Comparing lesion and feature selections to predict progression in newly diagnosed DLBCL patients with FDG PET/CT radiomics features.

PURPOSE: Biomarkers that can accurately predict outcome in DLBCL patients are urgently needed. Radiomics features extracted from baseline [18F]-FDG PET/CT scans have shown promising results. This study aims to investigate which lesion- and feature-selection approaches/methods resulted in the best prediction of progression after 2 years. METHODS: A total of 296 patients were included. 485 radiomics features (n = 5 conventional PET, n = 22 morphology, n = 50 intensity, n = 408 texture) were extracted for all individual lesions and at patient level, where all lesions were aggregated into one VOI. 18 features quantifying dissemination were extracted at patient level. Several lesion selection approaches were tested (largest or hottest lesion, patient level [all with/without dissemination], maximum or median of all lesions) and compared to the predictive value of our previously published model. Several data reduction methods were applied (principal component analysis, recursive feature elimination (RFE), factor analysis, and univariate selection). The predictive value of all models was tested using a fivefold cross-validation approach with 50 repeats with and without oversampling, yielding the mean cross-validated AUC (CV-AUC). Additionally, the relative importance of individual radiomics features was determined. RESULTS: Models with conventional PET and dissemination features showed the highest predictive value (CV-AUC: 0.72-0.75). Dissemination features had the highest relative importance in these models. No lesion selection approach showed significantly higher predictive value compared to our previous model. Oversampling combined with RFE resulted in highest CV-AUCs. CONCLUSION: Regardless of the applied lesion selection or feature selection approach and feature reduction methods, patient level conventional PET features and dissemination features have the highest predictive value. Trial registration number and date: EudraCT: 2006-005174-42, 01-08-2008.

18F-FDG PET baseline radiomics features improve the prediction of treatment outcome in diffuse large B-cell lymphoma.

PURPOSE: Accurate prognostic markers are urgently needed to identify diffuse large B-Cell lymphoma (DLBCL) patients at high risk of progression or relapse. Our purpose was to investigate the potential added value of baseline radiomics features to the international prognostic index (IPI) in predicting outcome after first-line treatment. METHODS: Three hundred seventeen newly diagnosed DLBCL patients were included. Lesions were delineated using a semi-automated segmentation method (standardized uptake value ≥ 4.0), and 490 radiomics features were extracted. We used logistic regression with backward feature selection to predict 2-year time to progression (TTP). The area under the curve (AUC) of the receiver operator characteristic curve was calculated to assess model performance. High-risk groups were defined based on prevalence of events; diagnostic performance was assessed using positive and negative predictive values. RESULTS: The IPI model yielded an AUC of 0.68. The optimal radiomics model comprised the natural logarithms of metabolic tumor volume (MTV) and of SUVpeak and the maximal distance between the largest lesion and any other lesion (Dmaxbulk, AUC 0.76). Combining radiomics and clinical features showed that a combination of tumor- (MTV, SUVpeak and Dmaxbulk) and patient-related parameters (WHO performance status and age > 60 years) performed best (AUC 0.79). Adding radiomics features to clinical predictors increased PPV with 15%, with more accurate selection of high-risk patients compared to the IPI model (progression at 2-year TTP, 44% vs 28%, respectively). CONCLUSION: Prediction models using baseline radiomics combined with currently used clinical predictors identify patients at risk of relapse at baseline and significantly improved model performance. TRIAL REGISTRATION NUMBER AND DATE: EudraCT: 2006-005,174-42, 01-08-2008.

Machine learning-based analysis of [18F]DCFPyL PET radiomics for risk stratification in primary prostate cancer.

