Jeffrey S. Ross

BACKGROUND AND PURPOSE: Studies show a modest yield for image-guided biopsy of suspected vertebral osteomyelitis-discitis. Many studies evaluate factors to improve diagnostic yield, and few studies assess how biopsy results impact clinical management. We aim to evaluate the impact of biopsy results on clinical management in suspected vertebral osteomyelitis-discitis. MATERIALS AND METHODS: We performed a retrospective study of patients who underwent image-guided biopsy for suspected vertebral osteomyelitis-discitis. Data collected included risk factors, imaging findings, laboratory values, antibiotics, biopsy procedure details, microbiology and pathology results and clinical course. Factors assessed for management change included whether biopsy results affected antibiotic type or course, decision to start or stop antibiotics, surgical decisions or if an alternate diagnosis was determined. RESULTS: 310 biopsies were included. Biopsy yield with true positive culture results was 34% (104/310) and similar for patients on antibiotics (36%, 34/94) and off antibiotics (32%, 66/204). Yield was greater when disc was sampled (36%, 82/228) versus bone only (8%, 2/24) and with aspiration of disc and/or bone (42%, 39/92) versus core only (29%, 56/193). With positive blood cultures prior to biopsy, biopsy yield was 50% (22/44) with concordance and discordance rates of 75% (18/24) and 17% (4/24) respectively, and 8% (2/24) of positive biopsy results deemed contaminants. Management was affected in 36% (113/310) of all biopsies and in 78% (81/104) of biopsies with a positive culture result. No management change occurred in 57% (177/310) of biopsies. Management change was unclear in 6% (20/310). Biopsy results changed antibiotics in 27% (85/310). Management change occurred in 23% (10/44) of cases with prior positive blood culture compared to 41% (93/233) without a prior culture source (P = 0.024). Negative culture results influenced management in 16% (32/194). CONCLUSIONS: Image-guided biopsy for vertebral osteomyelitis-discitis has a meaningful impact on management despite modest yield. Greatest management impact is seen with positive culture results, no prior culture source and patients not on antibiotics at the time of biopsy. Biopsy culture yield is not affected by preceding antibiotics, and yield is greater with disc sampling and aspiration. ABBREVIATIONS: VO = vertebral osteomyelitis-discitis; WBC = white blood cell count; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; MC = management change; NMC = no management change; UC = unclear management change; Abx = antibiotics

BACKGROUND AND PURPOSE: We aimed to evaluate GPT-49s ability to write radiology editorials and to compare these with human-written counterparts, thereby determining their real-world applicability for scientific writing. MATERIALS AND METHODS: Sixteen editorials from eight journals were included. To generate the AI-written editorials, the summary of 16 human-written editorials was fed into GPT-4. Six experienced editors reviewed the articles. First, an unpaired approach was used. The raters were asked to evaluate the content of each article using a 1-5 Likert scale across specified metrics. Then, they determined whether the editorials were written by humans or AI. The articles were then evaluated in pairs to determine which article was generated by AI and which should be published. Finally, the articles were analyzed with an AI detector and for plagiarism. RESULTS: The human-written articles had a median AI probability score of 2.0%, whereas the AI-written articles had 58%. The median similarity score among AI-written articles was 3%. 58% of unpaired articles were correctly classified regarding authorship. Rating accuracy was increased to 70% in the paired setting. AI-written articles received slightly higher scores in most metrics. When stratified by perception, human-written perceived articles were rated higher in most categories. In the paired setting, raters strongly preferred publishing the article they perceived as human-written (82%). CONCLUSIONS: GPT-4 can write high-quality articles that iThenticate does not flag as plagiarized, which may go undetected by editors, and that detection tools can detect to a limited extent. Editors showed a positive bias toward human-written articles. ABBREVIATIONS: AI = Artificial intelligence; LLM = large language model; SD = standard deviation.

Myelopathy and radiculopathy can be due to extrinsic causes, most often degenerative in origin. Imaging can elucidate an often-confusing clinical picture to guide management and provide prognostic information. For intrinsic causes of myelopathy, the differential diagnosis categories include demyelination, inflammation, infection, vascular, and neoplasm. The combination of clinical symptoms, timing of presentation, and imaging features can help narrow the differential diagnosis.

