Focal Nodular Hyperplasia—A Review of Myths and Truths
- First Online:
- Cite this article as:
- Nahm, C.B., Ng, K., Lockie, P. et al. J Gastrointest Surg (2011) 15: 2275. doi:10.1007/s11605-011-1680-x
- 625 Views
Focal nodular hyperplasia (FNH) is a benign hyperplastic lesion of the liver with no known malignant potential. It has generated much interest due to the frequency with which it presents with atypical features on radiological imaging. Often resulting in misdiagnosis. Moreover, the understanding of particular subtypes of this lesion at a molecular level has changed in recent years. This may have implications on how certain subtypes should be managed.
This review aims to analyse current literature pertaining to FNH and to provide clinically relevant advice regarding diagnosis and management.
KeywordsDiagnosisFocal nodular hyperplasiaLiverNeoplasmsTreatment
Focal nodular hyperplasia (FNH) is a benign tumour of the liver with an indolent course, no known potential for malignant transformation, and an extremely low rate of rupture or haemorrhage. It is thought to arise as a hyperplastic lesion in response to a preexisting arterial malformation. The widespread use of imaging has resulted in FNH being found occasionally as an incidental finding. Although this lesion has a well-documented natural history, management decisions may be difficult because of atypical radiological features and because there is still uncertainty as to whether growth is influenced by the hormonal milieu. Moreover, with recent developments in the understanding of certain FNH subtypes at a molecular level, management options are still evolving. The aim of this review is to analyse the current literature regarding FNH in order to provide clinically relevant advice about the diagnosis and management of these lesions.
FNH is the second most common benign liver tumour after liver haemangiomas.1 In a large autopsy study of 2,270 cadaveric specimens there was an incidence of 0.31%1 which fits with our clinical experience and confirms that this diagnosis is far less common than liver haemangioma. Although FNH may affect both women and men of all ages,1,2 it is most often found in females between the ages of 30 and 501,3 with a reported female-to-male ratio of between 8:1 and 12:1.2,3 This diagnosis is rarely made in the paediatric population. In fact, one case series claiming to describe 21 paediatric cases of FNH should be interpreted with caution as the microscopic findings of a fibrous tumour in many of the lesions in this study do not match with the currently accepted histopathological descriptions of FNH.4 Accordingly, the diagnosis of FNH in a child should be questioned. Equally puzzling is the fact that this lesion is almost never seen in elderly patients, suggesting that focal nodular hyperplastic lesions may involute as women pass through menopause.
From the literature, it is difficult to determine what proportion of patients with FNH suffers from symptoms directly attributable to the lesion itself. Most cases are discovered incidentally on abdominal imaging.3 Some authors have suggested that large subcapsular lesions may cause stretching of Glisson’s capsule or displacement of adjacent organs and that this might lead to vague abdominal pain.5 Many reports attributing abdominal pain to the presence of an FNH do not describe rigorous attempts to exclude alternative causes for the patients’ symptoms.6,7
Clinicians should be cautious when attributing symptoms to FNH as the physiological mechanisms of pain in the context of this lesion have not been specifically identified. Nevertheless, FNH may grow extremely large, up to 190 mm,2 and occasionally present with an abdominal mass or hepatomegaly.2,3 Liver function test abnormalities are uncommon unless the FNH is large enough to cause extrinsic intrahepatic biliary duct compression, which usually manifests as a mildly elevated serum gamma-glutamyl transferase level.
There are several case reports describing a first presentation of FNH in patients with acute abdominal pain due to intratumoral haemorrhage and a subsequent haemoperitoneum.8,9 However, this is rare and is most often in patients with large exophytic tumours or in those with multiple FNH lesions.10 In this latter case report, the histopathological description of the culprit lesion describes the presence of “portal triads” which many experts believe do not occur in true FNH. This is a typical example of how the literature has become confused on this topic and why there needs to be a consensus regarding diagnostic criteria for FNH.