PURPOSE: Quantitative prostate-specific membrane antigen (PSMA) PET analysis may provide for non-invasive and objective risk stratification of primary prostate cancer (PCa) patients. We determined the ability of machine learning-based analysis of quantitative [18F]DCFPyL PET metrics to predict metastatic disease or high-risk pathological tumor features. METHODS: In a prospective cohort study, 76 patients with intermediate- to high-risk PCa scheduled for robot-assisted radical prostatectomy with extended pelvic lymph node dissection underwent pre-operative [18F]DCFPyL PET-CT. Primary tumors were delineated using 50-70% peak isocontour thresholds on images with and without partial-volume correction (PVC). Four hundred and eighty standardized radiomic features were extracted per tumor. Random forest models were trained to predict lymph node involvement (LNI), presence of any metastasis, Gleason score ≥ 8, and presence of extracapsular extension (ECE). For comparison, models were also trained using standard PET features (SUVs, volume, total PSMA uptake). Model performance was validated using 50 times repeated 5-fold cross-validation yielding the mean receiver-operator characteristic curve AUC. RESULTS: The radiomics-based machine learning models predicted LNI (AUC 0.86 ± 0.15, p < 0.01), nodal or distant metastasis (AUC 0.86 ± 0.14, p < 0.01), Gleason score (0.81 ± 0.16, p < 0.01), and ECE (0.76 ± 0.12, p < 0.01). The highest AUCs reached using standard PET metrics were lower than those of radiomics-based models. For LNI and metastasis prediction, PVC and a higher delineation threshold improved model stability. Machine learning pre-processing methods had a minor impact on model performance. CONCLUSION: Machine learning-based analysis of quantitative [18F]DCFPyL PET metrics can predict LNI and high-risk pathological tumor features in primary PCa patients. These findings indicate that PSMA expression detected on PET is related to both primary tumor histopathology and metastatic tendency. Multicenter external validation is needed to determine the benefits of using radiomics versus standard PET metrics in clinical practice.

The quantitative assessment of interstitial lung disease with positron emission tomography scanning in systemic sclerosis patients.

OBJECTIVES: The reversibility of interstitial lung disease (ILD) in SSc is difficult to assess by current diagnostic modalities and there is clinical need for imaging techniques that allow for treatment stratification and monitoring. 18F-Fluorodeoxyglucose (FDG) PET/CT scanning may be of interest for this purpose by detection of metabolic activity in lung tissue. This study aimed to investigate the potential role of 18F-FDG PET/CT scanning for the quantitative assessment of SSc-related active ILD. METHODS: 18F-FDG PET/CT scans and high resolution CT scans of eight SSc patients, including five with ILD, were analysed. For comparison, reference groups were included: eight SLE patients and four primary Sjögren's syndrome (pSS) patients, all without ILD. A total of 22 regions of interest were drawn in each patient at apical, medial and dorsobasal lung levels. 18F-FDG uptake was measured as mean standardized uptake value (SUVmean) in each region of interest. Subsequently, basal/apical (B/A) and medial/apical (M/A) ratios were calculated at patient level (B/A-p and M/A-p) and at tissue level (B/A-t and M/A-t). RESULTS: SUVmean values in dorsobasal ROIs and B/A-p ratios were increased in SSc with ILD compared with SSc without ILD (P = 0.04 and P = 0.07, respectively), SLE (P = 0.003 and P = 0.002, respectively) and pSS (P = 0.03 and P = 0.02, respectively). Increased uptake in the dorsobasal lungs and increased B/A-t ratios corresponded to both ground glass and reticulation on high resolution CT. CONCLUSION: Semi-quantitative assessment of 18F-FDG PET/CT is able to distinguish ILD from non-affected lung tissue in SSc, suggesting that it may be used as a new biomarker for SSc-ILD disease activity.

Quantification of 18F-fluorodeoxyglucose uptake to detect residual nodal disease in locally advanced head and neck squamous cell carcinoma after chemoradiotherapy: results from the ECLYPS study.

