Multi-reader multiparametric DECT study evaluating different strengths of iterative and deep learning-based image reconstruction techniques.

OBJECTIVES: To perform a multi-reader comparison of multiparametric dual-energy computed tomography (DECT) images reconstructed with deep-learning image reconstruction (DLIR) and standard-of-care adaptive statistical iterative reconstruction-V (ASIR-V). METHODS: This retrospective study included 100 patients undergoing portal venous phase abdominal CT on a rapid kVp switching DECT scanner. Six reconstructed DECT sets (ASIR-V and DLIR, each at three strengths) were generated. Each DECT set included 65 keV monoenergetic, iodine, and virtual unenhanced (VUE) images. Using a Likert scale, three radiologists performed qualitative assessments for image noise, contrast, small structure visibility, sharpness, artifact, and image preference. Quantitative assessment was performed by measuring attenuation, image noise, and contrast-to-noise ratios (CNR). For the qualitative analysis, Gwet's AC2 estimates were used to assess agreement. RESULTS: DECT images reconstructed with DLIR yielded better qualitative scores than ASIR-V images except for artifacts, where both groups were comparable. DLIR-H images were rated higher than other reconstructions on all parameters (p-value < 0.05). On quantitative analysis, there was no significant difference in the attenuation values between ASIR-V and DLIR groups. DLIR images had higher CNR values for the liver and portal vein, and lower image noise, compared to ASIR-V images (p-value < 0.05). The subgroup analysis of patients with large body habitus (weight ≥ 90 kg) showed similar results to the study population. Inter-reader agreement was good-to-very good overall. CONCLUSION: Multiparametric post-processed DECT datasets reconstructed with DLIR were preferred over ASIR-V images with DLIR-H yielding the highest image quality scores. CLINICAL RELEVANCE STATEMENT: Deep-learning image reconstruction in dual-energy CT demonstrated significant benefits in qualitative and quantitative image metrics compared to adaptive statistical iterative reconstruction-V. KEY POINTS: Dual-energy CT (DECT) images reconstructed using deep-learning image reconstruction (DLIR) showed superior qualitative scores compared to adaptive statistical iterative reconstruction-V (ASIR-V) reconstructed images, except for artifacts where both reconstructions were rated comparable. While there was no significant difference in attenuation values between ASIR-V and DLIR groups, DLIR images showed higher contrast-to-noise ratios (CNR) for liver and portal vein, and lower image noise (p value < 0.05). Subgroup analysis of patients with large body habitus (weight ≥ 90 kg) yielded similar findings to the overall study population.

Implementation of an Institution-Wide Rules-Based Automated CT Protocoling System.

BACKGROUND. Examination protocoling is a noninterpretive task that increases radiologists' workload and can cause workflow inefficiencies. OBJECTIVE. The purpose of this study was to evaluate effects of an automated CT protocoling system on examination process times and protocol error rates. METHODS. This retrospective study included 317,597 CT examinations (mean age, 61.8 ± 18.1 [SD] years; male, 161,125; female, 156,447; unspecified sex, 25) from July 2020 to June 2022. A rules-based automated protocoling system was implemented institution-wide; the system evaluated all CT orders in the EHR and assigned a protocol or directed the order for manual radiologist protocoling. The study period comprised pilot (July 2020 to December 2020), implementation (January 2021 to December 2021), and postimplementation (January 2022 to June 2022) phases. Proportions of automatically protocoled examinations were summarized. Process times were recorded. Protocol error rates were assessed by counts of quality improvement (QI) reports and examination recalls and comparison with retrospectively assigned protocols in 450 randomly selected examinations. RESULTS. Frequency of automatic protocoling was 19,366/70,780 (27.4%), 68,875/163,068 (42.2%), and 54,045/83,749 (64.5%) in pilot, implementation, and postimplementation phases, respectively (p < .001). Mean (± SD) times from order entry to protocol assignment for automatically and manually protocoled examinations for emergency department examinations were 0.2 ± 18.2 and 2.1 ± 69.7 hours, respectively; mean inpatient examination times were 0.5 ± 50.0 and 3.5 ± 105.5 hours; and mean outpatient examination times were 361.7 ± 1165.5 and 1289.9 ± 2050.9 hours (all p < .001). Mean (± SD) times from order entry to examination completion for automatically and manually protocoled examinations for emergency department examinations were 2.6 ± 38.6 and 4.2 ± 73.0 hours, respectively (p < .001); for inpatient examinations were 6.3 ± 74.6 and 8.7 ± 109.3 hours (p = .001); and for outpatient examinations were 1367.2 ± 1795.8 and 1471.8 ± 2118.3 hours (p < .001). In the three phases, there were three, 19, and 25 QI reports and zero, one, and three recalls, respectively, for automatically protocoled examinations, versus nine, 19, and five QI reports and one, seven, and zero recalls for manually protocoled examinations. Retrospectively assigned protocols were concordant with 212/214 (99.1%) of automatically protocoled versus 233/236 (98.7%) of manually protocoled examinations. CONCLUSION. The automated protocoling system substantially reduced radiologists' protocoling workload and decreased times from order entry to protocol assignment and examination completion; protocol errors and recalls were infrequent. CLINICAL IMPACT. The system represents a solution for reducing radiologists' time spent performing noninterpretive tasks and improving care efficiency.

