Synopsis

Keywords: Diffusion/other diffusion imaging techniques, DSC & DCE Perfusion, IVIM

DCE-MRI can provide quantitative estimates of perfusion-related parameters in tumors, such as blood flow and blood volume fraction. It has also been proposed that IVIM MRI can be used to characterize perfusion in addition to microstructure. This proposal remains controversial and requires further investigation. In this study, we investigate the relationship between perfusion-related parameters measured by IVIM and DCE-MRI in a cohort of patients with breast cancer imaged before and after one and two cycles of neoadjuvant chemotherapy.

Introduction

Dynamic contrast-enhanced (DCE)-MRI has a key role in the response monitoring of patients with breast cancer to treatment with neoadjuvant chemotherapy (NACT)¹. DCE-MRI enables estimation of perfusion-related parameters of breast tumors, including blood flow (Fb) and blood volume fraction (vb)². An alternative technique proposed to probe tumor perfusion is intravoxel incoherent motion (IVIM) diffusion-weighted imaging (DWI)³. It has been suggested that IVIM enables simultaneous assessment of tissue diffusion (Dt), pseudo-diffusion (Dp), perfusion fraction (f), and blood flow (f×Dp)⁴. However, previous studies that have explored the relationships between perfusion-related parameters measured by IVIM and DCE-MRI in a range of tumors have produced contradictory results⁵. The biophysical interpretation of the IVIM parameters f and f×Dp and their relationship to the DCE parameters Fb and vb remain controversial and require further investigation. This study aims to assess the correlations between these parameters in a cohort of patients with breast cancer imaged before and after one and two cycles of neoadjuvant chemotherapy.

Methods

The study comprised 40 patients with primary breast cancer (mean age 46, range 25-69) due to undergo NACT. Patients were imaged on a 1.5 T MRI scanner (Aera; Siemens) before treatment, following one and three cycles of NACT. The scanning sequences included DWI with 6 b-values (0, 50, 100, 200, 400 and 800 s/mm²), a 3D IR-prepared spoiled gradient echo sequence to estimate T1 values (pre-and post-contrast) and interleaved high spatial and high temporal resolution (HSR and HTR , 2 s per volume) 3D spoiled gradient echo DCE sequences for tumor delineation and tracer kinetic analysis, respectively^6,7. In addition to a 16-channel breast coil, a flexible array coil placed on the back was used to increase the signal from the descending aorta to enable accurate measurement of the arterial input function (AIF)⁸. Gd-DOTA was administered (0.1 mmol/kg) followed by saline (20 ml) at a rate of 3 ml/s.

All MRI data were processed via in-house programs developed in MATLAB (Mathworks, USA). The DWI images (including ADC maps) were aligned to the corresponding HTR images and HTR and HSR subtraction images were generated. For each patient, a whole-volume region of interest (ROI) was generated using a 3D seeded region growing algorithm based on the threshold signal intensity (SI) of enhancing tumor in the HSR subtraction images. To reduce the possibility of tumor heterogeneity compromising the analysis, two sub-ROIs (5×5 pixels) of the whole-volume ROI were generated; the region with the lowest values on the ADC map (cold-spot ROI) and the region with the highest SI on the HTR subtraction (hot-spot ROI). All 3 ROIs were copied to the corresponding DWI, ADC, IR and HTR images.

For DCE, these three ROIs were used to generate SI-time curves from the HTR images and estimate T1 relaxation-times from the IR data. Further ROIs were drawn in the descending aorta for generating SI-time curves and estimating T1 before and after Gd-DOTA administration for AIF calculation⁸. SI-time data were converted to Gd-DOTA concentration-time using a bookend T1 correction^8,9. A two-compartment exchange model was fitted to the DCE data and Fb, vb, capillary permeability-surface area product (PS) and extracellular-extravascular volume fraction (ve) were estimated¹⁰.

For DWI, the mean SI for each b‑value was extracted from the three tumor ROIs and fitted using a monoexponential and an IVIM model. The IVIM parameters were estimated using an over-segmented approach (estimating Dt and f from the high b-value data, then fixing Dt and f and estimating Dp)¹¹. A Spearman rank test was used to assess the correlation between the diffusion and DCE parameters.

Results

When data from all 3 visits were combined there were moderate positive correlations between the diffusion parameters ADC & Dt and the DCE parameter ve estimated in whole-tumor ROIs (r = 0.493 and 0.505 for ADC & Dt, respectively; n = 74, P<0.001) (Figure 1). However, there were no significant correlations between the IVIM parameters f & f×Dp and the DCE parameters Fb & vb. Significant correlations between these parameters were only seen in selected subsets of the combined data (e.g., a moderate correlation between f and Fb following cycle 1 in hot-spot ROIs; r = 0.559, n = 32, P = 0.001, Figure 2). This correlation was not maintained across other tumor ROIs or treatment time-points

Discussion and conclusion

Our findings of the moderate correlations between the diffusion parameters (ADC, Dt) and ve suggest that water diffusion increases as the interstitial fluid volume increases, an expected characteristic of the diffusion signal but one that has been challenged¹². However, our data did not identify any consistent correlations between perfusion parameters estimated by IVIM and DCE. We recognize that the precision with which parameters such as vb and f×Dp are estimated is poor but estimates of f and Fb are more precise and our data support the suggestion that a single diffusion coefficient may not reflect the complex diffusion properties of the vascular signal¹³.

Acknowledgements

The study was funded by Breast Cancer Now (award 2014MayPR241).

References

1. Shenoy H, Peter M, Masannat Y, Dall B, Dodwell D, Horgan KJSO (2009) Practical advice on clinical decision making during neoadjuvant chemotherapy for primary breast cancer. Surgical Oncology 1:65-71