Performance of an Automated Deep Learning Algorithm to Identify Hepatic Steatosis Within Noncontrast Computed Tomography Scans Among People with and Without HIV

Overview

Journal Pharmacoepidemiol Drug Saf

Publisher Wiley

Specialties Pharmacology
Public Health

Date 2023 Jun 5

PMID 37276449

Authors

Jessie Torgersen

Scott Akers

Yuankai Huo

James G Terry

J Jeffrey Carr

Alexander T Ruutiainen

Melissa Skanderson

Woody Levin

Joseph K Lim

Tamar H Taddei

Kaku So-Armah

Debika Bhattacharya

Christopher T Rentsch

Li Shen

Rotonya Carr

Russell T Shinohara

Michele McClain

Matthew Freiberg

Amy C Justice

Vincent Lo Re 3rd

Affiliations

Soon will be listed here.

Abstract

Purpose: Hepatic steatosis (fatty liver disease) affects 25% of the world's population, particularly people with HIV (PWH). Pharmacoepidemiologic studies to identify medications associated with steatosis have not been conducted because methods to evaluate liver fat within digitized images have not been developed. We determined the accuracy of a deep learning algorithm (automatic liver attenuation region-of-interest-based measurement [ALARM]) to identify steatosis within clinically obtained noncontrast abdominal CT images compared to manual radiologist review and evaluated its performance by HIV status.

Methods: We performed a cross-sectional study to evaluate the performance of ALARM within noncontrast abdominal CT images from a sample of patients with and without HIV in the US Veterans Health Administration. We evaluated the ability of ALARM to identify moderate-to-severe hepatic steatosis, defined by mean absolute liver attenuation <40 Hounsfield units (HU), compared to manual radiologist assessment.

Results: Among 120 patients (51 PWH) who underwent noncontrast abdominal CT, moderate-to-severe hepatic steatosis was identified in 15 (12.5%) persons via ALARM and 12 (10%) by radiologist assessment. Percent agreement between ALARM and radiologist assessment of absolute liver attenuation <40 HU was 95.8%. Sensitivity, specificity, positive predictive value, and negative predictive value of ALARM were 91.7% (95%CI, 51.5%-99.8%), 96.3% (95%CI, 90.8%-99.0%), 73.3% (95%CI, 44.9%-92.2%), and 99.0% (95%CI, 94.8%-100%), respectively. No differences in performance were observed by HIV status.

Conclusions: ALARM demonstrated excellent accuracy for moderate-to-severe hepatic steatosis regardless of HIV status. Application of ALARM to radiographic repositories could facilitate real-world studies to evaluate medications associated with steatosis and assess differences by HIV status.

References

Zeb I, Li D, Nasir K, Katz R, Larijani V, Budoff M . Computed tomography scans in the evaluation of fatty liver disease in a population based study: the multi-ethnic study of atherosclerosis. Acad Radiol. 2012; 19(7):811-8. PMC: 3377794. DOI: 10.1016/j.acra.2012.02.022. View

Tang Y, Gao R, Lee H, Chen Y, Gao D, Bermudez C . Phase identification for dynamic CT enhancements with generative adversarial network. Med Phys. 2021; 48(3):1276-1285. PMC: 9053868. DOI: 10.1002/mp.14706. View

Perez-Matute P, Perez-Martinez L, Blanco J, Oteo J . Role of mitochondria in HIV infection and associated metabolic disorders: focus on nonalcoholic fatty liver disease and lipodystrophy syndrome. Oxid Med Cell Longev. 2013; 2013:493413. PMC: 3736404. DOI: 10.1155/2013/493413. View

Younossi Z, Koenig A, Abdelatif D, Fazel Y, Henry L, Wymer M . Global epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of prevalence, incidence, and outcomes. Hepatology. 2015; 64(1):73-84. DOI: 10.1002/hep.28431. View

Belghiti J, Hiramatsu K, Benoist S, Massault P, Sauvanet A, Farges O . Seven hundred forty-seven hepatectomies in the 1990s: an update to evaluate the actual risk of liver resection. J Am Coll Surg. 2000; 191(1):38-46. DOI: 10.1016/s1072-7515(00)00261-1. View

Shah A, Lydecker A, Murray K, Tetri B, Contos M, Sanyal A . Comparison of noninvasive markers of fibrosis in patients with nonalcoholic fatty liver disease. Clin Gastroenterol Hepatol. 2009; 7(10):1104-12. PMC: 3079239. DOI: 10.1016/j.cgh.2009.05.033. View

Justice A, Dombrowski E, Conigliaro J, Fultz S, Gibson D, Madenwald T . Veterans Aging Cohort Study (VACS): Overview and description. Med Care. 2006; 44(8 Suppl 2):S13-24. PMC: 3049942. DOI: 10.1097/01.mlr.0000223741.02074.66. View

