On the Use of DSC-MRI for Measuring Vascular Permeability

Abstract

Background and Purpose

Contrast agent (CA) extravasation has been shown to confound brain tumor perfusion measurements using DSC-MRI, necessitating the use of correction techniques (e.g. Weisskoff, Bjornerud). Model parameters (K₂ and K_a) postulated to reflect vessel permeability can be extracted from these correction methods, however the biophysical interpretation of these parameters and their relationship to commonly used MR measures of vascular permeability (e.g. K^trans) remains unclear. Given that vascular density, as assessed by blood volume, and vascular permeability, as reflected by K^trans (and potentially K₂ or K_a), report on unique and clinically informative vascular characteristics, there is a compelling interest to simultaneously assess these features.

Materials and Methods

Multi-echo DSC-MRI data was acquired, allowing the simultaneous computation and voxel-wise comparison of single- and dual-echo derived measures of K₂, K_a and K^trans in glioma patients. This acquisition enabled the investigation of competing T₁ and T₂^* leakage effects and echo time dependency on these parameters.

Results

K₂ and K_a displayed non-significant (p = 0.150 and p = 0.060, respectively) voxel-wise linear correlations with K^trans, while a significant (p < 0.0001) inverse relationship was observed between K₂ and K_a [r² = 0.466–0.984]. Significantly different (p < 0.005) mean estimates were found between voxels exhibiting predominately T₁ or T₂^* leakage effects for K₂ and K_a. K^trans, however, was observed to be similar between these voxels (0.109 min⁻¹ vs 0.092 min⁻¹). Significant differences (p < 0.0005) in v_e (0.285 vs 0.167) were also observed between cohorts. Additionally, K₂ and K_a were found to have a significant quadratic relationship (p = 0.031 and p = 0.005, respectively) with v_e.

Conclusion

Estimates of vascular permeability in brain tumors may be simultaneously acquired from multiple-echo DSC-MRI via K^trans, however caution should be used in assuming a similar relationship for K₂ and K_a.

Introduction

Brain tumors are characterized by abnormal, poorly constructed vasculature that is often permeable¹, making them identifiable on contrast-enhanced MR images. With dynamic contrast enhanced (DCE)-MRI methods, contrast agent (CA) wash-in and extravasation alters the tissue T₁ relaxation time and kinetic analysis of the associated signal change permits the computation of the volume transfer constant, K^trans, which reflects vascular permeability and perfusion. In dynamic susceptibility contrast (DSC)-MRI studies, CA flowing through blood vessels decreases tissue T₂^* and the acquired signal changes can be used to estimate tumor blood volume. However, CA extravasation has been shown to confound measurements of tissue perfusion (e.g. underestimation of blood volume), particularly in high-grade brain tumors^2–4. When corrected for CA leakage effects, DSC-MRI measures of blood volume correlate with brain tumor grade and may be useful for monitoring treatment response^{2, 5}.

CA extravasation leads to simultaneous and competing T₁ and T₂^* effects that can substantially alter the temporal dynamics of DSC-MRI signals^{2, 6}, necessitating the use of correction techniques. One such technique, developed by Weisskoff et al.⁷ and Boxerman et al.², incorporates knowledge of the average signal time-course across the brain in non-enhancing voxels to model and correct time-courses in tumor voxels. As a result, a parameter termed ‘K₂’ can be extracted that reflects the degree of CA extravasation. Though initially developed to correct T₁ leakage effects, the Weisskoff method has been adapted to also account for T₂^* leakage effects⁸. A known limitation of this method, however, is that it assumes the mean transit times (MTT) of both healthy and diseased tissue to be equal, which has been observed to not be true in gliomas⁹. To address this issue, Bjornerud et al. recently developed a MTT insensitive approach for correcting both T₁ and T₂^* leakage effects on DSC-MRI signals^{10, 11}. In this method, the tissue residue function, which describes the CA passage through a voxel, is separated into an intravascular and an extravascular component, from which a parameter ‘K_a’ (similar to K₂) can be estimated. A third technique aims to remove T₁-based CA leakage effects, through the use of multiple gradient-echo acquisitions^{3, 12–14}. A feature of this approach is that dynamic T₁-weighted information can be separated and quantified^15–17. Traditional pharmacokinetic modeling^{18, 19} can then be applied to this data to extract a measure of K^trans, in a manner similar to DCE-MRI. This approach has been validated in animal brain tumor models and has been recently applied in high-grade glioma patients^{16, 17, 20}. To collect both DCE-MRI and DSC-MRI datasets, an alternative strategy is to acquire traditional DCE-MRI data during a pre-load injection of contrast agent, which is a technique also commonly used to reduce T₁ leakage effects in single-echo based DSC-MRI data³.

