Wednesday , December 8 2021

AI may predict a response to blockade control in patients with metastatic melanoma



[ad_1]

A computational method that combines clinical-demographic variables with in-depth study of pre-treatment histology imaging may predict a response to blockade of immune control in patients with advanced melanoma. Cancer Clinical Research, Journal of Cancer Research for the American Association.

“Although immunological control inhibitors have profoundly altered the melanoma treatment landscape, many tumors do not respond to treatment and many patients have treatment-related toxicity,” said the study’s author, Iman Osman, a physician in the Department of Medicine, Dermatology and Medicine. (Oncology) at New York University (NYU) Grossman School of Medicine and Director of the Interdisciplinary Melanoma Program at NYU Langone’s Perlmutter Cancer Center. “The unmet need is the ability to accurately predict which tumors will respond to therapy. This would allow for personalized treatment strategies that maximize the potential for clinical benefit and minimize exposure to unnecessary toxicities.”

“Some recent attempts to predict immunotherapy responses are made with strong accuracy but with technology, such as RNA sequencing, they are not easily generalized to the clinical environment,” said Dr. Aristotelis Tsirigos. Member of the Computational Medicine NYU Grossman School of Medicine and NYU Langone’s Perlmutter Cancer Center. “Our approach shows that responses can be predicted using standard care clinical information, such as pre-treatment histology images and other clinical variables.”

The researchers used data from a training cohort of 121 patients with metastatic melanoma who received point-to-point block treatment between 2004 and 2018. All patients were treated with first-line CTLA-4 therapy, PD-1 therapy, or a combination. both clinical outcomes and progression of disease or response were recorded, including full or partial responses (patients with stable disease were excluded for this study of evidence for the principle). The researchers used computer algorithms called deep combinatorial neural networks (DCCNs) to study digital images of melanoma metastatic tumors and to identify patterns associated with treatment response. Through this approach, they developed a classifying response aimed at predicting whether an untreated tumor in the patient would respond to immune control blockade or post-treatment advances. This DCCN response was validated in an independent cohort of 30 patients with metastatic melanoma treated at the Vanderbilt-Ingram Cancer Center between 2010 and 2017.

The performance of the DCCN classifier response was assessed by calculating the area under the curve (AUC), a measure of the accuracy of the model, where a score of 1 corresponds to the perfect prediction. The DCCN prediction model achieved an AUC of about 0.7 in the training and validation cohorts.

To increase the accuracy of the model prediction, the researchers performed multivariate logistic regressions by combining the DCCN prediction with conventional clinical features. The latest model included the DCCN forecast, the performance status and treatment regimen (CTLA-4 monotherapy, PD-1 monotherapy, or combination therapy) of the Eastern Cooperative Oncology Group (ECOG). In the training and validation cohorts, the multivariate classifier achieved an AUC of about 0.8. In the validation cohort, the classifier can stratify patients with high and low risk of disease progression, with significantly different outcomes for progression-free survival between the two groups.

While the majority of patients in the training cohort received monotherapy against CTLA-4 (approximately 64% of patients), the majority of patients in the validation cohort received anti-PD-1 agents (approximately 53% of patients). The results suggested that some predictive models are not determined for the purpose of point control, Osman noted. Class activation maps, which can identify regions within the digital imagery used by neural networks to create predictions, suggested that cell nuclei were important for DCCN predictions, where larger and more abundant nuclei were related to disease progression. “These results suggest that ploidy may be one of the biological determinants perceived by the DCCN,” he added.

“It is possible to use computer algorithms to analyze histology images and predict treatment response, but more work needs to be done using larger sets of training and analysis data, along with additional validation parameters, to determine if a clinical-achieving algorithm can be developed. “Tsirigos said.

Among the limitations of the study is the relatively small number of images used to work on the computer algorithm, with 302 images in the training cohort and 40 images in the validation cohort. “There are data that suggest that thousands of images are needed to train models that achieve clinical-level performance,” Tsirigos said.


Research has shown how interferon-gamma is targeted against cancer immunotherapy


More information:
Cancer Clinical Research (2020). DOI: 10.1158 / 1078-0432.CCR-20-2415

Granted by the American Association for Cancer Research

Mention: AI can predict control immune response in patients with metastatic melanoma (2020, November 18) Retrieved November 18, 2020 https://medicalxpress.com/news/2020-11-ai-response-immune-checkpoint-blockade.html

This document is subject to copyright. In addition to any exceptions intended for private examination or investigation, no part may be reproduced without written permission. Content is for informational purposes only.



[ad_2]
Source link