BACKGROUND: The Hopkins criteria were introduced for nodal response evaluation after therapy in head and neck cancer, but its superiority over quantification is not yet confirmed. METHODS: SUVbody weight thresholds and lesion-to-background ratios were explored in a prospective multicenter study of standardized FDG-PET/CT 12 weeks after CRT in newly diagnosed locally advanced head and neck squamous cell carcinoma (LAHNSCC) patients (ECLYPS). Reference standard was histology, negative FDG-PET/CT at 12 months after treatment or ≥ 2 years of negative follow-up. Area under the receiver operator characteristics curves (AUROC) were estimated and obtained thresholds were validated in an independent cohort of HNSCC patients (n = 127). RESULTS: In ECLYPS, 124 patients were available for quantification. With a median follow-up of 20.4 months, 23 (18.5%) nodal neck recurrences were observed. A SUV70 threshold of 2.2 (AUROC = 0.89; sensitivity = 79.7%; specificity = 80.8%) was identified as optimal metric to identify nodal recurrence within 1 year after therapy. For lesion-to-background ratios, an SUV50/SUVliver threshold of 0.96 (AUROC = 0.89; sensitivity = 79.7%; specificity = 82.8%) had the best performance. Compared with Hopkins criteria (AUROC = 0.81), SUV70 and SUV50/SUVliver provided a borderline significant (p = 0.040 and p = 0.094, respectively) improvement. Validation of thresholds yielded similar AUROC values (SUV70 = 0.93, SUV50/SUVliver = 0.95), and were comparable to the Hopkins score (AUROC = 0.91; not statistically significant). CONCLUSION: FDG quantification detects nodal relapse in LAHNSCC patients. When using EARL standardized PET acquisitions and reconstruction, absolute SUV metrics (SUV70 threshold 2.2) prove robust, yet ratios (SUV50/SUVliver, threshold 0.96) may be more useful in routine clinical care. In this setting, the diagnostic value of quantification is comparable to the Hopkins criteria. TRIAL REGISTRATION: US National Library for Medicine, NCT01179360. Registered 11 August 2010, https://clinicaltrials.gov/ct2/show/NCT01179360.

Predictive value of interim positron emission tomography in diffuse large B-cell lymphoma: a systematic review and meta-analysis.

PURPOSE: Diffuse large B-cell lymphoma (DLBCL) represents the most common subtype of non-Hodgkin lymphoma. Most relapses occur in the first 2 years after diagnosis. Early response assessment with 18F-fluoro-2-deoxy-D-glucose (18F-FDG) positron emission tomography (PET) may facilitate early change of treatment, thereby preventing ineffective treatment and unnecessary side effects. We aimed to assess the predictive value of visually-assessed interim 18F-FDG PET on progression-free survival (PFS) or event-free survival (EFS) in DLBCL patients treated with first-line immuno-chemotherapy regimens. METHODS: For this systematic review and meta-analysis Pubmed, Embase, and the Cochrane Library were searched until July 11, 2017. Prospective and retrospective studies investigating qualitative interim PET response assessment without treatment adaptation based on the interim PET result were eligible. The primary outcome was two-year PFS or EFS. Prognostic and diagnostic measures were extracted and analysed with pooled hazard ratios and Hierarchical Summary Receiver Operator Characteristic Curves, respectively. Meta-regression was used to study covariate effects. RESULTS: The pooled hazard ratio for 18 studies comprising 2,255 patients was 3.13 (95%CI 2.52-3.89) with a 95% prediction interval of 1.68-5.83. In 19 studies with 2,366 patients, the negative predictive value for progression generally exceeded 80% (64-95), but sensitivity (33-87), specificity (49-94), and positive predictive values (20-74) ranged widely. CONCLUSIONS: These findings showed that interim 18F-FDG PET has predictive value in DLBCL patients. However, (subgroup) analyses were limited by lack of information and small sample sizes. Some diagnostic test characteristics were not satisfactory, especially the positive predictive value should be improved, before a successful risk stratified treatment approach can be implemented in clinical practice.

Interobserver reproducibility of tumor uptake quantification with 89Zr-immuno-PET: a multicenter analysis.

