Hepatic Complications of Oral Contraceptive Pills and Estrogen: Controversies and Update- Adenoma and Beyond

Hepatic complications of oral contraceptive pills and estrogen: Controversies and update- adenoma and beyond.

Abstract

Hepatic complications of the oral contraceptive pills and estrogen may include intrahepatic canalicular cholestasis, neoplasm formation and vascular pathologies. Adenomas have classically described with exposure to oral contraceptive pills and estrogen. Controversial literature exists to support role of oral contraceptive pills and estrogen in focal nodular hyperplasia, hemangioma, and hamartoma. Four different subgroups of adenomas have been described: Inflammatory, HNF-1α-mutated, β-catenin-mutated and unclassified. β-catenin–activation could be a secondary event along the pathway to malignant transformation of any histologic subtype. Vascular complications may include Budd-Chiari syndrome, vascular thrombosis, dilated sinusoids and peliosis.

Introduction

Oral contraceptive agents (OCP) are the most commonly used reversible means of birth control in women. Studies have shown an association between use of oral contraceptive pills and cholestasis, hepatic neoplasms, as well as vascular pathologies. Acute intrahepatic canalicular cholestasis usually develops 2 to 3 months after starting OCPs and resolves after cessation. Patients with OCP-induced cholestasis often have a history of idiopathic cholestasis of pregnancy, and there is an underlying genetic component involving the bile salt export pump (BSEP) and ATP-binding cassette sub-family B 11 member gene (ABC B11)[1].

Women have been cautioned to avoid use of oral contraceptive pills in certain clinical conditions, which may include[2]: a) Previously experienced cholestatic jaundice of pregnancy, b) Those who have first-degree relatives with cholestasis of pregnancy or oral contraceptive-induced cholestasis, c) Current or previous benign or malignant hepatic tumors, d)  Active hepatitis, e) Diagnosed with familial defects of biliary excretion (Dubin-Johnson syndrome, Rotor’s syndrome), and f) In individuals with benign intrahepatic recurrent cholestasis.

OCPs have a known association with benign hepatic neoplasms such as adenomas and may rarely induce malignant transformation. Vascular complications are often seen with coexisting  risk factors like Protein C, Protein S deficiency or Factor V Leiden mutations predisposing to venous thrombosis.

Fortuitously, these complications are rare and may be decreasing due to lower-dose OCPs used now compared to decades ago. Nevertheless, it is important to recognize complications in a timely fashion, to avoid life-threatening events. Complications may still occur, however; they need to be recognized by imaging and appropriate treatment and follow-up initiated including cessation of drugs.

Cholestasis due to OCP:

Cholestasis appears to be related to estrogen effects on orphan nuclear receptors that modulate bilirubin metabolism, resulting in inhibition of bilirubin secretion into the biliary tree[3].

The reported frequency of cholestasis is 2.5 occurrences per 10,000[4]. Moderate increase in risk, up to 50%, is seen in patients with a prior history of cholestasis of pregnancy[4]. Clinical presentation includes mild prodromal symptoms such as anorexia and nausea, followed by pruritus, which usually develop 2 to 3 months after starting OCPs. Moderately elevated serum alkaline phosphatase is often revealed on lab investigation. Progression to chronic cholestasis is extremely rare and generally recovery occurs within days to weeks after cessation of this medication. Hormone replacement therapy (HRT) may similarly increase bilirubin levels and monitoring LFTs is advisable. The primary role of imaging here is to exclude gallbladder stones and other structural causes of obstructive jaundice.

Dubin-Johnson Syndrome:

Dubin-Johnson syndrome is an autosomal recessive inheritance with mutation in ABCC2 gene resulting in defective transport of bilirubin from the hepatocytes into bile ducts, and consequently leads to a rise in serum conjugated bilirubin levels. It is often asymptomatic but may manifest only during pregnancy or with use of oral contraceptives. OCP use can impair hepatic excretory function and can transform a mild hyperbilirubinemia into frank jaundice[5]. Typically, imaging shows no cause of obstructive jaundice.

