INTRODUCTION

Multi-parametric MRI (mpMRI) plays an important role in detection and grading of prostate cancer (PCa) [1]. Although T2-weighted imaging and diffusion-weighted imaging are the two main components for prostate mpMRI, dynamic contrast enhanced (DCE) MRI is included in mpMRI [2]. Prostate DCE-MRI is often analyzed quantitatively using pharmacokinetic models, such as the standard Tofts model to extract the volume transfer rate constant (K^trans) (exchange between blood plasma and the extravascular extracellular space (EES)) and fractional volume of EES (v_e) [3]. However, the standard Tofts model may not be compatible with the heterogeneous characteristics of tumor micro-environment that results in an initial rapid uptake of contrast agent followed by a less rapid, but prolonged, uptake of the contrast agent [4]. As a result, tumor heterogeneity at the microscopic level could cause poor fits to DCE-MRI data for the standard Tofts model and errors in extracting K^trans and v_e. This would limit the diagnostic accuracy of using the standard Tofts model to analyzing DCE-MRI data.

In this study, the two-tissue compartment model (2TCM) [5] was used to analyze prostate DCE-MRI and the results were compared with those from the standard Tofts model. In contrast to the standard Tofts model with only one tissue compartment, the 2TCM has one slow and one fast exchanging tissue compartment.

METHODS

A total of 29 patients with biopsy-confirmed prostate cancer were included in this IRB-approved study. MRI data were acquired on a Philips Achieva 3T-TX scanner. After T2-weighted and diffusion-weighted imaging, baseline T1 mapping was performed with using the variable flip angle method [6]. Subsequently, DCE data using 3D T1-FFE mDIXON sequence were acquired pre- and post-contrast media injection (0.1 mmol/kg Multihance; TR/TE1/TE2=4.6/1.7/3.3 ms, FOV=250×385 mm², matrix size=200×308, flip angle=10°, slice thickness=3.5 mm, typical number of slices=24, SENSE factor=1.67, half scan factor=0.675) for 60 dynamic scans with typical temporal resolution of 8.3 sec/image.

The tissue contrast agent concentration (C(t)) as a function of time (t) was calculated using the gradient echo signal equation [7]. Arterial input functions (AIF) (C_p(t)) were extracted by manually segmenting the left iliac artery on a slice with cancer. The DCE-MRI data was analyzed by using the standard Tofts model:
$$C(t)=K^{trans}\int_{a}^{b}C_p(\tau)\exp(-(t-\tau)k_{ep})d\tau,------(1)$$
as well as using the 2TCM:
$$C(t)=\int_{a}^{b}C_p(\tau)[K_1^{trans}\exp(-(t-\tau)k^1_{ep})+K_2^{trans}\exp(-(t-\tau)k^2_{ep})]d\tau,------(2)$$
where k_ep=K^trans/v_e, $$$k^i_{ep}=K_i^{trans}/v_e^i$$$ (i=1,2). In order to obtain unique results for fits of C(t) using MATLAB, Eq. 2 was written in an asymmetric form as:
$$C(t)=K_1^{trans}\int_{a}^{b}C_p(\tau)\exp(-(t-\tau)k^1_{ep})\cdot[1+\epsilon\cdot\exp((t-\tau)\lambda)]d\tau,------(3)$$
where $$$K_2^{trans}=\epsilon\cdot{K_1^{trans}}$$$ and $$$k_{ep}^2=k_{ep}^1-\lambda$$$.

The regions-of-interest (ROIs) for prostate cancer (n=54) and normal tissue (n=83) in different prostate zones were drawn on T2W images and transferred to DCE images. Pearson’s correlation coefficient was calculated between physiological parameters obtained from the standard Tofts model and 2TCM. The Student’s T-test was performed to determine whether there was significant difference between cancer and normal tissue for all six physiological parameters. A p-value less than 0.0083 (= 0.05/6) with Bonferroni adjustment for multiple testing was considered statistically significant.

RESULTS

Figure 1 shows a prostate DCE image, plot of the AIF (purple line) and plots of measured C(t) (black dots), as well as corresponding fits with the standard Tofts model (red line) and 2TCM (green line) for three tumor pixels and one normal tissue pixel. The 2TCM fits are much better than those of the standard Tofts model.

Figures 2 and 3 compare a histology slice, the corresponding T2W image and ADC map with physiological parametric maps obtained from the standard Tofts model (K^trans and k_ep) and the 2TCM (K_i^trans and kⁱ_ep, i=1,2). K₁^trans is similar to K^trans with fewer false positives and K₂^trans is more specific for cancer because K₂^trans is very small for normal tissue.

There are strong correlations (r=0.82 to 0.94, p<0.001) between K^trans and K_i^trans (i=1,2) for cancer (Fig.4 (a)), and moderate to strong correlations (r=0.69 to 0.93, p<0.001) for normal tissue (Fig. 4 (b)). There was weak correlation between k_ep and k¹_ep, but strong correlation between k_ep and k²_ep for cancer (Fig. 4 (c)). This indicates poor fitting with the Tofts model as shown in Fig. 1, suggesting advantages for the 2TCM. There were moderate correlations between k_ep and kⁱ_ep (i=1,2) for normal tissue (Fig. 4(d)).

T-tests show significant difference (p < 0.006) for all the parameters between cancer and normal tissue (Fig. 5). Receiver operating characteristics (ROC) analysis shows that the parameters K^trans, K₁^trans and K₂^trans have the area under the curve (AUC) of 0.74, 0.79 and 0.69, respectively.

DISCUSSION

The results demonstrated that prostate cancer is heterogeneous involving both the fast (K₁^trans) and the slow (K₂^trans) exchanging compartments. The strong correlation between K^trans and K₁^trans but weak correlation between k_ep and k¹_ep for cancer suggest that the 2TCM is needed in order to reduce false positive produced in using the Tofts model. Since the contribution of the second compartment (K₂^trans) is close to zero in healthy tissue, the parametric maps derived from the 2TCM showed fewer false positives, suggesting potential advantages for diagnosis of prostate cancer.

CONCLUSION

Our study demonstrated that the 2TCM of DCE-MRI may be useful for quantitative analysis of prostate DCE-MRI.

Acknowledgements

This research is supported by National Institutes of Health (R01CA218700, U01CA142565, R01CA172801 and S10OD018448.).