Cholankeril G, Perumpail R, Pham E, Ahmed A, Harrison S . Nonalcoholic Fatty Liver Disease: Epidemiology, Natural History, and Diagnostic Challenges. Hepatology. 2016; 64(3):954. DOI: 10.1002/hep.28719. View

Lake J, Overton T, Naggie S, Sulkowski M, Loomba R, Kleiner D . Expert Panel Review on Nonalcoholic Fatty Liver Disease in Persons With Human Immunodeficiency Virus. Clin Gastroenterol Hepatol. 2020; 20(2):256-268. PMC: 9069630. DOI: 10.1016/j.cgh.2020.10.018. View

10.

Lemoine M, Assoumou L, De Wit S, Girard P, Valantin M, Katlama C . Diagnostic Accuracy of Noninvasive Markers of Steatosis, NASH, and Liver Fibrosis in HIV-Monoinfected Individuals at Risk of Nonalcoholic Fatty Liver Disease (NAFLD): Results From the ECHAM Study. J Acquir Immune Defic Syndr. 2018; 80(4):e86-e94. DOI: 10.1097/QAI.0000000000001936. View

11.

Cholankeril G, Wong R, Hu M, Perumpail R, Yoo E, Puri P . Liver Transplantation for Nonalcoholic Steatohepatitis in the US: Temporal Trends and Outcomes. Dig Dis Sci. 2017; 62(10):2915-2922. DOI: 10.1007/s10620-017-4684-x. View

12.

Kullberg J, Hedstrom A, Brandberg J, Strand R, Johansson L, Bergstrom G . Automated analysis of liver fat, muscle and adipose tissue distribution from CT suitable for large-scale studies. Sci Rep. 2017; 7(1):10425. PMC: 5585405. DOI: 10.1038/s41598-017-08925-8. View

13.

Speliotes E, Massaro J, Hoffmann U, Foster M, Sahani D, Hirschhorn J . Liver fat is reproducibly measured using computed tomography in the Framingham Heart Study. J Gastroenterol Hepatol. 2008; 23(6):894-9. PMC: 3057524. DOI: 10.1111/j.1440-1746.2008.05420.x. View

14.

Bohte A, van Werven J, Bipat S, Stoker J . The diagnostic accuracy of US, CT, MRI and 1H-MRS for the evaluation of hepatic steatosis compared with liver biopsy: a meta-analysis. Eur Radiol. 2010; 21(1):87-97. PMC: 2995875. DOI: 10.1007/s00330-010-1905-5. View

15.

Armah K, Chang C, Baker J, Ramachandran V, Budoff M, Crane H . Prehypertension, hypertension, and the risk of acute myocardial infarction in HIV-infected and -uninfected veterans. Clin Infect Dis. 2013; 58(1):121-9. PMC: 3864500. DOI: 10.1093/cid/cit652. View

16.

Behrns K, Tsiotos G, DeSouza N, Krishna M, Ludwig J, Nagorney D . Hepatic steatosis as a potential risk factor for major hepatic resection. J Gastrointest Surg. 1998; 2(3):292-8. DOI: 10.1016/s1091-255x(98)80025-5. View

17.

Agarwal N, Iyer D, Gabbi C, Saha P, Patel S, Mo Q . HIV-1 viral protein R (Vpr) induces fatty liver in mice via LXRα and PPARα dysregulation: implications for HIV-specific pathogenesis of NAFLD. Sci Rep. 2017; 7(1):13362. PMC: 5645472. DOI: 10.1038/s41598-017-13835-w. View

18.

Huo Y, Terry J, Wang J, Nair S, Lasko T, Freedman B . Fully automatic liver attenuation estimation combing CNN segmentation and morphological operations. Med Phys. 2019; 46(8):3508-3519. PMC: 6692233. DOI: 10.1002/mp.13675. View

19.

Bush K, Kivlahan D, McDonell M, Fihn S, Bradley K . The AUDIT alcohol consumption questions (AUDIT-C): an effective brief screening test for problem drinking. Ambulatory Care Quality Improvement Project (ACQUIP). Alcohol Use Disorders Identification Test. Arch Intern Med. 1998; 158(16):1789-95. DOI: 10.1001/archinte.158.16.1789. View

20.

Graffy P, Sandfort V, Summers R, Pickhardt P . Automated Liver Fat Quantification at Nonenhanced Abdominal CT for Population-based Steatosis Assessment. Radiology. 2019; 293(2):334-342. PMC: 6822771. DOI: 10.1148/radiol.2019190512. View