Investigating optimal methods for assessing early response of head and neck cancer xenografts (HNCX) to hypofractionated radiation therapy (RT) using F-18 FDG-microPET-CT (PET)

Objectives To investigate optimal methods to obtain parameters from PET for assessing early response of HNCX to single 15 Gy RT.

Methods Human head & neck squamous cancer cells (UT14) were injected to the flank of female nu/nu mice to generate xenografts. RT or sham (control) started at tumor size of 400-500cc, Tumor volumes were measured twice a week. At different time points after RT or sham, 2-3 mice were first assessed by PET with dynamic acquisition over 2 hrs, and then immediately acrificed to harvest tumors, which were oriented to correlate with PET image and stained with HE for histology. T-tests and ROC analysis was used to compared standard uptake value (SUV) without and with correction by glucose (SUVglc) at 30, 60, 90 and 120 min, rate constants (K1, k2, K3) and transfer constants by kinetic modeling (Ki-mod) and Patlak analysis from 0-120 min (Ki-120) and 12-70 min (Ki-70) for their ability to detect tumor growth and viability.

Results 30 mice were implanted with HNCX and 27 were successfully imaged by PET (11 control; 16 RT). Tumor growth curve for controls was linear but that for RT was arrested for the first 21 days after RT and resumed at a linear rate thereafter. Comparing to control, all time point SUVs (p<0.05), Ki-70 and Ki-mod (p<0.01) were significantly lower during the tumor-arrest phase after RT but not during the regrowth phase (p=NS). The corresponding SUVglc, K1, k2 or k3 alone and Ki-120 failed to show significant differences between control and RT in both phases. ROC analysis showed best performace of Ki-mod, Ki-70, SUV at 60 amd 90 min (p<0.04) but not others (p=NS).

Conclusions SUV without glucose correction and Ki-70 or Ki-mod appear to correlate well for tumor growth and viability. Prolonged acquisition beyond 70 minutes may not be necessary for assessing early response after RT