Histopathology and Pathogenesis
The classic histopathological description of FNH is that of a non-encapsulated nodule with a central fibrous body from which there are radiating septa dividing nodules of hyperplastic hepatocytes sometimes forming plates that are two cells thick. The central fibrous regions usually contain abnormal vessels of different sizes, as well as proliferating bile ductules. However, FNH is typically void of any formal portal triads.1 In addition to this classical FNH description, Nguyen et al., in a study of 305 FNH lesions, defined three “nonclassical” histological subtypes. These accounted for 20% of their series, and there was no central fibrous scar in any of these cases.2 The subtypes described include the telangiectatic FNH (tFNH), the mixed hyperplastic and adenomatous forms, and FNH with cytologic atypia. tFNH is characterised by dilated sinusoids and a gross resemblance to hepatic adenoma. Mixed hyperplastic and adenomatous forms essentially contain separate regions resembling tFNH and hepatic adenoma but with some lesions showing transitional morphological features between the two. This transitional appearance is an interesting observation consistent with recent molecular studies of FNH discussed below. Only eight of 305 lesions in this series were categorised as FNH with cytologic atypia. This group of lesions was characterised by the presence of atypical hepatocytes with enlarged hyperchromatic nuclei with irregular contours frequently demonstrating cytoplasmic–nuclear inclusions. Five of these lesions demonstrated classical FNH morphology, two lesions demonstrated tFNH, and one lesion was of mixed type.
FNH nodules are usually solitary but may be multiple in approximately 20% of cases.1,2 Reports of up to 30 lesions in one patient have been documented.2 The size of the lesion can vary between 1 and 190 mm in diameter, but most are in the range of 40–50 mm at the time of presentation.11,12
Of great clinical interest is the debate about the exact role of the oestrogen or progesterone milieu in the development or growth of FNH. The clinical dilemma about whether pregnancy or the oral contraceptive pill has any impact on incidental or suspiciously symptomatic lesions is ongoing. However, a review of the relevant literature does allow some sensible conclusions. Importantly, there has been a lack of prospective comparative series of decent sample size to make any firm statement regarding the impact of the hormonal milieu on FNH. There are numerous series and reports suggesting that oral contraceptives are a contributing factor in the development of FNH, especially with long-term use.14,15 Some of these reports, although small in sample size and lacking rigorous study design, attained histological confirmation for the majority of the lesions that were diagnosed as FNH.14,16 However, much of the supportive literature describes patients presenting in the 1960s and 1970s when oral contraceptive formulations contained much higher doses of oestrogen (>50 mcg) than at present. Also, radiological and even histological diagnosis was not as accurate as it is today. More recent literature, albeit the majority only with a radiological diagnosis of FNH, seems to refute the argument for an association.7,17,18 However, it is also possible that modern low-dose oestrogen formulations of oral contraceptives have averted the risk of patients developing FNH. Regardless, and to be definitive, more prospective studies of larger sample sizes and with histopathological confirmation of the diagnosis of FNH would need to be undertaken.
The effect of pregnancy on the behaviour of FNH has only been observed in a few small case series and individual case reports, most of which describe stable appearances of the lesion and uncomplicated deliveries.7,19
More recent developments in the study of FNH pathogenesis have been at a molecular level, focussing on clonal analysis. Most of the literature dealing with this subject supports a polyclonal origin of FNH,20–22 which is consistent with the most accepted theory that it is a hyperplastic lesion. Only one series of ten lesions in nine patients has proposed that most FNH lesions are monoclonal.23 This last series describes the histopathology of each studied FNH lesion as demonstrating a thin fibrous capsule demarcating it from the surrounding liver parenchyma, which is not in keeping with the current understanding of FNH. It is possible that the lesions studied in this paper were not true FNH. Others have demonstrated a lack of somatic gene mutations, especially of those involved in liver tumorigenesis, further supporting the theory that FNH is truly a nonneoplastic lesion.21,22,24 However, several small but recent pathological series found parts of the FNH nodules were monoclonal, suggesting, at least, partial neoplastic transformation.25,26 These intriguing findings raise further questions about malignant potential but may be specific only for certain subtypes of FNH. In this regard, Bioulac-Sage et al., in a series that compares 13 telangiectatic FNH (tFNH) with 28 classical FNH and 17 hepatic adenomas, showed that 100% of tFNH are monoclonal in origin and more closely resemble hepatic adenomas than classical FNH.21 Paradis et al. made similar conclusions in a series studying the patterns of X-chromosome inactivation in ten tFNH, six typical FNH, and six hepatocellular adenoma lesions. Seventy-five percent of tFNH lesions in this series were monoclonal in nature, whereas 100% of typical FNH lesions were polyclonal and 100% of hepatocellular adenoma lesions were monoclonal.27 This confirms that tFNH should be considered as a separate subtype when classifying FNH.
