Drug-Induced Liver Injury in Pregnancy: The U.S. Drug-Induced Liver Injury Network Experience

Abstract

There are limited data on the etiologic agents and characteristics of drug-induced liver injury (DILI) in pregnant individuals. DILI cases enrolled in the ongoing multicenter DILI Network (DILIN) between 2004–2022 and occurring during pregnancy, or 6 months postpartum, were reviewed and compared to DILI in non-pregnant women of childbearing age. Among 325 individuals of childbearing age in the DILIN, 16 (5%) occurred during pregnancy or postpartum. Compared to non-pregnant female DILI controls, pregnancy-related DILI was more severe (p<0.05). One elective termination and three miscarriages were documented; there were no maternal deaths. We recommend that isoniazid for latent tuberculosis be deferred to the postpartum period whenever feasible and that beta-blockers/calcium channel blockers be used for hypertension management rather than methyldopa during pregnancy.

Précis:

Drug-induced liver injury should be suspected during pregnancy, particularly when methyldopa is used, as severe liver injury requiring liver transplantation can occur.

INTRODUCTION

Data on the etiology, clinical presentation, and outcomes of drug-induced liver injury (DILI) in the context of pregnancy are limited ¹. We evaluated all cases from the U.S. DILI Network (DILIN) that occurred during pregnancy or within 6-months postpartum ¹ and compared them with non-pregnant women of childbearing age with DILI.

METHODS

The DILIN is a NIH-funded multicenter study that enrolls patients with suspected DILI (https://dilin.org/) and collects demographic features (including age, race, ethnicity that may be associated with DILI outcomes and clinical features of individuals with DILI ². We identified all cases enrolled from September 2004 to March 2022 that occurred during pregnancy or the immediate 6-month postpartum period using pregnancy-related keywords (see Appendix 1, available online at http://links.lww.com/xxx). All cases were reviewed and adjudicated by the DILIN Causality Committee using consensus expert opinion, ³ as the structured DILIN expert opinion process produces higher agreement rates and likelihood scores than RUCAM in assessing causality.⁴ Available biopsies were centrally reviewed by an expert pathologist (DEK) ⁵. Selected patient vignettes were highlighted to provide more in-depth details of clinical courses (see Appendix 2, available online at http://links.lww.com/xxx). The study was approved by the Institutional Review Boards at each clinical site and data coordinating center.

Demographic and clinical characteristics were compared between pregnant DILI cases and non-pregnant DILI cases in women of childbearing age (i.e, ages 18-45 y). All analyses were carried out with SAS version 9.4.

RESULTS

Of the 2626 cases enrolled in the DILIN database, 393 were in women of childbearing ages. Of these, 21 cases were in women who were pregnant or less than 6 months postpartum at DILI onset (see Appendix 3, available online at http://links.lww.com/xxx, and Table 1). Sixteen of the 21 were high confidence DILI cases based on a causality score of 1, 2, or 3 (Table 1) for an approximate 5% prevalence in the DILIN database. The remaining 5 adjudicated cases were deemed due to non-DILI etiologies (Appendix 4, available online at http://links.lww.com/xxx). Table 2 shows baseline features of the study population.

Table 1.

Selected Characteristics of High Causality Pregnancy-associated DILI Cases

Case #	Age (y)	Race/Ethnicity	Implicated Drug	DILI Onset	Weeks Gestation at onset	Pattern of Injury	Causality Score	Severity Score	Maternal Outcome	Fetal Outcome	Peak ALT U/L	Peak TB mg/dL
1	43	W	Methyldopa	Pregnancy	10	HC	Definite	Moderate Hospitalized	Resolution	Unknown	2459	11
2	29	B	Methyldopa	Pregnancy	20	HC	Highly Likely	Mild	Resolution	Miscarriage	800	0.8
3	30	B	Methyldopa	Pregnancy	8	HC	Highly Likely	Moderate Hospitalized	Resolution	Unknown	1315	17.8
4	24	B	Infliximab	Pregnancy	22	HC	Highly Likely	Severe	Resolution	Delivered	1155	25.41
5	20	W	Quetiapine	Pregnancy	29	HC	Highly Likely	Mild	Resolution	Delivered	1078	0.4
6	34	H	Isoniazid	Pregnancy	1st trimester	HC	Probable	Severe/Fatal	Transplant	Miscarriage	2095	13.8
7	28	W	Letrozole	Pregnancy	NA	HC	Probable	Moderate	Resolution	Miscarriage	827	7
8	28	B	Methyldopa	Postpartum	-	HC	Highly Likely	Severe	Resolution	Delivered	579	21.7
9	20	W	OCP	Postpartum	-	HC	Highly Likely	Moderate Hospitalized	Resolution	Delivered	445	14.9
10	32	H	Methyldopa	Postpartum	-	HC	Definite	Severe	Resolution	Delivered	1415	29.9
11	28	W	Cefazolin	Postpartum	-	C	Highly Likely	Moderate Hospitalized	Resolution	Terminated	201	7.3
12	25	B	Methyldopa	Postpartum	-	HC	Highly Likely	Moderate Hospitalized	Resolution	Delivered	1292	24.9
13	39	B	Methyldopa	Postpartum	-	HC	Highly Likely	Moderate Hospitalized	Resolution	Delivered	1414	19.5
14	26	W	Methyldopa	Postpartum	-	HC	Highly Likely	Moderate Hospitalized	Resolution	Delivered	1577	15.3
15	31	H	PTU	Pregnancy	36	M	Probable	Severe	Resolution	Delivered	549	8.4
16	31	B	Methyldopa	Pregnancy	11	HC	Probable	Moderate Hospitalized	Unknown	Unknown	1481	3.2