Hepatocellular adenoma (HCA):

Pathogenesis and genetics:

Hepatocellular adenomas have been traditionally described as neoplasm that occur in women on oral contraceptives. Adenomas are now understood to represent a heterogeneous set of genetically disparate tumors with different genetic abnormalities and variable clinical and prognostic features. Lately, the Bordeaux group recognized a new molecular and pathologic classification of HCA. According to genotypic and phenotypic characteristics and clinical features, estrogen induced HCA are broadly divided into four different subgroups including inflammatory, Hepatocyte Nuclear Factor (HNF)-1α-mutated, β-catenin-mutated and unclassified[6,7].

Diagnosis of adenoma is usually made on imaging. Biopsy is usually neither recommended to make a diagnosis of adenoma (unless imaging is nonconclusive) nor to determine histological subtype [8,9].

Inflammatory subtype:

Inflammatory HCA is the most common subtype of all HCA, making up 40-50% of all adenomas. These are classically seen in young women with a history of OCP usage (>90%), in obese patients, and in those with underlying hepatic steatosis, diabetes mellitus, alcohol abuse and glycogen storage disease (type 1) (GSD). About 10% of inflammatory HCAs can also show mutations involving the β-catenin gene[10,11].

Pathogenesis is the result of mutations leading to altered or overexpressed glycoprotein 130 which activates the JAK-STAT3 pathway, leading to increased levels of acute phase reactants and this in turn, allows inflammatory cell infiltration into the adenoma[12].

Patients present with “systemic inflammatory syndrome” characterized by fever, leukocytosis, and elevated serum CRP levels. They are associated with increased risk of bleeding (>30%) and a risk of malignant transformation (5-10%)[13,14].

Pathological hallmark includes sinusoidal prominence, infiltration of polymorphous inflammatory cells, scattered areas of peliosis, tortuous and thickened arteries and diffuse ductal reaction. Lipid within these types of nodule can be occasionally seen, however is less widespread compared with HNF-1α HCA.

A subset of atypical focal nodular hyperplasia with prominent telangiectasias or sinusoidal dilatation (Telangiectatic focal nodular hyperplasia), characterized by the absence of nodular architecture and by its mild architectural distortion, is now recognized as an inflammatory adenoma based on molecular genetics and proteomic profiling.

Imaging findings:

MRI demonstrates T2 intermediate or hyperintense signal with a characteristic “atoll sign”, which shows a peripheral rim of T2 intermediate or hyperintensity with isointense center. The appearance of the atoll sign is thought to be due to sinusoidal dilatation within inflammatory HCA[15]. T2 hyperintense peripheral areas may reflect high water content in these blood-filled regions. These areas enhance in the venous phase, whereas the rest of the tumor enhances in the arterial phase[16].

On gadolinium enhanced images, Inflammatory HCA shows a strong arterial enhancement that may persist on delayed phases, though is not always seen.  Also, inflammatory HCA usually appear hypointense to the background liver parenchyma on the hepatobiliary phase on gadoxetic acid–enhanced MR. Occasionally, inflammatory HCA it may also show hyperintensity on the hepatobiliary phase and thereby mimic the appearance of focal nodular hyperplasia.

HNF1α-mutated subtype:

HNF 1A is the second most common subtype of adenomas and account for 30-35% of all adenomas).   This type is exclusively found in the female population.  Ninety percent of these women are (on OCPs. They can manifest as multiple lesions in about 50% of cases. Association has been shown with maturity-onset diabetes of the young (MODY) and familial hepatic adenomatosis.

Pathogenesis of this subtype of HCA comprises genetic alterations directly from exogenous estrogens resulting, leading to a nonfunctional HNF-1α transcription factor, ultimately resulting in to intracellular lipid deposition/steatosis due to a complex mechanism involving suppressed gluconeogenesis-activated glycolysis, and promotion of fatty acid biosynthesis. Defective transport of fatty acids and to intracellular deposition of fat can be attributed to the decreased fatty acid binding protein-1.

