Scientists interested in the physiological and histological basis of T1-DCE MRI as well as clinicians applying T1-DCE in neuroradiology.To elucidate if kinetic parameters of T1-DCE MRI based on the extended Tofts model (ETK) have a foundation in the microvascular anatomy of meningioma [1].

In T1-DCE MRI, kinetic parameters, such as transfer constants K^trans and k_ep but also the extracellular-extravascular (v_e) and the plasma volume fractions (v_p), relate to vascular density and surface area. They are presumed to characterize tissue microvasculature and cellularity [2]. T1-DCE MRI is thus increasingly clinically applied to deduce characteristics of tissue composition. However, few studies investigate if this correlation with microvascular anatomy actually exists [3,4].

Bearing in mind that models to derive kinetic parameters are known to be vulnerable to factors such as vascular permeability, global vascularization and perfusion, correlations depend on both tissue and applied model [5,6]. Meningioma is a usually highly perfused and vascularized extra-axial tumor entity that seems ideal to elucidate whether and respectively to what extent ETK-derived kinetic parameters are related to microvascular anatomy based on precisely matching tissue specimens [1,7].

19 patients with confirmed meningiomas (15 WHO I, 3 WHO III, 1 WHO II) underwent a whole-brain T1-DCE MRI [3.0T Achieva TX (Philips Healthcare, Best/NL); 8-channel head coil; axial orientation; 36 slices; 1.57x1.6x3mm acq. voxel size; dual FA technique (5°/15°; TE 2.3ms); 50 dynamics at 5.3s each (FA 8°; 1.7ms); contrast: 0.1mmol/kg BW Gadobutrol (Bayer Healthcare, Leverkusen/D); 24ml saline flush; flow rate 3ml/s].

The ETK, defined by C(t) = v_p* C_a(t) + K^trans*e^−tkep* C_a(t), was applied to derive K^trans, k_ep, v_e and v_p (fig. 1, Intellispace Portal 5.0, Philips Healthcare) [1]. Arterial vessels from internal and external carotid artery feed a meningioma, but cannot be distinguished on imaging. However, as meningiomas certainly drain into large venous sinuses the ROI for vascular input function (VIF) calculation was placed in the superior sagittal sinus. VIF considered hematocrit and contrast relaxivity 5.0 L/(mmol*s)].

In each patient 1 to 5 biopsy regions were marked (5x5mm) and tissue based kinetic parameters were measured. T1-DCE MRI data sets were exported to neuronavigation software (Brainlab, Feldkirchen/D). The marked biopsy specimens were collected during total resection of the meningioma (64 specimens in total). These were fixated, sliced and immunohistochemically stained with CD34 (marker for endothelial cells). The slices were scanned with tissue quantification software (Tissue Studio, Definiens, fig. 2). Vessels were divided into three size groups: small<2mm², intermediate<5mm², large>5mm². Software was trained to reliably detect vessels with and without lumen. For subsequent automated analysis the tissue was marked, then divided into 0.6*0.6mm ROI grids. Representative ROIs were checked for correct registration of vessel wall, lumen and size segmentation (fig. 3). Data retrieved was: 1. vessel density (vessels/mm²), 2. relative area of vasculature in total slice area (%), 3. number of vessels by size group, 4. distribution of vessel sizes per slice and 5. mean vessel wall thickness.

All data was split into subgroups of vessels with and without lumen. Kinetic parameter data (K^trans, k_ep, v_e, v_p) was inter- and intra-individually analyzed for correlation with the microvascular data (mixed linear model; SAS9.4, SAS Institute Inc., Cary/USA.)

The vascular fraction ranged from 0.5 to 14.9% of the total tissue area. It did not significantly correlate with any of the kinetic parameters on an inter-individual basis (e.g.: v_p: p=0.954, correlation r=0.01; K^trans: p=0.642; correlation r=0.09). Vascular size, relative fractions of vascular sizes and thickness of vascular lumina were also not significantly correlated to any of the kinetic parameters.

The inter-individual variability of kinetic parameters was expectedly high (K^trans: 0.28±0.29/min; k_ep: 0.99±0.79/min; v_p: 36.3±57.0%; v_e: 27.9±21.9%). Therefore an intra-individual correlation was performed despite the low number of 3 to 5 biopsy points per patient. This rendered inconsistent results (e.g.: v_p vs. vascular fraction of total tissue: r=-0.99 to +0.89; mean r: 0.11; K^trans vs. vascular fraction of total tissue: r= -0.99 to+0.99; mean r: 0.04).

Quite different from initial expectations kinetic parameters of T1-DCE MRI did not correlate with any of the microvascular tissue factors anylyzed in this multimodal setting. Pronounced inter-individual variability may not explain this lack of correlation, as intra-individual results were highly inconsistent. Future studies applying different kinetic and VIF models are needed to rule out a bias from ill fitting of the models applied in this piece of research.Kinetic parameters of T1-DCE MRI cannot directly be deduced from the microvascular architecture in meningioma. Clinical users of T1-DCE MRI should therefore be careful to derive anatomical tissue aspects from their kinetic data.