Click here to


Are you sure ?

Yes, do it No, cancel

Role of Mid-Treatment Imaging Biomarkers in Phase II: Adaptive De-Escalation of Radiation Therapy Dose in HPV-Positive Oropharyngeal Carcinoma (ART)

P Galavis*, M Tam , S Kim , E Zan , W Wang , K Hu , NYU Langone Health, New York, NY


(Monday, 7/15/2019) 1:15 PM - 1:45 PM

Room: Exhibit Hall | Forum 2

Purpose: Concurrent chemotherapy with radiotherapy is the standard of care for locally advanced oropharyngeal cancer patients. However, the main drawback of this approach is the high toxicities experienced by the patients. This has motivated new clinical trials that investigate the role of imaging biomarkers in dose de-escalation to mitigate the side effects of treatments.

Methods: Ten patients from an institutional phase II clinical trial were CE-CT (Contrast-Enhanced-CT) simulated prior to starting radiotherapy treatment and at week-four as part of the protocol. A radiation oncologist manually contoured the GTVn (primary nodal disease) on both scans. Based on GTVn volume variation (≥40%) patients were eligible/ineligible for dose de-escalation. CE-CT scans and contours were transfer to IBEX for texture-feature calculation. The relative net change for 77 texture-features was calculated. The Pearson correlation coefficient (r) was used to correlate the volume change with the feature changes. Texture-features that presented an r >0.5 are possible candidates for treatment assessment. Significance level was evaluated using the t-test (p<0.05)

Results: Eight patients met criteria for mid-treatment nodal response and were de-escalated. For the two patients who proceeded with standard treatment, shape texture-features variation were low, ranging [-6%-14%] when compared to de-escalated patients range [-10%-60%]. Across all the patients two shape features (surface area and surface area density) showed high correlation with node-tumor volume changes, with r-values of 0.81 (p<0.05) and -0.66 (p<0.05). Histogram-based-like-skewness showed a medium correlation with r-value of 0.52 (p>0.05), whereas dissimilarity feature from the Gray-Level-Occurrence-Matrix showed correlation of 0.63 (p<0.05).

Conclusion: Features with high Pearson correlation values are potential candidates to be used as additional metrics for treatment assessment. The study includes other imaging modalities (e.g MRI and PET) which will be included as a future work. More analysis will be added to the study as more patients are continually enrolled in the protocol.


Not Applicable / None Entered.


TH- response assessment : CT imaging-based

Contact Email