Machine learning in the differentiation of follicular lymphoma from diffuse large B-cell lymphoma with radiomic [18F]FDG PET/CT features.

BACKGROUND: One of the challenges in the management of patients with follicular lymphoma (FL) is the identification of individuals with histological transformation, most commonly into diffuse large B-cell lymphoma (DLBCL). [18F]FDG-PET/CT is used for staging of patients with lymphoma, but visual interpretation cannot reliably discern FL from DLBCL. This study evaluated whether radiomic features extracted from clinical baseline [18F]FDG PET/CT and analyzed by machine learning algorithms may help discriminate FL from DLBCL. MATERIALS AND METHODS: Patients were selected based on confirmed histopathological diagnosis of primary FL (n=44) or DLBCL (n=76) and available [18F]FDG PET/CT with EARL reconstruction parameters within 6 months of diagnosis. Radiomic features were extracted from the volume of interest on co-registered [18F]FDG PET and CT images. Analysis of selected radiomic features was performed with machine learning classifiers based on logistic regression and tree-based ensemble classifiers (AdaBoosting, Gradient Boosting, and XG Boosting). The performance of radiomic features was compared with a SUVmax-based logistic regression model. RESULTS: From the segmented lesions, 121 FL and 227 DLBCL lesions were included for radiomic feature extraction. In total, 79 radiomic features were extracted from the SUVmap, 51 from CT, and 6 shape features. Machine learning classifier Gradient Boosting achieved the best discrimination performance using 136 radiomic features (AUC of 0.86 and accuracy of 80%). SUVmax-based logistic regression model achieved an AUC of 0.79 and an accuracy of 70%. Gradient Boosting classifier had a significantly greater AUC and accuracy compared to the SUVmax-based logistic regression (p≤0.01). CONCLUSION: Machine learning analysis of radiomic features may be of diagnostic value for discriminating FL from DLBCL tumor lesions, beyond that of the SUVmax alone.

Diagnostic performance of FDG-PET/CT of post-transplant lymphoproliferative disorder and factors affecting diagnostic yield.

PURPOSE: Post-transplant lymphoproliferative disorder (PTLD) is a serious complication after solid organ and hematopoietic stem cell transplantation, requiring a timely and accurate diagnosis. In this study, we evaluated the diagnostic performance of FDG-PET/CT in patients with suspected PTLD and examined if lactate dehydrogenase (LDH) levels, Epstein-Barr virus (EBV) load, or timing of FDG-PET/CT relate to detection performance of FDG-PET/CT. METHODS: This retrospective study included 91 consecutive patients with clinical suspicion of PTLD and a total of 97 FDG-PET/CT scans within an 8-year period. Pathology reports and a 2-year follow-up were used as the reference standard. Diagnostic performance of FDG-PET/CT for detection of PTLD as well as logistic regression analysis for factors expected to affect diagnostic yield were assessed. RESULTS: The diagnosis of PTLD was established in 34 patients (35%). Fifty-seven FDG-PET/CT scans (59%) were true negative, 29 (30%) were true positive, 6 (6%) false positive, and 5 (5%) false negative. Sensitivity of FDG-PET/CT for the detection of PTLD was 85%, specificity 90%, positive predictive value 83%, and negative predictive value 92%, with good inter-observer variability (k = 0.78). Of the parameters hypothesized to be associated with a true positive FDG-PET/CT result for the diagnosis of PTLD, only LDH was statistically significant (OR 1.03, p = 0.04). CONCLUSION: FDG-PET/CT has a good diagnostic performance in patients suspected of PTLD, with a good inter-observer agreement. Only LDH levels seemed to influence the detection performance of FDG-PET/CT. EBV-DNA load and timing of FDG-PET/CT after transplantation did not affect FDG-PET/CT diagnostic yield.

Relationship between semiquantitative 18F-fluorodeoxyglucose positron emission tomography metrics and necrosis in classical Hodgkin lymphoma.

