A Multicenter Assessment of Interreader Reliability of LI-RADS Version 2018 for MRI and CT.

Background Various limitations have impacted research evaluating reader agreement for Liver Imaging Reporting and Data System (LI-RADS). Purpose To assess reader agreement of LI-RADS in an international multicenter multireader setting using scrollable images. Materials and Methods This retrospective study used deidentified clinical multiphase CT and MRI and reports with at least one untreated observation from six institutions and three countries; only qualifying examinations were submitted. Examination dates were October 2017 to August 2018 at the coordinating center. One untreated observation per examination was randomly selected using observation identifiers, and its clinically assigned features were extracted from the report. The corresponding LI-RADS version 2018 category was computed as a rescored clinical read. Each examination was randomly assigned to two of 43 research readers who independently scored the observation. Agreement for an ordinal modified four-category LI-RADS scale (LR-1, definitely benign; LR-2, probably benign; LR-3, intermediate probability of malignancy; LR-4, probably hepatocellular carcinoma [HCC]; LR-5, definitely HCC; LR-M, probably malignant but not HCC specific; and LR-TIV, tumor in vein) was computed using intraclass correlation coefficients (ICCs). Agreement was also computed for dichotomized malignancy (LR-4, LR-5, LR-M, and LR-TIV), LR-5, and LR-M. Agreement was compared between research-versus-research reads and research-versus-clinical reads. Results The study population consisted of 484 patients (mean age, 62 years ± 10 [SD]; 156 women; 93 CT examinations, 391 MRI examinations). ICCs for ordinal LI-RADS, dichotomized malignancy, LR-5, and LR-M were 0.68 (95% CI: 0.61, 0.73), 0.63 (95% CI: 0.55, 0.70), 0.58 (95% CI: 0.50, 0.66), and 0.46 (95% CI: 0.31, 0.61) respectively. Research-versus-research reader agreement was higher than research-versus-clinical agreement for modified four-category LI-RADS (ICC, 0.68 vs 0.62, respectively; P = .03) and for dichotomized malignancy (ICC, 0.63 vs 0.53, respectively; P = .005), but not for LR-5 (P = .14) or LR-M (P = .94). Conclusion There was moderate agreement for LI-RADS version 2018 overall. For some comparisons, research-versus-research reader agreement was higher than research-versus-clinical reader agreement, indicating differences between the clinical and research environments that warrant further study. © RSNA, 2023 Supplemental material is available for this article. See also the editorials by Johnson and Galgano and Smith in this issue.

Contrast-enhanced ultrasound for abdominal image-guided procedures.

INTRODUCTION: Since FDA approval for contrast-enhanced ultrasound (CEUS), clinical applications have increased to include diagnostic imaging of hepatic, renal, and other abdominal lesions. The modality has also demonstrated utility in certain image-guided procedures. Intravascular ultrasound contrast agents use microbubbles to improve visibility of solid tumors. Lesions not well seen on grayscale or Doppler ultrasound may become amenable to CEUS-guided biopsy or ablation. MATERIALS AND METHODS: This pictorial essay provides eleven examples to illustrate the current use of CEUS in a variety of abdominal image-guided procedures. Hepatic, renal, peritoneal, and soft tissue cases are presented. CONCLUSION: CEUS can improve visualization and targeting in abdominal image-guided procedures, without nephrotoxicity or radiation exposure.

Effectiveness of Molecular Testing Techniques for Diagnosis of Indeterminate Thyroid Nodules: A Randomized Clinical Trial.

