In advanced epithelial ovarian cancer (EOC), peritoneal and omental deposits are often sub-centimetre, ill-defined or confluent in nature on T₂-weighted imaging,¹ which makes their response to chemotherapy difficult to assess by Response Evaluation Criteria In Solid Tumours version 1.1 (RECIST) criteria. Preliminary single-centre studies have demonstrated the potential role of DW-MRI in EOC, not only because of its superior soft-tissue contrast in defining peritoneal disease extent but also as a quantitative biomarker for early treatment response assessment.^2,3 However, the utility of the technique has not been exploited in relapsed disease for defining response by lesion volume reduction or for determining the time-course of apparent diffusion coefficient (ADC) changes indicative of response when re-challenged with platinum-based chemotherapy.To evaluate change in volume and ADC in lesions classified by RECIST criteria as responding and non-responding in patients with relapsed EOC or primary peritoneal cancer (PPC) re-challenged with platinum-based chemotherapy.

Study protocol: 21 patients with relapsed EOC or PPC, scheduled to receive platinum-based chemotherapy, were recruited from three institutions as part of an ongoing prospective multi-centre clinical trial (DISCOVAR, NCRN portfolio number 11182) and gave their written consent to participate. Each patient underwent an MRI prior to commencement of chemotherapy and following their first and third cycles of chemotherapy.

Imaging protocol: Following administration of hyoscine butylbromide (20mg i.m.), free-breathing axial single-shot echo-planar DW-MRI (b-values 0,100,500,900smm^-2, slice thickness 6mm, pixel size 3x3mm) and slice-matched T₁-weighted and T₂-weighted anatomical imaging were acquired in the abdomen and pelvis on a 1.5T scanner.

Analysis: Up to ten lesions were analysed per patient using in-house software.⁴ Regions of interest (ROI) were drawn by region growing around lesions (Figure 1) on consecutive computed high b-value axial DW-MRI images (b=1000smm^-2 derived from b-values:100,500,900smm^-2) encompassing the whole lesion, generating a volume of interest (VOI). The ROI included regions of impeded diffusion arising from residual tumor. Cystic or necrotic areas were excluded by visual matching with the morphological MRI sequences. ADC parameters (median, 25th percentile) and volume of each lesion were calculated from the individual pixel ADC values and total pixel number respectively within the VOI on the baseline and post-treatment imaging. The post-treatment change after one and three cycles of chemotherapy was expressed as percentage change from baseline or previous cycle, and paired t-tests were performed taking p<0.05 to indicate a statistically significant difference. Per-lesion chemotherapeutic response was determined on the morphological MR images at baseline and after three cycles of chemotherapy according to RECIST criteria, whereby a reduction in longest diameter of at least 30% denoted response.⁵ Per disease site ADC histograms in RECIST responding and non-responding lesions were produced after summation of individual pixel ADCs of all lesions falling into each disease site and response group. The histograms were normalised for total disease burden to account for changes in volume post-treatment.

55 lesions were classified as responding on RECIST (42 peritoneal nodules, 12 lymph nodes, 1 liver metastasis), and 28 as non-responding (25 peritoneal nodules, 3 lymph nodes). Three lesions were non-measurable after the first cycle of chemotherapy, and a further nine were non-measurable after the third cycle of chemotherapy and were excluded from the analysis at these timepoints. Volume decrease after the first cycle of chemotherapy was five-times greater in responding lesions than non-responding lesions, this difference reduced to three-times after three cycles. The percentage increase in median ADC after the first cycle of chemotherapy in residual tumor from responding and non-responding lesions was equivalent. However, the rate of volume reduction in responding lesions slowed after cycle one, and ADC values in residual tumor did not change further. In contrast, the volume in the non-responding lesions continued to decrease at the same rate between cycle one and three and the level of ADC change in non-responding lesions was sustained after three cycles of chemotherapy indicating a continued delayed response. On histogram analysis, responding peritoneal disease showed a greater shift to the right of the ADC histogram (Figure 2) than nodal disease (Figure 3).In relapsed EOC/PPC, lesions classified as “responding” by RECIST criteria show greater change in volume and equivalent change in ADC to RECIST “non-responding” lesions after one cycle of chemotherapy. However, in “non-responding” lesions, the change in these parameters continued at the same rate after the first cycle of chemotherapy, indicating a delayed response. ADC histograms of individual lesions may be a more sensitive biomarker of response in peritoneal lesions than nodal disease and may be useful to evaluate differences in site-specific response. DW-MRI may be useful in longitudinal studies to detect platinum resistant lesions early.