In the case of brain tumors, K^trans is largely considered to reflect vascular permeability¹⁹ and has demonstrated promise in tumor grading^{21, 22} and identifying disease progression and treatment response^23–26. It has been postulated that measures of K₂ and K_a may also directly report on vascular permeability, however their relationship with imaging biomarkers such as K^trans is not entirely clear and may be dependent on CA kinetics, tissue microstructure, and imaging parameters. Preliminary studies have also investigated the use of K₂ and K_a for assessing tumor type²⁷, grade^{28, 29}, and treatment response¹¹.

Inherent to the aforementioned DSC-MRI correction techniques, estimates of K₂ and K_a may assume positive or negative values depending on whether T₁ (+K₂, −K_a) or T₂^* (−K₂, +K_a) leakage effects are the dominating source of signal error. Unlike K₂ and K_a, estimates of K^trans assume the use of a ‘purely’ T₁-weighted signal and, therefore, presume insensitivity to competing T₁ and T₂^* leakage effects. In this regard, a previous simulation study reported a non-linear relationship between K_a and K^trans when large flip angles (>70) were used¹⁰. In a follow up in vivo study¹¹, a positive quadratic relationship between K_a and K^trans was observed. A more recent study found a positive linear correlation between K₂ and K^trans when comparing maximum whole tumor values across patients³⁰. These studies, however, were limited to ROI-based estimates and measures of K^trans acquired from separate DCE-MRI acquisitions, and did not take into consideration the dominating CA leakage effect.

As suggested by previous works, the presence of simultaneous T₁ and T₂^* leakage effects within a tumor may influence the magnitude and interpretation of K₂ and K_a. The overarching goal of this study, therefore, was to investigate the contribution of both T₁ and T₂^* effects on K₂ and K_a, while evaluating these parameters as imaging biomarkers of vascular permeability in brain tumors. This was achieved through voxel-wise comparisons of DSC-MRI derived measures of K₂, K_a, and K^trans using the previously described methods. The multi-echo nature of this study allowed simultaneous measurement of these parameters from the same data set, permitting a more accurate comparison free of registration errors and/or sequence specific differences. In addition, the multi-echo data allowed further exploration of potential echo time dependencies of both Weisskoff and Bjornerud correction techniques.

Methods

MRI data were acquired in high-grade glioma patients (n = 7, See Table 1) under Vanderbilt University Institutional Review Board guidelines at 3T (Achieva, Philips Healthcare, Best, The Netherlands) using a 32-channel head coil. Multiple flip angle (MFA) data (TR = 7.6ms, TE = 4.6ms, FA = 2°-20° in 2° increments) were acquired to compute pre-contrast R₁ (R₁₀) maps. Dual-echo (DE) DSC-MRI data were then acquired using either a dual gradient-echo (DGE) EPI or SAGE EPI protocol^{17, 31} with: TR = 1.5s (DGE) or 1.8s (SAGE), TE₁/TE₂ = 7.0/31.0ms (DGE) or 8.3/25ms (SAGE), SENSE = 2, FOV = 240 × 240mm², Reconstructed Voxel Size = 2.5 × 2.5 × 5.0mm³, and slices = 15. For SAGE data, only the first two echoes were used in the analysis. Measurements were made before, during, and after administration of Gd-DTPA (0.1 mmol/kg, 4ml/s infusion rate followed by a 20ml saline flush). The scan duration was 7.5 minutes, including 80s of pre-bolus baseline data. A high-resolution T₁-weighted data set was collected following the DSC-MRI experiment. Dynamic estimates of ΔR₂^* were computed for each echo ($\mathrm{\Delta }{R}_{2,TE1}^{*}$ and $\mathrm{\Delta }{R}_{2,TE2}^{*}$) and for the dual-echo data ($\mathrm{\Delta }{R}_{2,DE}^{*}$) as previously described^{12, 13}.