IMPORTANCE: Approximately 20% of thyroid nodules display indeterminate cytology. Molecular testing can refine the risk of malignancy and reduce the need for diagnostic hemithyroidectomy. OBJECTIVE: To compare the diagnostic performance between an RNA test (Afirma genomic sequencing classifier) and DNA-RNA test (ThyroSeq v3 multigene genomic classifier). DESIGN, SETTING, AND PARTICIPANTS: This parallel randomized clinical trial of monthly block randomization included patients in the UCLA Health system who underwent thyroid biopsy from August 2017 to January 2020 with indeterminate cytology (Bethesda System for Reporting Thyroid Cytopathology category III or IV). INTERVENTIONS: Molecular testing with the RNA test or DNA-RNA test. MAIN OUTCOMES AND MEASURES: Diagnostic test performance of the RNA test compared with the DNA-RNA test. The secondary outcome was comparison of test performance with prior versions of the molecular tests. RESULTS: Of 2368 patients, 397 were eligible for inclusion based on indeterminate cytology, and 346 (median [interquartile range] age, 55 [44-67] years; 266 [76.9%] women) were randomized to 1 of the 2 tests. In the total cohort assessed for eligibility, 3140 thyroid nodules were assessed, and 427 (13.6%) nodules were cytologically indeterminate. The prevalence of malignancy was 20% among indeterminate nodules. The benign call rate was 53% (95% CI, 47%-61%) for the RNA test and 61% (95% CI, 53%-68%) for the DNA-RNA test. The specificities of the RNA test and DNA-RNA test were 80% (95% CI, 72%-86%) and 85% (95% CI, 77%-91%), respectively (P = .33); the positive predictive values (PPV) of the RNA test and DNA-RNA test were 53% (95% CI, 40%-67%) and 63% (95% CI, 48%-77%), respectively (P = .33). The RNA test exhibited a higher PPV compared with the prior test version (Afirma gene expression classifier) (54% [95% CI, 40%-67%] vs 38% [95% CI, 27%-48%]; P = .01). The DNA-RNA test had no statistically significant difference in PPV compared with its prior version (ThyroSeq v2 next-generation sequencing) (63% [95% CI, 48%-77%] vs 58% [95% CI, 43%-73%]; P = .75). Diagnostic thyroidectomy was avoided in 87 (51%) patients tested with the RNA test and 83 (49%) patients tested with the DNA-RNA test. Surveillance ultrasonography was available for 90 nodules, of which 85 (94%) remained stable over a median of 12 months follow-up. CONCLUSIONS AND RELEVANCE: Both the RNA test and DNA-RNA test displayed high specificity and allowed 49% of patients with indeterminate nodules to avoid diagnostic surgery. Although previous trials demonstrated that the prior version of the DNA-RNA test was more specific than the prior version of the RNA test, the current molecular test techniques have no statistically significant difference in performance. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT02681328.

Association of tumor grade, enhancement on multiphasic CT and microvessel density in patients with clear cell renal cell carcinoma.

PURPOSE: Clear cell renal cell carcinoma (ccRCC) comprises nearly 90% of all diagnosed RCC subtypes and has the worst prognosis and highest metastatic potential. The strongest prognostic factors for patients with ccRCC include histological subtype and Fuhrman grade, which are incorporated into prognostic models. Since ccRCC is a highly vascularized tumor, there may be differences in enhancement patterns on multidetector CT (MDCT) due to the hemodynamics and microvessel density (MVD) of the lesions. This may provide a noninvasive method to characterize incidentally detected low- and high-grade ccRCCs on MDCT. The purpose of our study was to determine the correlation between MDCT enhancement parameters, ccRCC MVD, and Fuhrman grade to determine its utility and value in assessing tumor vascularity and grade in vivo. METHODS: In this retrospective, HIPAA-compliant, institutional review board-approved study with waiver of informed consent, 127 consecutive patients with 89 low-grade (LG), and 43 high-grade (HG) ccRCCs underwent preoperative four-phase MDCT. A 3D volume of interest (VOI) was obtained for every tumor and absolute enhancement and the wash-in/wash-out of enhancement for each phase was assessed. Immunohistochemistry on resected specimens was used to quantify MVD. Linear regression and Pearson correlation were used to investigate the strength of the association between 3D VOI enhancement and MVD. Stepwise logistic regression analysis determined independent predictors of HG ccRCC. Cut-off values and odds Ratio (OR) with 95% CIs were reported. The clinical, radiomic, and pathologic features with the highest performance in the stepwise logistic regression analysis were evaluated using receiver operator characteristics (ROC) and area under the curve (AUC). RESULTS: Absolute enhancement in the nephrographic phase < 52.1 Hounsfield Units (HU) (HR 0.979, 95% CI 0.964-0.994, p value = 0.006), lesion size > 4.3 cm (HR 1.450, 95% CI 1.211-1.738, p value < 0.001), and an intratumoral MVD < 15% (HR 0.932, 95% CI 0.867-1.002, p value = 0.058) were independent predictors of HG ccRCC with an AUC of 0.818 (95% CI 0.725-0.911). HG ccRCCs had a significant association between 3D VOI enhancement and MVD in each post-contrast phase (r2 = 0.238 to 0.455, p < 0.05). CONCLUSIONS: Absolute enhancement of the entire lesion obtained from a 3D VOI in the nephrographic phase on preoperative MDCT can provide quantitative data that are a significant, independent predictor of a high-grade clear cell RCC and can be used to assess tumor vascularity and grade in vivo.

Clinical features of pseudocirrhosis in metastatic breast cancer.