The Natural History of FNH
Apart from scattered case reports of the synchronous discovery of FNH and hepatocellular carcinoma,28,29 there has never been a documented case of transformation of a histologically proven FNH into a malignant lesion. In support of this lack of risk, numerous series describe long-term follow-up imaging of histologically proven FNH where no malignant transformation occurred.30–32
Changes in the size of FNH lesions are common and may simply reflect dynamic growth fluctuations known to occur in normal liver tissue. In an ultrasound study of 16 cases of FNH in which 14 cases were histologically proven, size reduction was noted over a mean follow-up of 33 months in seven cases, with complete disappearance of the lesion in one patient. A larger study following 53 patients with conservatively managed FNH over a mean of 32 months demonstrated that a minority of lesions may increase in size.32 Another imaging study using ultrasound that followed 30 patients with 34 FNH nodules followed for a mean period of 42 months showed 71% were stable in size, 3% increased in diameter by more than 30%, and 27% even regressed in size over the follow-up period.33 It has been postulated that lesion regression occurs as a result of thrombosis of the feeding artery, a pathological finding that was confirmed in one large autopsy study.1 The stability of most FNH lesions along with the likelihood of regression with age, the lack of potential for malignant transformation, and the extremely low risk of rupture and haemorrhage supports a conservative approach to the management of most patients with radiologically convincing or histologically proven FNH.
Imaging of FNH
The similarity on imaging between FNH and other hepatic lesions may cause diagnostic dilemmas. FNH, on multiple imaging modalities, may mimic a liver cell adenoma34 and even a hepatocellular carcinoma (particularly the fibrolamellar subtype).35 Obviously, it is important to differentiate FNH from both of these diagnoses which usually mandate resection. The typical appearance of FNH as seen with different imaging modalities has been described. When these findings are atypical more invasive diagnostic measures may be required to confirm or exclude the diagnosis.
Although ultrasound is highly sensitive for diagnosing FNH, frequently FNH cannot be characterised well, and therefore, further imaging may be required to make a confident diagnosis. The typical ultrasound (US) appearance is a well-demarcated homogeneous hypo- or isoechoic lesion, although confusingly FNH can sometimes be hyperechoic. A hyper- or hypoechoic central scar is particularly helpful for a diagnosis, but is seen in less than 20% of cases.36,37 Usually, there is absence of a peripheral hypoechoic rim, compatible with the absence of a true capsule.
Colour Doppler US may assist by identifying a central arterial structure, with a “spoke-wheel” pattern of peripherally radiating smaller aberrant vessels, or even a basket pattern of vessels in some cases.37,38 However, similar vascular patterns can also be seen in hepatic malignancy and therefore these findings should be interpreted with caution.38
Overall, US is attractive because it avoids the radiation risks of computed tomography (CT) scanning, and is more accessible than magnetic resonance imaging. However, US is operator-dependent, and image quality may be limited by large body habitus or overlying bowel gas.
Computed Tomography (CT) Scanning
A radiological examination of 13 patients with histologically proven tFNH found features on CT that distinctly differed from those of classical FNH. Comparatively, tFNH lesions were more likely to be multiple in number (62%), heterogeneous (43%), without a central scar (92%), and persistently enhancing on delayed phase imaging.44 Lesions with this appearance on CT should be approached with suspicion, especially in light of new molecular evidence that supports the classification of tFNH as a variant of hepatic adenoma rather than of FNH, as previously discussed.