Society of Radiologists in Ultrasound Consensus Conference Recommendations for Incidental Gallbladder Polyp Management: Interreader Agreement Among 10 Radiologists.

BACKGROUND. The 2022 Society of Radiologists in Ultrasound (SRU) consensus conference recommendations for small gallbladder polyps support management that is less aggressive than earlier approaches and may help standardize evaluation of polyps by radiologists. OBJECTIVE. The purpose of the present study was to assess the interreader agreement of radiologists in applying SRU recommendations for management of incidental gallbladder polyps on ultrasound. METHODS. This retrospective study included 105 patients (75 women and 30 men; median age, 51 years) with a gallbladder polyp on ultrasound (without features highly suspicious for invasive or malignant tumor) who underwent cholecystectomy between January 1, 2003, and January 1, 2021. Ten abdominal radiologists independently reviewed ultrasound examinations and, using the SRU recommendations, assessed one polyp per patient to assign risk category (extremely low risk, low risk, or indeterminate risk) and make a possible recommendation for surgical consultation. Five radiologists were considered less experienced (< 5 years of experience), and five were considered more experienced (≥ 5 years of experience). Interreader agreement was evaluated. Polyps were classified pathologically as nonneoplastic or neoplastic. RESULTS. For risk category assignments, interreader agreement was substantial among all readers (k = 0.710), less-experienced readers (k = 0.705), and more-experienced readers (k = 0.692). For surgical consultation recommendations, inter-reader agreement was substantial among all readers (k = 0.795) and more-experienced readers (k = 0.740) and was almost perfect among less-experienced readers (k = 0.811). Of 10 readers, a median of 5.0 (IQR, 2.0-8.0), 4.0 (IQR, 2.0-7.0), and 0.0 (IQR, 0.0-0.0) readers classified polyps as extremely low risk, low risk, and indeterminate risk, respectively. Across readers, the percentage of polyps classified as extremely low risk ranged from 32% to 72%; as low risk, from 24% to 65%; and as indeterminate risk, from 0% to 8%. Of 10 readers, a median of zero change to 0 (IQR, 0.0-1.0) readers recommended surgical consultation; the percentage of polyps receiving a recommendation for surgical consultation ranged from 4% to 22%. Of a total of 105 polyps, 102 were nonneo-plastic and three were neoplastic (all benign). Based on readers' most common assessments for nonneoplastic polyps, the risk category was extremely low risk for 53 polyps, low risk for 48 polyps, and indeterminate risk for one polyp; surgical consultation was recommended for 16 polyps. CONCLUSION. Ten abdominal radiologists showed substantial agreement for polyp risk categorizations and surgical consultation recommendations, although areas of reader variability were identified. CLINICAL IMPACT. The findings support the overall reproducibility of the SRU recommendations, while indicating opportunity for improvement.