PURPOSE: Pseudocirrhosis has been demonstrated to mimic cirrhosis radiographically, but studies evaluating the pathophysiology and clinical features are lacking. To better understand the incidence, risk factors, clinical course, and etiology of pseudocirrhosis, we performed a retrospective analysis of consecutively treated patients with metastatic breast cancer (MBC). METHODS: Of 374 patients treated for MBC from 2006 to 2012, 199 had imaging available for review. One radiologist evaluated computed tomography scans for evidence of pseudocirrhosis. Features of groups with and without pseudocirrhosis were compared by Kaplan-Meier product-limit survival estimates and log-rank tests. Wilcoxon Rank-Sum testing evaluated if patients more heavily treated were more likely to develop pseudocirrhosis. Univariate and multivariate Cox proportional hazard models investigated factors associated with mortality. RESULTS: Pseudocirrhosis developed in 37 of 199 patients (19%). Of the patients with liver metastases, 55% developed pseudocirrhosis. Liver metastases were demonstrated in 100% of patients with pseudocirrhosis. Survival in the subset with liver metastases favored those without pseudocirrhosis, 189 versus 69 months (p = 0.01). The number of systemic regimens received were higher in patients with pseudocirrhosis (p = 0.01). Ascites was demonstrated in 68%, portal hypertension in 11%, and splenomegaly in 8% of patients with pseudocirrhosis. CONCLUSIONS: Pseudocirrhosis does not occur in the absence of liver metastases, can manifest as hepatic decompensation, and appears to be associated with poorer survival amongst patients with hepatic metastases. Higher cumulative exposure to systemic therapy may be causative, instead of the previously held belief of pseudocirrhosis as an adverse effect of a particular systemic agent/class.

Components of Radiologic Progressive Disease Defined by RECIST 1.1 in Patients with Metastatic Clear Cell Renal Cell Carcinoma.

Background Progression-free survival (PFS) determined by Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST 1.1) is the reference standard to assess efficacy of treatments in patients with clear cell renal cell carcinoma. Purpose To assess the most common components of radiologic progressive disease as defined by RECIST 1.1 in patients with clear cell renal cell carcinoma and how the progression events impact PFS. Materials and Methods This secondary analysis of the phase III METEOR trial conducted between 2013 and 2014 included patients with metastatic clear cell renal cell carcinoma, with at least one target lesion at baseline and one follow-up time point, who were determined according to RECIST 1.1 to have progressive disease. A chest, abdominal, and pelvic scan were acquired at each time point. Kruskal-Wallis analysis was used to test differences in median PFS among the RECIST 1.1 progression events. The Holm-Bonferroni method was used to compare the median PFS of the progression events for the family-wise error rate of 5% to adjust P values for multiple comparisons. Results Of the 395 patients (296 men, 98 women, and one patient with sex not reported; mean age, 61 years ± 10), 73 (18.5%) had progression due to non-target disease, 105 (26.6%) had new lesions, and 126 (31.9%) had progression of target lesions (defined by an increase in the sum of diameters). Patients with progression of non-target disease and those with new lesions had shorter PFS than patients with progression defined by the target lesions (median PFS, 2.8 months [95% confidence interval {CI}: 1.9 months, 3.7 months] and 3.6 months [95% CI: 3.3 months, 3.7 months] vs 5.4 months [95% CI: 5.0 months, 5.5 months], respectively [P < .01]). Conclusion The most common causes for radiologic progression of renal cell carcinoma were based on non-target disease and new lesions rather than change in target lesions, despite this being considered uncommon in the Response Evaluation Criteria in Solid Tumors version 1.1 literature. © RSNA, 2019 See also the editorial by Kuhl in this issue.

Deep learning and radiomics: the utility of Google TensorFlow™ Inception in classifying clear cell renal cell carcinoma and oncocytoma on multiphasic CT.

