Oden Institute for Computational Engineering and Sciences, The University of Texas at Austin, Austin, TX, 78712, USA; Department of Imaging Physics, The University of Texas MD Anderson Cancer Center, Houston, TX, 77030, USA; Department of Breast Imaging, The University of Texas MD Anderson Cancer Center, Houston, TX, 77030, USA; Department of Biostatistics, The University of Texas MD Anderson Cancer Center, Houston, TX, 77030, USA; Institute for Data Science in Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, 77030, USA.
Oden Institute for Computational Engineering and Sciences, The University of Texas at Austin, Austin, TX, 78712, USA; Livestrong Cancer Institutes, The University of Texas at Austin, Austin, TX, 78712, USA.
Comput Biol Med. 2024 Sep;179:108889. doi: 10.1016/j.compbiomed.2024.108889. Epub 2024 Jul 19.
Proper catheter placement for convection-enhanced delivery (CED) is required to maximize tumor coverage and minimize exposure to healthy tissue. We developed an image-based model to patient-specifically optimize the catheter placement for rhenium-186 (Re)-nanoliposomes (RNL) delivery to treat recurrent glioblastoma (rGBM).
The model consists of the 1) fluid fields generated via catheter infusion, 2) dynamic transport of RNL, and 3) transforming RNL concentration to the SPECT signal. Patient-specific tissue geometries were assigned from pre-delivery MRIs. Model parameters were personalized with either 1) individual-based calibration with longitudinal SPECT images, or 2) population-based assignment via leave-one-out cross-validation. The concordance correlation coefficient (CCC) was used to quantify the agreement between the predicted and measured SPECT signals. The model was then used to simulate RNL distributions from a range of catheter placements, resulting in a ratio of the cumulative RNL dose outside versus inside the tumor, the "off-target ratio" (OTR). Optimal catheter placement) was identified by minimizing OTR.
Fifteen patients with rGBM from a Phase I/II clinical trial (NCT01906385) were recruited to the study. Our model, with either individual-calibrated or population-assigned parameters, achieved high accuracy (CCC > 0.80) for predicting RNL distributions up to 24 h after delivery. The optimal catheter placements identified using this model achieved a median (range) of 34.56 % (14.70 %-61.12 %) reduction on OTR at the 24 h post-delivery in comparison to the original placements.
Our image-guided model achieved high accuracy for predicting patient-specific RNL distributions and indicates value for optimizing catheter placement for CED of radiolabeled liposomes.
为了最大限度地提高肿瘤覆盖率并最大限度地减少对健康组织的暴露，对流增强递送（CED）的导管放置必须正确。我们开发了一种基于图像的模型，用于针对铼-186（Re）-纳米脂质体（RNL）的递送来治疗复发性胶质母细胞瘤（rGBM），以优化导管放置。
该模型由 1）通过导管输注产生的流体场，2）RNL 的动态传输以及 3）将 RNL 浓度转换为 SPECT 信号组成。通过预递 MRI 分配患者特定的组织几何形状。使用以下两种方法之一对模型参数进行个性化处理：1）通过纵向 SPECT 图像进行个体化校准，或 2）通过留一法交叉验证进行群体分配。使用一致性相关系数（CCC）来量化预测和测量的 SPECT 信号之间的一致性。然后，使用该模型模拟来自一系列导管放置的 RNL 分布，从而产生肿瘤内外累积 RNL 剂量的比率，即“脱靶比”（OTR）。通过最小化 OTR 来确定最佳导管放置位置。
从 I/II 期临床试验（NCT01906385）中招募了 15 名患有 rGBM 的患者进行研究。我们的模型，无论是使用个体校准还是群体分配参数，都能在输送后 24 小时内准确预测高达 24 小时的 RNL 分布（CCC>0.80）。与原始放置相比，使用该模型确定的最佳导管放置位置在 24 小时后递送时，OTR 的中位数（范围）降低了 34.56％（14.70％-61.12％）。
我们的图像引导模型能够准确预测患者特定的 RNL 分布，并表明对于优化放射性标记脂质体的 CED 导管放置具有价值。
