(P 221) EARLY RESULTS IN NOVEL USE OF CONTRAST-ENHANCED SUSCEPTIBILITY-WEIGHTED IMAGING IN THE ASSESSMENT OF RESPONSE AND PROGRESSION IN DESMOID FIBROMATOSIS: PILOT STUDY IN A SPECIALIZED CANCER INSTITUTION

Objective: To provide initial evidence outlining the novel use of CE-SWI as a single MRI sequence capable of providing insight regarding the underlying biological changes of responding and progressing desmoid tumors using an improved tridimensional volume assessment allowing improved separation of T2 hypointense collagenized mature tumor from T2 hyperintense, T1 shortened enhancing, immature, progressive tumor components.

Methods: An IRB-approved waiver of consent was obtained for this retrospective study, including an initial pilot analysis of ten single-lesion extremity desmoid fibromatosis patients undergoing standard-of-care MRI examinations in our institution. The scans included advanced imaging sequences, including diffusion-weighted imaging (DWI/ADC), perfusion-weighted imaging with dynamic contrast enhancement (PWI/DCE), and contrast-enhanced susceptibility-weighted imaging (CE-SWI), performed between March 2021 and May 2023.
CE-SWI and T2-STIR data was collected from each patient across multiple time points in their treatment. A three-dimensional manual tumor segmentation was performed in 48 VOIs by an imaging-specialized research assistant and a 27-year-experienced skeletal radiologist using MIM commercial software.

Data obtained from each VOI included maximum diameter, volume, and modified Choi (mChoi) measurements from CE-SWI and T2-STIR sequences. In-house developed software (CARPI-AF: Cancer Radiomic and Perfusion Imaging Automated Framework) was used to automatically extract five first-order radiomic features (mean, skewness, kurtosis, 10th and 90th percentiles) from the SWI and T2 images. Patient response was assessed using the conventional Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 as a reference standard and compared against T2-STIR and CE-SWI volumetric, mChoi, first-order radiomic features, and routine radiologic reporting (RRR). Thresholds for progression and response were set at 20% and 30% for unidimensional RECIST and volumetric mChoi assessments. An increase of 25% and a decrease of 50% were considered thresholds for progression and response, respectively, for 3-D volumetric assessment. mChoi values were estimated as a normalization quotient obtained by dividing an entire desmoid lesion's mean 3D intensity value by the adjacent standard muscle intensity value.

Results: This study included eight female and two male patients with an average age of 42 years (range 19-61 years). Five patients were treated with Sorafenib, two with Pazopanib, and three were undergoing active surveillance without therapy. Of the ten patients, two (2) were considered true progression by T2-based RECIST, while the other eight (8) were considered stable. No true T2-RECIST-based positive responses were included in the pilot study.

True T2-based RECIST progression: Two patients displayed true progression by the T2-based RECIST reference standard. This same result was detected by CE-SWI volume and mChoi at an average of 4.5 months earlier than T2-based RECIST. Both cases were assessed by routine radiologic reporting (RRR) as progression, at the same time as CE-SWI volume and mChoi. In true progression cases, SWI detected an average of 23% more voxels within the 90th percentile relative to T2, while T2 demonstrated an average of 8.5% more voxels within the 10th percentile than SWI.

T2-based RECIST Stability:
RECIST Stable with discrepant progression: Four patients presented as stable using T2-based RECIST. Progression was determined in these patients by an increase greater than 25% of both T2-based volume (in three patients) and CE-SWI-based volume (in two patients). Routine radiologic reporting (RRR) determined only one of the four to be progression.

RECIST stable with a discrepant positive response: Four patients displayed stable T2-based RECIST with discrepant positive response predicted by CE-SWI mChoi (in three patients) and T2 mChoi (in three patients). All cases assessed by mChoi as responding patients were also considered positive responses by routine radiology reporting (RRR).

Conclusion: This pilot study suggests that radiologists (RRR) assessing desmoid tumors over time are relatively insensitive to volumetric tumor changes that precede RECIST detectable progression. While, they tend to be better tuned with changes in T2 hypo intensity as a measure of collagenization and T1 shortening-enhancement as a measure of progressive disease, which is manifested in parallel with the mChoi values, with respect to both positive mChoi responses in stable RECIST and to mChoi progression in RECIST progression.
All eight cases of T2-RECIST stability (100%) had a discrepant evaluation of either progression by volume or positive response by mChoi, suggestive of the insensitivity of RECIST to detect the biologic changes displayed by desmoid tumors undergoing systemic treatment or active surveillance.
Trends, including a higher % of hyperintense voxel capture (90 percentile) and earlier detection of response by mChoi, suggest an overall performance advantage of CE-SWI volumetric and mChoi assessments over conventional T2-weighted equivalents and traditional RECIST.
The expected trend of increasing 10th percentile hypointense voxels, decreasing 90th percentile hyperintense voxels, decreasing mean, or increasing skewness in responding patients, and the opposite in progressive disease, was present at least in one of the four features [Table 1] in nine patients (90%), while all four combined, only in three (30%). Such results align with the statistical power expected from a small sample, outlining the need for an extension to a larger population analysis. Our early results suggest that the novel use of CE-SWI-based volumetric and mChoi measurements could improve the prediction of response/progression in desmoid tumors by providing a better assessment of multi-dimensional tumor size and increased discrimination between the collagenized mature component and the enhancing immature components, respectively predominant in mature-responding and immature-progressive disease.