Semiquantitative 18F-fluoro-2-deoxy-D-glucose positron emission tomography (18F-FDG PET) parameters have been proposed as prognostic markers in classical Hodgkin lymphoma (cHL). In non-Hodgkin lymphoma necrosis as assessed by 18F-FDG PET or computed tomography (CT) (necrosisvisual) correlates with an adverse prognosis. We investigated whether semiquantitative 18F-FDG PET metrics correlate with necrosisvisual, determined the incidence of necrosisvisual and explored the prognostic impact of these factors in cHL. From 87 cHL cases treated with ABVD, (escalated) BEACOPP or CHOP chemotherapy between 2010 and 2017, 71 had both a NEDPAS/EARL accredited 18F-FDG PET and a contrast enhanced CT scan. Semiquantitative 18F-FDG PET parameters were determined using Hermes Hybrid 3D software. Necrosisvisual, defined by photopenic tumor areas on 18F-FDG PET and attenuation values between 10 and 30 Hounsfield units (HUs) on CT, was assessed blinded to outcome. Univariate Cox regression survival analyses of progression free survival (PFS) were performed. Necrosisvisual was observed in 18.3% of cHL patients. Bulky disease (tumor mass >10 cm in any direction) (P = 0.002) and TLG (P = 0.041) but no other semiquantitative parameters were significantly associated with necrosisvisual. In exploratory univariate survival analysis for PFS the covariates IPS, bulky disease, MTV and TLG were prognostic, while necrosisvisual was not.

Performance of advanced imaging modalities at diagnosis and treatment response evaluation of patients with post-transplant lymphoproliferative disorder: A systematic review and meta-analysis.

INTRODUCTION AND AIM: Post-transplant lymphoproliferative disorder (PTLD) is a serious complication after solid organ and hematopoietic stem cell transplantation, associated with significant morbidity and mortality. In this systematic review we evaluated the clinical performance of advanced imaging modalities at diagnosis and treatment response evaluation of PTLD patients after solid organ and hematopoietic stem cell transplantation. METHODS: We have carried out a literature search until December 15, 2017 using PubMed/Medline, Embase, "Web of Science" and Cochrane Library databases concerning the performance of computed tomography (CT), magnetic resonance imaging (MRI) and 18F-flurodeoxyglucose positron emission tomography/computed tomography (FDG-PET/CT) at diagnosis or treatment response evaluation of PTLD patients. RESULTS: A total of 11 studies were included comprising 368 patients, from which FDG-PET(/CT) was the primary imaging modality investigated. The methodological quality according to QUADAS-2 of the reviewed studies was moderate-poor. Subgroup analysis of imaging results for detection and staging in patients with PTLD indicated that FDG-PET/(CT) identified additional lesions not detected by CT and/or MRI in 27.8%, (95% confidence interval [95%CI]) 17.0%-42.0% (I2 = 51.1%), from which extra-nodal sites in 23.6% (95%CI: 7.9%-52.4%) (I2 = 76.6%). False negative results occurred in 11.5% (95%CI: 4.9%-24.5%) (I2 = 73.4%), predominantly in physiological high background activity regions and in early PTLD lesions. False positive results occurred in 4.8% (95%CI: 2.6%-8.6%) (I2 = 0%) predominantly due to inflammatory conditions. Subgroup analysis of imaging results at treatment response evaluation indicated that FDG-PET(/CT) findings altered or guided treatment in 29.0% (95%CI: 14.0%-50.5%) (I2 = 40.1%). False positive results during treatment response evaluation were reported in 20.0% (95%CI: 10.7%-34.2%) (I2 = 0%), predominantly due to inflammatory conditions. CONCLUSION: FDG-PET(/CT) is currently the most frequently investigated imaging modality in PTLD patients. Available studies report promising results in detection, staging and therapy evaluation but suffer from methodological shortcomings. Concerns remain with regard to occurrence of false negatives due to physiological high background activity and early PTLD lesions as well as false positives due to inflammatory conditions.