Benign Liver Tumors

Due to the frequent use of medical imaging including ultrasonography, the incidence of benign liver tumors has increased, most of them being asymptomatic. There is a large variety of solid benign liver tumors of different cellular origin (Table 1). However, the most frequent lesions are hepatic hemangioma (HH), focal nodular hyperplasia (FNH), and hepatocellular adenoma (HCA). As imaging techniques improve, a precise diagnosis is possible in most cases reducing the need for a percutaneous biopsy. However, atypical lesions may require more than one imaging modality. Usually, clinical and radiological findings are sufficient to pinpoint an accurate diagnosis, and additional biopsy is only needed under few circumstances. In the last 10 years, several developments in radiological techniques have been described. A summary of well-established imaging features for benign liver tumors is depicted in Table 2.

The most common benign liver tumor is HH. Its incidence varies from 3 to 20% [13]. The female/male ratio is 5:1 [14]. Often, HH is an incidental finding during radiological imaging [15]. HH are mostly solitary lesions, and sizes range from a few millimeters to up to 30 cm, but most HH are <5 cm in diameter. HH >10 cm is referred to as giant hemangiomas (Fig. 1). In general, HH are well-demarcated tumors with a spongy red-brown cut surface. Long-term lesions frequently contain white fibrotic areas (Fig. 2). Based on microscopic findings, HH can be divided into 3 main subtypes: cavernous, capillary, and anastomosing HH. The cavernous HH is the most common type. It is characterized by widely dilated vascular channels with fibrous walls lined by a single layer of flat endothelial cells. Cytological atypia or mitosis are absent. Usually, the blood-filled channels can show fresh or organized thrombi (Fig. 2). Large und long-existing lesions with partial or complete sclerosis are known as sclerosed hemangiomas. It is known that HH arises from vascular malformation; however, its exact pathogenesis remains to be elucidated. Recently, it has been shown that an increase in vascular endothelial growth factor (VEGF) plays an essential role in the development of hemangioma [16]. This concept was supported by case reports showing hemangioma shrinkage following anti-VEGF treatment [17]. However, Lee et al. [18] found no effect in 21 anti-VEGF-treated patients. Although very few hemangiomas have estrogen receptors, growth of hemangiomas has been observed after hormone replacement therapy, oral contraceptives, and pregnancy [19]. The direct mechanisms of those hormone effects are still unknown. Current evidence does not support a contraindication for oral contraceptives or hormone substitution in hemangioma patients [2].

Even though most hemangiomas are asymptomatic, large lesions may cause pressure on adjacent structures, including compression of the vena cava, portal vein, and the bile duct, leading to abdominal pain, discomfort, fullness in the right upper quadrant, nausea, and early satiety [20]. Rare symptoms are fever, jaundice, dyspnea, high-output cardiac failure, and hemobilia [21, 22]. Spontaneous bleeding of HH is uncommon [23]. Donati et al. [23] found only 46 cases with spontaneous bleeding in a literature review between 1898 and 2010. Low-grade fever, weight loss, abdominal pain, anemia, thrombocytosis, and increased fibrinogen level can be caused by inflammation due to partial thrombosis of the hemangioma. The Kasabach-Merritt syndrome represents a rare complication of giant HH [24]. This coagulopathy, which consists of intravascular coagulation, clotting, and fibrinolysis within the hemangioma, may progress to secondary increased systemic fibrinolysis and thrombocytopenia. The Kasabach-Merritt syndrome has an incidence ranging from 0.3% in all hemangiomas up to 26% in hemangiomas >15 cm [20, 25]. It is life-threatening but reversible after removal of the hemangioma.

Usually, liver tests including alkaline phosphatase and γ-GT are normal in patients with hemangioma. If patients present with an active cholestasis, it is due to the pressure on the bile duct system caused by the hemangioma.

