Characterizing the spatiotemporal displacement of fiducial markers for cyberknife radiosurgery in liver cancer and identifying predictive factors

Fiducial markers are essential for accurate image-guided stereotactic body radiotherapy (SBRT) in hepatocellular carcinoma (HCC). However, post-implantation positional variation introduces geometric uncertainty. This study aims to quantify the spatiotemporal observed displacement of fiducial markers in HCC patients treated with CyberKnife SBRT and to identify predictive factors for significant displacement. We retrospectively reviewed data from 185 HCC patients who underwent percutaneous implantation of 518 fiducial markers between January 2018 and December 2022. Serial CT scans were acquired at baseline (Day 1) and on days 3, 5, 7, 10, and 14 post-implantation. Fiducial displacement was measured in three dimensions (craniocaudal, anteroposterior, mediolateral) relative to the baseline scan after rigid registration to vertebral bodies. Significant displacement was defined as a 3D vector displacement > 3 mm, encompassing true migration and geometrical uncertainties. To account for data clustering (multiple markers per patient), univariate and multivariable Generalized Estimating Equations (GEE) models were used to identify predictors of significant displacement. Internal validation was performed using bootstrap resampling. Furthermore, a dosimetric analysis was conducted to evaluate the clinical impact of significant marker displacement. The mean 3D vector displacement was greatest at Day 3 (2.8 ± 1.5 mm) and progressively decreased, plateauing after Day 7 (Day 7: 1.9 ± 1.1 mm; Day 14: 1.8 ± 1.0 mm; p < 0.001 for trend). The craniocaudal axis exhibited the largest magnitude of displacement. By Day 7, 88.6% of markers achieved stability (displacement ≤ 3 mm). Dosimetric evaluation revealed that simulated displacement > 3 mm significantly degraded target coverage (PTV V100%: 98.5% vs. 92.1%, p < 0.001). Multivariate GEE analysis identified tumor location within 2 cm of the diaphragm (Odds Ratio [OR] = 3.05, 95% CI: 1.75–5.20, p < 0.001) and tumor diameter > 5 cm (OR = 1.98, 95% CI: 1.15–3.30, p = 0.012) as independent predictors for significant fiducialisplacement. The optimism-corrected area under the curve (AUC) for the predictive model was 0.76. Fiducial markers in HCC patients undergo significant early displacement, primarily within the first week, and stabilize thereafter. Proximity to the diaphragm and larger tumor size are associated with a higher risk of displacement, which can negatively impact delivered dosimetry in simulated scenarios. These findings support a recommended 7-day waiting period between implantation and CT simulation as a reasonable safety guideline and suggest that patients with high-risk features may benefit from extended monitoring or adaptive radiotherapy strategies.