KEY POINTS
- This retrospective Maastro study evaluated an integrated patient-specific QA program covering 18,558 pretreatment plans from 2020–2024 and 1,817 plans undergoing in vivo verification in 2024. The system combined independent secondary dose calculations, machine log-file analysis, and electronic portal dosimetry performed before and during treatment.
- Pretreatment secondary dose calculation failed in 239 of 18,558 plans (1.29%). Of the total cohort, 72 plans (0.39%) showed AAA-related dose calculation inaccuracies, 39 (0.21%) involved omitted immobilization devices, and 26 (0.14%) showed Acuros dose underestimation in small jaw-tracked fields.
- Among 14,380 plans assessed using log files, 26 showed dose-delivery discrepancies. Pretreatment electronic portal dosimetry identified an additional 107 failed plans, often involving complex fluence patterns, demonstrating that calculations, machine records and independent measurements identify overlapping but nonidentical failure modes.
- In vivo CBCT-based dose reconstruction detected at least one failing fraction in 385 of 1,817 plans (21.2%). The most frequent clinically relevant issue was patient alignment error (166 plans; 9.1%), particularly in breast treatments, followed by systematic anatomical changes in 3.9% and random anatomical variations in 2.9%.
- Some failed QA results arose from measurement or reconstruction limitations rather than actual treatment errors. CBCT-related false positives occurred in approximately 5.1% of in vivo plans, particularly because of limited field of view, HU differences and prosthesis artifacts, emphasizing the need for physicist review before corrective action.
- Positive predictive values differed across QA methods: approximately 57% for planning-CT secondary dose calculation and 76% for CBCT-based in vivo dose reconstruction, while transit portal dosimetry showed lower and more variable values depending on failure category. Combining independent methods improved the interpretation of failures compared with relying on any single QA result.
- The integrated program detected nearly all assessed high-priority failure categories, but machine output miscalibration was not reliably detected, because the portal measurement workflow normalized away output variation. The authors therefore emphasize that patient-specific QA must remain complementary to independent machine QA and that gamma-based results alone do not directly quantify clinical dose consequences.
CLINICAL TAKEAWAY
A comprehensive patient-specific QA program should combine independent dose calculation, log-file verification and measured in vivo dosimetry because these methods detect different classes of treatment errors. This large institutional experience supports integrated QA and structured investigation of failures, but individual gamma thresholds and positive predictive values require local interpretation, and machine output verification remains essential.