Development and validation of time-to-event models to predict metastatic recurrence of localized cutaneous melanoma
(2023) In Journal of the American Academy of Dermatology 90(2). p.288-298- Abstract
Background: The recent expansion of immunotherapy for stage IIB/IIC melanoma highlights a growing clinical need to identify patients at high risk of metastatic recurrence and, therefore, most likely to benefit from this therapeutic modality. Objective: To develop time-to-event risk prediction models for melanoma metastatic recurrence. Methods: Patients diagnosed with stage I/II primary cutaneous melanoma between 2000 and 2020 at Mass General Brigham and Dana-Farber Cancer Institute were included. Melanoma recurrence date and type were determined by chart review. Thirty clinicopathologic factors were extracted from electronic health records. Three types of time-to-event machine-learning models were evaluated internally and externally in... (More)
Background: The recent expansion of immunotherapy for stage IIB/IIC melanoma highlights a growing clinical need to identify patients at high risk of metastatic recurrence and, therefore, most likely to benefit from this therapeutic modality. Objective: To develop time-to-event risk prediction models for melanoma metastatic recurrence. Methods: Patients diagnosed with stage I/II primary cutaneous melanoma between 2000 and 2020 at Mass General Brigham and Dana-Farber Cancer Institute were included. Melanoma recurrence date and type were determined by chart review. Thirty clinicopathologic factors were extracted from electronic health records. Three types of time-to-event machine-learning models were evaluated internally and externally in the distant versus locoregional/nonrecurrence prediction. Results: This study included 954 melanomas (155 distant, 163 locoregional, and 636 1:2 matched nonrecurrences). Distant recurrences were associated with worse survival compared to locoregional/nonrecurrences (HR: 6.21, P < .001) and to locoregional recurrences only (HR: 5.79, P < .001). The Gradient Boosting Survival model achieved the best performance (concordance index: 0.816; time-dependent AUC: 0.842; Brier score: 0.103) in the external validation. Limitations: Retrospective nature and cohort from one geography. Conclusions: These results suggest that time-to-event machine-learning models can reliably predict the metastatic recurrence from localized melanoma and help identify high-risk patients who are most likely to benefit from immunotherapy.
(Less)
- author
- organization
- publishing date
- 2023
- type
- Contribution to journal
- publication status
- in press
- subject
- keywords
- clinicopathologic factors, locoregional recurrence, metastatic recurrence, stage I/II melanoma, time-to-event prediction
- in
- Journal of the American Academy of Dermatology
- volume
- 90
- issue
- 2
- pages
- 11 pages
- publisher
- Elsevier
- external identifiers
-
- pmid:37797836
- scopus:85176236181
- ISSN
- 0190-9622
- DOI
- 10.1016/j.jaad.2023.08.105
- language
- English
- LU publication?
- yes
- id
- 9d9c136a-c48e-40cc-84d2-5a3cdc637224
- date added to LUP
- 2024-01-12 11:39:33
- date last changed
- 2024-04-27 07:03:43
@article{9d9c136a-c48e-40cc-84d2-5a3cdc637224,
abstract = {{<p>Background: The recent expansion of immunotherapy for stage IIB/IIC melanoma highlights a growing clinical need to identify patients at high risk of metastatic recurrence and, therefore, most likely to benefit from this therapeutic modality. Objective: To develop time-to-event risk prediction models for melanoma metastatic recurrence. Methods: Patients diagnosed with stage I/II primary cutaneous melanoma between 2000 and 2020 at Mass General Brigham and Dana-Farber Cancer Institute were included. Melanoma recurrence date and type were determined by chart review. Thirty clinicopathologic factors were extracted from electronic health records. Three types of time-to-event machine-learning models were evaluated internally and externally in the distant versus locoregional/nonrecurrence prediction. Results: This study included 954 melanomas (155 distant, 163 locoregional, and 636 1:2 matched nonrecurrences). Distant recurrences were associated with worse survival compared to locoregional/nonrecurrences (HR: 6.21, P < .001) and to locoregional recurrences only (HR: 5.79, P < .001). The Gradient Boosting Survival model achieved the best performance (concordance index: 0.816; time-dependent AUC: 0.842; Brier score: 0.103) in the external validation. Limitations: Retrospective nature and cohort from one geography. Conclusions: These results suggest that time-to-event machine-learning models can reliably predict the metastatic recurrence from localized melanoma and help identify high-risk patients who are most likely to benefit from immunotherapy.</p>}},
author = {{Wan, Guihong and Leung, Bonnie W. and DeSimone, Mia S. and Nguyen, Nga and Rajeh, Ahmad and Collier, Michael R. and Rashdan, Hannah and Roster, Katie and Zhou, Xu and Moseley, Cameron B. and Nirmal, Ajit J. and Pelletier, Roxanne J. and Maliga, Zoltan and Marko-Varga, Gyorgy and Németh, István Balázs and Tsao, Hensin and Asgari, Maryam M. and Gusev, Alexander and Stagner, Anna M. and Lian, Christine G. and Hurlbert, Marc S. and Liu, Feng and Yu, Kun Hsing and Sorger, Peter K. and Semenov, Yevgeniy R.}},
issn = {{0190-9622}},
keywords = {{clinicopathologic factors; locoregional recurrence; metastatic recurrence; stage I/II melanoma; time-to-event prediction}},
language = {{eng}},
number = {{2}},
pages = {{288--298}},
publisher = {{Elsevier}},
series = {{Journal of the American Academy of Dermatology}},
title = {{Development and validation of time-to-event models to predict metastatic recurrence of localized cutaneous melanoma}},
url = {{http://dx.doi.org/10.1016/j.jaad.2023.08.105}},
doi = {{10.1016/j.jaad.2023.08.105}},
volume = {{90}},
year = {{2023}},
}