Review Article

Hypoxia in Head and Neck Cancer in Theory and Practice: A PET-Based Imaging Approach

Table 4

Stochastic tumour models utilising PET oxygenation data to predict the efficacy of nonstandard treatment solutions.

ReferenceModelling methodsDetails and outcomes

(Toma-Daşu et al. 2009) [33]F18[MISO] and Cu[ATSM] distribution functions modelled in a 108 cell tumour growth and O2 transport model followed by uniform or central boost RT.Tracer binding versus pO2 functions used (higher uptake at intermediate O2 for Cu) to generate tracer uptake maps for each tracer on a 2D slice of heterogeneous spherical tumour. Convolution function used to describe finite resolution of the imaging modality. Local temporal changes in cellular O2 accounted for. Virtual image maps generated to predict LQ survival and Poisson tumour control using 2 different circular dose distributions with central boost doses. Redistribution of dose (same integral dose but hotter in the centre) was possible for each tracer without decreasing the target tumour control (90%).

(Titz and Jeraj 2008, Titz et al. 2012)  [34, 35]Simulating effects of antiangiogenic treatment using F18[FDG], F18[FLT] and Cu61[ATSM] PET data in a tumour proliferation and therapy model (2008—where RT is modelled as the treatment modality) with an added vascular and PET/drug PK/PD component (open 2-compartment) (2012).BvMb plasma concentration-time-profiles cbev(t) utilised, with model parameters adapted to population-based values (e.g., MVD to determine BvMb PD). A linear relationship between VEGF expression and endothelial cell (EC) proliferation used. Nonnormal distributions manipulated the raw data O2 PET data (~4 mm pixels) for cellular level input to generate 2D oxygenation maps considering multiple diameters and angles of the vessels. 8 HNSCC PET scans used (phase 1 trial data), before and after RT for input and comparison to model predictions. A decrease in SUVs (i.e., reduction in vasculature) after BvMb agreed with follow-up PET. Increase in hypoxia due to BvMb observed, peaking at week 2 after treatment, but decreasing with increasing baseline levels of hypoxia and increasing CCT. Due to pO2 and proliferation interdynamics, simulations could provide estimates of optimal drug administration times (i.e., every 2 or 3 weeks). Expansion planned for the use of voxel-based kinetic parameters to model drug uptake more precisely and vessel “remodelling” in response drugs.

[PD: pharmacodynamic; CCT: cell cycle time; BvMb: bevacizumab].