PURPOSE: Currently, all solid enhancing renal masses without microscopic fat are considered malignant until proven otherwise and there is substantial overlap in the imaging findings of benign and malignant renal masses, particularly between clear cell RCC (ccRCC) and benign oncocytoma (ONC). Radiomics has attracted increased attention for its utility in pre-operative work-up on routine clinical images. Radiomics based approaches have converted medical images into mineable data and identified prognostic imaging signatures that machine learning algorithms can use to construct predictive models by learning the decision boundaries of the underlying data distribution. The TensorFlow™ framework from Google is a state-of-the-art open-source software library that can be used for training deep learning neural networks for performing machine learning tasks. The purpose of this study was to investigate the diagnostic value and feasibility of a deep learning-based renal lesion classifier using open-source Google TensorFlow™ Inception in differentiating ccRCC from ONC on routine four-phase MDCT in patients with pathologically confirmed renal masses. METHODS: With institutional review board approval for this 1996 Health Insurance Portability and Accountability Act compliant retrospective study and a waiver of informed consent, we queried our institution's pathology, clinical, and radiology databases for histologically proven cases of ccRCC and ONC obtained between January 2000 and January 2016 scanned with a an intravenous contrast-enhanced four-phase renal mass protocol (unenhanced (UN), corticomedullary (CM), nephrographic (NP), and excretory (EX) phases). To extract features to be used for the machine learning model, the entire renal mass was contoured in the axial plane in each of the four phases, resulting in a 3D volume of interest (VOI) representative of the entire renal mass. We investigated thirteen different approaches to convert the acquired VOI data into a set of images that adequately represented each tumor which was used to train the final layer of the neural network model. Training was performed over 4000 iterations. In each iteration, 90% of the data were designated as training data and the remaining 10% served as validation data and a leave-one-out cross-validation scheme was implemented. Accuracy, sensitivity, specificity, positive (PPV) and negative predictive (NPV) values, and CIs were calculated for the classification of the thirteen processing modes. RESULTS: We analyzed 179 consecutive patients with 179 lesions (128 ccRCC and 51 ONC). The ccRCC cohort had a mean size of 3.8 cm (range 0.8-14.6 cm) and the ONC cohort had a mean lesion size of 3.9 cm (range 1.0-13.1 cm). The highest specificity and PPV (52.9% and 80.3%, respectively) were achieved in the EX phase when we analyzed the single mid-slice of the tumor in the axial, coronal and sagittal plane, and when we increased the number of mid-slices of the tumor to three, with an accuracy of 75.4%, which also increased the sensitivity to 88.3% and the PPV to 79.6%. Using the entire tumor volume also showed that classification performance was best in the EX phase with an accuracy of 74.4%, a sensitivity of 85.8% and a PPV of 80.1%. When the entire tumor volume, plus mid-slices from all phases and all planes presented as tiled images, were submitted to the final layer of the neural network we achieved a PPV of 82.5%. CONCLUSIONS: The best classification result was obtained in the EX phase among the thirteen classification methods tested. Our proof of concept study is the first step towards understanding the utility of machine learning in the differentiation of ccRCC from ONC on routine CT images. We hope this could lead to future investigation into the development of a multivariate machine learning model which may augment our ability to accurately predict renal lesion histology on imaging.

Association of qualitative and quantitative imaging features on multiphasic multidetector CT with tumor grade in clear cell renal cell carcinoma.

PURPOSE: The purpose of the study was to determine if enhancement features and qualitative imaging features on multiphasic multidetector computed tomography (MDCT) were associated with tumor grade in patients with clear cell renal cell carcinoma (ccRCC). METHODS: In this retrospective, IRB approved, HIPAA-compliant, institutional review board-approved study with waiver of informed consent, 127 consecutive patients with 89 low grade (LG) and 43 high grade (HG) ccRCCs underwent preoperative four-phase MDCT in unenhanced (UN), corticomedullary (CM), nephrographic (NP), and excretory (EX) phases. Previously published quantitative (absolute peak lesion enhancement, absolute peak lesion enhancement relative to normal enhancing renal cortex, 3D whole lesion enhancement and the wash-in/wash-out of enhancement within the 3D whole lesion ROI) and qualitative (enhancement pattern; presence of necrosis; pattern of; tumor margin; tumor-parenchymal interface, tumor-parenchymal interaction; intratumoral vascularity; collecting system infiltration; renal vein invasion; and calcification) assessments were obtained for each lesion independently by two fellowship-trained genitourinary radiologists. Comparisons between variables included χ2, ANOVA, and student t test. p values less than 0.05 were considered to be significant. Inter-reader agreement was obtained with the Gwet agreement coefficient (AC1) and standard error (SE) was reported. RESULTS: No significant differences were observed between the LG and HG ccRCC cohorts with respect to absolute peak lesion enhancement and relative lesion enhancement ratio. There was a significant inverse correlation between low and high grade ccRCC and tumor enhancement the NP (71 HU vs. 54 HU, p < 0.001) and EX (52 HU vs. 39 HU, p < 0.001) phases using the 3D whole lesion ROI method. The percent wash-in of 3D enhancement from the UN to the CM phase was also significantly different between LG and HG ccRCCs (352% vs. 255%, p = 0.003). HG lesions showed significantly more calcification, necrosis, collecting system infiltration and ill-defined tumor margins (p < 0.05). Overall agreement between the two readers had a mean AC1 of 0.8172 (SE 0.0235). CONCLUSIONS: Quantitatively, high grade ccRCC had significantly lower whole lesion enhancement in the NP and EX phases on MDCT. Qualitatively, high grade ccRCC were significantly more likely to be associated with calcifications, necrosis, collecting system infiltration, and an ill-defined tumor margin.

