Background: This study investigated the performance of ensemble learning holomic models for the detection of breast cancer, receptor status, proliferation rate, and molecular subtypes from [18F]FDG-PET/CT images with and without incorporating data pre-processing algorithms. Additionally, machine learning (ML) models were compared with conventional data analysis using standard uptake value lesion classification. Methods: A cohort of 170 patients with 173 breast cancer tumors (132 malignant, 38 benign) was examined with [18F]FDG-PET/CT. Breast tumors were segmented and radiomic features were extracted following the imaging biomarker standardization initiative (IBSI) guidelines combined with optimized feature extraction. Ensemble learning including five supervised ML algorithms was utilized in a 100-fold Monte Carlo (MC) cross-validation scheme. Data pre-processing methods were incorporated prior to machine learning, including outlier and borderline noisy sample detection, feature selection, and class imbalance correction. Feature importance in each model was assessed by calculating feature occurrence by the R-squared method across MC folds. Results: Cross validation demonstrated high performance of the cancer detection model (80% sensitivity, 78% specificity, 80% accuracy, 0.81 area under the curve (AUC)), and of the triple negative tumor identification model (85% sensitivity, 78% specificity, 82% accuracy, 0.82 AUC). The individual receptor status and luminal A/B subtype models yielded low performance (0.46–0.68 AUC). SUVmax model yielded 0.76 AUC in cancer detection and 0.70 AUC in predicting triple negative subtype. Conclusions: Predictive models based on [18F]FDG-PET/CT images in combination with advanced data pre-processing steps aid in breast cancer diagnosis and in ML-based prediction of the aggressive triple negative breast cancer subtype.
Few data exist regarding the prognostic value of L-[S-methyl-11C]methionine (MET) PET for treatment-naïve gliomas. A total of 160 glioma patients (89 men, 71 women; mean age: 45, range 18–84 y) underwent a MET PET prior to any therapy. The PET scans were evaluated visually and semiquantitatively by tumor-to-background (T/N) ratio thresholds chosen by analysis of receiver operating characteristics. Additionally, isocitrate dehydrogenase 1–R132H (IDH1-R132H) immunohistochemistry was performed. Survival analysis was done using Kaplan–Meier estimates and the Cox proportional hazards model. Significantly shorter mean survival times (7.2 vs 8.6 y; P = 0.024) were seen in patients with amino acid avid gliomas (n = 137) compared with visually negative tumors (n = 33) in MET PET. T/N ratio thresholds of 2.1 and 3.5 were significantly associated with survival (10.3 vs 7 vs 4.3 y; P < 0.001). Mean survival differed significantly using the median T/N ratio of 2.4 as cutoff, independent of histopathology (P < 0.01; mean survival: 10.2 ± 0.8 y vs 5.5 ± 0.6 y). In the subgroup of 142 glioma patients characterized by IDH1-R132H status, METT/N ratio demonstrated a significant prognostic impact in IDH1-R132H wildtype astrocytomas and glioblastoma (P = 0.001). Additionally, multivariate testing revealed semiquantitative MET PET as an independent prognostic parameter for treatment-naïve glioma patients without (P = 0.031) and with IDH1-R132H characterization of gliomas (P = 0.024; odds ratio 1.57). This retrospective analysis demonstrates the value of MET PET as a prognostic parameter on survival in treatment-naïve glioma patients.
The influence of thyroid hormones on the immune system is controversial. We analyzed the cytokine expression pattern of T lymphocytes in patients with severe nonimmune hypothyroidism in order to establish the role of thyroid hormones on the immune system. The study comprised 7 patients (1 male and 6 females) 20 to 76 years of age (mean age, 53 years), without signs of chronic thyroiditis, verified by histology and laboratory data. The patients were studied 5 weeks after total thyroidectomy. Peripheral blood mononuclear cell (PBMC) cultures and intracellular cytokine detection by flow cytometry before and after thyroid hormone replacement therapy were performed (free thyroxine [FT4] 0.030 +/- 0.034 ng/dL, versus FT4: 2.16 +/- 0.68 ng/dL). The control group consisted of 7 healthy subjects (FT4: 1.20 +/- 0.21 ng/dL). We found no significant differences in the cytokine pattern (interferon [IFN]-gamma, IL-2, interleukin [IL]-4, IL-5, IL-6, IL-10, IL-13, tumor necrosis factor [TNF]-alpha, TNF-beta) of CD4+ and CD8+ between the matched groups (hypothyroid subjects versus controls, levothyroxine treated subjects versus controls). Our data show no change in the type 1/type 2 balance of peripheral CD4+ and CD8+ T lymphocytes from patients with nonimmune hypothyroidism. According to our results, the hypothyreotic state does not contribute to the reported changes in cytokine production patterns in Hashimoto's thyroiditis.