CT radiomic features of photodynamic priming in clinical pancreatic adenocarcinoma treatment

Abstract

Photodynamic therapy (PDT) offers localized focal ablation in unresectable pancreatic tumors while tissues surrounding the treatment volume experience a lower light dose, termed photodynamic priming (PDP). While PDP does not cause tissue damage, it has been demonstrated to promote vascular permeability, improve drug delivery, alleviate tumor cell density, and reduce desmoplasia and the resultant internal pressure in pre-clinical evaluation. Preclinical data supports PDP as a neoadjuvant therapy beneficial to subsequent chemotherapy or immunotherapy, yet it is challenging to quantify PDP effects in clinical treatment without additional imaging and testing. This study investigated the potential of radiomic analysis using CT scans acquired before and after PDT to identify areas experiencing PDT-induced necrosis as well as quantify PDP effects in the surrounding tissues. A total of 235 CT tumor slices from seven patients undergoing PDT for pancreatic tumors were examined. Radiomic features assessed included intensity metrics (CT number in Hounsfield Units) and texture analysis using several gray-level co-occurrence matrix (GLCM) parameters. Pre-treatment scans of tumor areas that resulted in PDT-induced necrosis showed statistically significant differences in intensity and texture-based features that could be used to predict the regions that did respond (paired t-test, response versus no response,p < 0.001). Evaluation of PDP effects on the surrounding tissues also demonstrated statistically significant differences, in tumor mean value, standard deviation, and GLCM parameters of contrast, dissimilarity and homogeneity (t-test, pre versus post,p < 0.001). Using leave-one-out cross validation, six intensity and texture-based features were combined into a support-vector machine model which demonstrated reliable prediction of treatment effects for six out of seven patients (ROC curve, AUC = 0.93). This study provides pilot evidence that texture features extracted from CT scans could be utilized as an effective clinical diagnostic prediction and assessment of PDT and PDP effects in pancreatic tumors. (clinical trial NCT03033225).

Keywords: CT texture analysis; pancreatic cancer; photodynamic therapy; radiomics.

Creative Commons Attribution license.

Figures

Figure 1.

Overview of texture analysis in assessment of photodynamic therapy treatment on PDAC patients. Each patient received a pre-treatment and a post-treatment CT scan. Pancreatic tumor segmentation was performed, and texture analysis was carried out to extract underlying features that showed significant differences after the treatment.

Figure 2.

Overview of PDT-induced changes observed between pre- and post-treatment CT scans. (A) Tumor size measured from 3D segmentation of tumors shows no significant change (paired t-test, p-value = 0.36). (B) All patients show a reduction in mean tumor CT number, indicating the tumor density has decreased after PDT treatment (paired t-test, p-value = 0.01). (C) Six out of seven patients show a reduction in standard deviation of tumor CT number, suggesting that the intratumor variation has decreased after treatment (paired t-test, p-value = 0.04). (D) Range of CT number is decreased for five out of seven patients which indicates the reduction of intratumor variation after PD (paired t-test, p-value = 0.47).

Figure 3.

Treatment-induced necrosis areas could be predicted by texture analysis in the Pre-Tx scan. (A) Visualization of tumor in Post-Tx (top row) and Pre-Tx (bottom row) CT scans. Observable necrosis area (PDT) in pink and remaining tumor region (PDP) in yellow are delineated in Post-Tx scans, which were then mapped onto Pre-Tx scans for texture analysis. GLCM Contrast feature for both regions in the Pre-Tx scans is illustrated to show that necrosis occurred at the area with lower contrast. (B) Other first-order texture features that are found to be statistically significant when comparing the PDT and the PDP regions in Pre-Tx scans (n = 40 tumor slices).

Figure 4.

Percentage of changes after light treatment between pre-treatment and post-treatment scans in the PDP regions of tumors, as compared to the normal liver and spleen values. For each patient, the percentage change in CT mean was reported for the non-necrotic regions of PDAC tumor (14 ± 7.3%), liver (5.6 ± 1.3%) and spleen (7.7 ± 3.3%). Mean CT number decreased in all ROIs from all patients, with the tumor regions consistently expressing the biggest reduction in CT number. This observation suggested that there were treatment-induced effects on the tumor CT numbers across the patient cohort.

Figure 5.

PDP effects after light treatment are seen with texture analysis. (A) A visualization of intratumor changes due to light treatment quantified by GLCM texture analysis. Means of tumor CT number before and after treatment are 74 ± 23 HU and 67 ± 17 HU, respectively. Tumor homogeneity is increased, while levels of contrast and dissimilarity are decreased, indicating a more uniform tumor attenuation profile. (B) Texture features that showed a statistical significance between Pre-Tx and Post-Tx scans are reported (n = 235 tumor slices) (C) Mean ROC curve for each reported feature calculated for all patients except for Patient 01, showing AUC values in the range of 0.75–0.80. (D) Mean ROC curve of combined classifiers using SVM model in blue yielded a better performance than any single classifier (AUC value is 0.93 ± 0.07). Leave-one-patient-out cross validation ROC curves are showed for all patients except Patient 01, and the shaded area indicated the ± 1 standard deviation of this cross validation.