Prevalence, Risk Profiles, and National Implications of Nonalcoholic Fatty Liver Disease in Pregnant Individuals

Abstract

The prevalence of nonalcoholic fatty liver disease (NAFLD) in women of childbearing age in the United States has been estimated at 9%,1Long M.T. et al.Liver Int. 2018; 38: 1495-1503Crossref PubMed Scopus (22) Google Scholar but there are limited data on prevalence in pregnant people.2Sarkar M. et al.J Hepatol. 2020; 73: 516-522Abstract Full Text Full Text PDF PubMed Scopus (61) Google Scholar Predominantly retrospective studies have suggested that NAFLD during pregnancy is associated with multiple adverse pregnancy complications, including gestational diabetes, preeclampsia, preterm birth, and postpartum hemorrhage.2Sarkar M. et al.J Hepatol. 2020; 73: 516-522Abstract Full Text Full Text PDF PubMed Scopus (61) Google Scholar, 3Hershman M. et al.Gastroenterol Hepatol (N Y). 2019; 15: 221-228PubMed Google Scholar, 4Hagström H. et al.Liver Int. 2016; 36: 268-274Crossref PubMed Scopus (71) Google Scholar Prior studies have been limited primarily by reliance on diagnostic codes to identify women with NAFLD, potentially resulting in under-reporting and misclassification of NAFLD in pregnant women.5Sarkar M. et al.Hepatology. 2021; 73: 318-365Crossref PubMed Scopus (37) Google Scholar To delineate the burden of NAFLD among pregnant individuals, we established a prospective cohort study in a large urban obstetrics center to estimate the radiographically confirmed prevalence of NAFLD in pregnant individuals, and to determine demographic and clinical factors associated with NAFLD in pregnancy. Pregnant individuals 18 years of age and older and between 18 to 26 weeks’ gestation presenting for routine fetal anatomy ultrasounds were invited to participate in the study. Individuals with reported heavy alcohol use (ie, >14 drinks/wk) within 2 years before pregnancy were excluded. Consented individuals underwent a liver ultrasound at the time of their routine anatomy scan. Four experienced obstetric sonographers were trained to perform liver sonography by a diagnostic radiologist with expertise in abdominal imaging. One of 3 ultrasound units with a 2- to 5-MHz convex transducer was used to capture ultrasound images of the maternal liver. Four to 6 images were obtained for each patient in a combination of both long and transverse views. Maternal liver ultrasound images were deidentified and then sent to a radiologist blinded to clinical data for review. The presence and grading of hepatic steatosis were determined qualitatively by the radiologist.6Hamaguchi M. et al.Am J Gastroenterol. 2007; 102: 2708-2715Crossref PubMed Scopus (632) Google Scholar,7Li Q. et al.World J Hepatol. 2018; 10: 530-542Crossref PubMed Scopus (138) Google Scholar To evaluate for inter-rater reliability in assessment of steatosis on ultrasound, 2 additional radiologists were asked to evaluate 114 ultrasounds (estimated to be the number needed to achieve a kappa of 0.7 considering a NAFLD prevalence of 15%). Mild (grade 1) steatosis was defined as diffusely increased hepatic echogenicity with appreciable periportal and diaphragmatic echogenicity. Moderate (grade 2) steatosis was defined as diffusely increased hepatic echogenicity obscuring periportal echogenicity but with appreciable diaphragmatic echogenicity. Severe (grade 3) steatosis was defined by diffusely increased hepatic echogenicity obscuring both periportal and diaphragmatic echogenicity. NAFLD diagnosis was determined by ultrasound imaging with hepatic steatosis (grades 1–3), exclusion of significant alcohol consumption during pregnancy, and the absence of known co-existing chronic liver disease. Additional data collected included self-identified race and/or ethnicity, most recent height and weight before pregnancy (later confirmed during review of electronic health record), and history of alcohol use. We reviewed medical charts to collect data on prior medical, obstetric, and gynecological history. Demographic and clinical characteristics were described using frequencies and percentages for categoric data and medians with interquartile ranges (IQRs) or means with SDs for continuous variables. We calculated the 95% CI using the binomial