Estimating hypoxic status in human tumours: a simulation using eppendorf oxygen probe data in cervix cancer patients

Abstract

Purpose: To define the minimum number of pO2 measurements needed to categorize human cervix tumours as either hypoxic or oxic, and to estimate the error resulting from using this minimum number of measurements. Materials and Methods: Using Eppendorf oxygen probe data from our ongoing prospective trial, we simulated the measurement of tumour oxygenation with a small number of data points. The Eppendorf pO2 histograph was used to measure tumour pO2 in 135 patients with locally advanced cancer of the cervix. The majority of subjects had 5 tracks sampled, and the median number of measurements per track was 32. The hypoxic proportion, defined as the percentage of pO2 values <5mm Hg (HP5), was calculated for each tumour. Hypoxic tumours were defined as those with a median HP5 >50%, and oxic tumours as those with median HP5 ≤50%. In assessing the individual track measurements, 1-7 pO2 measurements were randomly selected from each of 4 tracks from each tumour.PO2 measurements were randomly selected from the Eppendorf measurements in each tumour, and the tumour was defined as either hypoxic or oxic, based on the above definitions. Sensitivity and specificity were calculated, considering the classification as given by the Eppendorf measurements as the “gold standard”. The number of measurements chosen randomly was varied, and the process repeated two thousand times. Results: The probability of falsely classifying the tumour decreased as the pO2 measurements per tumour increased from 4 to 28, and at 16 measurements reached a value of approximately 10%. At 24 measurements per tumour, the probability of misclassification of the tumour was approximately 8%. Further measurements per tumour beyond 24 did not improve the ability to accurately classify the tumour. The probability of falsely classifying the tumour decreased as pO2 measurements per track increased. This appeared to plateau at 5 measurements per track, giving a probability of false classification of the tumour of approximately 9%. Conclusion: Approximately 16-24 measurements per tumour, or 5 measurements per track, using the Eppendorf pO2 histograph, are sufficient to categorize cervix tumours as either hypoxic or oxic. The results of this study will serve as a guide for research clinicians in the use of this and other systems in the assessment of tumour oxygenation in humans.