Improvement in the Accuracy of PET Signals by Analysis of Textural Heterogeneity Parameters

Abstract

Introduction: PET-CT is emerging to be the most efficient tool for tumor diagnosis and therapy monitoring. Some newer techniques like analysis of tumoral textural heterogeneity via PET are proving to be more effective than conventional PET parameters for diagnosis, patient selection and treatment planning. We carried out this study to analyze the role of textural heterogeneity parameters in improving the specificity of PET scans. The study was divided in three parts. In the first part, we explored the role of textural features in FET-PET for early detection of pseudoprogression in high grade gliomas as timely detection of pseudoprogression is crucial for the management of patients with HGG. In the second part, the objective was to assess the predictive ability of tumor textural heterogeneity parameters from baseline 68Ga-PSMA PET for patient selection prior to 177Lu-PSMA therapy. This could prove essential for response prediction and risk stratification of patients before the start of therapy resulting in better treatment outcome. Purpose of the third part of study was to investigate the role of tumor heterogeneity in pre and post therapy 68Ga-PSMA scans for early response prediction and estimation of over-all survival in patients with 177Lu-PSMA therapy. Materials and methods: For distinction between PsP and actual tumor progress fourteen patients with HGG and suspected of PsP underwent FET-PET. A set of 19 conventional and textural FET-PET features were evaluated and subjected to unsupervised consensus clustering and cluster stability assessment. The nearest shrunken centroids method was applied to determine the most relevant features underlying each cluster. The final diagnosis of true progression vs. PsP was based on follow-up MRI using RANO criteria. In the second part of study, retrospective analysis of 70 patients with mCRPC was performed. Five PET based textural heterogeneity parameters (COV, entropy, homogeneity, contrast, size variation) were determined in baseline 68Ga-PSMA scan. Results obtained were then compared with clinical parameters including pre and post therapy PSA, alkaline phosphate, bone specific alkaline phosphate levels and ECOG criteria. Spearman correlation was used to determine statistical dependence among variables. ROC analysis was performed to estimate the optimal cutoff value and AUC. In the third part of study, retrospective analysis of 50 patients undergoing 177Lu-PSMA therapy was performed. Pre-therapy, mid-therapy and post-therapy scans were used for analysis. In addition to conventional parameters, 5 PET based textural heterogeneity parameters were determined. ROC and Kaplan-Meier analyses were used for response assessment, time to progression and survival. Results: For differentiation of PsP, three robust clusters were identified. None of the patients with PsP fell into cluster 2, which was associated with high values for textural FET-PET markers. Three out of 4 patients with PsP were assigned to cluster 3 that was largely associated with low values of textural FET PET features. In comparison, tumor-to-normal ratio (TNRmax) at optimal cut-off 2.1 was less predictive of PsP (negative predictive value 57% for detecting true progression, p=0.07 vs. 75% with cluster 3, p=0.04). Furthermore, patients in cluster 2 were associated with a comparably lower progression-free survival. In the second part of study, in bone lesions entropy showed a negative correlation (rs = -0.327, p = 0.006, AUC = 0.695) and homogeneity showed a positive correlation (rs = 0.315, p = 0.008, AUC = 0.683) with change in pre and post therapy PSA levels. Other parameters did not show statistically significant correlations. It suggested that the more heterogeneous the tumor was in PSMA expression the more responsive it was to PSMA therapy. For the third part of study, in bone lesions entropy, homogeneity and intensity variation (AUC 0.725, 0.679, 0.716 respectively) showed statistically significant ability to predict response prediction. Entropy showed highest statistically significant capability to evaluate disease progression and to predict survival. In pre-therapy analysis the lesions with higher textural heterogeneity showed better response to treatment, however after 3 therapies patients having lesions with persistently high textural heterogeneity showed poor prognosis and survival. Conclusions: Textural heterogeneity parameters helped in distinguishing PsP from actual progress thus plying an essential role in therapy planning and patient outcome. For 177Lu-PSMA therapy our study showed a potential for response prediction through one baseline Ga-68-PSMA scan only. It also predicted which patients could respond better to the therapy thus forming selection criteria for patients that can help in better treatment planning for individual patients. In analysis of pre and post therapy data for the third part of study, tumor heterogeneity analysis proved to be superior to the investigated conventional parameters, as an important predictive factor in determining the therapy response and overall survival of patients.