Small, asymptomatic hemangiomas do not need treatment or follow-up. In HHs >5 cm, follow-up at 6–12 months has been suggested to assess the growth rate [26]. There is no need to make a change in specific lifestyle measures for patients with asymptomatic hemangiomas, and there is no evidence that the use of oral contraceptives or pregnancy has a negative impact on patients with HHs. Regardless of the size, treatment should be restricted only to symptomatic patients with pressure to adjacent organs or complications such as the Kasabach-Merritt syndrome and rupture with intraabdominal bleeding. Inability to exclude malignancy is also an indication for resection.

However, abdominal pain in hemangioma patients should be carefully analyzed. Other possible causes should be ruled out before definitive treatment is decided upon.

Surgery remains the most common treatment for symptomatic HHs (Fig. 1). In general, the surgical armamentarium consists of liver resection – including hypertrophy concepts like preoperative portal vein embolization, enucleation, hepatic artery ligation, and liver transplantation [27-32]. The choice of the procedure depends on HH size, number, and location, the surgeon’s experience, and the institutional resources. Several studies comparing enucleation with resection showed that enucleation is associated with lower morbidity, shorter operation time, less blood loss, and fewer complications [28, 29]. Furthermore, enucleation can preserve more hepatic parenchyma [28]. However, a single-center study in 86 patients with hemangiomas >10 cm showed no difference between enucleation and resection regarding operation time, blood loss, complications, and hospital stay [33].

The enucleation of centrally located hemangiomas requires a significantly longer vascular inflow occlusion and operating time, and more blood transfusions than enucleation of peripherally located hemangiomas [31].

Recently, laparoscopic liver resection has gained widespread acceptance and is considered to be a safe approach for the management of benign liver lesions. A propensity score matching analysis between laparoscopic and open HH surgery showed that the laparoscopic approach was associated with less blood loss, shorter postoperative hospital stay, and lower complication rates than the open approach [34]. However, short-term “quality of life” outcomes did not differ between the different treatment groups [34].

On rare occasions, even liver transplantations as an extreme surgical approach have been performed in patients suffering from HHs [35]. Published studies are mainly limited to small case series. However, Sundar Alagusundaramoorthy et al. [30] analyzed the results of liver transplantations for benign solid tumors of the United Network of Organ Sharing. Of 87,280 transplantations, 25 have been performed for HH, and their overall survival rates were 87.8, 81.5, and 74.8% at 1-, 3-, and 5-year follow-ups, respectively. Due to the significant postoperative morbidity and shortage of donor livers, the indication for such an aggressive therapy has yet to be defined. So far, liver transplantation should be reserved for unresectable giant HHs causing severe symptoms, failure of previous interventions, or life-threatening complications such as the Kasabach-Merritt syndrome [32].

Since surgical treatment can be associated with massive intraoperative blood loss, long operating and hospitalization times and perioperative complications, such as bile leakage for example, nonsurgical treatment options should be considered. However, no consensus exists among surgeons and interventional radiologists. Transcatheter arterial embolization (TAE) has been performed to control acute hemorrhage or to shrink HHs prior to surgery, which was confirmed in case reports [36, 37]. Substances used for TAE are, for example, metallic coils, polyvinyl alcohol, Gelfoam particles, and liquid agents such as N-butyl-2-cyanoacrylate and bleomycin-lipiodol [36-38]. A recent study including 23 patients and 29 HHs treated with TAE using bleomycin-lipiodol showed a 50% reduction in the hemangioma volume in 17 of 23 patients [39].

Available studies are limited due to retrospective design and small interventional sample sizes, thus prospective studies and longer clinical follow-ups are necessary to determine the exact role of TAE in the treatment armamentarium of HHs.

Radiofrequency ablation (RFA) has been reported as an alternative treatment of hemangiomas [40-42]. RFA can be used percutaneously, laparoscopically, or during open surgery. RFA is usually performed under ultrasound (US) guidance, but CT guidance for percutaneous RFA is also suitable for hemangiomas located deeply in liver parenchyma [40]. Laparoscopic RFA with US guidance is preferred for subcapsular HHs [41]. Compared with open resection, laparoscopic RFA is associated with shorter operative time, less pain, and shorter hospital stay [42].