Clinical Intricacies and Advances in Neuroendocrine Tumors: An Organ-Based Multidisciplinary Approach.

ABSTRACT: Neuroendocrine neoplasms (NENs) are rare neoplasms originating from neuroendocrine cells, with increasing incidence due to enhanced detection methods. These tumors display considerable heterogeneity, necessitating diverse management strategies based on factors like organ of origin and tumor size. This article provides a comprehensive overview of therapeutic approaches for NENs, emphasizing the role of imaging in treatment decisions. It categorizes tumors based on their locations: gastric, duodenal, pancreatic, small bowel, colonic, rectal, appendiceal, gallbladder, prostate, lung, gynecological, and others. The piece also elucidates the challenges in managing metastatic disease and controversies surrounding MEN1-neuroendocrine tumor management. The article underscores the significance of individualized treatment plans, underscoring the need for a multidisciplinary approach to ensure optimal patient outcomes.

Prevalence of Malignancy in Adrenal Nodules With Heterogeneous Microscopic Fat on Chemical-Shift MRI: A Multiinstitutional Study.

BACKGROUND. Homogeneous microscopic fat within adrenal nodules on chemical-shift MRI (CS-MRI) is diagnostic of benign adrenal adenoma, but the clinical relevance of heterogeneous microscopic fat is not well established. OBJECTIVE. This study sought to determine the prevalence of malignancy in adrenal nodules with heterogeneous microscopic fat on dual-echo T1-weighted CS-MRI. METHODS. We performed a retrospective study of adult patients with adrenal nodules detected on MRI performed between August 2007 and November 2020 at seven institutions. Eligible nodules had a short-axis diameter of 10 mm or larger with heterogeneous microscopic fat (defined by an area of signal loss of < 80% on opposed-phase CS-MRI). Two radiologists from each center, blinded to reference standard results, determined the signal loss pattern (diffuse, two distinct parts, speckling pattern, central loss, or peripheral loss) within the nodules. The reference standard used was available for 283 nodules (pathology for 21 nodules, ≥ 1 year of imaging follow-up for 245, and ≥ 5 years of clinical follow-up for 17) in 282 patients (171 women and 111 men; mean age, 60 ± 12 [SD] years); 30% (86/282) patients had prior malignancy. RESULTS. The mean long-axis diameter was 18.7 ± 7.9 mm (range, 10-80 mm). No malignant nodules were found in patients without prior cancer (0/197; 95% CI, 0-1.5%). Four of the 86 patients with prior malignancy (hepatocellular carcinoma [HCC], renal cell carcinoma [RCC], lung cancer, or both colon cancer and RCC) (4.7%; 95% CI, 1.3-11.5%) had metastatic nodules. Detected patterns were diffuse heterogeneous signal loss (40% [114/283]), speckling (28% [80/283]), two distinct parts (18% [51/283]), central loss (9% [26/283]), and peripheral loss (4% [12/283]). Two metastases from HCC and RCC showed diffuse heterogeneous signal loss. Lung cancer metastasis manifested as two distinct parts, and the metastasis in the patient with both colon cancer and RCC showed peripheral signal loss. CONCLUSION. Presence of heterogeneous microscopic fat in adrenal nodules on CS-MRI indicates a high likelihood of benignancy, particularly in patients without prior cancer. This finding is also commonly benign in patients with cancer; however, caution is warranted when primary malignancies may contain fat or if the morphologic pattern of signal loss may indicate a collision tumor. CLINICAL IMPACT. In the absence of prior cancer, adrenal nodules with heterogeneous microscopic fat do not require additional imaging evaluation.

Adrenal Neoplasms: Lessons from Adrenal Multidisciplinary Tumor Boards.