PURPOSE: In-vivo quantification of tumor uptake of 89-zirconium (89Zr)-labelled monoclonal antibodies (mAbs) with PET provides a potential tool in strategies to optimize tumor targeting and therapeutic efficacy. A specific challenge for 89Zr-immuno-PET is low tumor contrast. This is expected to result in interobserver variation in tumor delineation. Therefore, the aim of this study was to determine interobserver reproducibility of tumor uptake measures by tumor delineation on 89Zr-immuno-PET scans. METHODS: Data were obtained from previously published clinical studies performed with 89Zr-rituximab, 89Zr-cetuximab and 89Zr-trastuzumab. Tumor lesions on 89Zr-immuno-PET were identified as focal uptake exceeding local background by a nuclear medicine physician. Three observers independently manually delineated volumes of interest (VOI). Maximum, peak and mean standardized uptake values (SUVmax, SUVpeak and SUVmean) were used to quantify tumor uptake. Interobserver variability was expressed as the coefficient of variation (CoV). The performance of semi-automatic VOI delineation using 50% of background-corrected ACpeak was described. RESULTS: In total, 103 VOI were delineated (3-6 days post injection (D3-D6)). Tumor uptake (median, interquartile range) was 9.2 (5.2-12.6), 6.9 (4.0-9.6) and 5.5 (3.3-7.8) for SUVmax, SUVpeak and SUVmean. Interobserver variability was 0% (0-12), 0% (0-2) and 7% (5-14), respectively (n = 103). The success rate of the semi-automatic method was 45%. Inclusion of background was the main reason for failure of semi-automatic VOI. CONCLUSIONS: This study shows that interobserver reproducibility of tumor uptake quantification on 89Zr-immuno-PET was excellent for SUVmax and SUVpeak using a standardized manual procedure for tumor segmentation. Semi-automatic delineation was not robust due to limited tumor contrast.

Lesion detection by [89Zr]Zr-DFO-girentuximab and [18F]FDG-PET/CT in patients with newly diagnosed metastatic renal cell carcinoma.

PURPOSE: The main objective of this preliminary analysis of the IMaging PAtients for Cancer drug selecTion (IMPACT)-renal cell cancer (RCC) study is to evaluate the lesion detection of baseline contrast-enhanced CT, [89Zr]Zr-DFO-girentuximab-PET/CT and [18F]FDG-PET/CT in detecting ccRCC lesions in patients with a good or intermediate prognosis metastatic clear cell renal cell carcinoma (mccRCC) according to the International Metastatic Database Consortium (IMDC) risk model. METHODS: Between February 2015 and March 2018, 42 newly diagnosed mccRCC patients with good or intermediate prognosis, eligible for watchful waiting, were included. Patients underwent CT, [89Zr]Zr-DFO-girentuximab-PET/CT and [18F]FDG-PET/CT at baseline. Scans were independently reviewed and lesions of ≥10 mm and lymph nodes of ≥15 mm at CT were analyzed. For lesions with [89Zr]Zr-DFO-girentuximab or [18F]FDG-uptake visually exceeding background uptake, maximum standardized uptake values (SUVmax) were measured. RESULTS: A total of 449 lesions were detected by ≥1 modality (median per patient: 7; ICR 4.25-12.75) of which 42% were in lung, 22% in lymph nodes and 10% in bone. Combined [89Zr]Zr-DFO-girentuximab-PET/CT and CT detected more lesions than CT alone: 91% (95%CI: 87-94) versus 56% (95%CI: 50-62, p = 0.001), respectively, and more than CT and [18F]FDG-PET/CT combined (84% (95%CI:79-88, p < 0.005). Both PET/CTs detected more bone and soft tissue lesions compared to CT alone. CONCLUSIONS: The addition of [89Zr]Zr-DFO-girentuximab-PET/CT and [18F]FDG-PET/CT to CT increases lesion detection compared to CT alone in newly diagnosed good and intermediate prognosis mccRCC patients eligible for watchful waiting.