FNH is the second most common benign liver tumor (Fig. 3). Its incidence is between 0.3 and 3% [13], and it is mostly found at 30–40 years of age [43]. The role of hormones in FNH development is supported by its high female predominance (female:male ratio  of  9:1) and high prevalence amongst women taking oral contraceptives [44]. Recently, estrogen receptor expression was found in 12 of 13 FNH tissue samples [45]. The grossly well-demarcated lesion does not show a capsule, and on cut section, lobulated, firm parenchyma with slightly lighter color than the surrounding liver presents. Eye catching and very characteristic is the presence of one or more stellate scars (Fig. 3). The liver parenchyma, by definition, is noncirrhotic. Histologically, FNH looks like a focal cirrhosis because of fibrous bands encompassing groups of benign hepatocytes, which could be a potential pitfall in needle biopsy (Fig. 4). True portal tracts are absent. Bile ducts are replaced by ductular proliferation along the fibrous septa, and, characteristically, there are arteries with medial hypertrophy und intimal fibrosis, so-called dysplastic arteries. FNH occurs solitary in 80–95% and is usually <5 cm in diameter. In up to 20%, multiple FNH occur, sometimes in association with hemangioma or HCA [46]. FNH is considered as a hyperplastic, regenerative response to arterial hyperperfusion and shunting with characteristic anomalous arteries found in the center of the nodules [47]. FNH might also be associated with vascular disorders of the liver, as Gincul et al. [48] found FNH nodules in 13.7% of patients with hereditary hemorrhagic telangiectasia, and Cazals-Hatem et al. [49] identified them in 9 of 17 explanted livers with Budd-Chiari syndrome. Moreover, congenital absence of portal flow or portal vein thrombosis with subsequent hepatic arterialization is associated with the development of FNH [50, 51].

There is no evidence for malignant transformation of FNH [52]. The majority of patients with FNH is asymptomatic. Only a few patients with large FNH show symptoms, such as abdominal pain or discomfort, due to pressure effects on adjacent organs [53]. Rupture, necrosis, or bleeding is exceptional [54]. As there is most often no change in blood parameters, laboratory tests are not helpful for making the diagnosis of FNH. The natural course of FNH is uneventful, with very few complications and changes over time. However, enlargement of FNHs during oral contraceptive medication and pregnancy have been reported [55]. Therefore, asymptomatic patients should be managed conservatively if the diagnosis is firmly established [56]. Follow-up controls at 6 months are sufficient to prove the stability of the lesion and its benign nature, and no long-term follow-up is routinely required afterwards. Surgery should strictly be considered for patients with symptomatic or highly suspicious lesions, where malignancy cannot be ruled out with modern imaging or even biopsy. In some symptomatic patients with large FNH, where a risky resection would be needed, TAE has been successful [10, 57]. Experience with TAE shows that it could be an attractive treatment option to decrease the size of the lesions and, therefore, symptom relief [58, 59]. Furthermore, RFA has been described as an effective treatment for symptomatic FNH [60].

As numerous reports regarding FNH during pregnancy have been published, the association with endogenous and/or exogenous estrogens is very likely. Small lesions seem not to be a significant risk for a successful pregnancy, although observation is strongly recommended [56].

HCA is a benign hepatic tumor which occurs predominantly in young and middle-aged women who take oral contraceptives or other steroid medications (Fig. 5) [61]. The annual incidence is 3–4 per 100,000 in women who have used oral contraceptives over a sustained period [62].

Baum et al. [63] first described the association between oral contraceptives and HCA occurrence in 1973. In the 1960s, benign liver tumors were in general relatively rare until the introduction of oral contraceptives [64]. Between 1918 and 1954, only 2 HCA cases were found in 48,900 autopsies performed in the Los Angeles General Hospital [65]. Estrogens seem to be a predominant factor for the development of HCA. The positive correlation between oral contraceptives and HCA incidence is dose dependent, as spontaneous regression of HCAs after estrogen withdrawal has been observed [66]. Androgen intake is also associated with the development of HCA [67]. Furthermore, the influence of obesity on the development of HCA is suggested [68], as recently a higher incidence of HCAs in patients suffering from nonalcoholic steatohepatitis has been reported [1].