Gene Expression Classifier vs Targeted Next-Generation Sequencing in the Management of Indeterminate Thyroid Nodules.

Context: Molecular testing has reduced the need for diagnostic hemithyroidectomy for indeterminate thyroid nodules. No studies have directly compared molecular testing techniques. Objective: Compare the diagnostic performance of Afirma Gene Expression Classifier (GEC) with that of ThyroSeq v2 next-generation sequencing assay. Design: Parallel randomized trial, monthly block randomization of patients with Bethesda III/IV cytology to GEC or ThyroSeq v2. Setting: University of California, Los Angeles. Participants: Patients who underwent thyroid biopsy (April 2016 to June 2017). Intervention: Testing with GEC or ThyroSeq v2. Main Outcome Measure: Molecular test performance. Results: Of 1372 thyroid nodules, 176 (13%) had indeterminate cytology and 149 of 157 eligible indeterminate nodules (95%) were included in the study. Of nodules tested with GEC, 49% were suspicious, 43% were benign, and 9% were insufficient. Of nodules tested with ThyroSeq v2, 19% were mutation positive, 77% were mutation negative, and 4% were insufficient. The specificities of GEC and ThyroSeq v2 were 66% and 91%, respectively (P = 0.002); the positive predictive values of GEC and ThyroSeq v2 were 39% and 57%, respectively. Diagnostic hemithyroidectomy was avoided in 28 patients tested with GEC (39%) and 49 patients tested with ThyroSeq v2 (62%). Surveillance ultrasonography was available for 46 nodules (45 remained stable). Conclusions: ThyroSeq v2 had higher specificity than Afirma GEC and allowed more patients to avoid surgery. Long-term surveillance is necessary to assess the false-negative rate of these particular molecular tests. Further studies are required for comparison with other available molecular diagnostics and for newer tests as they are developed.

Quantitative computer-aided diagnostic algorithm for automated detection of peak lesion attenuation in differentiating clear cell from papillary and chromophobe renal cell carcinoma, oncocytoma, and fat-poor angiomyolipoma on multiphasic multidetector computed tomography.

OBJECTIVE: To evaluate the performance of a novel, quantitative computer-aided diagnostic (CAD) algorithm on four-phase multidetector computed tomography (MDCT) to detect peak lesion attenuation to enable differentiation of clear cell renal cell carcinoma (ccRCC) from chromophobe RCC (chRCC), papillary RCC (pRCC), oncocytoma, and fat-poor angiomyolipoma (fp-AML). MATERIALS AND METHODS: We queried our clinical databases to obtain a cohort of histologically proven renal masses with preoperative MDCT with four phases [unenhanced (U), corticomedullary (CM), nephrographic (NP), and excretory (E)]. A whole lesion 3D contour was obtained in all four phases. The CAD algorithm determined a region of interest (ROI) of peak lesion attenuation within the 3D lesion contour. For comparison, a manual ROI was separately placed in the most enhancing portion of the lesion by visual inspection for a reference standard, and in uninvolved renal cortex. Relative lesion attenuation for both CAD and manual methods was obtained by normalizing the CAD peak lesion attenuation ROI (and the reference standard manually placed ROI) to uninvolved renal cortex with the formula [(peak lesion attenuation ROI - cortex ROI)/cortex ROI] × 100%. ROC analysis and area under the curve (AUC) were used to assess diagnostic performance. Bland-Altman analysis was used to compare peak ROI between CAD and manual method. RESULTS: The study cohort comprised 200 patients with 200 unique renal masses: 106 (53%) ccRCC, 32 (16%) oncocytomas, 18 (9%) chRCCs, 34 (17%) pRCCs, and 10 (5%) fp-AMLs. In the CM phase, CAD-derived ROI enabled characterization of ccRCC from chRCC, pRCC, oncocytoma, and fp-AML with AUCs of 0.850 (95% CI 0.732-0.968), 0.959 (95% CI 0.930-0.989), 0.792 (95% CI 0.716-0.869), and 0.825 (95% CI 0.703-0.948), respectively. On Bland-Altman analysis, there was excellent agreement of CAD and manual methods with mean differences between 14 and 26 HU in each phase. CONCLUSION: A novel, quantitative CAD algorithm enabled robust peak HU lesion detection and discrimination of ccRCC from other renal lesions with similar performance compared to the manual method.