The radiologic diagnosis of adrenal disease can be challenging in settings of atypical presentations, mimics of benign and malignant adrenal masses, and rare adrenal anomalies. Misdiagnosis may lead to suboptimal management and adverse outcomes. Adrenal adenoma is the most common benign adrenal tumor that arises from the cortex, whereas adrenocortical carcinoma (ACC) is a rare malignant tumor of the cortex. Adrenal cyst and myelolipoma are other benign adrenal lesions and are characterized by their fluid and fat content, respectively. Pheochromocytoma is a rare neuroendocrine tumor of the adrenal medulla. Metastases to the adrenal glands are the most common malignant adrenal tumors. While many of these masses have classic imaging appearances, considerable overlap exists between benign and malignant lesions and can pose a diagnostic challenge. Atypical adrenal adenomas include those that are lipid poor; contain macroscopic fat, hemorrhage, and/or iron; are heterogeneous and/or large; and demonstrate growth. Heterogeneous adrenal adenomas may mimic ACC, metastasis, or pheochromocytoma, particularly when they are 4 cm or larger, whereas smaller versions of ACC, metastasis, and pheochromocytoma and those with washout greater than 60% may mimic adenoma. Because of its nonenhanced CT attenuation of less than or equal to 10 HU, a lipid-rich adrenal adenoma may be mimicked by a benign adrenal cyst, or it may be mimicked by a tumor with central cystic and/or necrotic change such as ACC, pheochromocytoma, or metastasis. Rare adrenal tumors such as hemangioma, ganglioneuroma, and oncocytoma also may mimic adrenal adenoma, ACC, metastasis, and pheochromocytoma. The authors describe cases of adrenal neoplasms that they have encountered in clinical practice and presented to adrenal multidisciplinary tumor boards. Key lessons to aid in diagnosis and further guide appropriate management are provided. © RSNA, 2023 Online supplemental material is available for this article. Quiz questions for this article are available through the Online Learning Center.

Hereditary and Sporadic Pheochromocytoma: Comparison of Imaging, Clinical, and Laboratory Features.

BACKGROUND. A considerable fraction of pheochromocytomas initially suspected to be sporadic, whether or not symptomatic, are a result of germline mutations. OBJECTIVE. The purpose of this article is to compare imaging features between hereditary and sporadic pheochromocytomas. METHODS. This retrospective study included 71 patients (39 women, 32 men; median age, 48 years) who underwent adrenal pheochromocytoma resection from January 2002 to October 2021 after preoperative CT or MRI. Two radiologists independently reviewed examinations to assess features of the largest resected pheochromocytoma. Interreader agreement was assessed by prevalence-adjusted bias-adjusted kappa coefficients; a third radiologist resolved discrepancies for further analysis. Genetic testing was used to classify pheochromocytomas as hereditary or sporadic and to classify hereditary pheochromocytomas by germline mutation clusters. Symptoms associated with pheochromocytomas and preoperative biochemical laboratory values were recorded. Groups were compared using Kruskal-Wallis, Fisher exact, and chi-square tests, and false-discovery rate-adjusted p values were computed to account for multiple comparisons. RESULTS. Hereditary pheochromocytoma (n = 32), compared with sporadic pheochromocytoma (n = 39), was associated with younger median age (38 vs 52 years, p = .001) and smaller median size (24 vs 40 mm, p < .001). Interreader agreement for CT and MRI features, expressed as kappa, ranged from 0.44 to 1.00. Hereditary and sporadic pheochromocytoma showed no difference in frequency of calcifications, hemorrhage, cystic change/necrosis, or macroscopic fat on CT, or in frequency of hemorrhage, cystic change/necrosis, macroscopic fat, or microscopic fat on MRI (p > .05). When combining CT and MRI, cystic change/necrosis was observed in 35% of hereditary versus 67% of sporadic pheochromocytomas (p = .10). Hereditary pheochromocytoma, compared with sporadic, had lower frequency of symptoms (31% vs 74%; p = .004) and lower 24-hour urinary normetanephrines (1.1 vs 5.1 times upper limits of normal, p = .006). Among hereditary pheochromocytomas, cystic change/necrosis (when assessable on imaging) was present in 18% and 45% of those with cluster 1 (n = 11) and cluster 2 (n = 21) germ-line mutations, respectively. CONCLUSION. Hereditary pheochromocytomas, compared with sporadic, are detected at a younger age and smaller size, produce lower 24-hour urinary normetanephrines, are less often symptomatic, and may less frequently show cystic change/necrosis. CLINICAL IMPACT. Imaging findings may complement clinical and biochemical features in raising suspicion for a previously unsuspected germline mutation in patients with pheochromocytoma.

