Background

Immunotherapy has changed the landscape of metastatic melanoma treatment improving survival¹. Due to the mechanism of action, atypical response patterns have been described including disease shrinkage after initial increase in tumour burden and shrinkage after appearance of new lesions; the reported incidence of ‘pseudoprogression’ is up to 10%². Conventional size-based response criteria such as RECIST 1.1³ can lead to premature termination of treatment.

Whole-body MRI (WB-MRI) includes morphological and functional images. Compared with CT this offers increased sensitivity for disease detection at sites such as brain, connective tissues, bones, liver and peritoneum. The inclusion of diffusion-weighted imaging (DWI), also has the potential to provide a more accurate assessment of treatment response⁴. The DWI technique enables the calculation of the apparent diffusion coefficient (ADC), which reflects tissue microarchitecture including tissue cell density, hence giving an indication of changes on a cellular scale. Calculation of ADC is affected by the inherently low signal-to-noise-ratio (SNR) of DWI; therefore, it is important to optimise b-values to minimise the effects of noise.

Purpose

In order to develop a WB-MRI protocol for patients with metastatic melanoma undergoing treatment, the main aim of this study is to establish optimal b-values for calculating ADC of metastatic disease and to establish the distribution of ADC.

Methods

Patients:
This was a retrospective single-institution study with local institutional review board approval. Eleven patients with previously treated metastatic melanoma underwent MRI including DWI: 4 patients had baseline and post-immunotherapy scans; 6 patients had baseline scans only; one patient had a single post-treatment scan.

Imaging:
Images were acquired using 1.5T and 3T Siemens scanners (Aera, Avanto and Skyra) with 8 patients scanned at 1.5T and 3 patients scanned at 3T. Of these patients, 5 were scanned with whole body coverage and 6 with abdominal and pelvic coverage. Axial DWI was obtained with b-values = 50,600,900 s/mm² for 5 patients, b-values = 50,600,1050 s/mm² for 5 patients and b-values = 50,600,900,1050 s/mm² for one patient. ADC maps were calculated from these images using a mono-exponential decay model.

Analysis:
A minimum tumour diameter of 1 cm was chosen for regions of interest (ROIs) to avoid partial volume effects. ROIs were drawn around the tumour on the central slice on b = 50 s/mm² images and a mask was created from this and applied to the ADC map to extract the corresponding ADC values of each voxel in the tumour, as shown in Figure 1. It is possible to show that two b-values should be acquired with an optimal separation, b_opt, which can be found by solving^{5, 6} :
$$b_{opt}D_{0} ≈ 1.25$$
where D₀ represents the ADC of the tissue. An optimal low b-value is 0 s/mm² but many authors choose 50 s/mm² to reduce unwanted flow artefacts⁷, and the number of signal averages for low/high b-values should have ratio 1:3.

The distribution of ADC estimates was assessed for (i) all tumours combined, (ii) according to whether the scan was acquired as a baseline or post-treatment, and (iii) according to tumour location. The distributions were weighted by 1/L, where L is the number of voxels in the tumour, in order to remove bias towards larger tumours.

An independent t-test was performed on mean tumour ADC values from the baseline and post-treatment scans, not assuming equal variance in the two distributions.

Results and Discussion

The mean ADC values and associated optimal b-value separations are presented in Table 1. There is a large range of values for all locations of metastases and the mean b_opt for both baseline and post-treatment are high, indicating generally lower ADC values in this tumour type than observed in many other tumour groups⁸. Based on these results, it is suggested that to calculate the most accurate value of the ADC for metastatic melanoma, the protocol should include a low b-value of 50 s/mm² and a high b-value of 1250 s/mm², with the latter having thrice the number of signal averages as the former. An additional acquisition at 600 s/mm² may be valuable for the assessment of lesions presenting ADCs in the upper quartile.

The distribution of ADC values, weighted according to tumour size, are show in Figure 2. Mean ADC was higher in post-treatment tumours compared with pre-treatment (p = 0.02, independent t-test, Figure 2.ii). For single-site distributions (Figure 2.iii), only ADC values derived from tumours in the bone and liver are displayed as sample sizes at the other locations are too small.

Conclusion

An optimal WB-MRI protocol for examination of patients with metastatic melanoma should use the optimal b-values 50 and 1250 s/mm². The practicalities of achieving robust imaging at b = 1250 s/mm² in a clinical setting need to be explored. ADC estimates may provide further evidence which suggests that average ADC could provide a potent response biomarker to treatment.

Acknowledgements

We acknowledge CRUK and EPSRC support to the Cancer Imaging Centre at ICR and RMH in association with MRC and Department of Health C1060/A10334, C1060/A16464 and NHS funding to the NIHR Biomedical Research Centre and the NIHR Royal Marsden Clinical Research Facility. This report is independent research funded 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.