The DeepTCR system has shown its effectiveness as a predictive clinical tool and has provided insights into the biological mechanisms underlying patient response to immunotherapy.
A machine learning algorithm predicted which melanoma patients would respond better to immunotherapy and which would not, according to a study by researchers at Johns Hopkins Kimmel Cancer Center and its Bloomberg Kimmel Institute for Cancer Immunotherapy.
The DeepTCR system has shown its effectiveness as a predictive clinical tool, but has also provided insight into the biological mechanisms underlying patient response to immunotherapy.
“The predictive power of DeepTCR is exciting, but what I found most fascinating was that we were able to see what the model learned about the immune system’s response to immunotherapy,” said John- William Sidhom, MD, PhD, first author of the study, in a press release. “We can now leverage this information to develop more robust models, and possibly better treatment approaches, for many diseases, even those outside of oncology.”
DeepTCR uses deep learning to recognize patterns in large volumes of data from amino acid sequences of proteins called T cell receptors (TCRs). These sit outside the T cells of the immune system waiting to engage a protein from an enemy, such as a bacteria or virus.
Current immunotherapy drugs, or checkpoint inhibitors, involve proteins that scramble this ability in tumors, causing the T-cell response to cancer; however, these drugs have proven useful for a limited number of patients, according to the investigators.
The current study used materials collected during the CheckMate 038 clinical trial, which evaluated the effectiveness of an immunotherapy drug (nivolumab) versus a combination of 2 (nivolumab and ipilimumab) for 43 patients with inoperable melanoma.
Tumor biopsies were taken before and during treatment. In the study, no significant difference was seen in patients given the single drug versus the 2-drug combination. Some patients in both groups responded, others did not.
Investigators used high-tech genetic sequencing to assess the collection of TCRs surrounding each tumor by determining the type and number of TCRs in each biopsy. This data was entered into the DeepTCR program along with what datasets belonged to responders versus non-responders, and then the algorithm looked for patterns.
The researchers assessed pretreatment differences between the immunotherapy TCR collection in responders and nonresponders. The identified differences were as predictive of patient response as known biomarkers, according to the study. The researchers said these results need to be confirmed in a larger patient population before the algorithm can be used to guide treatment.
“Precision immunotherapy based on the tumor immune microenvironment is essential to guide the optimal choice of treatment options for each patient,” said Drew Pardoll, MD, PhD, professor of oncology and director of the Bloomberg Kimmel Institute for Cancer Immunotherapy, in a “These findings from DeepTCR set a new dimension for predicting tumor response to immune checkpoint blockade by applying a novel artificial intelligence strategy to deconvolute the vast array of receptors expressed by T cells. infiltrating the tumour, the main immune components responsible for the direct destruction of tumor cells.
The researchers also sought to determine the differences between responders and non-responders using data from another study that linked specific TCRs to the enemy proteins that activated them. They found that those who responded to immunotherapy had higher numbers of virus-specific T cells in their tumors, and non-responders had more tumor-specific T cells.
The team learned that non-responders had higher T-cell turnover.
“Responders and non-responders had approximately the same number of tumor-specific T cells before and during treatment,” Sidhom said in a press release. “The identity of these T cells remained the same in the responders, but in the non-responders there was a different variety of T cells before and during treatment. Our hypothesis is that the non-responders had a high number tumor-specific T cells that were ineffective from the start.When immunotherapy started, their immune system sent out a new batch of T cells, trying to find an effective one, but the dysfunction remained.In contrast, the responders had effective T-cells early on, but their anti-tumor activity was blocked by the tumor.When immunotherapy started, it lifted the blockade and allowed them to do their job.
REFERENCE
Machine learning can help predict patient response to cancer immunotherapy. Johns Hopkins Medicine. November 16, 2022. Accessed November 17, 2022. https://www.hopkinsmedicine.org/news/newsroom/news-releases/machine-learning-can-help-predict-patient-response-to-cancer-immunotherapy
#Machine #learning #helps #predict #patients #respond #immunotherapy