Amygdalar activity measured using FDG-PET/CT at head and neck cancer staging independently predicts survival

Journal article

Aim/Introduction: The process of determining time-activity curves (TACs) for radiopharmaceutical therapy (RPT) relies heavily on user- and site-specifc steps, impacting time-integrated activity (TIA) and, efectively, absorbed dose calculations. Despite TIA’sclinical signifcance, there is no consensus on data processing methods nor an understanding of how user-dependent TAC calculation afects personalized RPT dosimetry. In 2023, the TACTIC AAPM Grand Challenge was created to address these challenges. This work presents results and insights from the challenge. Materials and Methods: Launched in January 2023, the TACTIC challenge consisted of three phases: warm-up (P0), individual patient-based TAC ftting (P1), and population-based TAC ftting (P2). Participants were provided with pre-processed synthetic biokinetic data of [177Lu]Lu-PSMA-617 (kidney, blood, and tumor) and tasked with modeling the TAC and calculating TIA values for each target organ. Additionally, participants submitted information about the TAC type and parameters used for ft optimization. The best-performing team in P1 and P2 was determined by the lowest total root mean squared error (RMSE) error over the organs. Results: A total of 132 individuals from over 30 countries registered for the challenge, representing a diverse mix of highly experienced dosimetry groups, industryprofessionals, and newcomers to RPT dosimetry. Among them, 73 participants requested data, of which 35 (P0), 35 (P1) and 28 (P2) submitted their results. Across the three phases, 13 diferentft functions were utilized, with varying advanced model selection criteria and levels of uncertainty incorporation. Notably, the biexponential function was most prevalent, utilized in 51% (P1) and 32% (P2) of submissions, while the least square objective functionwas the primary choice for 40% of submissions (P1). Despite the challenge’s nature, only a minority of participants—6 in P1 and 8 in P2—incorporated uncertainty budgets into their TIAC calculations. Population-based information was utilized in only 7 submissions during P2. Interestingly, no correlations were found between choice of ft function, objective function, uncertainty i ncorporation, or population information use and participants’ performance. Winners in each phase employed diverse models and objective functions. However, the top-performing participants consistently integrated uncertainty information when selectingthe most suitable TAC model. A decrease in some participants’ performance from P1 to P2 when including uncertainty or population-based information suggests that more guidance and training is needed to use them efectively. Conclusion: The TACTIC challenge results ofer insights into global TAC modeling practices, revealing signifcant variations in result quality. This underscores the importance of education in TAC ftting methodologies

Authors: Hassan, MZO; Tawakol, A; Wang, Y; Alvi, RM; Awadalla, M

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: Public Library of Science
• Journal: PLoS ONE
• Volume: 18
• Issue: 8
• Pages: e0279235-e0279235
• Publication date: 2023-08-04
• DOI: 10.1371/journal.pone.0279235
• EISSN: 1932-6203
• ISSN: 1932-6203
• Language: English
• Keywords: Nuclear medicine; Retrospective cohort study; Internal medicine; Standardized uptake value; Medicine; Oncology; Head and neck cancer; Cohort; Positron emission tomography; Cancer