distribution. Univariate and multivariate assessment of factors associated with NAFLD during pregnancy was conducted using logistic regression with forward selection. All analyses were performed using STATA software, version 16.1 (StataCorp LLC). Among 750 individuals approached, 637 were eligible for participation, and 540 consented to participation from December 2020 to January 2021. Seventy-eight (14%; 95% CI, 11%–17%) had evidence of hepatic steatosis: the prevalence of hepatic steatosis was 19% among those of Hispanic race, 17% among those with a body mass index (BMI) of 25 or higher, and 20% among those with a BMI of 30 or higher (Supplementary Figure 1). Overall, 65 (83%; 95% CI, 75%–92%), 11 (14%; 95% CI, 6%–22%), and 2 (3%; 95% CI, 0%–6%) had steatosis grades 1, 2, and 3, respectively. There was moderate agreement between radiologists in evaluating for steatosis, which was in line with prior data on inter-rater reliability for ultrasound steatosis assessment outside of the pregnancy setting (kappa, 0.45–0.49).8Hernaez R. et al.Hepatology. 2011; 54: 1082-1090Crossref PubMed Scopus (999) Google Scholar,9McHugh M.L. Biochem Med (Zagreb). 2012; 22: 276-282Crossref PubMed Google Scholar Among patients included, 10 (<2%) had steatosis seen previously on other imaging and 3 (0.55%) had a diagnosis of NAFLD in their medical record. Individuals with steatosis in pregnancy were more likely to be Hispanic or from Mexico, and less likely to be Black or from Africa (Table 1) (P < .05). They also had a higher median BMI than those without hepatic steatosis (29 kg/m2 [IQR, 25–34 kg/m2] vs 23 kg/m2 [IQR, 21–31 kg/m2]), with 79% of individuals with steatosis having a BMI of 25 or higher. Patients with steatosis were more likely to have chronic hypertension, diabetes (DM), and autoimmune thyroid disease (P < .05). The hepatic steatosis index, based on first trimester laboratory results (available in 155 [29%] study participants), was higher among those with steatosis. Pregnancy characteristics, such as parity and pregnancy type (ie, spontaneous vs via assisted reproductive methods), were similar between groups. Regarding prior pregnancy history, among those who had history of prior pregnancies (n = 121) and who had steatosis on current pregnancy were more likely to have a history of gestational diabetes (29% vs 5%; P = .023) (Table 1). There were no differences in history of cholestasis of pregnancy, gestational hypertension, preterm delivery, hemolysis elevated liver enzymes, low platelet count, acute fatty liver of pregnancy, or other reported pregnancy complications. In univariate analysis, prepregnancy BMI (odds ratio [OR], 1.06; 95% CI, 1.02–1.10), history of DM (OR, 4.52; 95% CI, 1.67–12.27), hypertension (OR, 2.31; 95% CI, 1.11–4.82), Hispanic ethnicity (OR, 2.59; 95% CI, 1.49–4.54), and history of gestational DM in a prior pregnancy (OR, 2.81; 95% CI, 1.11–7.13) were associated significantly with steatosis in current pregnancy. In multivariate analysis, only a BMI greater than 30 (OR, 1.98; 95% CI, 1.09–3.61) and Hispanic ethnicity (OR, 2.56; 95% CI, 1.31–5.00) remained associated independently with steatosis. Supplementary Figure 1 shows a range of prevalence estimates in distinct subpopulation risk groups.Table 1Demographic and Clinical Characteristics of Fatty Liver in Pregnancy (FLIP) CohortNAFLD (n = 78; 14%)No NAFLD (n = 462; 86%)P valueDemographic characteristics Age, y28 (24–33)28 (24–33).758 Race, n (%).010White1 (1)32 (7)Black16 (21)171 (37)Hispanic48 (62)203 (44)Asian4 (5)11 (3)Other8 (10)40 (9) Ethnicity, n (%).002Hispanic/Latino60 (77)250 (56)Not Hispanic/Latino18 (23)195 (44)Unknown3 (1)0 (0) Country/region of origin, n (%).044Europe14 (5)0 (0)Asia3 (6)16 (6)Latin AmericaMexico20 (41)53 (20)Puerto Rico12 (25)66 (25)Dominican Republic9 (18)49 (19)Other4 (8)31 (12)Africa1 (2)24 (9)Other0 (0)7 (3) Insurance, n (%).450Government72 (92)412 (85)Private6 (7)49 (11)Clinical and metabolic comorbiditiesbNo patients had autoimmune hepatitis, alcohol-associated liver disease, Wilson’s disease, or primary biliary cirrhosis/PSC. Chronic hypertension11 (14)31 (7).022 Diabetes type II7 (9)10 (2).001 Dyslipidemia07 (2).278 PCOS22 (5)5 (7).507 HBV0 (0)5 (1).360 Renal disease0 (0)2 (0.5).562 Hematologic