There are some features on multiphase CT which can assist in the differentiation between FNH and hepatocellular adenoma. On non-contrast scans, hepatocellular adenomas may exhibit hyperdensity if there has been prior haemorrhage and may contain low attenuating focal fat deposits. This contrasts with FNH which tends to be relatively homogenously iso- or hypodense. In the arterial phase, hepatocellular adenomas also exhibit hyperdense enhancement, but in a more heterogeneous pattern and usually to a lesser degree than FNH. Adenomas then gradually lose their heterogeneity and hyperdensity, eventually becoming homogenously hypodense during the delayed phase due to arteriovenous shunting. The presence of a central hypodense area in FNH, representing a fibrous central scar, is rare in hepatocellular adenoma.45
Differentiation of FNH from the fibrolamellar subtype of hepatocellular carcinoma (FLHCC) may be difficult on CT, particularly if FNH exhibits atypical findings such as the presence of intralesional calcification.46 In a retrospective analysis of FNH in 295 patients, calcification was demonstrated in only 1.4% of cases, and this prompted resection. This pathological feature may also be seen in cholangiocarcinoma, hepatocellular carcinoma, as well as in some hepatic metastases.47,48 Obviously, the presence of a very large lesion (such as greater than 100 mm) or any associated vascular or biliary tract invasion or locoregional lymphadenopathy would go against a diagnosis of FNH. Nonetheless, the fact that FNH and FLHCC occur in the same age groups and in those patients with no underlying liver disease may lead to confusion if the lesions have an atypical appearance.35
Magnetic Resonance Imaging
As a modality with no radiation risk and excellent tumour characterisation capabilities, magnetic resonance (MR) imaging is a valuable tool in the diagnosis of FNH and other liver lesions that occur in women of child-bearing age. The typical findings on MR are that of a homogeneous lesion that is isointense or slightly hypointense on T1-weighted images, and isointense or slightly hyperintense on T2-weighted images. The central scar is typically hypointense on T1 and hyperintense on T2. During arterial phase enhancement, a typical FNH appears homogenously hyperintense apart from a hypointense central scar. During the portal phase, the lesion returns to isointensity or occasionally remains slightly hyperintense. On delayed phase images, the FNH is either isointense or slightly hyperintense, with the central scar often exhibiting avid enhancement due to slow diffusion of contrast through the myxomatous stroma.49,50 In one series of 41 patients with FNH reported by Cherqui et al., a central scar was visualised on MR imaging in 78% of cases.51
Although MR may be very helpful for diagnosing FNH, unusual or atypical features are frequently found. In one series of 37 patients with 48 histologically proven FNH lesions, the typical appearance of isointensity on T1- and T2-weighted images and a central hyperintense scar on T2 was seen only in 43% of cases. However, this particular series may represent surgical bias as these cases were resected presumably because of unusual radiological features.34
Unusual or atypical features of FNH on MR, including the absence of a central scar, a strong hyperintense signal on T1 (which may occur due to steatosis), a hypointense central scar on T2, the presence of a complete or incomplete pseudocapsule (which may occur due to dilated vessels and sinusoids around the lesion), or a strong hyperintense lesion on T2 make diagnosis difficult and, from a clinical viewpoint, suggest either biopsy, resection, or a period of observation. In particular, a T2 hypointense central scar is often suggestive of fibrolamellar HCC.
As in CT imaging, telangiectatic FNH demonstrates different imaging features on MR as compared with classical FNH. tFNH, as detected by MR, is more likely to be heterogeneous (43% of lesions), hyperintense on T1 (53%), strongly hyperintense on T2 (44%), without a central scar (92%), and to demonstrate persistent enhancement in delayed phase imaging.44
One study using delayed phase contrast-enhanced MR imaging showed that FNH could be successfully differentiated from other benign or malignant liver lesions with an overall accuracy of 98%.52 Unfortunately, as with many studies related to FNH imaging, only 58 of 100 patients in this series had a histopathological diagnosis, and therefore, the results are difficult to interpret.
Nuclear medicine imaging has an important role in differentiating FNH from other liver lesions. Uptake of radiolabelled colloids (technetium Tc 99 m sulphur colloid) occurs because of Kupffer cell activity within many FNH.53 Although this is considered highly diagnostic, unfortunately only 7–30% of cases show increased uptake.54 While frustrating, this test is still important because uptake demonstrated as a “hot spot” excludes a malignant lesion or a liver cell adenoma.55
There are conflicting reports regarding the benefits of needle biopsy in the diagnostic work-up of FNH. Due to the high specificity of CT and MRI in diagnosing most FNH, there is usually no indication for biopsy in the presence of typical radiological features. However, needle biopsy may be appropriate in selected cases if imaging is indeterminate.56 In these situations, the risks of a biopsy have to be balanced with the potential benefits. These risks include the small but real chance of bleeding or seeding of malignant cells if the lesion is not benign.
