Quantification of CCI-103F labeling heterogeneity in canine solid tumors

Purpose: The purposes of this study were to assess sources of variation in the distribution of nitroimidazole-labeled hypoxic cells in canine tumors and to quantify the reliability of estimating overall nitroimidazole-labeled area fraction from biopsies. Methods and Materials: Hypoxic cells were labeled in 24 canine tumors by immunostaining of the nitroimidazole hypoxia marker CCI-103F. In tumors with a volume <100cm³, each cubic centimeter of tumor was examined; in larger tumors 100 randomly selected 1 cm³ samples were examined. These data were used to estimate the overall CCI-103F-labeled area fraction in the tumor. A variance components model was used to quantify intertumoral, intratumoral, and within slide (residual) sources of variation. The ability to estimate intratumoral CCI-103F-labeled area fraction based on information obtained from biopsies was assessed by randomly selecting two or four samples from the dataset for each tumor and comparing the mean CCI- 103F-labeled area fraction from this limited sample to the labeled area fraction based on each cubic centimeter; this simulation process was repeated 1000 times. Results: Intratumoral (27% of total) and intertumoral (30% of total) variation in CCI-103F-labeled area fraction were similar. Residual variation (variation at the microscopic level) accounted for 43% of total variation in CCI-103F labeling. Intratumoral variation in labeling decreased as the intratumoral CCI-103F mean labeled area fraction decreased. The accuracy of estimating the intratumoral CCI-103F-labeled area fraction in a tumor from limited sampling increased as the number of samples increased or the intratumoral labeled area fraction decreased. When four random samples were used to estimate overall CCI-103F-labeled area fraction in the tumor, estimates from approximately 90% of the 1000 simulations were within 0.10 of the intratumoral CCI-103F-labeled area fraction. Classifying a minimally labeled tumor as unlabeled based on limited sampling was unlikely. Conclusion: Despite intratumor variation, acceptable estimates of nitroimidazole-labeled cells in a tumor may be obtained from a clinically feasible number of biopsies.