The Prevalence and Radiologic Features of Renal Cancers Associated with FLCN, BAP1, SDH, and MET Germline Mutations.

Purpose To investigate the prevalence of FLCN, BAP1, SDH, and MET mutations in an oncologic cohort and determine the prevalence, clinical features, and imaging features of renal cell carcinoma (RCC) associated with these mutations. Secondarily, to determine the prevalence of encountered benign renal lesions. Materials and Methods From 25 220 patients with cancer who prospectively underwent germline analysis with a panel of more than 70 cancer-predisposing genes from 2015 to 2021, patients with FLCN, BAP1, SDH, or MET mutations were retrospectively identified. Clinical records were reviewed for patient age, sex, race/ethnicity, and renal cancer diagnosis. If RCC was present, baseline CT and MRI examinations were independently assessed by two radiologists. Summary statistics were used to summarize continuous and categorical variables by mutation. Results A total of 79 of 25 220 (0.31%) patients had a germline mutation: FLCN, 17 of 25 220 (0.07%); BAP1, 22 of 25 220 (0.09%); SDH, 39 of 25 220 (0.15%); and MET, one of 25 220 (0.004%). Of these 79 patients, 18 (23%) were diagnosed with RCC (FLCN, four of 17 [24%]; BAP1, four of 22 [18%]; SDH, nine of 39 [23%]; MET, one of one [100%]). Most hereditary RCCs demonstrated ill-defined margins, central nonenhancing area (cystic or necrotic), heterogeneous enhancement, and various other CT and MR radiologic features, overlapping with the radiologic appearance of nonhereditary RCCs. The prevalence of other benign solid renal lesions (other than complex cysts) in patients was up to 11%. Conclusion FLCN, BAP1, SDH, and MET mutations were present in less than 1% of this oncologic cohort. Within the study sample size limits, imaging findings for hereditary RCC overlapped with those of nonhereditary RCC, and the prevalence of other associated benign solid renal lesions (other than complex cysts) was up to 11%. Keywords: Familial Renal Cell Carcinoma, Birt-Hogg-Dubé Syndrome, Carcinoma, Renal Cell, Paragangliomas, Urinary, Kidney © RSNA, 2024.

Deep learning for [18F]fluorodeoxyglucose-PET-CT classification in patients with lymphoma: a dual-centre retrospective analysis.

BACKGROUND: The rising global cancer burden has led to an increasing demand for imaging tests such as [18F]fluorodeoxyglucose ([18F]FDG)-PET-CT. To aid imaging specialists in dealing with high scan volumes, we aimed to train a deep learning artificial intelligence algorithm to classify [18F]FDG-PET-CT scans of patients with lymphoma with or without hypermetabolic tumour sites. METHODS: In this retrospective analysis we collected 16 583 [18F]FDG-PET-CTs of 5072 patients with lymphoma who had undergone PET-CT before or after treatment at the Memorial Sloa Kettering Cancer Center, New York, NY, USA. Using maximum intensity projection (MIP), three dimensional (3D) PET, and 3D CT data, our ResNet34-based deep learning model (Lymphoma Artificial Reader System [LARS]) for [18F]FDG-PET-CT binary classification (Deauville 1-3 vs 4-5), was trained on 80% of the dataset, and tested on 20% of this dataset. For external testing, 1000 [18F]FDG-PET-CTs were obtained from a second centre (Medical University of Vienna, Vienna, Austria). Seven model variants were evaluated, including MIP-based LARS-avg (optimised for accuracy) and LARS-max (optimised for sensitivity), and 3D PET-CT-based LARS-ptct. Following expert curation, areas under the curve (AUCs), accuracies, sensitivities, and specificities were calculated. FINDINGS: In the internal test cohort (3325 PET-CTs, 1012 patients), LARS-avg achieved an AUC of 0·949 (95% CI 0·942-0·956), accuracy of 0·890 (0·879-0·901), sensitivity of 0·868 (0·851-0·885), and specificity of 0·913 (0·899-0·925); LARS-max achieved an AUC of 0·949 (0·942-0·956), accuracy of 0·868 (0·858-0·879), sensitivity of 0·909 (0·896-0·924), and specificity of 0·826 (0·808-0·843); and LARS-ptct achieved an AUC of 0·939 (0·930-0·948), accuracy of 0·875 (0·864-0·887), sensitivity of 0·836 (0·817-0·855), and specificity of 0·915 (0·901-0·927). In the external test cohort (1000 PET-CTs, 503 patients), LARS-avg achieved an AUC of 0·953 (0·938-0·966), accuracy of 0·907 (0·888-0·925), sensitivity of 0·874 (0·843-0·904), and specificity of 0·949 (0·921-0·960); LARS-max achieved an AUC of 0·952 (0·937-0·965), accuracy of 0·898 (0·878-0·916), sensitivity of 0·899 (0·871-0·926), and specificity of 0·897 (0·871-0·922); and LARS-ptct achieved an AUC of 0·932 (0·915-0·948), accuracy of 0·870 (0·850-0·891), sensitivity of 0·827 (0·793-0·863), and specificity of 0·913 (0·889-0·937). INTERPRETATION: Deep learning accurately distinguishes between [18F]FDG-PET-CT scans of lymphoma patients with and without hypermetabolic tumour sites. Deep learning might therefore be potentially useful to rule out the presence of metabolically active disease in such patients, or serve as a second reader or decision support tool. FUNDING: National Institutes of Health-National Cancer Institute Cancer Center Support Grant.