Imaging findings:

On MR examination, HNF-1α-mutated subtype of HCA frequently shows microscopic fat deposition as loss of intensity on opposed phase T1W sequences. Significant signal drop in T1 out-phased imaging for predictive HNF-1α-mutated HCA is reported to be 85% sensitive, 100% specific, with a 100% positive predictive value and a 94% negative predictive value, respectively[17].

They show moderate arterial enhancement that does not persist during the delayed phase,but without washout on delayed phases either.

β-catenin-mutated subtype:

β-catenin-mutated subtype is the least common subtype, accounting for 10-15% of all HCAs. It is more frequent in men. This subtype of HCA originates from activating mutations of the catenin β1 gene (CTNNB1), which leads to unrestrained hepatocyte proliferation. β-Catenin has a key role in hepatocyte development, differentiation, proliferation, and regeneration. Mutation of β-catenin itself or mutation of cytoplasmic degradation complex causes sustained activation of β-catenin and excessive nuclear accumulation, which results in autonomous growth of hepatocyte and HCA formation[18]. Avid, diffuse, and homogeneous glutamine synthetase (GS) staining has been shown in β-catenin–mutated HCA.

β-catenin-mutated adenomas are associated with glycogen storage disease (GSD), familial adenomatous polyposis, and androgen administration. They also have a greater risk of malignant transformation to hepatocellular carcinoma (HCC)[19]. Risk of malignant transformation of HCA to HCC may range from 4.2% to 13.0%[20]. However, HCC associated with adenoma was found in 46% of β-catenin-mutated tumors[21].

Imaging findings:

β-catenin-mutated adenomas do not have any particularly specific imaging features. They may appear homogeneous or heterogeneous with lack of intracellular fat and demonstrate a vaguely defined T2-hyperintense scar and ill-defined  areas of T2-weighted hyperintensity[10]. After administration of intravenous contrast, an early hyperenhancement can be seen, which may (or may not) persist on the venous and equilibrium-phase, though some can demonstrate washout which is non-specific.

Atoll sign is classically described in inflammatory HCA, whereas scar in β-catenin mutated subtypes[10,22]. Presence of the atoll sign with a scar on imaging may differentiate between β-catenin–positive inflammatory HCA and β-catenin–negative inflammatory HCA[10]. It is essential to identify  β-catenin–positive inflammatory HCA whenever possible as there is additional chance of malignant transformation of this subset of β-catenin–positive inflammatory HCA[10].

Unclassified subtype:

Unclassified subtype accounts for approximately10% of cases and encompasses all other adenomas, which do not belong to the previously described categories. They are poorly understood but may become malignant.

Imaging findings:

No specific MR imaging patterns or other imaging features have yet been proposed to definitively characterize these unclassified HCAs[23].

Recent developments and expanded system of classification

Recently an additional subtype of HCA has been recognized, the sonic hedgehog–activated adenoma, which accounts for 4% of all HCA and previously considered to be unclassified. Sonic hedgehog–activated hepatocellular adenomas show a fusion of the Inhibin-Beta E Subunit (INHBE) and GLI1 genes[24,19]. Sonic hedgehog genetic pathway activated adenoma has a higher propensity to bleed, therefore remain clinically important[24].

Additionally, a group of heterogenous adenomas has demonstrated high propensity to develop malignant transformation. This group includes androgen-associated adenomas, pigmented adenomas, and myxoid adenomas. The pigmented adenomas are characterized by the presence of heavy lipofuscin deposition and myxoid adenomas are characterized by abundant myxoid material dissecting through the tumors[25,26]. Further, it has been suggested that beta-catenin–activation and pigmentation could be secondary events along the pathway to malignant transformation of any histologic subtype[26].