F8-IL10: A New Potential Antirheumatic Drug Evaluated by a PET-Guided Translational Approach.

Antibody fragment F8-mediated interleukin 10 (IL10) delivery is a novel treatment for rheumatoid arthritis (RA). F8 binds to the extra-domain-A of fibronectin (ED-A). In this study, in vivo biodistribution and arthritis targeting of radiolabeled F8-IL10 were investigated in RA patients, followed by further animal studies. Therefore, three RA patients (DAS28 > 3.2) received 0.4 mg of 30-74 megabecquerel [124I]I-F8-IL10 for PET-CT and blood sampling. In visually identified PET-positive joints, target-to-background was calculated. Healthy mice, rats, and arthritic rats were injected with iodinated F8-IL10 or KSF-IL10 control antibody. Various organs were excised, weighed, and counted for radioactivity. Tissue sections were stained for fibronectin ED-A. In RA patients, [124I]I-F8-IL10 was cleared rapidly from the circulation with less than 1% present in blood after 5 min. PET-CT showed targeting in 38 joints (11-15 per patient) and high uptake in the liver and spleen. Mean target-to-background ratios of PET-positive joints were 2.5 ± 1.2, 1.5 times higher for clinically active than clinically silent joints. Biodistribution of radioiodinated F8-IL10 in healthy mice showed no effect of the radioiodination method. [124I]I-F8-IL10 joint uptake was also demonstrated in arthritic rats, ∼14-fold higher than that of the control antibody [124I]I-KSF-IL10 ( p < 0.001). Interestingly, liver and spleen uptake were twice as high in arthritic than in healthy rats and were related to increased (∼7×) fibronectin ED-A expression in these tissues. In conclusion, [124I]I-F8-IL10 uptake was observed in arthritic joints in RA patients holding promise for visualization of inflamed joints by PET-CT imaging and therapeutic targeting. Patient observations and, subsequently, arthritic animal studies pointed to awareness of increased [124I]I-F8-IL10 uptake in the liver and spleen associated with moderate systemic inflammation. This translational study demonstrated the value of in vivo biodistribution and PET-CT-guided imaging in development of new and potential antirheumatic drugs'.

Is C-11 Methionine PET an alternative to 18-F FDG-PET for identifying recurrent laryngeal cancer after radiotherapy?

OBJECTIVE: 18F FDG-PET is superior to other imaging techniques in revealing residual laryngeal cancer after radiotherapy. Unfortunately, its specificity is low, due to FDG uptake in inflammation and in anaerobic conditions. PET imaging with the amino acid-based radiopharmaceutical C11-methionine (MET) should be less influenced by post-radiation conditions. The aim of this study was to investigate the potential of MET in diagnosing recurrent laryngeal cancer after radiotherapy as compared to 18F-FDG. METHODS: Forty-eight patients with a clinical suspicion of local residual disease at least 3 months after completion of radiotherapy or chemoradiotherapy for a T2-4 laryngeal carcinoma, along with an indication for direct laryngoscopy, were included. They received MET-PET and FDG-PET prior to the direct laryngoscopy. One senior nuclear medicine physician assessed both the FDG-PET and MET-PET images visually for the degree of abnormal uptake. The gold standard was a biopsy-proven recurrence 12 months after PET. The nuclear physician had no access to the medical charts and was blinded to the results of the other PET. Sensitivity, specificity and positive and negative predictive value were calculated. RESULTS: The sensitivity of FDG was 77.3% and the specificity 56.0% after the conservative reading, with these values equalling 54.5% and 76.0% for MET. The positive predictive value of FDG was 60.7% and the negative predictive value 73.7%. The PPV of MET was 66.7%, and the NPV was 65.5%. The McNemar test within diseased (sensitivity comparison) shows a p-value of 0.125, and the McNemar test within non-diseased (specificity comparison) shows a P-value of 0.180. CONCLUSION: MET-PET is not superior to FDG-PET in terms of identifying recurrent laryngeal cancer.