Table 1.

Patient Demographics

Patient	Age (yr)	Sex	Prior Resection	Pathology	OS (mo)
1	61	Female	Yes	Grade IV Glioblastoma	17.9
2	66	Male	Yes	Grade IV Glioblastoma	18.2
3	65	Male	Yes	Grade III Anaplastic Astrocytoma	N/A
4	51	Male	Yes	Grade IV Glioblastoma	4.3
5	55	Male	No	Grade III Oligodendroglioma	13.1
6	40	Male	Yes	Grade IV Glioblastoma	11.0
7	42	Female	Yes	Grade IV Glioblastoma	N/A

OS = overall survival after radiologically confirmed tumor recurrence/progression

K₂ Computation

The method proposed by Weisskoff et al.⁷ allows the extraction of K₂ from Eq 1,

$$\mathrm{\Delta }\widetilde{R_2^{*}}(t)\approx K_1·\overline{\mathrm{\Delta }{R}_2^{*}(t)}-K_2{\int }_0^t\overline{\mathrm{\Delta }{R}_2^{*}(t\prime )}dt\prime$$ (1)

where $\overline{\mathrm{\Delta }{R}_2^{*}}$ is the average ΔR₂^* from a mask of non-enhancing brain voxels and $\mathrm{\Delta }\widetilde{R_2^{*}}$ is the leakage affected estimate of ΔR₂^*. A voxel-wise least squares fit to Eq. 1 was performed to extract K₂ using 80s of pre-bolus baseline data and 70s of post-bolus data (2.5 min total) consistent with previous reports^{2, 3, 29}.

K_a Computation

In the presence of CA extravasation, the tissue concentration time-course, C_t(t), can be represented as

$$C_t(t)=f{\int }_0^{t}R(t)·C_p(t-\tau )d\tau +K_a{\int }_{T_c}^{t\prime }{\mathrm{C}}_p(\mathrm{t}\prime -\mathrm{\tau })·\text{exp}(-K_a(\tau -T_c)/{\nu }_e)d\tau$$ (2)

where f is proportional to tissue blood flow, R(t) is defined as the tissue-specific residue function, T_c is the capillary transit time of the CA, v_e is the extracellular-extravascular volume fraction, and C_p is the tracer [CA] in plasma (computed from an AIF extracted from the dual-echo data using an automated selection process^{32, 33}). In DSC-MRI, C_t(t) is estimated in relative terms through measurements of ΔR_2,t^*(t)¹⁰, where ΔR_2,t^*(t) ∝ r₂^*·C_t(t) and r₂^* is the effective transverse relaxivity. Circular deconvolution of Eq. 2 with the AIF³⁴ (over the same time-course used in the Weisskoff correction), results in a composite residue function H(t) described by an early vascular phase (0 ≤ t < T_c) and an extravasation phase (t ≥ T_c)¹⁰:

$$H(t)\approx f·R(t)\hspace{1em}0\le t<T_c$$ (3)

$$H(t)\approx K_a·\text{exp}(-K_a(t-T_c)/{\nu }_e)\hspace{1em}t\ge T_c$$

In the context of a single-echo DSC-MRI acquisition, H(t) ≈ K_a for t >> T_c. In this study, K_a was estimated as the mean value of H(t) following a time T_c, equivalent to 1.5× the mean transit time (MTT), up to H(t=60s).

K^trans Computation

To compute an estimate of K^trans from multi-echo DSC-MRI data, a T₁-weighted signal time-course (S_T1w(t)) was first extracted from dual-echo data via Eq. 4^{15, 16, 35}.

$$S_{T1w}(t)=S_{T{E}_1}(t)·e^{\mathit{\text{ln}}\left(\frac{S_{T{E}_1}(t)}{S_{T{E}_2}(t)}\right)·\left(\frac{T{E}_1}{T{E}_2-T{E}_1}\right)}$$ (4)

A pre-contrast R₁ (R₁₀) map was combined with the S_T1w(t) data to produce dynamic R₁ time-courses (R_1t(t)) for each voxel^{36, 37}. K^trans and v_e were estimated by fitting R_1t(t) and C_p(t) (AIF) with the standard Tofts model^{18, 19}.