These descriptions do not fit well into the existing schema of classification. Therefore, an expanded system for classifying hepatic adenomas has been proposed based on identifiable risk factors like exposure to estrogen, androgens, glycogen storage disease or no identifiable factors[26]. Based on this classification, estrogen induced adenomas include HNF1-Alpha Inactivated (40%), inflammatory (50%), and unclassified (10%) adenomas (Table.xx). Myxoid type of adenomas have been described as Protein Kinase A–Associated adenomas in this classification, and exclusively consist of HNF1-Alpha inactivated variety. Beta-catenin–activation and pigmentation can be seen with any of these subtypes and may represent a risk factor for malignant transformation.

Adenomatosis

Adenomatosis is a term coined by Flejou et al in 1985[27]. It is a distinct entity and is defined as presence of multiple (>10) HCA without known risk factors, including history of steroid use, OCP intake or underlying GSD. Absence or occlusion of portal vein or portohepatic venous shunts is seen in these patients[28].  Patient taking OCPs and presenting with multiple adenomas should not be classified as having “adenomatosis”.

Management guidelines of HCA are discussed in the flowchart.

Focal nodular hyperplasia

Focal nodular hyperplasia (FNH) is the 2^nd most frequent benign hepatic neoplasm. FNH is a hyperplastic response to a localized vascular abnormality. It can be considered as a benign congenital hamartomatous malformation.

FNH is mostly seen in the same group of patients as HCA, especially with history of OCP consumption. However, association of FNH and OCPs remains controversial. In a 9-year study of 216 women, Mathieu et al suggested that neither the size nor the number of lesions is influenced by OCP use[29]. Unlike in HCA, discontinuation of OCPs therefore will not lead to a reduction in size or number of FNH.

In 2008, the WHO expert working group stated that women with FNH could use hormonal contraception, as the advantages to use outweigh any potential (yet to be established) risks.

FNH with sinusoidal dilatation and FNH with cytological atypia are two subtypes of FNH. On pathology FNHs > 5 cm in diameter often show a central scar composed of fibrous connective tissue, cholangiocellular proliferation, and vascular deformity. Hemorrhage and necrosis occur within the lesion, which is unique, unlike HCA. No malignant degeneration has been observed[30,14]. As mentioned earlier, telangiectatic focal nodular hyperplasia (FNH) is now recognized as an inflammatory adenoma.

Imaging findings:

On MR imaging, classic FNH appears as iso to intermediate on T2w images and iso to slightly hypointense on T1w images. On contrast study they show homogeneous arterial enhancement and wash out during the portal phase. FNH appear isointense (apart from the scar) to the liver parenchyma during equilibrium phase. A characteristic scar, when present, appears as a hyperintense stellate area on T2 weighed images which is hypointense on fat saturated T1 weighted images during the arterial and portal-venous phases, but which demonstrates slight or gradual hyperintensity during the equilibrium phase.

After Gd-BOPTA and Gd-EOB-DTPA administration, FNH is isointense or slightly hyperintense to the surrounding liver parenchyma on delayed and hepatobiliary phase[31].

This unique feature differentiates them from HCAs, which display hypointensity of the solid, non-hemorrhagic components of the lesions (apart from the inflammatory subtype). The diagnostic accuracy of gadoxetic acid–enhanced MR imaging to differentiate HCA vs. FNH is presumed high but given the heterogenous behavior of inflammatory subtype, it might be overestimated and need further supporting studies[32,33].

Heterogeneity and hyperintensity on T1-weighted images, lack of central scar, strong hyperintensity on T2-weighted images, and persistent contrast enhancement on delayed images have been described with telangiectatic focal nodular hyperplasia, which is now described as a variant of inflammatory HCA[34].

Table 1 shows the difference between hepatocellular adenoma, hemangioma and focal nodular hyperplasia.

Hemangioma

Liver hemangiomas are the most common benign liver lesions. On cut section they show large, well-defined blood-filled spaces lined by a single layer of endothelium and separated by fibrous septae[35]. They are usually found incidentally on imaging and are often asymptomatic. Infrequently, large liver hemangiomas present with pain when it bleeds requiring surgical intervention.  Klippel-Trenaunay-Weber disease, Osler-Rendu-Weber disease, and von Hippel–Lindau disease are all associated with giant hemangiomas[36].

