PV QA 3 - Poster Viewing Q&A 3
TU_25_3113 - Multi-Parametric Imaging Defines Primary Tumor Subregions Enriched at Diagnosis and in Response to Therapy in Patients with Rhabdomyosarcoma
Tuesday, October 23
1:00 PM - 2:30 PM
Location: Innovation Hub, Exhibit Hall 3
John Lucas, MD, MS
St. Jude Children's Research Hospital
St Jude Children's Research Hospital: Assistant Member: Employee; University of Memphis: Adjunct Professor: Employee
Multi-Parametric Imaging Defines Primary Tumor Subregions Enriched at Diagnosis and in Response to Therapy in Patients with Rhabdomyosarcoma
J. T. Lucas Jr, C. Y. Hsu, B. McCarville, A. M. Davidoff, A. Pappo, B. L. Shulkin, and M. Krasin; St. Jude Children's Research Hospital, Memphis, TN
Purpose/Objective(s): Studies evaluating the utility of anatomic (CT or MRI) & metabolic (FDG-PET) imaging for response assessment in rhabdomyosarcoma have shown conflicting results. In light of the uncertain role of imaging in defining prognosis and characterizing response, we evaluated the potential of multi-parametric imaging to define distinct phenotypic primary (1o) tumor subregions in patients with rhabdomysarcoma at diagnosis & in response to therapy.
Materials/Methods: 56 patients with newly diagnosed rhabdomyosarcoma underwent multimodality imaging including 11C-Methionine PET/CT, & MRI (T1+Gd, T2, ADC) at diagnosis, & response evaluations at weeks 6, & 12 or 19. 18FDG-PET/CT was performed only at diagnosis. The 1o tumor was serially segmented in an experimental cohort composed of the first 19 patients with local Group III disease at all available time points (N=53, N=4 not collected). The 1o tumor was split into discrete supervoxels based on relative intensities from co-registered imaging. Intensity values from supervoxels from each tumor, patient & time point underwent unsupervised hierarchical clustering to identify groups of supervoxels with similar intensity profiles which represented well-defined 1o tumor subregions. The proportion of 1o tumor represented by each subregion was correlated with clinical features & response to therapy.
Results: Unsupervised clustering revealed six 1o tumor subregions with distinct imaging profiles. FDG & MET avidity at diagnosis were concordant across all but one tumor subregion which was characterized by facilitated diffusion & T2 hyper-intensity. MET & FDG avidity were highest in a subregion with restricted diffusion & reduced enhancement on T1Gd. The prior predominated at the 1o site in 100% of patients who had metastatic disease at diagnosis. The proportion of 1o tumor defined by each tumor subregion showed no relationship to histology, site, size, or nodal status. At week 6, in the non-parameningeal patients who had received chemotherapy alone, only 3 of 6 tumor subregions predominated (N=12), while only one tumor subregion (hypo-intense on T1 & T2WI with facilitated diffusion) predominated after the 12 or 19 week time point (10 of 17 patients) suggesting enrichment for these regions after systemic therapy. In 2 patients with parameningeal tumors who were imaged post-RT, only one cluster predominated (T1, T2 hyper-intense, low MET avidity, facilitated diffusion). Validation of the inter-relationship between imaging defined subregions, tumor response & risk category is ongoing.
Conclusion: MET & FDG highlight a distinct tumor subregion which corresponds to an area of restricted diffusion which be enriched in those with metastatic dissemination. The incorporation of multimodality imaging may improve our understanding of treatment response and targeting & faclitate risk adapted local & systemic therapy stratification.
Author Disclosure: J.T. Lucas: Employee; University of Memphis. C. Hsu: None. B. McCarville: None. A. Pappo: None. B.L. Shulkin: None. M. Krasin: None.