2924 - PET Imaging for Modeling of Skeletal Related Events (SREs) in Patients with Asymptomatic High-Risk Bone Metastases
Presenter(s)
P. Yeh1, Y. Kim1, Y. Kim1, J. Leu2, A. G. Amin3, A. Som4, P. C. Thurlow4, H. Khan5, R. Raychaudhuri5, E. Y. Yu5, J. Kang1, M. H. Blau1, O. Y. Mian1, S. S. Lo1, A. I. Iravani4, D. Chen4, J. T. Yang6, E. F. Gillespie1, C. Grassberger1, and W. Vuong1; 1Department of Radiation Oncology, University of Washington/Fred Hutchinson Cancer Center, Seattle, WA, 2University of Washington School of Medicine, Seattle, WA, 3Department of Neurological Surgery, University of Washington/Fred Hutchinson Cancer Center, Seattle, WA, 4Department of Radiology, University of Washington/Fred Hutchinson Cancer Center, Seattle, WA, 5Division of Hematology and Oncology, University of Washington/Fred Hutchinson Cancer Center, Seattle, WA, 6Department of Radiation Oncology, NYU Grossman School of Medicine, New York, NY
Purpose/Objective(s): SREs are a significant source of morbidity in patients with metastatic cancer and predictive models are needed to identify potential SREs for early prevention. Asymptomatic high-risk bone-metastatic lesions (aHRLs) are suggested to be precursors of SREs. We investigated the hypothesis that among patients with prostate (PCa) or non-small cell lung cancer (NSCLC) who develop SREs, greater PET SUVmax of aHRLs is associated with development of SREs.
Materials/Methods: A single-institution retrospective analysis was conducted on patients with radiographic progression and aHRLs identified on PET imaging between Feb-Mar 2022. aHRLs were defined as bulky disease >2cm or lesions involving long bones, junctional/posterolateral spine (HR-spine), or shoulder/hip/sacroiliac joints. Patients were included if =1 SRE, defined as pathologic fractures (path fx), cord compression (ESCC), surgery for instability, or radiation for pain (RT), occurred =1 year after the assessed PET and excluded if any SRE was present at time of imaging. Patients were divided by PSMA-PET vs. FDG-PET. aHRLs were clustered by patient and modeled using univariate and multivariate Bayesian multilevel logistic regression, fitting lesion-specific SUVmax with covariates of age, gender, histology, and lesion location to predict occurrence of SRE within 1 year. Time-to-SRE was modeled using Cox regression.
Results: A total of 138 patients with PCa and NSCLC were screened within the study period. Of 77 patients with PET scans, 10 (prostate: 6, lung: 4; PSMA: 5, FDG: 5) developed =1 SRE within 1 year. 137 aHRLs were identified with 53 HR-spine, 51 bulky, 20 hip, 8 long bone, and 5 shoulder lesions. All SREs were symptomatic, occurring in 21/137 aHRLs (15.3%) including 11 RT events, 6 path fx, and 4 ESCC. SUVmax of aHRLs was a positive predictor for SREs within 1 year on both univariate and multivariate analyses (multivariate OR per unit SUVmax = 1.055 [PSMA-PET, 95% CI: 1.026-1.086] and 1.079 [FDG-PET, 95% CI: 1.007-1.157]). Time-to-SRE had a significant concordance with SUVmax in the PSMA cohort of 0.81 (HR per unit SUVmax = 1.05, p<0.005) but not in the FDG cohort (0.63, HR per unit SUVmax = 1.07, p=0.44).
Conclusion: This lesion-level study of aHRLs from PCa or NSCLC suggests that for a given patient, aHRLs with higher PET SUVmax may be more likely to develop symptomatic SREs within 1 year, although its impact on time-to-SRE may vary by tracer type, highlighting potential differences across sub-populations which warrants validation in the context of bone-modifying agent use, lytic vs. blastic lesions, and other potential confounders. Optimal patient selection for prevention of SREs remains an ongoing challenge although PET imaging may provide guidance with further refinement in larger cohorts.