In a series of 155 patients with benign liver tumours (including 42 FNH), 30 patients underwent percutaneous needle biopsy, 19 of which were incorrect or indeterminate when compared with histopathology of the resected specimen. It is not stated whether the specimens were fine needle aspirates or core biopsies. This likely relates to sampling error, and the authors concluded that percutaneous needle biopsy rarely influences management of benign liver tumours.57 In another study retrospectively analysing patients with FNH with atypical imaging, only 14 of 24 patients (58.3%) who underwent ultrasound-guided needle core biopsy had a correct diagnosis of FNH.56 A large collated series of 100 archival cases of needle biopsy specimens of FNH from a number of different hospitals confirmed that a confident diagnosis of FNH was made by the initial pathologist in only 24 of 100 cases. Overall, these data suggest that there is still a need for consensus about diagnostic criteria regarding needle biopsy features of FNH, and that this investigation has a limited role to play. An alternative approach for managing patients with atypical imaging features is to “observe” the lesions over a defined time period (usually 3–4 months) and then to repeat the imaging to exclude a change in the size or characteristics of the lesion.
Considering the indolent natural history of focal nodular hyperplasia, the rare chance of acute complications, and the fact that there is no potential for malignant change, patients with asymptomatic FNH should be treated conservatively. If there is a typical radiological appearance of FNH, then patients should be reassured that no further imaging is required. Importantly, they should also be strongly encouraged to keep the films safe for future reference in the event that this liver lesion is identified again.
Even with multiple imaging modalities (ultrasound, multiphase CT, multiphase MR, and nuclear scintigraphy), some patients have lesions that are likely to be benign but have atypical features of FNH. This typically occurs in patients over 40 years of age and may reflect fluctuations in the hormonal milieu as women reach menopause. It is interesting that FNH is almost never seen in elderly women despite the widespread use of imaging in this age group. Presumably, this is because many focal nodular hyperplastic lesions involute following menopause. In patients with atypical but likely benign imaging features, and when there are no other worrying features to the presentation, then a period of observation is reasonable with repeat imaging approximately 3 to 4 months later. This is long enough to detect any change but not too long to cause undue anxiety for the patient or for the clinician. The choice of the most appropriate imaging modality depends on the location of the lesion as well as the available local resources. A simple transabdominal ultrasound or a contrast-enhanced MR scan will avoid the need for further radiation dosage. If the lesion changes in imaging characteristics or enlarges significantly or becomes symptomatic over that observation period, then surgical intervention should be considered for both therapeutic and diagnostic purposes.58
Although the majority of patients with FNH are completely asymptomatic, there is a subgroup who do develop symptoms that may require resection. This is usually in patients with large and subcapsular lesions. It is appropriate to consider resection in these patients but only after other causes for these patients’ symptoms are ruled out. In our practice, we perform a biliary ultrasound to exclude gallstones, a HIDA scan to exclude chronic acalculous cholecystitis, and an upper endoscopy to rule out oesophageal or gastroduodenal pathology. The indications for resection include patients with persistent symptoms, those with atypical lesions that have increased in size or changed their imaging characteristics, or when symptoms develop after a period of observation. Also, patients should be offered immediate resection when there are such atypical radiological features that it is not possible to confidently rule out a malignant process. In those patients who are offered a period of observation, the oral contraceptive pill should be stopped in the first instance even though there is only limited data supporting an association between the low-dose oral contraceptive pill and FNH. In women who are pregnant or in those who wish to become pregnant, there is no need to resect these lesions, and most patients can be observed carefully with serial ultrasounds throughout the pregnancy.
In conclusion, our understanding of FNH has evolved over the past 30 years. The rapid advances in imaging have allowed more accurate diagnoses without the need for biopsy or resection. The natural history of FNH is still not fully understood, although multiple different pathological subtypes have now been described, and this may explain the heterogeneous clinical behaviour. The fact that the diagnosis of focal nodular hyperplasia is almost never made in postmenopausal women strongly supports the role of the hormonal milieu in influencing the development and growth of these benign lesions. Longitudinal studies are required to better understand the natural history of this condition.