ESMO 2021 | ESMO highlights: methylation-based MCED and the predictive role of ctDNA in uveal melanoma

Hi, my name is Christian Rolfo. I’m the Associate Director for Clinical Research at the Center of Thoracic Oncology at Mount Sinai in New York. It’s my pleasure to discuss today about my participation in ESMO 2021. In this important congress, I had the opportunity to discuss two interesting abstracts. The first one was the early detection using ctDNA whole genome methylation in a pan-cancer approach. And the second one was the use of ctDNA in the scenario of immunotherapeutic drugs for uveal melanoma, using also the comparison between the ctDNA and the RECIST criteria for the assessment of the overall survival in patients.

The first abstract, that is the whole genome methylation, is part of an important trajectory study that was performed in discovery, training and validation. The training and validation were already published in Annals of Oncology in 2020 and 2021. And we have here, in ESMO 2021, the opportunity to know why whole genome methylation was preferred over other multiomics approach, including allelic fraction, and also fragmentomics and SNP.

This important study was including a high number of patients, more than 2800 patients with an inclusion of solid tumors and hematological disease. And also we saw that we were including most patients with early stage and advanced stage. In this study, the sensitivity for whole genome sequencing was around 34% with a high specificity that was 98%. And looking for the next step that we’re doing in the validation, this sensitivity actually was increasing, actually it was on a higher specificity; 99.5%, sensitivity in general population was around 55%. So, that means that the whole genome sequencing was performing better than other technologies, even when we are putting these other technologies of the ctDNA all together.

Some of the questions that we have here obviously is the differences that we saw in different tumor types and also in the different stages. In some of the stages, in for example, in lung cancer, the performance for early stage is very low, and we are analyzing the general population was around 16% for stage one. So, that means that we are losing some patients in this early detection.

Another thing that is important to recognize of this study is the site of cancer origin was able to be detected in high performance. Actually, in the validation was higher; 88.7%, 75% in the discovery cohort, and also the opportunity to have a low limit of detection. That was also important because they were working with allelic fraction variation and this technology also have the opportunity to… Doesn’t require, we would say, the clonal hematopoiesis [inaudible]. So we don’t need to remove that. And this is also improving the technology for a low limit of detection. So, this technology was able to identify patients in some stages of some of the tumors but specifically, we need still some answer for different isotypes and we need also answers for patients in early stage.

This technology was not on this discovery, and then the validations were not including high risk populations for different tumor types, so we don’t know what is the role of this technology in screening. For example, in the case of lung cancer, the association with low dose CT scan for our screening to improve this and have more patients saved.

The other study, the second study, was another study presented by the colleagues from Memorial Sloan Kettering and, in this case, the use of very specific analysis in uveal melanoma using ctDNA and it seems that the ctDNA was able to perform better than the RECIST criteria for the outcomes of the immunotherapy in this disease. So, there is a very nice correlation between the overall survival, even though the drug was not giving an important response rate. The overall survival compared with historically was importantly increased. And when we see the correlation between the ctDNA and the overall survival; it’s very linear.

Obviously, then we need to have in consideration that the comparator here was RECIST criteria coming from CT scans. My question for them was, do we need really to have a RECIST criteria as a comparator, or we need to have other models? For example, immunoPET for seeing also the metabolic activity compared with the biological activity that is detected by ctDNA. In the future we would see that a little bit more complemented. And also for example, other techniques like proteomics could be interesting to include here. What is true, that this is an interesting exercise in a very specific disease, it’s a unique approach for overall survival. So, that’s what’s really important in this trial.