Benjamin Rosen, PhD
University of Michigan
Head and Neck Cancer
PV QA 2 - Poster Viewing Q&A 2
Purpose/Objective(s): To compare the performance of lymph-node specific and conventional total node FDG-PET metabolic tumor volume (MTV) for failure-free survival (FFS) prediction in low-risk oropharyngeal cancer patients
Materials/Methods: Pretreatment FDG-PET scans of 144 HPV-associated oropharyngeal cancer patients with ≤10 pack-year smoking history and treated with definitive chemoradiation therapy (chemo-RT) to 70 Gy IMRT were retrospectively analyzed. Patients with upfront surgical excision of the primary tumor or planned neck dissection were excluded. The primary tumor and each PET-avid lymph node (totaling 481) were contoured and labeled by a radiation oncologist. Conventional nodal MTV (MTVC) was calculated using a combined lymph node 50% maximum standardized uptake value (SUV50%) threshold. Lymph-node-specific MTV (MTVI) was calculated using SUV50% from each lymph node. Finally, modified nodal MTV (MTVM) was calculated as the sum of all MTVI for each patient. Recurrence patterns including nodal failure location and time to biopsy-proven failure were analyzed. Nodal failure was defined as positive pathology in an unplanned neck dissection. ROC analysis was performed to assess the predictive performance of MTVC and MTVM for patient-level nodal failure within one year as a binary endpoint right censored by follow-up time, distant metastasis, new primary tumor, or death within one year. ROC analysis was also used on MTVI for node-specific failure prediction. Using Kaplan-Meier analysis, Mantel-Haenszel hazard ratios (HRs) were calculated for three evenly distributed quartile cut points to compare relative risks for patient-level and node-specific FFS. Statistical significance was assessed at a confidence level of 95% using Bonferroni correction for multiplicity. Confidence intervals of the ROC area under the curve (AUC) were calculated using bootstrap resampling.
Results: Eleven nodal failures (7.6% of patients, 3.3% of lymph nodes) occurred within a median (interquartile) of 5.0 (4.0-11.2) months from chemo-RT initiation. For one-year patient-level failure, modified MTV did not improve prediction performance (MTVC-AUC = 0.758 (95% CI: 0.636-0.882), MTVM-AUC = 0.682 (95% CI: 0.517-0.847)). MTVC>11.2 cc was associated with statistically significant HR of 7.6 (95% CI: 1.76-32.8), p<0.019; however, MTVM was not statistically significant for any of the tested cut points. For node-specific failure, MTVI-AUC= 0.833 (95% CI: 0.697-0.968), and MTVI > 6.6 cc was associated with HR of 14.43 (95% CI: 3.62-57.6), p<0.001.
Conclusion: Modified MTV using additive node-specific thresholding did not improve patient-level nodal FFS prediction, indicating that conventional nodal MTV may be adequate. However, node-specific MTV was strongly predictive of node-specific failures, which may guide differential dosing regimens for oropharyngeal RT in low-risk patients. As such, radiation dose de-intensification may not be indicated for some highly metabolic tumor sub-volumes.
University of Michigan
The asset you are trying to access is locked. Please enter your access key to unlock.