conditionaIncludes anemia, sickle cell trait, β-thalassemia, and α-thalassemia traits.10 (13)65 (14).725 Autoimmune diseaseThyroid7 (9)13 (3).007Other210 Smoking, n (%)16 (21)88 (19).775 Prepregnancy alcohol, n (%)Yes (unspecified quantity)5 (6)39 (6)>7 drinks/wk (<14 drinks)1 (0)3 (0.5)None/social27 (35)147 (23) Bariatric surgery history, n (%)1 (1)14 (3).347Laboratory data, median (IQR) HSI (n = 155)HSI value44 (39–50)38 (33–43)<.001HSI > 36, n (%)25 (89)72 (57).001HSI < 30, n (%)1 (4)21 (17).075FIB-4, median (IQR) (n = 361)0.53 (0.43–0.64)0.54 (0.38–0.72)662Prior pregnancy characteristics, median (IQR) Term pregnancies1 (1–2)1 (0–2).181 Preterm pregnancies0 (0–0)0 (0–0).730 Prior abortions/pregnancy losses4 (7)20 (6).638 Gravida3 (1–4)3 (2–4).824 Multiple gestation pregnancy3 (4)8 (2).214 Pregnancy type, n (%).358Spontaneous78 (100)460 (99)IVF0 (0)5 (1) Primigravida19 (24)103 (22).638 Prior pregnancy complications (n, %) Gestational diabetes7 (29)17 (5).023 Gestational hypertension4 (7)20 (6).638 Preeclampsia10 (18)34 (10).065 Intrahepatic cholestasis of pregnancy0 (0)7 (2).291 Preterm history6 (11)52 (15).385 HELLP history1 (2)1 (0.3).130 AFLP00AFLP, acute fatty liver of pregnancy; BMI, body mass index; FIB-4, Fibrosis-4; HBV, hepatitis B virus; HELLP, hemolysis elevated liver enzymes, low platelet count; HSI, hepatic steatosis index; IVF, in vitro fertilization; IQR, interquartile range; NAFLD, nonalcoholic fatty liver disease; PCOS, polycystic ovary syndrome; PSC, primary sclerosing cholangitis.a Includes anemia, sickle cell trait, β-thalassemia, and α-thalassemia traits.b No patients had autoimmune hepatitis, alcohol-associated liver disease, Wilson’s disease, or primary biliary cirrhosis/PSC. Open table in a new tab AFLP, acute fatty liver of pregnancy; BMI, body mass index; FIB-4, Fibrosis-4; HBV, hepatitis B virus; HELLP, hemolysis elevated liver enzymes, low platelet count; HSI, hepatic steatosis index; IVF, in vitro fertilization; IQR, interquartile range; NAFLD, nonalcoholic fatty liver disease; PCOS, polycystic ovary syndrome; PSC, primary sclerosing cholangitis. In summary, in this large prospective cohort study, we identified an overall NAFLD prevalence of 14% among pregnant individuals screened during early pregnancy, with the highest prevalence in obese Mexican Americans. Our study was unique in its use of ultrasonographic assessment for the detection of NAFLD among consecutive individuals seeking prenatal care, as well as in its prospective design ,which allows for well-adjudicated outcomes. Although ultrasound assessment of the presence and staging of steatosis may be subjective, our inter-reader reliability was similar to that reported previously in ultrasound assessment outside of the pregnancy setting. Given how few pregnant individuals had a documented history of NAFLD in our study, prior retrospective studies relying on diagnosis codes for NAFLD likely were highly insensitive. This was a contemporary prospective cohort of NAFLD in pregnancy in the United States. A smaller Canadian prospective study from 2016 identified a 16% prevalence of at least 1 sonographic feature of NAFLD in a multiethnic cohort.10De Souza L.R. et al.Am J Gastroenterol. 2016; 111: 665-670Crossref PubMed Scopus (46) Google Scholar With an estimated 4 million annual pregnancy deliveries in the United States, more than 500,000 of those deliveries may occur in individuals with NAFLD,11Michelle J.K. et al.https://www.cdc.gov/nchs/data/nvsr/nvsr72/nvsr72-01.pdfDate accessed: August 20, 2022Google Scholar and this number is projected to increase further in the coming decade.12Wang M.C. et al.J Am Heart Assoc. 2021; 10e020717Google Scholar Importantly, fewer than 5% of individuals in our cohort with NAFLD had a prior evaluation for or recognition of NAFLD before study participation, highlighting an important opportunity to detect NAFLD, not only from the standpoint of counseling regarding associated pregnancy outcomes but also for further risk stratification and linkage to ongoing care. Contributors as part of the FLIP Study Group: Emma Rosenbluth, Carin Carroll, Theresa Worthington, Shaelyn O’Hara, Joanne Stone, Scott Friedman, Sonam Rosberger, Deborah Feldman, Rhoda Sperling.

Topics & Concepts