Comparison of four methods to measure haemoglobin concentrations in whole blood donors (COMPARE): A diagnostic accuracy study.

OBJECTIVE: To compare four haemoglobin measurement methods in whole blood donors. BACKGROUND: To safeguard donors, blood services measure haemoglobin concentration in advance of each donation. NHS Blood and Transplant's (NHSBT) customary method have been capillary gravimetry (copper sulphate), followed by venous spectrophotometry (HemoCue) for donors failing gravimetry. However, NHSBT's customary method results in 10% of donors being inappropriately bled (ie, with haemoglobin values below the regulatory threshold). METHODS: We compared the following four methods in 21 840 blood donors (aged ≥18 years) recruited from 10 NHSBT centres in England, with the Sysmex XN-2000 haematology analyser, the reference standard: (1) NHSBT's customary method; (2) "post donation" approach, that is, estimating current haemoglobin concentration from that measured by a haematology analyser at a donor's most recent prior donation; (3) "portable haemoglobinometry" (using capillary HemoCue); (4) non-invasive spectrometry (using MBR Haemospect or Orsense NMB200). We assessed sensitivity; specificity; proportion who would have been inappropriately bled, or rejected from donation ("deferred") incorrectly; and test preference. RESULTS: Compared with the reference standard, the methods ranged in test sensitivity from 17.0% (MBR Haemospect) to 79.0% (portable haemoglobinometry) in men, and from 19.0% (MBR Haemospect) to 82.8% (portable haemoglobinometry) in women. For specificity, the methods ranged from 87.2% (MBR Haemospect) to 99.9% (NHSBT's customary method) in men, and from 74.1% (Orsense NMB200) to 99.8% (NHSBT's customary method) in women. The proportion of donors who would have been inappropriately bled ranged from 2.2% in men for portable haemoglobinometry to 18.9% in women for MBR Haemospect. The proportion of donors who would have been deferred incorrectly with haemoglobin concentration above the minimum threshold ranged from 0.1% in men for NHSBT's customary method to 20.3% in women for OrSense. Most donors preferred non-invasive spectrometry. CONCLUSION: In the largest study reporting head-to-head comparisons of four methods to measure haemoglobin prior to blood donation, our results support replacement of NHSBT's customary method with portable haemoglobinometry.

Interreader Concordance of the TI-RADS: Impact of Radiologist Experience.

OBJECTIVE. The objective of this article is to assess radiologist concordance in characterizing thyroid nodules using the American College of Radiology Thyroid Imaging Reporting and Data System (TI-RADS), focusing on the effect of radiologist experience on reader concordance. MATERIALS AND METHODS. Three experienced and three less experienced radiologists assessed 150 thyroid nodules using the TI-RADS lexicon. Percent concordance was determined for various endpoints. RESULTS. Interreader concordance for the five TI-RADS categories was 87.2% for shape, 81.2% for composition, 76.1% for echogenicity, 72.9% for margins, and 69.8% for echogenic foci. Concordance for individual features was 96.3% for rim calcifications, 90.8% for macrocalcifications, 90.1% for spongiform, 83.5% for comet tail artifact, and 77.7% for punctate echogenic foci. Concordance for the TI-RADS level and recommendation for fine-needle aspiration (FNA) were 50.4% and 78.9%, respectively. Concordance was significantly (p < 0.05) higher for less experienced readers in identifying margins (84.3% vs 67.4%), echogenic foci (76.9% vs 69.3%), comet tail artifact (89.6% vs 79.2%), and punctate echogenic foci (85.3% vs 75.5%), and lower for peripheral rim calcifications (95.0% vs 97.8 %), but was not different (p > 0.05) for the remaining categories and features. CONCLUSION. A range of TI-RADS categories, features, and recommendations for FNA had generally moderate interreader agreement among six radiologists. Our results show that concordance for numerous characteristics was significantly higher for the less experienced versus the more experienced readers. These results suggest that less experienced readers relied more on the explicit TI-RADS criteria, whereas the experienced radiologists partially relied on their accumulated experience when forming impressions. However, the overall TI-RADS level and recommendation for FNA were unaffected, supporting the robustness of the TI-RADS lexicon and its continued use in practice.