HCA development is associated with several genetic syndromes, including glycogen storage diseases, particularly with type Ia glycogenesis, in which more than half of patients have multiple tumors [69]. Hepatic vascular abnormalities, including portal deprivation or both intra- and extrahepatic portosystemic venous shunts, are involved in the development of HCAs [1].

HCA is a heterogeneous entity that includes several tumor subtypes, which are associated with the risk of various complications (Table 3). First HCA was divided into 4 subtypes according to genomic analysis: inflammatory HCA (I-HCA), hepatocyte nuclear factor (HNF) 1A-mutated HCA (H-HCA), β-catenin-activated HCA (β-HCA), and unclassified HCA [70]. Later, Nault et al. [71] described 6 molecular subtypes of HCA (Table 3). The β-HCA group was additionally subdivided into 2 subgroups, exon 3 β-catenin-mutated and exons 7–8 β-catenin-mutated HCA [71]. The different HCA subtypes carry specific types of risks (Table 3). For the sonic hedgehog HCA, Nault et al. [71] found a higher risk for symptomatic bleeding. Independent of size, in males, HCAs bear an increased risk of malignant transformation.

Histologically, HCAs arise in noncirrhotic liver and appear as circumscribed nonencapsulated lesions. The yellow to brown color and soft to dense texture depends on the degree of hemorrhage and cystic changes. Microscopic appearance differs depending on the subtype, but the lack of portal tracts and isolated naked arteries is common in all subtypes (Fig. 6). To specify a certain subtype, there are several characteristics in combination with immunohistochemistry and molecular features. Diffuse steatosis is commonly seen in HNF1α-inactivated HCA. A marked, predominantly lymphocytic inflammation resembling portal tracts at low magnification is the main feature of the inflammatory subtype. It is associated with alcohol consumption and/or an increased BMI and has an increased risk of hemorrhage. β-Catenin-activated HCAs do not show any special morphological features. Nuclear β-catenin expression and diffuse glutamine synthetase reactivity as a surrogate marker of β-catenin activation are their hallmarks. β-Catenin HCA is more likely to show cytological atypia. A small subset of adenomas remains unclassified even after immunostaining. Key differentiating features of HCA and HCC are a preserved reticulin framework and absence of mitosis in HCA.

The majority of patients with HCA are asymptomatic and diagnosed incidentally during liver ultrasound examinations. In symptomatic patients, the most common symptom is abdominal pain. The main complication of HCA is hemorrhage; it occurs in approximately 25% of patients [72]. In large tumors (>5 cm in diameter), visualization of arteries within the HCA, location in the left lateral liver, and exophytic growth are risk factors for hemorrhage [72, 73].

The overall risk of malignant transformation in HCA is reported to be 5–6%, but it is probably lower, as it is difficult to distinguish between HCA and well-differentiated hepatocellular carcinoma (HCC) in imaging and even histopathologically. HCA may transform into HCC (adenoma-carcinoma sequence), but the natural history of this progression is not well defined and understood [74, 75].

The risk of malignant transformation can vary from virtually 0 to almost 50%, depending on patient-related factors, tumor size, and pathology classification, which makes the relevance of its overall estimation questionable. Risk factors for malignant transformation include sex, androgen use, β-catenin HCA subtype, and tumor diameter >5 cm. A substantial number of I-HCAs can also contain β-catenin mutations and are at risk of malignant progression.

Up to one third of patients diagnosed with HCA have multiple lesions. Multiple HCAs (>3 lesions) of various sizes are termed liver adenomatosis [76]. In patients with adenomatosis, the risk of complications is not higher than in patients with a solitary HCA; therefore, adenomatosis patients do not have to be treated differently.

