ideaya, Pierre Fabre, Eisai, Nektar, Regeneron. A.M. Menzies: Financial Interests, Personal, Advisory Board, advisory board: BMS, MSD, Novartis, Roche, Pierre Fabre, QBiotics. R.A. Scolyer: Financial Interests, Personal, Advisory Role: SkylineDx BV, IO Biotech ApS, MetaOptima Technology Inc, F. Hoffmann-La Roche Ltd, Evaxion, Provectus Biopharmaceuticals Australia, Qbiotics, Novartis, Merck Sharp & Dohme, NeraCare, AMGEN Inc, Bristol Myers Squibb, Myriad Genetics, GSK. G.V. Long: Financial Interests, Personal, Other, Consultant Advisor: Agenus Inc, Amgen Inc, Array Biopharma Inc, AstraZeneca UK Limited, Bayer Healthcare Pharmaceuticals, BioNTech SE, Boehringer Ingelheim Ingelheim International GmbH, Bristol Myers Squibb, Evaxion Biotech A/S, Hexal AG, Highlight Therapeutics S.L, IOBiotech, Immunocore Ireland Limited, Innovent Bioilogics USA Inc, Merck Sharp & Dohme, Novartis Pharma AG, PHMR Limited, Pierre Fabre, Regeneron Pharmaceuticals Inc, Scancell Limited, SkylineDX B.V; Non-Financial Interests, Principal Investigator, GL is PI on over 30 clinical trials: GL is PI on over 30 clinical trials. All other authors have declared no conflicts of interest.