Unmet need
“We currently treat stage 3 non-small cell lung cancer (NSCLC) with chemoradiation followed by durvalumab consolidation. However, we don’t know who will benefit most from this regimen, and there are no robust biomarkers available at this time. All patients with inoperable stage 3 non-small cell lung cancer get the same treatment: chemoradiation, followed by a year-long course of durvalumab immunotherapy. Durvalumab is an anti PD-1 agent... this is the standard of care for everybody, regardless. There is nothing really that has been shown to predict who will do well with this and who will not do well with this.
Of course, there can be toxicity associated with immunotherapy. So you are talking about a year long course of maintenance therapy with associated risks and we do not know who is going to respond to it and who does not. So, in an attempt to try to avoid harm -really, that is what it is- can we see who will do well with the durvalumab and who is going to recur and probably needs another alternative approach to treatment? This was the whole point of this was to really identify who would do well with durvalumab maintenance therapy and who would not, basically to predict response to durvalumab.
We asked ourselves whether the radiomics data collected from CT scans of these patients might assist us in generating an algorithm to identify those patients who do really well for this regimen, but also, importantly, seek alternative strategies for patients who will not respond to this particular therapeutic approach.”
The study
“We took 118 patients with stage 3 NSCLC treated at our center who were treated with chemoradiation and durvalumab between July 2017 - July 2019, and for whom we had good imaging data available. We randomly split them into 2 groups, a training set (n=59) and a test set (n=59). The images were evaluated by the Center of Computational Imaging and Personalized Diagnostics (CCIPD) at Case Western University, who performed analyses to distinguish between images of patients who recurred and those who did not. That formed the basis of algorithm consisting of radiomic features capturing both intra- and peritumoral texture patterns, essentially radiomic features that were more present in the images of patients who had recurred. And then based on the extent of expression of these radiomic features associated with recurrence, they gave patients either a high risk score or a low risk score based on the median overall score. Those who expressed more radiomic features were identified as having a high radiomics risk score and those who expressed fewer radiomic features, were given a lower radiomics risk score.
The primary endpoint of this study was progression-free survival (PFS), and the secondary objective was difference in PFS within high PD-L1 and low PD-L1 groups, using tumor PD-L1 expression of 50% as a cut-off point. We saw that the radiomic risk score showed that people who had high radiomics risk scores and people who had low radiomics risk scores had a statistically significant difference in progression-free survival, so our primary endpoint was met. The radiomics risk-score and PD-L1 expression were found to be significantly associated with PFS in both the training (HR 2.3, 95% CI: [1.46 – 3.63], P = 0.0003; PD-L1: HR 0.31, 95% CI 0.081 – 0.96, P=0.038 and test sets (HR 2.56, 95% CI 1.75 – 4, P= 8.7e-05; PD-L1 HR 0.27, 95% CI 0.048 – 0.58, P=0.005).”
What is the next step?
‘’We have been communicating very closely with our colleagues at the CCIPD lab at Case Western University about how to best proceed with this. We are happy with the results. What we are trying to do now is acquire more images and validate this on larger and larger subsets of patients to further validate this tool. In addition, at this point all we have is progression-free survival results. One clear step forward we are waiting for would be evaluating the differences in overall survival. Furthermore, we are trying to accrue more patients. We would definitely like to get more than just 118 patients and see this work on larger patient sets. If retrospectively it continues to show the same predictive abilities and the same predictive results where we can actually predict outcomes, both progression-free survival and ideally overall survival, based on the radiomics risk score, then the next step would be to validate it prospectively and see if moving forward, clinicians who still have not treated these patients can predict who will respond to therapy and who will not.
And then, basically ask: how do we use this to make decisions?”
Can this radiomics risk score predict safety events from durvalumab?
“That is a good question. Potentially. That is not something that we have looked into just yet. We have recorded patients who developed toxicity to the durvalumab. We have not actually looked into that, but we will probably need more patients.”
What were the limitations of the study?
“Limitations include the fact that it was single center, retrospective, cohort study with a relatively small sample size. Although the sample size was sufficient to show a statistical significance, we would definitely like more patients and we are trying to work on that. Other limitations of the study, at least with what we have presented at ASCO, is that we only had progression-free survival outcomes at that point.
We are working on getting the overall survival outcomes as well. To really be able to say that this is worth using clinically, we would need to validate it prospectively.”
Copyright ©2021 Medicom Medical Publishers
Posted on
site created by:
© 2024 Medicom Medical Publishers. All rights reserved. Terms and Conditions | Privacy Policy