Prediction of the treatment outcome using machine learning with FDG-PET image-based multiparametric approach in patients with oral cavity squamous cell carcinoma.

AIM: To investigate the value of machine learning-based multiparametric analysis using 2-[18F]-fluoro-2-deoxy-d-glucose positron-emission tomography (FDG-PET) images to predict treatment outcome in patients with oral cavity squamous cell carcinoma (OCSCC). MATERIALS AND METHODS: Ninety-nine patients with OCSCC who received pretreatment integrated FDG-PET/computed tomography (CT) were included. They were divided into the training (66 patients) and validation (33 patients) cohorts. The diagnosis of local control or local failure was obtained from patient's medical records. Conventional FDG-PET parameters, including the maximum and mean standardised uptake values (SUVmax and SUVmean), metabolic tumour volume (MTV), and total lesion glycolysis (TLG), quantitative tumour morphological parameters, intratumoural histogram, and texture parameters, as well as T-stage and clinical stage, were evaluated by a machine learning analysis. The diagnostic ability of T-stage, clinical stage, and conventional FDG-PET parameters (SUVmax, SUVmean, MTV, and TLG) was also assessed separately. RESULTS: In support-vector machine analysis of the training dataset, the final selected parameters were T-stage, SUVmax, TLG, morphological irregularity, entropy, and run-length non-uniformity. In the validation dataset, the diagnostic performance of the created algorithm was as follows: sensitivity 0.82, specificity 0.7, positive predictive value 0.86, negative predictive value 0.64, and accuracy 0.79. In a univariate analysis using conventional FDG-PET parameters, T-stage and clinical stage, diagnostic accuracy of each variable was revealed as follows: 0.61 in T-stage, 0.61 in clinical stage, 0.64 in SUVmax, 0.61 in SUVmean, 0.64 in MTV, and 0.7 in TLG. CONCLUSION: A machine-learning-based approach to analysing FDG-PET images by multiparametric analysis might help predict local control or failure in patients with OCSCC.

FDG positron emission tomography in isolated limb perfusion therapy in patients with locally advanced melanoma: preliminary results.

PURPOSE: Isolated limb perfusion (ILP) with high-dose chemotherapy and tumor necrosis factor is being tested in clinical trials as a treatment for locally advanced extremity melanoma. The authors investigated the ability of F-18 fluorodeoxyglucose positron emission tomography (FDG PET) to determine the true extent of disease in patients with this condition, whose distribution of lesions differs from that seen in previous studies. METHODS: Nine patients with locally advanced melanoma were selected for imaging of the entire body and extremities using FDG PET from a group of participants in a clinical trial of ILP with melphalan +/- tumor necrosis factor. Scans were obtained without attenuation correction. Post-treatment scans were obtained in three patients 1 month after ILP. The findings in the FDG-PET scans were compared with those of a standard protocol (SP) that included anatomic images and physical examinations. RESULTS: Eighty lesions (74 malignant, 6 benign) were detected with FDG PET and the SP combined. Only malignant lesions were detected by both methods in the perfused limbs. Of the malignant lesions, FDG PET detected 65 lesions (sensitivity rate, 88%). In contrast, 48 lesions were detected with the SP (sensitivity rate, 65%). Twenty-six malignant lesions were seen only with FDG PET (35%), whereas nine malignant lesions were seen only with SP (12%). The six benign lesions included three false-positive mediastinal lymph nodes in one patient. The accuracy rates of FDG PET and the SP were 83% and 65%, respectively. These results are comparable to those seen in previous studies with patients who had disease confined primarily to the torso. All post-therapy FDG-PET scans showed a reduction in the number of visualized limb lesions, and diffuse uptake throughout the perfused limbs. The diffuse uptake correlated with post-therapy limb inflammation. CONCLUSIONS: Non-attenuation-corrected FDG PET is more sensitive than the SP in detecting the extent of disease in candidates for ILP. The FDG uptake associated with post-therapy inflammation may reduce the contrast resolution of this technique and must be evaluated further.