Biases incurred from nonrandom repeat testing of haemoglobin levels in blood donors: Selective testing and its implications.

To help prevent anaemia, it is a requisite for blood donors to undergo a haemoglobin test to ensure levels are not too low before donation. It is therefore important to have an accurate testing device and strategy to ensure donors are not being inappropriately bled. A recent study in blood donors used a selective testing strategy where if a donor's haemoglobin level is below the level required for donation, then another reading is taken and if this occurs again, a third and final reading is used. This strategy can reduce the average number of readings required per donor compared to taking three measurements for all donors. However, the final decision-making measurement will on average be higher than a single measurement. In this paper, a selective testing strategy is compared against other strategies. Individual-level biases are derived for the selective strategy and are shown to depend on how close a donor's true haemoglobin level is to the donation threshold and the magnitude of error in the testing device. A simulation study was conducted using the distribution of haemoglobin levels from a large donor population to investigate the effects different strategies have on population performance. We consider scenarios based on varying the measurement device bias and error, including differential biases that depend on the underlying haemoglobin level. Discriminatory performance is shown to be affected when using the selective testing strategies, especially when measurement error is large and when differential bias is present in the device. We recommend that the average of a number of readings should be used in preference to selective testing strategies if multiple measurements are available.

Synthesis of Benzodihydrofurans by Asymmetric C-H Insertion Reactions of Donor/Donor Rhodium Carbenes.

Metal carbenes appended with two electron-donating groups, known as "donor/donor" carbenes, undergo diastereo- and enantioselective rhodium-catalyzed C-H insertion reactions with ether substrates to form benzodihydrofurans. Unlike the reactions of metal carbenes with electron-withdrawing groups attached, the attenuated electrophilicity enables these reactions to be conducted in Lewis basic solvents (e.g., acetonitrile) and in the presence of water. The diazo precursors for these species are prepared in situ from hydrazone using a mild and chemoselective oxidant (MnO2 ). Although this sequence often can be performed in one-pot, control experiments have elucidated why a "two-pot" process is often more efficient. A thorough screening of achiral catalysts demonstrated that sterically encumbered catalysts are optimal for diastereoselective reactions. Although efficient insertion into allylic and propargylic C-H bonds is observed, competing dipolar cycloaddition processes are noted for some substrates. The full substrate scope of this useful method of benzodihydrofuran synthesis, mechanisms of side reactions, and computational support for the origins of stereoselectivity are described.

Synthesis of ß-Ketosulfonamides Derived from Amino Acids and Their Conversion to ß-Keto-α,α-difluorosulfonamides via Electrophilic Fluorination.

ß-Ketosulfonamides derived from Boc or Cbz-protected amino acids bearing hydrophobic side chains were prepared in good to excellent yield by treating N-allyl, N-alkyl methanesulfonamides with n-BuLi, followed by reaction of the resulting carbanion with methyl esters of N-protected l-amino acids. The analogous reaction using the dianion derived from an N-alkyl methanesulfonamide proceeded in much lower yield. Electrophilic fluorination of the ß-ketosulfonamides using Selectfluor in the presence of CsF in DMF at room temperature for 15-60 min provided ß-keto-α,α-difluorosulfonamides in good to excellent yields. The allyl protecting group could be removed in good yield using cat. Pd(PPh)3)4 and dimethyl barbituric acid. When the fluorination reaction was performed with Cs2CO3 as base, ß-ketosulfonamides derived from Val, Leu or Ile gave the expected ß-keto-α,α-difluorosulfonamides, while ß-ketosulfonamides derived from Ala, Phe, or hPhe gave the hydrates of the imino ß-keto-α,α-difluorosulfonamides.

