Quantitative analysis of [¹⁸F]FMISO PET for tumor hypoxia: Correlation with results using immunohistochemistry

Objectives Quantitative evaluation of tumor hypoxia based on [¹⁸F]FMISO PET is important for treatment planning in radiotherapy. However the capability of different analysis methods in revealing the underlying hypoxia status, is not clear. This study aims to assess these quantitative analysis methods with the support of histological data.

Methods 16 nude mice with xenografted human squamous cell carcinomas (FaDu or CAL-33) were scanned using 2h FMISO PET and 2 blood samples taken at the end. Tumors were sliced and stained with the hypoxia marker pimonidazole. The hypoxic area was segmented using a k-means clustering algorithm and the hypoxic fraction was calculated as the ratio of the hypoxic area in vital tissue area pooled over 3 tissue slices (top, middle, bottom). PET images were reconstructed using filtered-back projection and tumor and normal organs were outlined manually based on fused CT and PET images. PET images were analyzed using various methods including static analysis (SUV, tumor-to-blood ratio, SUVR) and kinetic modeling (Casciari αk_A, irreversible and reversible two-tissue compartment k₃, Thorwarth w_A, Patlak K, Logan DV, Cho slope) and correlated with the hypoxic fraction.

Results Significant correlations were observed for k₃ of the irreversible two compartment model (r=0.68, p=0.004), and tumor-to-brain ratio (r=0.59, p=0.02). The correlation to tumor-to-muscle ratio varies for 6 different muscles assessed here (r=0.59 to -0.28).

Conclusions With the support of histological data, this study shows that different analysis methods for FMISO PET perform differently for the assessment of tumor hypoxia. The irreversible two compartment model has the best performance among the investigated methods. FMISO PET for radiotherapy treatment planning needs to be carefully analyzed and further clinical assessment is needed