The natural history of hepatic hemangiomas is not well understood. A study by Conter et al showed use of OCPs could lead to enlargement of pre-existing hemangiomas or recurrence after resection, although this is controversial. However, there is no published data to support that OCPs cause formation of hemangiomas[37].

A case control study by Gemer et al of 40 women with hemangiomas and 109 controls concluded that liver hemangiomas were not associated with use of OCPs or influenced by menstrual or reproductive hormones[38]. There is no clear association between OCPs and hepatic hemangioma has been established.

Imaging findings:

Hemangiomas typically demonstrate T1W hypointensity, T2W hyperintensity, and peripheral nodular discontinuous enhancement, which progress centripetally on delay images when compared to the background liver. They tend to hold on to contrast on delayed images. Generally, post gadoxetate administration, hemangiomas may appear hypointense on equilibrium and hepatobiliary phase relative to surrounding parenchyma. Hemangioma usually follow intensity following portal venous blood pool on all phases[39].

Hamartomas

In 1974, O’Sullivan et al described three cases of liver hamartomas in patients on OCPs, but later thought them more likely to represent adenomas[40]. The following year, in 1975, Christopherson et al described three cases of liver hamartomas while evaluating hepatic tumors in women on OCPs[41].Goldfarb et al performed a literature review in 1976 about the association of OCP use with benign hepatic neoplasia, but found a lack of consistency in histologic criteria for diagnosis[42]. All case reports of liver hamartomas in patients on OCPs have been dated back to timeworn literature and are uncertain of definitive causative association. Thus, currently there has been no association between OCP use and hepatic hamartomas development.

Hepatocellular carcinoma (HCC)

Most of the primary liver cancer is hepatocellular carcinoma, accounting for approximately 90% of the cases. The main risk factors for primary liver cancer include chronic infection with hepatitis B or C virus, alcohol consumption, and exposure to aflatoxin B1[43]. The risk of developing HCC in women taking OCPs is very low and remains speculative.

In 2015 Ning An performed a dose–response meta-analysis of observational studies to quantitatively summarize the existing evidence of OCPs and liver cancer. It summarized the evidence of fourteen case–control and 3 cohort studies and concluded that OCP use does not have a significant positive effect on the risk of liver cancer[43].

Waetjen et al study showed no support for a measurable effect of OCPs in primary HCC based on data from the USA, Sweden, and Japan.

An Armed Forces Institute of Pathology review of 128 cases of HCC by Goodman et al reported no convincing association of HCC development in women with history of OCP use[44]. Conversely a study by Yu et al demonstrated that use of OCPs for >5 years was linked to the increased risk of HCC in women, independent of seropositive viral hepatitis.  Given discrepant results in the literature, more research in this area may be warranted in the future.

Other studies have shown that the salient features of HCC in patients using OCPs are different, in that the HCC usually develops in a non-cirrhotic liver, does not infiltrate surrounding tissues and rarely metastasizes[45]. Patients often present with fewer symptoms and lower alpha-fetoprotein levels however they also commonly present with hemoperitoneum due to increased vascularity and intraperitoneal rupture[45]. Survival rate is usually longer in these patients.

Budd-Chiari syndrome and venous thrombosis

Budd-Chiari syndrome (BCS) is defined as obstructed hepatic venous outflow due to occlusion of the hepatic veins or inferior vena cava. In western societies BCS is caused mainly by thrombosis of hepatic veins.

OCPs have been linked with an increased incidence of hepatic vein thrombosis (Budd-Chiari syndrome) as well as portal vein thrombosis. It often occurs with an underlying hypercoagulable state including myeloproliferative disorder, acquired abnormalities of coagulation such as lupus or anti-cardiolipin antibodies, as well as due to inherited abnormalities including protein C and S deficiency or abnormalities of clotting factors V and II[46,47]. In addition, BCS has also been implicated in the development of HCA[26].

A study by Valla D et al indicated that the risk of hepatic thrombosis significantly increases in women having recently used oral contraceptives as compared with nonusers[48]. Hepatic venous thrombosis could be sequala of drug-induced hepatocellular injury and genetic predisposition.

