Abstract 805P
Background
Keynote 716 has recently shown improved recurrence-free survival (RFS) for sentinel node-negative Stage IIB/C melanoma patients given adjuvant pembrolizumab. Accordingly, identification of patients most likely to benefit from adjuvant systemic therapy requires accurate, personalised risk prediction. The current AJCC 8th Edition (AJCC-8) estimates melanoma-specific survival (MSS). We developed and externally validated a tool for predicting RFS in stage II melanoma patients.
Methods
Development cohort clinicopathological data were extracted from the Melanoma Institute Australia (MIA) database for patients diagnosed with stage II melanoma (n=3243). Survival prediction models were developed using multivariable Cox regression analyses and externally validated using datasets from the Dutch national cancer registry (PALGA) (n=8631) and the MD Anderson Cancer Center (n=703). Predictive performance of the models was assessed using C-statistics, decision curve analysis and calibration plots.
Results
The development and validation C-statistics for the MIA prediction tool (Table) demonstrated good calibration and large, statistically-significant discrimination gains over the AJCC-8 stage groups, ranging from 8.3% to 12.2%. The MIA tool showed good performance using external datasets and was thus validated. Table: 805P
| RFS | C-statistics (95%CI) | |||||
| Australian model | AJCC-8 | |||||
| Development | Validation | Development | Validation | |||
| Dutch | US | Dutch | US | |||
| 5-Year | 68.8% (66.8–70.8) | 71.8% (70.6-73.0) | 69.4% (65.2-73.6) | 60.5% (58.6-62.5) | 64.3% (63.2-65.5) | 61.9% (57.8-66.0) |
| 10-Year | 71.7% (69.3-74.0) | 74.0% (72.5-75.6) | 69.7% (64.3-75.2) | 59.5% (57.1-61.9) | 62.3% (60.7-64.0) | 61.3% (55.9-66.7) |
Conclusions
The MIA model for RFS demonstrated both good discrimination and calibration to predict survival for individual stage II melanoma patients at both 5 and 10 years, which held true on external validation using independent datasets from two different continents. This robust model provides clinicians with important data to guide recommendations for adjuvant immunotherapy, based on individual patients’ risk-benefit ratios. An online tool will be available at www.melanomarisk.org.au.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Melanoma Institute Australia.
Disclosure
A.H.R. Varey: Financial Interests, Personal, Invited Speaker: Novartis. J.E. Gershenwald: Financial Interests, Personal, Advisory Role: Merck; Financial Interests, Personal, Invited Speaker: Novartis, Syndax, Bristol Myers Squibb, Regeneron. G.V. Long: Financial Interests, Personal, Other, Consultant Advisor: Agenus Inc, Amgen Inc, Array Biopharma Inc, Boehringer Ingelheim International GmbH, Bristol Myers Squibb, Evaxion Biotech A/S, Hexal AG (Sandoz Company), Merck Sharpe & Dohme (Australia) Pty Limited, Novartis Pharma AG, OncoSec Medical Australia, Pierre Fabre, Provectus Australia, Qbiotics Group Limited, Regeneron Pharmaceuticals Inc; Financial Interests, Personal, Advisory Board, Consultant Advisor: Highlight Therapeutics S.L. R.A. Scolyer: Financial Interests, Personal, Advisory Role: F. Hoffmann-La Roche Ltd, Evaxion, Provectus, Biopharmaceuticals Australia, Qbiotics, Merck Sharp & Dohme, AMGEN Inc, Bristol Myers Squibb, Myriad Genetics, GlaxoSmithKline. J.F. Thompson: Financial Interests, Personal, Advisory Role: Bristol Myers Squibb, Merck-Shape-Dohme, Glaxo Smith Klein, Provectus. All other authors have declared no conflicts of interest.