Introduction

Magnetic resonance fingerprinting (MRF) generates co-registered quantitative maps from a single acquisition¹. These acquisitions are faster than typical quantitative MR measurements and may re-open the way to using T₁ and T₂ measurements for assessment of tumours and normal tissues.

Aim

This prospective study evaluates MRF-derived measures of treatment-induced T₁ and T₂ changes in prostate cancer patients with metastatic bone disease, by comparison with existing quantitative T₁ and T₂ measurements.

Patient cohort

A cohort of 13 patients with metastatic prostate cancer (mean age: 68 years, range: 57-76) with focal active pelvic bone lesions were recruited according to an ethics and research board approved protocol and were scanned before and after the start of a new line of treatment. The follow-up scan was usually performed after an average of 3.3 months (range: 2-6 months).

Methods

All imaging was performed on a 1.5T MAGNETOM Aera scanner (Siemens Healthcare, Erlangen, Germany) using a prototype MRF Fast Imaging with Steady State Precession (FISP) sequence² with Gadgetron reconstruction³ which yields maps of T₁ and T₂ relaxation times, and proton density M₀. Established sequences were utilised as comparative techniques for T₁ (inversion recovery turbo spin echo, IR-TSE) and T₂ (double spin echo, DSE) measurements. These sequences were validated against the ISMRM/NIST MRI system phantom (CaliberMRI, Inc, Boulder, CO), and were accurate to within 9% (T₁) and 12% (T₂) over the T₁ and T₂ ranges observed in the patients in this study. The MR parameters of MRF and standard quantitative T₁ and T₂ sequences were matched as best as possible, see Table 1. To limit the total acquisition time to 15 minutes, the T₁ (5 inversion times) and T₂ measurements (2 echo times) were acquired from a single axial slice covering the biggest active lesion in the pelvis. The MRF sequence covered 3 slices, with its central slice having matched location and field-of-view to the other measurements slice.

The MRF, IR-TSE and DSE added 15 minutes to the end of routine clinical scans that included diffusion-weighted imaging (DWI) and T₁-weighted Dixon sequences. DWI and Dixon sequences were used to identify the active target lesion (high signal on b=900 s/mm² images, low signal on apparent diffusion coefficient map, and low signal on post-processed fat-fraction Dixon images). No contrast agent was administrated during the scanning session.

Regions of interest (ROIs) drawn on the MRF M₀ image (using the DWI and Dixon images as a visual guide) were copied to MRF T₁ and T₂ images, as well as to the other T₁ and T₂ maps, which were calculated using an in-house software (MATLAB R2019a, The MathWorks, Inc.). Lesion ROIs were drawn for each patient, at each visit, using Horos software (Horosproject.org) and saved to DICOM-RT using a pyOsiriX plugin⁴. ROI-derived statistics (median and standard deviation values) for each map and Spearman correlation tests were obtained using MATLAB. Spearman correlation coefficients were considered according to the following scale: weak (0.0-0.39), moderate (0.4-0.59), strong (0.6-0.79) and very strong (0.8-1).

Results

Figure 1 shows example MRF- and IR-TSE T₁ and DSE T₂ maps acquired on the same patient before (first row) and after treatment (second row) with overlaid contours delineating the selected active lesion.

Cohort summary results (mean and standard deviation values) are plotted in Figure 2, showing no significant statistical differences between IR-TSE and MRF T₁ measurements at both time-points, while T₂ measurements showed significant differences at both time-points.

Figure 3 compares the treatment-related changes induced in pelvic bone metastases observed using the MRF and existing quatitative methods, where a very strong correlation was found for T₁ relative changes (Spearman R=0.84) and a moderate correlation for T₂ relative changes (R=0.46). In addition, 9/13 patients had relative MRF T₁ changes outside the limits of agreement (LoA) values previously obtained using this imaging sequence⁵, while 4/13 patients with significant changes were detected with MRF T₂.

Discussion

This cohort included patients with a high burden disease who had exhausted several lines of treatment, and so a variety of T₁ and T₂ treatment responses (magnitude and trend) were expected in this group. Nevertheless, the same trend of increased T₁ and T₂ values post-therapy was observed with both standard and MRF methods across the cohort (Figure 2). Whilst the range of relative changes in T₁ and T₂ measures were similar (0.87 – 1.2 for T₁; 0.84 – 1.3 for T₂), the tighter limits of agreement on the T₁ measures (relative repeatability coefficient: 6.2% for T₁, 28.4% for T₂) meant that more patients had significant changes detected with T₁ than with T₂. This is consistent with previous evaluations of the repeatability of T₁ and T₂ measures obtained with MRF in other tissue types⁶.

Conclusion

This study demonstrated a good correlation of MRF-derived T₁ and T₂ measurements with existing quantitative methods, supporting the use of MRF for faster measurements in bone lesions.

Acknowledgements

We acknowledge the support of NHS funding to the NIHR Biomedical Research Centre and NIHR Royal Marsden Clinical Research Facility. This report is independent research funded partially by the National Institute for Health Research. The views expressed in this publication are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health.