The use of single-timepoint images to link administered radioiodine activity (MBq) to a prescribed lesion radiation-absorbed dose (cGy): a regression-based prediction interval tool for the management of well-differentiated thyroid cancer patients.

PURPOSE: To introduce a biomarker-based dosimetry method for the rational selection of a treatment activity for patients undergoing radioactive iodine 131I therapy (RAI) for metastatic differentiated thyroid cancer (mDTC) based on single-timepoint imaging of individual lesion uptake by 124I PET. METHODS: Patients referred for RAI therapy of mDTC were enrolled in institutionally approved protocols. A total of 208 mDTC lesions (in 21 patients) with SUVmax > 1 underwent quantitative PET scans at 24, 48, 72, and 120 h post-administration of 222 MBq of theranostic NaI-124I to determine the individual lesion radiation-absorbed dose. Using a general estimating equation, a prediction curve for biomarker development was generated in the form of a best-fit regression line and 95% prediction interval, correlating individual predicted lesion radiation dose metrics, with candidate biomarkers ("predictors") such as SUVmax and activity in microcurie per gram, from a single imaging timepoint. RESULTS: In the 169 lesions (in 15 patients) that received 131I therapy, individual lesion cGy varied over 3 logs with a median of 22,000 cGy, confirming wide heterogeneity of lesion radiation dose. Initial findings from the prediction curve on all 208 lesions confirmed that a 48-h SUVmax was the best predictor of lesion radiation dose and permitted calculation of the 131I activity required to achieve a lesional threshold radiation dose (2000 cGy) within defined confidence intervals. CONCLUSIONS: Based on MIRD lesion-absorbed dose estimates and regression statistics, we report on the feasibility of a new single-timepoint 124I-PET-based dosimetry biomarker for RAI in patients with mDTC. The approach provides clinicians with a tool to select personalized (precision) therapeutic administration of radioactivity (MBq) to achieve a desired target lesion-absorbed dose (cGy) for selected index lesions based on a single 48-h measurement 124I-PET image, provided the selected activity does not exceed the maximum tolerated activity (MTA) of < 2 Gy to blood, as is standard of care at Memorial Sloan Kettering Cancer Center. TRIAL REGISTRATION: NCT04462471, Registered July 8, 2020. NCT03647358, Registered Aug 27, 2018.

New era of PET/CT in breast cancer: high-resolution PET/dfCT.

OBJECTIVE: To determine whether breast cancer staging differs between high-resolution (HR) and standard-resolution (SR) PET/computed tomography acquisition. PATIENTS AND METHODS: This retrospective study included 39 women with breast cancer referred for staging. Images were assessed for the number of primary breast lesions with the corresponding size and the average maximum standardized uptake value (SUVmax), the anatomical site of fluorine-18-fluorodeoxyglucose-avid lymph nodes (LNs) with their SUVmax, and the number and type (lytic/blastic) of metastatic bone lesions. RESULTS: On HR, 42 breast tumor lesions with a size range of 0.30 cm up to 12.00 cm were detected versus 34 breast tumor lesions on SR. One hundred and forty-one versus 90 axillary LNs were detected on HR versus SR, respectively. Pathology reports were available for 60 axillary LNs out of the total 141 identified on HR. Rates for HR versus SR sensitivity, true positivity, false positivity, and false negativity are as follows: 92 versus 75%, 92 versus 75%, 2 versus 0%, and 7 versus 25%. The higher detection rate of axillary LN on HR was significant (P<0.001), but not the number of nodes detected (P=0.091). SUVmax for breast tumor lesions (P=0.225) and axillary LNs (P=0.134) were not significant. Three (8%) patients had a change in staging and management. CONCLUSION: HR detected primary breast lesions and metastatic LNs missed on SR, which led to change in staging and management. In addition, HR images provided higher SUVmax, which enabled a more comfortable localization, especially when SR presented borderline values. Finally, HR images decreased the number of gray zone lesions, especially in axillary LN detection.