Intra-Treatment Tumor Apparent Diffusion Coefficient, a Quantitative Imaging Metric, is Associated with Neck Nodal Recurrence in De-Escalated Treatment of HPV-Positive Oropharyngeal Cancer (OPC)

Abstract

<h3>Purpose/Objective(s)</h3> De-escalation strategies for the treatment of HPV-associated OPC show promise for effective, less toxic treatment. However, identifying accurate markers of more aggressive subsets of HPV OPC remains challenging. The 30 Reduction in Radiation for OPC (ROC) trial prospectively used intra-treatment ¹⁸F-FMISO imaging to selectively de-escalate patients to 30Gy chemoradiotherapy, however, a subset of patients still developed nodal recurrence. We hypothesized that diffusion-weighted (DW) MRI-derived quantitative imaging metrics, such as apparent diffusion coefficient (ADC), a surrogate marker of tumor cellularity based on water diffusion patterns, might help identify patients at high risk of recurrence and enable broader use of de-escalation strategies<b>.</b> <h3>Materials/Methods</h3> The 30 ROC trial enrolled patients with p16-positive OPC, cT0-2, N1-2c (AJCC 7^th ed). All patients had pre-treatment ¹⁸F-FMISO (fluoromisonidazole) PET scan and those with baseline nodal hypoxia underwent a 2^nd intra-treatment scan. All patients underwent primary site surgery where negative margin was not required. All patients had gross neck nodal disease. Those with intra-treatment resolution of nodal hypoxia or without initial hypoxia were de-escalated to 30Gy (non-hypoxic group) with 2 cycles of concurrent chemotherapy, while those with persistent radiographic hypoxia (hypoxic group) were treated to 70Gy. Standard T1, T2 weighted, and multiple b-value DW-MRI data were acquired pre-treatment and weekly during the first 4 weeks of treatment. Two radiation oncologists delineated nodal tumor regions of interest (ROIs) in a blinded fashion with ITK-SNAP on DW images (b= 0 s/mm²). Multiple b-value DW data were fitted to a monoexponential model to calculate ADC (mm²/s). The data analysis was performed on ROIs and voxel-wise using the in-house developed tool MRI-QAMPER. Standard statistical analysis was performed using MRI and clinical response data based on RECISTv1.1 provided by a radiologist. <h3>Results</h3> 158 patients were enrolled, of whom 95 had pre-treatment DW-MRI data available with tumor ROI with ADC values. Based on overall ¹⁸F-FMISO imaging, 12.6% (12/95) exhibited persistent hypoxia and 87.4% (83/95) had resolved/no hypoxia. All patients with nodal recurrences were in the de-escalated non-hypoxic group (8/83). Pre-treatment ADC was not different between overall hypoxic and non-hypoxic groups. Among de-escalated patients, those who developed nodal recurrence exhibited significantly different intra-treatment ADC at week 3 (before completing 30Gy) relative to those who did not recur (P<0.05, Wilcoxon rank-sum test). <h3>Conclusion</h3> The quantitative imaging metric, ADC, reflects both the restricted and hindered Brownian motion of water molecules in tumor tissue. Thus, ADC may be a useful intra-treatment marker to identify patients at risk for nodal recurrence to benefit from the incorporation of an upfront planned neck dissection after 30Gy as part of the de-escalation strategy.