Imaging findings:

Ultrasound may identify presence of IVC or hepatic venous thrombosis and monophasic hepatic venous wave forms on doppler studies. Chronic BCS may be seen atrophic and echogenic veins and intraparenchymal venous collaterals. Hypertrophy of caudate lobe can be identified on cross sectional imaging. The liver parenchyma demonstrates early central and delayed peripheral enhancement.

Dilated sinusoids and peliosis hepatis

The first report of generalized peliosis hepatis as a complication of long-term use of oral contraceptives was made by Van Erpecum et al. in 1988[49].

Peliosis hepatis is a rare acquired vascular disorder characterized by randomly distributed blood cysts communicating with hepatic sinusoids. The pathogenesis of peliosis hepatis is unclear but possibly includes hepatocellular necrosis and injury to the sinusoidal endothelium[50].

However, oral contraceptives do not cause sinusoidal endothelial cell injury, in contrast to other agents such as azathioprine, 6-thioguanine and oxaliplatin. Hence, the mechanism by which oral contraceptives lead to peliosis hepatis remains uncertain. The lesions show regression after interruption of usage with incomplete response in some cases progressing to cirrhosis[49].

On gross inspection, the liver has a swiss cheese pattern with multiple blood-filled cystic spaces. Two pathological types have been described – parenchymal type is characterized by hemorrhagic parenchyma necrosis with blood cavities neither lined by sinusoids/ fibrosis and phlebectatic type based on aneurysmal dilatation of central vein with cavities lined by endothelium and fibrosis[51]. Peliosis secondary to OC pill use show profound sinusoidal dilatation with venous lakes[52]. The varied clinical manifestations range from incidental to abnormality in liver enzymes, massive hepatomegaly, liver failure and life threatening hemoperitoneum[53-55].

Chronic use of OCPs has been associated with sinusoidal dilatation as seen on liver biopsies. Cessation of OCPs has been shown to regress the severity of peliosis. Patients may present with abdominal pain due to hepatomegaly. Hepatic arteriography shows stretched and attenuated branches of the hepatic arteries with a patchy parenchymal pattern of enhancement.

The radiological presentation can be focal to diffuse with varying appearance depending on the size and presence of hemorrhage[56]. USG appearance is homogenous hypoechoic in patients with steatosis, hyperechoic in healthy liver and heterogeneous if complicated by hemorrhage. Doppler ultrasound can demonstrate perinodular and intranodal vascularity[57].

The lesions are typically hypoattenuating on noncontract CT and the appearance can vary secondary to presence of blood products. Characteristic pattern of CT enhancement includes early vessel like arterial enhancement with centrifugal fill in , centripetal fill in is also known [57,58]. Phlebectatic variant demonstrate early vessel like arterial enhancement with centrifugal progression and late diffuse hyperattenuation. Small lesions (<2 cm) can show hyperattenuation in both arterial and venous phases [58]. On MRI, the lesions have low signal intensity on T1 w Images and hyperintensity on T2 w and DWI images [56,57]. There is delayed and slow but intense enhancement on contrast-enhanced T1-weighted images. They are predominately hypointense on hepatobiliary phase with some lesions retaining contrast indicative of cholestasis [58]. On angiography, the lesion show contrast accumulation on late arterial phase which persists in portal venous phase[57].

Summary

Use of OCPs is associated with an increased risk of certain liver diseases and neoplasms. Recognizing these entities and understanding the potential influence of OCPs is of importance for diagnosis and management. Certain hepatic tumors such as hepatocellular adenoma and tumor-like lesions including peliosis hepatis have a proven association with long-term use of OCPs. Increased incidences of hepatic vein and portal vein thrombosis are often linked with an underlying hypercoagulable state. Association of FNH, hepatocellular carcinoma, hemangiomas and hamartomas with OCPs has randomly been described in the literature but is not clearly established.  Further research is needed, particularly with lower dose OCPs to determine their ultimate effect on patient health.