Voxel Selection

Voxels selected for this analysis were obtained from enhancing regions on the post-Gd T₁-weighted images, determined using a 50% signal threshold (based on the maximum signal intensity in tumor-containing slices) over a manually drawn tumor ROI. These voxels were further categorized by the predominate leakage effect (T₁ or T₂^*) exhibited in their dynamic ΔR₂^* time-course. In this study, ‘T₂^* voxels’ were defined by a positive mean ΔR₂^* over the last 20s of the time-course used for computation of K_a and K₂. ‘T₁ voxels’ were defined as those in which this estimate was negative.

Statistical Analysis

Voxel-wise measures of K₂ and K_a were compared with K^trans and v_e to examine the relationship between these parameters. Associations between the aforementioned parameters were first analyzed on an individual basis using simple linear regression and reported using the r-squared (r²) statistic. Unless otherwise noted, group voxel-wise comparisons were conducted using analysis of covariance in a generalized linear model for repeated measures. Generalized estimating equations (GEE) were used with an exchangeable covariance structure to model the correlation among voxels across patients.

Results

Fig. 1a shows a representative uncorrected tumor ΔR₂^* time-courses for each echo time and the dual-echo signal, along with the associated Weisskoff model fit. Fig. 1b shows the corresponding tissue residue functions used to compute K_a from the same patient. The computed K^trans, K₂, and K_a maps (overlaid on post-Gd T₁-weighted images) for this patient (at TE₂) can be seen in Fig. 2b–d, respectively, along with the corresponding post-Gd T₁-weighted image (Fig. 2a). Fig. 3a and 3b show an example voxel-wise comparison of K₂ and K_a (computed at TE₂) with the parameter K^trans. The range of correlations at TE₂ were r² = [0.014 – 0.430] for K₂ and r² = [0.0001 – 0.403] for K_a. Across patients, both K₂ and K_a were found to have non-significant (p = 0.150 and p = 0.060, respectively) linear correlations with K^trans. A significant (p < 0.0001) inverse relationship was observed (Fig. 3c), however, between K₂ and K_a [r² = 0.466–0.984]. To help elucidate these observed relationships, further analysis was performed.

Fig. 1.

a) Representative uncorrected tumor ΔR₂^* time-course and the associated Weisskoff model fit (solid) used to compute K₂ at TE₁ (square), TE₂ (dot), and DE (diamond). b) Corresponding tissue residue function used to compute K_a at TE₁, TE₂, and DE.

Fig. 2.

a) T₁-weighted post-Gd anatomical image showing a high-grade brain tumor. Example computed permeability maps (units in min⁻¹) for b) K^trans, c) K₂ and d) K_a.

Fig. 3.

a) Example voxel-wise comparison between K₂ at TE₂ and K^trans. b) Example voxel-wise comparison between K_a at TE₂ and K^trans. c) Voxel-wise comparison between K₂ (y-axis) and K_a (x-axis). Linear regression line shown in black.

With the availability of multi-echo data, the effect of echo time on K₂ and K_a was investigated. Fig. 4a and 4b show box plots using the median values of K₂ and K_a, across all patients. A statistically significant difference (Mann-Whitney U test) was observed between K₂ at TE₁ and TE₂ (p<0.001), K₂ at TE₁ and DE (p<0.001), and K₂ at TE₂ and DE (p<0.01) acquisitions. Similar differences were observed for K_a. For TE₂, voxel-wise estimates of K₂ were observed to be predominately positive for high-grade gliomas whereas K_a was predominately negative. A decrease in echo time (TE₁) resulted in a broader voxel-wise distribution of values across patients with estimates of K₂ becoming increasingly positive and K_a becoming increasingly negative. The computation of K₂ using the $\mathrm{\Delta }{R}_{2,DE}^{*}$ time-course resulted in a negative shift in the distribution of values, with an increase in the number of voxels near K₂ = 0. A similar shift in the distribution towards positive values was observed for K_a.

Fig. 4.

Box plots of median parameter estimates (from all patients) calculated at various echo times for K₂ (a) and K_a (b). Box plots display the median, 25th and 75th percentiles (edges of box), and extreme data points (whiskers). Outliers are plotted individually (+). Significance determined by Mann-Whitney U test.