Current genomic deep learning models display decreased performance in cell type specific accessible regions.

Background: A number of deep learning models have been developed to predict epigenetic features such as chromatin accessibility from DNA sequence. Model evaluations commonly report performance genome-wide; however, cis regulatory elements (CREs), which play critical roles in gene regulation, make up only a small fraction of the genome. Furthermore, cell type specific CREs contain a large proportion of complex disease heritability. Results: We evaluate genomic deep learning models in chromatin accessibility regions with varying degrees of cell type specificity. We assess two modeling directions in the field: general purpose models trained across thousands of outputs (cell types and epigenetic marks), and models tailored to specific tissues and tasks. We find that the accuracy of genomic deep learning models, including two state-of-the-art general purpose models - Enformer and Sei - varies across the genome and is reduced in cell type specific accessible regions. Using accessibility models trained on cell types from specific tissues, we find that increasing model capacity to learn cell type specific regulatory syntax - through single-task learning or high capacity multi-task models - can improve performance in cell type specific accessible regions. We also observe that improving reference sequence predictions does not consistently improve variant effect predictions, indicating that novel strategies are needed to improve performance on variants. Conclusions: Our results provide a new perspective on the performance of genomic deep learning models, showing that performance varies across the genome and is particularly reduced in cell type specific accessible regions. We also identify strategies to maximize performance in cell type specific accessible regions.

Current genomic deep learning models display decreased performance in cell type-specific accessible regions.

BACKGROUND: A number of deep learning models have been developed to predict epigenetic features such as chromatin accessibility from DNA sequence. Model evaluations commonly report performance genome-wide; however, cis regulatory elements (CREs), which play critical roles in gene regulation, make up only a small fraction of the genome. Furthermore, cell type-specific CREs contain a large proportion of complex disease heritability. RESULTS: We evaluate genomic deep learning models in chromatin accessibility regions with varying degrees of cell type specificity. We assess two modeling directions in the field: general purpose models trained across thousands of outputs (cell types and epigenetic marks) and models tailored to specific tissues and tasks. We find that the accuracy of genomic deep learning models, including two state-of-the-art general purpose models-Enformer and Sei-varies across the genome and is reduced in cell type-specific accessible regions. Using accessibility models trained on cell types from specific tissues, we find that increasing model capacity to learn cell type-specific regulatory syntax-through single-task learning or high capacity multi-task models-can improve performance in cell type-specific accessible regions. We also observe that improving reference sequence predictions does not consistently improve variant effect predictions, indicating that novel strategies are needed to improve performance on variants. CONCLUSIONS: Our results provide a new perspective on the performance of genomic deep learning models, showing that performance varies across the genome and is particularly reduced in cell type-specific accessible regions. We also identify strategies to maximize performance in cell type-specific accessible regions.

Cannabis Use and Head and Neck Cancer.