Imaging for differential diagnosis between HCC and HCA can be challenging. In rare cases, HCCs have been misinterpreted as HCAs. Modern CT and MRI enhanced by gadobenate dimeglumine or gadoxetate disodium are very effective in differentiating HCA from FNH or other lesions, and also in identifying the HCA subtypes, if interpreted by a radiologist experienced in hepatobiliary lesions (Table 2) [11, 77]. If not, a biopsy of the lesion can be indicated to define the subtype according to their specific genetic and molecular markers. However, biopsy can be risky due to the vascular nature of HH and should, therefore, be reserved for cases in which the histology will have an impact on treatment decision [15].

In male patients, HCAs should be resected regardless of the size due to the high risk of developing malignancy. In female patients with HCAs <5 cm, cessation of hormonal therapy with MRI surveillance is recommended as initial management. In small HCAs that cannot be characterized by MRI, percutaneous biopsy has been advocated by some groups. For female patients with persistent HCAs >5 cm after cessation of hormonal therapy, surgical management is indicated. The risk of vascular invasion or lymph node involvement is reduced in HCA patients; therefore, a wide resection margin or regional lymphadenectomy does not seem necessary. Thus, if resection of HCA is indicated, a laparoscopic approach should be preferred, because it seems to be superior to open surgery in terms of quality of life after surgery and operative outcomes [78].

In patients with multiple HCAs, only tumors >5 cm in diameter should be resected. For smaller lesions in female patients, pathological confirmation is not mandatory, and regular follow-up is recommended. Indeed, follow-up data of HCAs <5 cm show that most remain stable in size, and sometimes even decrease or disappear at all.

To manage bleeding HCAs, TAE is widely used in order to prevent recurrence and occasionally even to reduce tumor size. However, preoperative TAE can be an option to reduce intraoperative blood loss [79]. Although patients with a ruptured HCA may display severe abdominal pain and have free intraperitoneal hemorrhage, hemodynamic stability is common in the majority of these patients making a conservative approach possible. After bleeding, in the absence of tumor or in the presence of minimal residual tumor tissue on MRI, a conservative nonoperative approach with MRI surveillance might be considered. However, when HCA tissue is left in a large-sized HCA, surgical resection should be performed. The efficacy of TAE on tumor regression in hemorrhage lesions has led some groups to consider using TAE electively in nonbleeding HCAs [80]. Liver transplantation might only be considered in very few situations, including multiple unresectable lesions in men, a large HCA associated with intrahepatic venous shunt, and in patients with glycogen storage disorders unresponsive to medical treatment [81].

Benign liver tumors are often diagnosed randomly. HHs are the most frequent lesions. The majority of benign liver tumors does not need any kind of treatment, and even follow-up is only necessary in some patients. In asymptomatic HH and FNH regardless of their size, no intervention is required. Patients with symptomatic HH and FNH and an impaired quality of life can be referred for surgery or nonsurgical therapeutic modalities, such as radiofrequency or TAE. The situation for patients with HCA is different, as severe bleeding and malignant transformation are associated risk factors. HCA can be stimulated by metabolic or hormonal abnormalities. Therefore, oral contraceptives and anabolic steroids should be avoided after being diagnosed with HCA. Six subtypes of HCA are identified on the basis of genomic analysis. They display different risk profiles. Therefore, a precise diagnosis including the identification of the subtype is helpful for treatment decisions. This can be obtained by modern MRI technology in most cases. Biopsies are seldom necessary. In male patients, HCA should be resected regardless of the size. In women with HCA <5 cm, cessation of hormonal therapy is recommended with MRI surveillance. In women with HCA >5 cm after cessation of hormonal therapy, surgical management is recommended. Pregnancy is not generally contraindicated for patients with HCA <5 cm. Since the majority of patients with operated benign liver tumors are suffering from tumors of larger size and of difficult locations, diagnosis and surgery should be performed in experienced hepatobiliary centers.

Since this is a review article, institutional approval and patient consent were not required.

All authors have no conflicts of interest to declare.

Authors received no funding for this paper.

K.J.O. and K.C.W. wrote the manuscript, K.H. added the pathology part, and V.H. and G.M. critically revised the manuscript.