Fig. 5 shows the contribution of both T₁ and T₂^* leakage effects on the relaxation rate time-courses. Fig. 5a shows the mean ΔR₂^* time course (TE₂) for a tumor ROI from patient 2. The resulting ΔR₁ time-course from the same tumor can be seen in Fig. 5b. Though the ΔR₂^* time-course appears to show no appreciable signs of CA leakage, the ΔR₁ time-course exhibits large changes in R₁ with bolus passage. This indicates CA extravasation and results in a moderate estimate of K^trans. Similarly, focusing on the smallest 10% of all voxels (based on the magnitude of K_a) in a given patient, K_a = −0.043 ± 0.050 min⁻¹, K₂ = 0.113 ± 0.553 min⁻¹, and K^trans = 0.060 ± 0.099 min⁻¹ (weighted mean (mean_w) ± pooled std (std_p)). Fig. 5c and 5d show mean ΔR₂^* and ΔR₁ time-courses from the same tumor with voxels separated by predominate T₁ or T₂^* leakage effects. Note that, in Fig. 5c and 5d, voxels from the same tumor exhibited positive and negative values of K₂ and K_a, while K^trans was observed to be almost identical between the two cohorts.

Fig. 5.

a) Example mean ΔR₂^* time-course (TE = 31ms) for a tumor ROI and b) the resulting ΔR₁ time-course. c) Mean ΔR₂^* and d) ΔR₁ time-courses from the same tumor with voxels separated by whether they predominately exhibit T₂^* leakage effects (‘T₂^* voxels’) or T₁ leakage effects (‘T₁ voxels’).

Table 2 displays the mean estimates of K₂, K_a, and K^trans (separated by T₁ and T₂^* voxels) across all patients. On average, 63% of voxels in the high-grade gliomas were found to predominately exhibit T₁ leakage effects. In addition, a significant difference (p < 0.005, paired t-test) was observed, across patients, between mean estimates from T₁ and T₂^* voxel cohorts for both K₂ and K_a. While the difference between T₁ and T₂^* cohorts for K^trans trended toward significance (p ≈ 0.05), the weighted mean for each cohort across patients were similar (0.109 min⁻¹ vs 0.092 min⁻¹). In all voxels across patients, we observed v_e = 0.241 ± 0.207. When separated by leakage effect, a significant difference (p < 0.0005, paired t-test) in mean estimates of v_e was also observed. Additionally, both K₂ and K_a were found to have a significant quadratic relationship (p = 0.031 and p = 0.005, respectively) with v_e.

Table 2.

Patient specific estimates of DSC-MRI and DCE-MRI parameters separated by the predominant leakage effect

	# Voxels (%)		K2 (min−1)		Ka (min−1)		Ktrans (min−1)		ve
Patient	T1	T2*	T1	T2*	T1	T2*	T1	T2*	T1	T2*
1	44 (79%)	12 (21%)	1.807	1.205	−0.373	−0.250	0.223	0.066	0.221	0.072
2	214 (45%)	265 (55%)	1.229	−0.815	−0.342	0.026	0.169	0.163	0.359	0.258
3	126 (61%)	79 (39%)	2.374	0.822	−0.372	−0.117	0.089	0.038	0.328	0.150
4	368 (47%)	417 (53%)	1.767	0.700	−0.536	−0.469	0.104	0.078	0.228	0.140
5	187 (56%)	147 (44%)	1.975	0.787	−0.149	−0.025	0.069	0.044	0.284	0.107
6	734 (93%)	52 (7%)	3.726	0.240	−0.256	0.004	0.099	0.050	0.290	0.138
7	16 (64%)	9 (36%)	2.591	0.025	−0.418	0.024	0.200	0.179	0.203	0.107
meanw			2.627	0.289	−0.329	−0.208	0.109	0.092	0.285	0.167