Enhancing the relaxivity of paramagnetic coordination complexes through the optimization of the molecular electrostatic potential.

The low relaxivity of paramagnetic coordination complexes limits their use as contrast agents in magnetic resonance imaging (MRI). To address this problem, we study the relationship between the molecular structure of these complexes and their relaxivity. While others have investigated the vibrational modes as molecular determinants of the electronic spin relaxation time, we focus on the analysis of the molecular electrostatic potential (MEP) of the paramagnetic coordination complex. Electrostatic forces dominate the interaction between the coordination complex and water. Hence, in addition to steric forces, the molecular electrostatic potential should be a determinant of the lifetime of the water-metal link (tm), the internuclear distance between the water hydrogens and the metal (R), and the number of water molecules attached to the metal in the inner and outer spheres of coordination. We compute the molecular electrostatic potential for a series of model metalloporphyrins because their physical and biologic properties are well known, and they are putative magnetic resonance imaging contrast agents with affinity to neoplastic tissue. Replacing the sulfonato groups in MnTPPS4 with carboxylate groups in the ortho position of the phenyl rings attached to the meso carbons results in an electrostatic focusing field that should reduce R and increase tm. Similar substitutions involving polar groups, including one modeled after a well-known picket-fence porphyrin, are not strong enough to generate a focusing field. Instead, these polar groups should modulate the water-metal interactions through steric interactions. Molecular dynamic simulations show a large outer sphere of coordination around the paramagnet that extends almost three times the distance of the inner sphere of coordination.

On the molecular spin density and the electrostatic potential as determinants of the relaxivity of metalloporphyrins.

The origin of the unexpectedly large relaxivity of a manganese metalloporphyrin is explored through computer simulations that include a description of the molecular spin density and the molecular electrostatic potential. These molecular properties describe not only the distribution of unpaired electrons in the coordination complex, but also a major driving force responsible for the dynamic behavior of the water molecules. By comparing the computed properties for the manganese complex with those of its congeneric iron complex, we gain insight into the origin of the unusually large relaxivity of the manganese metalloporphyrin. In the process, we learn how to use the computed molecular properties to formulate rules on how to modify the chemical structure of the metalloporphyrins to improve their relaxivity. Specifically, we show how to spatially direct the molecular spin density by the splitting of d orbitals and by the delocalization of electronic spin across unsaturated rings. We also learn how to attract water protons to the areas of high spin density by designing electrostatic focusing fields.

Role of primary and secondary protein structure in neurotransmitter receptor activation mechanisms.

A proton transfer triggered by a ligand interacting with the receptor had been suggested as the initial step in the activation of a receptor for the neurotransmitter serotonin (5-hydroxy-tryptamine; 5-HT). To evaluate the role of the receptor macromolecule in modulating the primary molecular event in ligand-mediated activation, the process of proton transfer was analysed in the environment of a protein model for the 5-HT receptor. In the absence of a detailed receptor structure, the enzyme actinidin was chosen as the model for the receptor based on criteria obtained from structure-activity considerations on the ligands. The first simulation of a mechanism for receptor activation was performed on this model using methods of theoretical chemistry to study the effect of specific structural elements. The premise is that the role of the elements of secondary structure of soluble proteins (e.g. actinidin) in determining structure-function relations in these macromolecules is maintained when these elements are part of membrane-bound receptor proteins. Results from the calculations of the effects of the six alpha helices of actinidin on the proton transfer process from the imidazolium side chain of His 162 to the thiol side chain of Cys 25 in the protein show that the helices contribute in different ways to modulate the energy of proton transfer. The largest helix, A1, opposes the proton transfer through the effect of the helix dipole. The charged residues (primary structure) in helix A3 favor the proton transfer, and mask the effect of its helix dipole (secondary structure) which opposes the transfer. The direction of the proton transfer simulated for the activation mechanism is opposite to that assumed in the catalytic process of the thiol protease, and the entire protein environment opposes the transfer. This supports the specific role of the ligand in triggering the proton transfer as a response to its binding.