Abstract 2497: Modeling the hidden danger of invisible tumor growth in the clinic

Abstract Tumor dormancy, relapse, and resistance are not fully understood. Residual and resistant tumor cells often persist and remain invisible for months or years, despite clean MRI/CT scans following neoadjuvant and/or adjuvant systemic therapies. In current literature reviews we see that residual tumor sizes and tumor detection thresholds can vary widely, further complicating patient prognosis and risk stratification. To address this unmet need, we developed an algorithm to model the timing, frequency, and imaging detection limits of tumor measurements to augment tumor relapse prediction with improved accuracy for clinical application, especially when tumors are below the current detection limits in the clinic. To visualize how long it takes an invisible residual tumor to recur, we generated multiple tumor growth curves to model the remission and relapse time course of 1, 101, 102, 103, and 104 tumor cells, and calculated the time for these therapy-refractory invisible tumors to develop into large 3cm3 MRI/CT-visible tumors. Firstly, we used simple exponential growth curves to model an unimpeded cancer cell doubling time ranging from 2 days to 30 days. Secondly, we used the Spratt model, a modified logistic model, to model the complex tumor growth curves in vivo. Tumor doubling times were varied from 5 to 30 days, indicative of tumor aggressiveness, dynamic tumor/tumor microenvironment interaction, and other tempo-spatial-context-dependency. Treatment frequency and duration were varied as biweekly, monthly, and 6-month office visits to simulate the time course of tumor resistance development following multiple rounds of therapeutic interventions. Additionally, tumor remission and relapse were modeled utilizing smaller residual tumors with longer remission periods. Using these parameters, we generated multiple growth curves to model how invisible and resistant residual tumor sizes, ranging from 10-2 mm3 to 10-6 mm3, will determine the time of diagnosis to time of relapse. Our dynamic tumor progression models demonstrate a rapid tumor relapse time course to reach the detection threshold, highlighting the need for invisible residual tumor modeling for diagnosis and relapse prediction in the clinic. This work highlights the importance of minimizing residual tumor size and reducing drug resistance. We will develop an invisible tumor model as a companion prognostic machine learning tool to identify and quantify therapeutic efficacy, risk stratify patients, and improve outcome and survival in the future. Citation Format: Bryan Hawickhorst, Amy H. Tang. Modeling the hidden danger of invisible tumor growth in the clinic [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 2497.