Discussion

DCE-MRI estimates of vascular permeability, often reported via K^trans, have been shown to be helpful in deciphering brain tumor grade²¹ and in predicting disease prognosis^{25, 38}. Unlike DCE-MRI, DSC-MRI acquisitions can actually be confounded by the increased vascular permeability present in brain tumors, requiring strategies for leakage correction of the MR signal time-courses. Rate constants (K₂ and K_a) computed from these correction techniques have been suggested to reflect vessel permeability^{7, 28}. To evaluate this relationship, a simultaneous comparison between K^trans and the parameters K₂ and K_a was performed using multi-echo DSC-MRI. In general, the range of K₂ and K_a estimates in this study were observed to be larger than that of K^trans, though they were consistent with previous measures in brain tumors^{8, 10, 28}. Voxel-wise linear relationships between K₂ and K_a and the parameter K^trans were found to be non-significant when computed from the same data set. Though a non-linear relationship between K_a and K^trans was previously presented in simulations¹⁰, this work provides additional in vivo confirmation. The individual correlations observed here between K₂ and K^trans in gliomas were similar to that observed by Bonekamp et al. using max K^trans and K₂ values from whole tumor ROIs³⁰. Though the lack of a strong linear correlation with K^trans suggests potential limitations with extracting permeability estimates from DSC-MRI correction methods themselves, it should not, however, be interpreted as a failure of these techniques to reliably correct CBV measures for CA leakage.

The effect of echo time on K₂ and K_a was also studied. From Fig. 4 we observed a significant increase (decrease) in estimates of K₂ (K_a) with a shorter echo time. This is due, in part, to the decrease in T₂^* weighting with decreasing echo time and subsequent dominance of T₁ leakage effects. Liu et al. previously explored the effect of echo time on K₂ in numerical simulations⁸ and noted that changes in the actual vascular permeability should not affect the polarity of K₂, though changes in imaging parameters (e.g. echo time) could. Prior to the current study, a similar analysis with K_a had not yet been performed.

In addition to echo time, the intrinsic presence of competing and simultaneous T₁ and T₂^* leakage effects, within a given voxel, were integral in determining the value of K₂ and K_a. As shown in Fig. 5, competing T₁ and T₂^* leakage effects can produce a ΔR₂^* time-course that paradoxically appears to be free of CA extravasation effects. This is misleading, as the dynamic ΔR₁ information reveals appreciable CA leakage, resulting in moderate estimates of K^trans. As noted by Bjornerud et al., the presence of both T₁ and T₂^* relaxation effects in the extracellular-extravascular space may drive K_a (and K₂) towards 0, resulting in artifactually low estimates. As an example, in the smallest 10% of all voxels (based on the magnitude of K_a), the mean K^trans was observed to be 50% larger than |K_a|. Conversely, the magnitude of the mean K_a was ≈3× larger than K^trans when computed using all voxels. Additionally, the mean value of K₂ and K_a, computed from the aforementioned subset of voxels (smallest 10%), were almost an order of magnitude smaller than the respective mean K₂ and K_a computed using all voxels. These findings clearly have implications on the reliability of these parameters as measures of vascular permeability.

In general, the relationship of K₂ and K_a with K^trans may indicate an inaccurate assumption that these parameters solely reflect vessel permeability in brain tumors. When separated into T₁ and T₂^* voxel cohorts, the mean values of K₂ and K_a across patients were found to be significantly different from one another (Table 2). The same was true for v_e. Similar to the previous observation between K_a and K^trans in vivo¹¹, a significant quadratic relationship was observed between K₂ and K_a and v_e across all patients. To this end, a recent theoretical study by Liu et al. demonstrated a potential relationship between v_e and the ratio of the parameters K₁ and K₂ from the Weisskoff correction method³⁹. These results indicate that K₂ and K_a may also be influenced by the extravasation space of the CA.

The data in Table 2 also revealed that ‘T₁ voxels’ demonstrated larger v_e values than those found in ‘T₂^* voxels’. This likely originates from the underlying biophysical basis of T₁ and T₂^* leakage effects. As in DCE-MRI, T₁ leakage effects result from the direct interaction of CA with the extracellular-extravascular water. Accordingly, the physiological factors that drive the tissue [CA] (compartmental volume fractions, perfusion and vascular permeability) as well as physical properties (CA T₁ relaxivity, pre-contrast T₁) and pulse sequence parameters (TR, flip angle) all influence the shape and magnitude of T₁ leakage effects on DSC-MRI signals. In addition to physiological factors and imaging parameters, T₂^* leakage effects are influenced by intravoxel susceptibility differences created by the spatial distribution of the CA within a voxel. Recently, Semmineh et al. demonstrated that these effects are predominantly influenced by cellular properties including density, size, distribution and shape⁴⁰. Consistent with the results presented herein, stronger T₂^* leakage effects were observed for tissues with higher cell density (or lower v_e). In general, the dependency of T₂^* leakage effects on tumor cellularity manifests as changes in the CA’s effective T₂^* relaxivity. So unlike T₁ leakage effects, where the CA’s T₁ relaxivity is essentially constant within and across tumors, the T₂^* relaxivity may vary from voxel to voxel as the cellular properties change⁴¹.

