Volume transfer constants spatial distribution across breast tumors: evidence of interstitial fluid pressure?

INTRODUCTION It is well known that high interstitial fluid pressure (IFP) can be present in the central areas of a tumor. This can lead to tissue necrosis and causes a barrier to drug delivery. IFP can be assessed invasively with the wick-in-needle procedure, this is not a suitable technique in clinical practice. MRI is a non-invasive technique which has shown to be highly sensitive in the detection of breast tumors and has the potential of generating spatial distribution maps of pharmacokinetic parameters linked to IFP. In the previous work of Dadiani an attempt of imaging pressure driven parameters in inoculated mice tumors was made by fitting a T1-weighted dynamic contrast enhanced dataset to the standard Tofts’ model, this produced an imbalanced estimate of Kin and Kout across the tumor. These quantities represent the volume transfer constants from the intravascular space into the lesion leakage space (Kin) and in opposite direction (Kout). They were fitted assuming a fixed value for the fractional volume of interstitial space ve. In this work we have generated parametric maps of breast cancers obtained from patients showing the spatial distribution of the two transfer constants across the lesions. MATERIALS AND METHODS A group of 18 patients with large or locally advanced breast cancer was scanned at 1.5T (GE, Waukesha WI, USA) before chemotherapy treatment. DCE-MRI was performed with 10 seconds temporal resolution using a T1-weighted 2D fast spoiled gradient echo (FSPGR) sequence (TR/TE/α = 8.4 msec /4.2 msec /35°). A set of 9 slices covering the lesion area was acquired in the coronal plane. Dynamic images were acquired for over 6 minutes after the Gd-DTPA bolus injection (0.2 mmol/kg). Using the Tofts’ model the T1-weighted MR signal samples were fitted to the FLASH equation obtaining a tissue contrast concentration dynamic curve. Free parameters in the fitting were Kin and Kout according to the following model: