2468 - MIRDrpt: A Radiopharmaceutical-Specific Dosimetry and Radiobiology Calculation Workflow for 177 Lu-DOTATATE with Preliminary Independent Code-Based Validation
Presenter(s)
V. Gopalakrishnan1, G. Kayal2, L. M. Carter2, A. Kesner2, and R. F. Hobbs1; 1Department of Radiation Oncology and Molecular Radiation Sciences, Johns Hopkins University School of Medicine, Baltimore, MD, 2Department of Medical Physics, Memorial Sloan Kettering Cancer Center, New York, New York, New York, NY
Purpose/Objective(s): Radiopharmaceutical therapy (RPT) is increasingly adopted in radiation oncology, however patient-specific dosimetry lacks the standardization established in external beam radiation therapy. Reproducible cross-institutional/industry implementations are essential for multicenter trials where dosimetry informs stratification, endpoints, or dose-escalation protocols. MIRDrpt, developed within the MIRDsoft community software suite, provides a structured calculation workflow integrating organ and tumor absorbed dose (AD), biological effective doses (BED) and equieffective doses (EQDx) to support personalized, benchmarkable RPT dosimetry. We hypothesized that MIRDrpt would demonstrate high concordance with an independent Python implementation, emphasizing numerical accuracy and model-agreement.
Materials/Methods: MIRDrpt accommodates single- and multi-timepoint dosimetry for liver, kidneys, spleen, salivary glands (PSMA-targeted agents) and bone marrow (blood-/image-based) with support for up to 5 tumors, five SPECT/CT timepoints and ten ex-vivo blood measurements per administration. Akaike information criterion-guided fitting (trapezoidal, mono-, bi-exponential) generates time-integrated activity coefficients (TIACs) driving mass-scaled S-value calculations. Optional recovery-coefficient based partial-volume correction and sphere-model tumor self-doses are included. BED and EQDx are derived from dose-rate (ADR) curves using configurable and repair parameters. Validation applied an independent Python engine, developed at another institution, using representative SPECT/CT data with identical inputs. Agreement in TIAC, AD, BED, and EQD2 was quantified by median absolute percent difference (MAPD), stratified by dosimetry method and fitting model.
Results: Independent validation reproduced MIRDrpt across key dosimetry outputs. Mono-exponential fitting yielded TIAC/AD/BED/EQD2 agreement within < 2%. Bi-exponential fitting produced agreement = 11% across most VOIs; bone marrow differed by approximately 22%, attributable to model selection and fit algorithmic differences. ADR differences were < 1% per timepoint, with BED/EQD2 deviations mirroring AD-level differences.
Conclusion: MIRDrpt demonstrates strong concordance with an independent implementation for 177Lu-DOTATATE. Differences mainly arose from time-activity curve model selection, and propagated into AD, BED, and EQD2. Validation highlighted the importance of fit algorithm agreement, a factor that is not currently standardized across radiopharmaceutical software options. MIRDrpt aims to address a critical gap by providing a freely available, validated platform for standardized RPT dosimetry. Pending MIRD Committee endorsement, planned extensions to additional radiopharmaceuticals will further advance cross-institutional consistency and traceability in personalized dosimetry.