Definite causality score corresponds to >95% likelihood, high likely= 70-95%, probable= 50-<70%. Abbreviations: W=White, B=Black, H= Hispanic, HC=Hepatocellular, C=Cholestatic, PTU=propylthiouracil; OCP, oral contraceptive pills.

Table 2.

Comparison of baseline demographic and clinical characteristics between high causality DILI cases in pregnancy and non-pregnant DILI case controls.

CHARACTERISTICS	PREGNANT (n=16)	NON-PREGNANT (n=309)	P-VALUE
DEMOGRAPHICS
Age, median (IQR)	28.8 (26.1, 31.9)	35.9 (27.4, 41.8)	0.02
Race, n (%)			0.05
White or Caucasian	7 (43.8%)	220 (71%)
Black or AA	6 (37.5%)	54 (17.5%)
Asian	0 (0.0)	9 (2.9%)
Other/Multiracial	3 (18.8%)	25 (8.1%)
Ethnicity, Latino, n (%)	3 (18.8%)	49 (15.9%)	0.75
BMI, median (IQR)	32.5 (23.6, 39.0)	26.1 (22.7,31.6)	0.05
CLINICAL CHARACTERISTICS
Pre-existing Liver Disease, n (%)	0	6 (1.9%)	>0.99
Hepatitis C	0	6 (1.9%)	>0.99
NAFLD	0	6 (1.9%)	>0.99
Other
CLINICAL EVALUATION AT PRESENTATION
ADJUDICATED DRUGS, n (%)
Antimicrobial	2 (12.5%)	142 (46.0%)	0.009
Cardiovascular	9 (56.3%)	13 (4.2%)	<0.001
Methyldopa	9 (56.3%)	3 (0.8%)	<0.001
Endocrine	2 (4.5%)	14 (4.5%)	0.18
CNS	1 (6.3%)	47 (15.2%)	0.48
Antineoplastic	1 (6.3%)	21 (6.8%)	>0.99
Other	1 (6.3%)	72 (23.3%)	0.01
Laboratory Test Results, median (IQR) or n (%)
ALT (U/L)	1174 (670, 1364)	519 (245,1132)	0.01
AST (U/L)	1511 (461, 1890)	400 (155, 1035)	0.01
TB (mg/dL)	6.6 (2.3, 15.6)	3.4 (0.8, 7.7)	0.12
AP (U/L)	161.5 (125, 191)	177.0 (117, 262)	0.41
ANA positive	6 (46.2%)	93 (32.4%)	0.36
ASMA positive	4 (30.8%)	58 (21.6%)	0.49
Eosinophilia (>500/uL)	1 (6.7%)	44 (15.2%)	0.71
MELD score (IQR)	17 (12, 20)	15 (9, 20)	0.62
INR	1.2 (1.1,1.5)	1.2 (1.0,1.7)	0.92
Serum IgG (mg/dL)	1300 (1015-1987)	1221 (944-1645)	0.53
Liver Injury Pattern, n (%)
Hepatocellular	14 (87.5%)	190 (64%)	0.22
Mixed	1 (6.3%)	52 (17.5%)
Cholestatic	1 (6.3%)	55 (18.5%)
R-value, median (IQR)	17.6 (11.5, 26.4)	8.7 (2.8, 19.6)	0.02

Abbreviations: AA, African American; NAFLD, nonalcoholic fatty liver disease; AP, alkaline phosphatase; ANA, antinuclear antibody; ASMA, anti-smooth muscle antibody; MELD, model for endstage liver disease; TB; total bilirubin; IgG, immunoglobulin G

For all 16 DILI cases, a single agent was implicated. Of note, nine (56%) cases were attributed to methyldopa (Table 1). All other implicated drugs were known potential hepatotoxins per the LiverTox database with a likelihood score of A or B, with the exception of letrozole ¹.