Collective Thomson scattering is a powerful diagnostic technique that is capable of local measurements of various plasma parameters including electron density, electron temperature, ion temperature, and velocity. We have developed an optical Thomson scattering system on the shear flow Z-pinch machine FuZE at ZAP Energy inc. in Seattle Washington to measure at least the electron temperature and density. The system uses an Nd-YAG laser that is frequency-doubled to produce 8 J at 532 nm. Scattering is collected at an angle of 15° from 8 points within the plasma column and delivered by a fiber bundle to 2 spectrometers with gated CMOS cameras. One spectrometer records the high-frequency electron plasma wave (EPW) feature, while the other spectrometer records the low-frequency ion acoustic wave (IAW) feature. The recorded data can be compared to a theoretical scattering form factor to measure densities above 1 x 10 18 cm -3 , electron temperatures above 100 eV, and flow velocities above 20 km/s. Due to the high rep rate of the machine (1 shot every ~ 10 min) the time evolution of the plasma parameters over the long quiescent period of the pinch (~10 μs) can be studied by probing different times with the laser. Initial data will be presented.

604 Background: MTAP loss is represented across a wide variety of cancer types including PDAC and is an emerging target for synthetic lethality-based cancer therapies. Preclinically, MTAP loss leads to the accumulation of 2-methylthioadenosine, reduced protein arginine N-methyltransferase 5 (PRMT5) methylation activity and increased vulnerability to targeting of the methionine adenosyltransferase IIα (MAT2A)/ PRMT5 axis. In addition, 9p21 loss, homozygous co-deletion of MTAP/CDKN2A or homozygous deletion of either gene have been associated with an immunologically “cold” tumor microenvironment, primary resistance to anti PD(L)1 immunotherapy (IO) and poor prognosis phenotype (Han G, Nat Commun 2021). We investigated concurrent mutations and immune biomarkers in clinical PDAC samples with MTAP-loss versus -intact status. Methods: From a series of 177705 consecutive cases, we performed comprehensive genomic profiling on 9423 cases of PDAC using an FDA-approved assay (F1CDx) to evaluate all classes of genomic alterations (GA). Tumor mutational burden (TMB) was determined on up to 1.1 Mbp of sequenced DNA and microsatellite instability (MSI) was determined on 114 loci. PD-L1 expression was determined by immunohistochemistry (Dako 22C3). Furthermore, we correlated pertinent findings within a database of 16558 cases of clinically advanced cancer with MTAP loss. Results: 2003 (21.3%) of 9423 PDAC demonstrated MTAP-loss. Similar gender, age and number of GA per tumor were observed between MTAP-loss and -intact groups. Frequencies of TP53, CDKN2A/B, SMAD4, PTEN and ARID1A were significantly higher in MTAP-loss PDAC. However, previously-described biomarkers of IO efficacy (MSI, TMB, CD274 amplification and PD-L1 expression) and resistance ( STK11, KEAP1 and MDM2) were infrequent and similar in both groups. The frequencies of other potentially targetable GA including BRCA1/2, ATM, KRAS G12C, ERBB2, BRAF, FGFR1, NF1 and PIK3CA were also infrequent and similar in both groups of PDAC patients. Amongst a database of 16558 cases of clinically advanced cancer with MTAP loss, 1538 (9.3%) featured co-alterations in MTAP and SMAD4. 52% of the MTAP/SMAD4 co-altered cases were PDAC. Conclusions: MTAP loss is associated with a distinctive concurrent genomic profile in PDAC and represents a potential new synthetic lethality-based opportunity for treatment with PRMT5 and MAT2A inhibitors. Furthermore, MTAP loss may represent an independent negative predictive biomarker for immune checkpoint inhibition in PDAC.[Table: see text]

Applications of deep learning (DL) in medical imaging have proliferated in the last few years. DL systems have proved particularly effective for segmenting organs and lesions in MRI and CT image volumes. By their nature, DL systems tend to be “black boxes,” unable to provide insight into how their segmentation results were obtained. Consequently, a lingering issue is the reproduction and validation of the many compelling results.