Importance: Cannabis is the most commonly used illicit substance worldwide. Whether cannabis use is associated with head and neck cancer (HNC) is unclear. Objective: To assess the clinical association between cannabis use and HNC. Design, Setting, and Participants: This large multicenter cohort study used clinical records from a database that included 20 years of data (through April 2024) from 64 health care organizations. A database was searched for medical records for US adults with and without cannabis-related disorder who had recorded outpatient hospital clinic visits and no prior history of HNC. Propensity score matching was performed for demographic characteristics, alcohol-related disorders, and tobacco use. Subsequently, relative risks (RRs) were calculated to explore risk of HNC, including HNC subsites. This analysis was repeated among those younger than 60 years and 60 years or older. Exposure: Cannabis-related disorder. Main Outcomes and Measures: Diagnosis of HNC and any HNC subsite. Results: The cannabis-related disorder cohort included 116 076 individuals (51 646 women [44.5%]) with a mean (SD) age of 46.4 (16.8) years. The non-cannabis-related disorder cohort included 3 985 286 individuals (2 173 684 women [54.5%]) with a mean (SD) age of 60.8 (20.6) years. The rate of new HNC diagnosis in all sites was higher in the cannabis-related disorder cohort. After matching (n = 115 865 per group), patients with cannabis-related disorder had a higher risk of any HNC (RR, 3.49; 95% CI, 2.78-4.39) than those without HNC. A site-specific analysis yielded that those with cannabis-related disorder had a higher risk of oral (RR, 2.51; 95% CI, 1.81-3.47), oropharyngeal (RR, 4.90; 95% CI, 2.99-8.02), and laryngeal (RR, 8.39; 95% CI, 4.72-14.90) cancer. Results were consistent when stratifying by older and younger age group. Conclusions and Relevance: This cohort study highlights an association between cannabis-related disorder and the development of HNC in adult patients. Given the limitations of the database, future research should examine the mechanism of this association and analyze dose response with strong controls to further support evidence of cannabis use as a risk factor for HNCs.

Mechanisms of crosstalk between the oropharyngeal microbiome and human papillomavirus in oropharyngeal carcinogenesis: a mini review.

Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer globally. Notably, human papillomavirus (HPV)-positive oropharyngeal squamous cell carcinoma (OPSCC) is on the rise, accounting for 70% of all OPSCC cases. Persistent high-risk HPV infection is linked to various cancers, but HPV infection alone is not sufficient to cause cancer. Advances in next-generation sequencing have improved our understanding of changes in the human microbiome of cancerous environments. Yet, there remains a dearth of knowledge on the impact of HPV-microbiome crosstalk in HPV-positive OPSCC. In this review, we examine what is known about the oropharyngeal microbiome and the compositional shifts in this microbiome in HPV-positive OPSCC. We also review potential mechanisms of crosstalk between HPV and specific microorganisms. Additional research is needed to understand these interactions and their roles on cancer development and progression.

Age and sex specific thresholds for risk stratification of cardiovascular disease and clinical decision making: prospective open cohort study.

Objective: To quantify the potential advantages of using 10 year risk prediction models for cardiovascular disease, in combination with risk thresholds specific to both age and sex, to identify individuals at high risk of cardiovascular disease for allocation of statin treatment. Design: Prospective open cohort study. Setting: Primary care data from the UK Clinical Practice Research Datalink GOLD, linked with hospital admissions from Hospital Episode Statistics and national mortality records from the Office for National Statistics in England, 1 January 2006 to 31 May 2019. Participants: 1 046 736 individuals (aged 40-85 years) with no cardiovascular disease, diabetes, or a history of statin treatment at baseline using data from electronic health records. Main outcome measures: 10 year risk of cardiovascular disease, calculated with version 2 of the QRISK cardiovascular disease risk algorithm (QRISK2), with two main strategies to identify individuals at high risk: in strategy A, estimated risk was a fixed cut-off value of ≥10% (ie, as per the UK National Institute for Health and Care Excellence guidelines); in strategy B, estimated risk was ≥10% or ≥90th centile of age and sex specific risk distributions. Results: Compared with strategy A, strategy B stratified 20 241 (149.8%) more women aged ≤53 years and 9832 (150.2%) more men aged ≤47 years as having a high risk of cardiovascular disease; for all other ages the strategies were the same. Assuming that treatment with statins would be initiated in those identified as high risk, differences in the estimated gain in cardiovascular disease-free life years from statin treatment for strategy B versus strategy A were 0.14 and 0.16 years for women and men aged 40 years, respectively; among individuals aged 40-49 years, the numbers needed to treat to prevent one cardiovascular disease event for strategy B versus strategy A were 39 versus 21 in women and 19 versus 15 in men, respectively. Conclusions: This study quantified the potential gains in cardiovascular disease-free life years when implementing prevention strategies based on age and sex specific risk thresholds instead of a fixed risk threshold for allocation of statin treatment. Such gains should be weighed against the costs of treating more younger people with statins for longer.