The variable CA T₂^* relaxivity also has important implications on the interpretation of the extracted K₂ and K_a parameters. Though voxels were designated as predominantly exhibiting either T₁ or T₂^* leakage effects, each voxel’s signal is the summation of these competing effects, as previously discussed. In the limiting case where T₂^* leakage effects are absent and the signals only reflect T₁ leakage effects, the K₂ and K_a parameters are primarily driven by the underlying CA kinetics and the assumptions built into the correction models and can be understood accordingly. However, when there are competing T₁ and T₂^* effects K₂ and K_a represent a complex balance between the CA kinetics and the tissue microstructure. Practically, this implies that a positive and negative estimate of K₂ or K_a of the same absolute value may not reflect the same combination of vascular permeability, tissue compartment size, or microstructural geometry. Similarly, K₂ and K_a values that are equivalent within or across tumors may not reflect the same underlying physiological environment since they could originate from unique combinations of competing T₁ and T₂^* effects. This observation may help further explain the discrepancies in using K₂ and K_a to evaluate tumor grade and to assess treatment response^{11, 28, 29}. Computational studies that account for the underlying biophysical basis of the DSC-MRI signal could be used to systematically investigate and provide insight into the complex interaction between T₁ and T₂^* leakage effects and the derived K₂ and K_a values.

The use of multi-echo DSC-MRI in this study enabled measures of DCE-MRI signals and, subsequently, computation of the associated K^trans maps. As mentioned above, an alternative approach to collect both datasets in the same exam is to acquire DCE-MRI data during a pre-load of CA. This enables the use of traditional DCE-MRI pulse sequences, ones that typically have higher spatial (and lower temporal) resolution. For the purpose of the study, this approach would have enabled the comparison of more conventionally derived K^trans values to K₂ and K_a. It is interesting to note, however, that the addition of a pre-load to this study would have reduced T₁ leakage effects and increased T₂^* leakage effects. It is unclear how this would influence the correlation between K^trans, K₂ and K_a. Another limitation of this study is the small sample size. While the findings are likely to hold in a larger population of glioma patients, it would be valuable to expand the tumor types considered (e.g. primary central nervous system lymphoma and brain metastasis) since different histologic subtypes have been shown to express varying degrees of T₁ and T₂^* leakage effects.

Conclusion

This study investigated the use of DSC-MRI for estimating vascular permeability in brain tumors. Implementation of common DSC-MRI leakage corrections techniques afforded the computation of rate constants (K₂ and K_a) postulated to report on vessel permeability. Additionally, the acquisition of multi-echo data allowed the computation of the DCE-MRI pharmacokinetic parameter K^trans. A voxel-wise comparison between the parameters K₂, K_a and K^trans revealed non-significant linear correlations that may be attributed, in part, to competing T₁ and T₂^* leakage effects and the effect of echo time on K₂ and K_a. Further investigation also revealed a significant quadratic relationship between K₂ and K_a and the DCE-MRI parameter v_e. Based on these findings, caution should be used in assuming a direct relationship between K₂ and K_a and vascular permeability in brain tumors. Furthermore, the acquisition of K^trans from multi-echo DSC-MRI data may provide a convenient method for simultaneously measuring vascular permeability and perfusion in brain tumors.

Abbreviations

CA

contrast agent

DSC

dynamic susceptibility contrast

DCE

dynamic contrast enhanced

K^trans

CA volume transfer constant

v_e

extracellular-extravascular volume fraction

CBV

cerebral blood volume

CBF

cerebral blood flow

MTT

mean transit time

MFA

multiple flip angle

AIF

arterial input function