Target Audience

Radiologists, oncologist and basic researchers within the field of cancer research and radiation response prediction.

Purpose

Hypoxia has, through measurements with pO2 electrodes, been associated with aggressive tumors and poor outcome of chemoradiotherapy in cervical cancer. Dynamic contrast-enhanced MRI (DCE-MRI) has been suggested as a non-invasive method for assessing hypoxia in tumors. Several models describing the uptake of contrast agent exist, among these are the commonly used Tofts and Brix models. Two parameters with a large clinical potential are the K^trans and the A parameter, A_Brix, from the Tofts and Brix model, respectively, which have both shown relationships to treatment outcome in cervical cancer. A_Brix has been associated with hypoxia, as measured by the expression of hypoxia responsive genes ¹, but the underlying physiological factors behind neither of the parameters are fully understood. In this study, DCE-MRI of cervical cancer xenografts was combined with immunohistochemical analysis in order to better understand the physiological background of the K^trans and A_Brix parameters, and thus their prognostic potential.

Method

Totally 29 adult female BALB/c-nu/nu mice were used as host animals for 17 HeLa and 12 SiHa cervical cancer xenografts. The tumors were imaged using DCE-MRI, acquired with a 3D FLASH sequence (7.05 T, slice thickness 0.8 mm, in-plane resolution 0.23 x 0.23 mm, TE/TR=2.17/10.2 ms) with a temporal resolution of 14.6 s and a total scan time of ~15.5 min. Gd-DOTA was used as the contrast agent. The imaging protocol also included pre-contrast high-resolution T2-weighted images, used for defining the tumor region, and T1-mapping (TE=8.5 ms, TR=5000, 3000, 1500, 800, 400 and 347.39 ms), used for calculating the concentration of contrast agent in the tumor. Both the Tofts and Brix model were used in analysis of the DCE series. For the Tofts model a population based arterial input function was applied, as described by Benjaminsen et al ². K^trans and A_Brix were calculated for each voxel and presented in parametric maps (Figure 1). Immediately after imaging, the tumors were excised and fixed in formalin. Immunohistochemistry was performed with pimonidazole as marker for tumor hypoxia (Figure 1), while CD31 and hematoxylin were used for measurement of vascular- and cell density, respectively. The histological sections were imaged in a microscope and analysed with the MatLab software.

Results

A_Brix showed a strong negative correlation with hypoxic fraction (P<0.001), as determined by the area fraction of tumor tissue staining positive for pimonidazole, whereas K^trans showed a weaker, but significant (P<0.01) negative correlation. K^trans correlated strongly with vascular density (P<0.0001), but showed no correlation to cell density. A weak, but significant, correlation with vascular density was also found for A_Brix (P<0.05), but in contrast to K^trans, this parameter showed a much stronger negative correlation with cell density (P<0.01). Examination of the histological sections revealed a weak, but significant negative correlation between hypoxia and vascular density, while there was a strong correlation between hypoxia and cell density (P<0.05 and P<0.001, respectively).

Discussion & Conclusion

Hypoxic conditions arise when oxygen supply does not meet the oxygen demand of the cells. This can be a result of both high oxygen consumption, and low oxygen supply. The strong correlation between cell density and pimonidazole staining in our work indicates that cell density, and thus a high level of oxygen consumption, was a major cause of hypoxia in the xenografts. This may explain why K^trans, which was correlated with vascular density and not with cell density, displayed a weak correlation with hypoxia. Moreover, A_Brix, which showed a strong association with cell density in addition to a weaker correlation with vascular density, displayed a strong correlation with hypoxia. These results imply that hypoxia may be the most prominent underlying physiological factor of A_Brix, whereas K^trans probably merely reflects the vascular density of cervical tumors. In accordance with this hypothesis, A_Brix would be the preferred DCE-MRI parameter for predicting hypoxia related treatment resistance in cervical cancer.

Acknowledgements

No acknowledgement found.