Among the 9 cases arising during pregnancy, 4 (44%) had DILI onset in first trimester, 2 (22%) in 2^nd trimester, and 2 (22%) in third trimester; one case did not have data on trimester of onset. Among the 7 postpartum cases, onset was within the first 60 days postpartum in 5 (71%). In fourteen cases (88%), liver enzyme abnormalities were consistent with a hepatocellular pattern of injury; median R-value was significantly higher in the pregnancy group (p=0.016) (Table 2, Appendix 5 [Appendix 5 is available online at http://links.lww.com/xxx]). Autoimmune features of DILI cases are described in Appendix 6, available online at http://links.lww.com/xxx. Representative histologic findings are shown in Figure 1. Twelve of the 16 cases (80 %) were hospitalized with 3 of these cases considered severe; one case of DILI secondary to Isoniazid required liver transplantation (see Case #6 vignette).

Figure 1.

Histology of drug-induced liver injury associated with pregnancy is variable even due to the same agent. A. Case 12, liver biopsy obtained 5 months postpartum showing methyldopa-associated submassive necrosis, showing near-complete loss of hepatocytes from this field (hematoxylin and eosin stain, 200x). B. Case 1, liver biopsy obtained during pregnancy showing acute hepatitis associated with methyldopa, showing multiple foci of lobular inflammation, but no areas of necrosis (hematoxylin and eosin stain, 200x). C. Case 7, liver biopsy obtained letrozole-associated bland intrahepatic cholestasis with canalicular cholestasis (arrow) but no inflammation (hematoxylin and eosin stain, 400x).

Pregnancy-related DILI was associated with significantly higher initial serum AST/ALT and higher peak AST values (p<0.05) and a higher prevalence that developed jaundice (81% vs 59%, p= 0.07) compared to non-pregnant controls (Table 2, Table S3).

Of the 16 DILI cases, 9 (56%) reported delivery of healthy infants, 3 (19%) ended in pregnancy losses, and 1 (6%) reported an infant with developmental delays. All pregnancy losses were reported in individuals who had not been hospitalized during the DILI event. None of the losses were directly attributed to DILI.

DISCUSSION

This large case series collected over an 18-year period provides insights into the types and features of DILI occurring in pregnant women in the United States. Although the traditional limited use of medications during pregnancy might be anticipated to lead to a lower incidence of DILI in pregnancy, our 5% prevalence, though low, was comparable to the expected percent of fertility (5.6%) for this age group ⁶. Although the RUCAM causality assessment traditionally allocates an extra point to pregnancy, our findings suggest that pregnancy in itself should not necessarily increase DILI causality/frequency⁷.

DILI appeared to be more severe in pregnancy, which might be due to delayed recognition of DILI among pregnant individuals or underutilization of liver biopsy during pregnancy. A liver biopsy may prove helpful in excluding other etiologies such as AIH that may mandate corticosteroids ⁵ but biopsies are generally avoided in pregnant individuals ⁸. Methyldopa cases predominated in our cohort. Unfortunately methyldopa is still recommended as a first line antihypertensive agent in some practices, but per our findings we would recommend to avoid Methyldopa use in pregnancy if alternate agents are available. ^{9, 10 11} Be aware that pregnancy related abnormal liver tests may be due to medications started within the prior 3 months and can initially occur postpartum such as seen in AIH, consult a liver specialist, exclude other common causes of pregnancy related abnormal liver tests, use the NIH “Livertox” website to identify the most likely causative medications (https://www.ncbi.nlm.nih.gov/books/NBK547852/), and discontinue likely causative medications promptly. ¹².

Abbreviations:

AA

African-American

AIH

Autoimmune hepatitis

ALT

Serum alanine aminotransferase

AMA

Anti-mitochondrial antibodies

ANA

Anti-nuclear antibodies

AP

Serum alkaline phosphatase

ASMA

Anti-smooth muscle antibodies

AST

Serum aspartate aminotransferase

C

Cholestatic-type liver injury [R ≤2]

DILI

Drug-induced liver injury

DILIN

U.S. Drug Induced Liver Injury Network

DRESS

Drug reaction with eosinophilia and systemic symptoms

HC

Hepatocellular-type liver injury [R ≥ 5]

HDS

Herbal and dietary supplements

INH

Isoniazid

INR

International normalized ratio

IQR

inter quartile range

M

Mixed-type liver injury [2<R <5]

MD

Alpha methyldopa

NA

Not available

NIH

National Institutes of Health

R

Ratio of serum ALT/AP both expressed as multiples of ULN

RUCAM

Roussel-Uclaf causality assessment method for DILI

SJS

Stevens - Johnson syndrome

SLE

Systemic lupus erythematosus

TEN

Toxic epidermal necrolysis

ULN

Upper limit of normal

U.S.

United States of America

VDRL

Venereal diseases research laboratory