Hepatic focal nodular hyperplasia (FNH) is a nonmalignant condition rarely affecting children previously treated for cancer, especially those who received hematopoietic SCT (HSCT). Some aspects of its pathogenesis still remain unclear and a strong association with specific risk factors has not yet been identified. We report here a single institution's case series of 17 patients who underwent HSCT and were diagnosed with FNH, analyzing retrospectively their clinical features and the radiological appearance of their hepatic lesions. We aimed to compare the diagnostic accuracy of ultrasound (US) and magnetic resonance imaging (MRI) and to explore the role of transient elastography (FibroScan) to evaluate the degree of hepatic fibrosis in FNH patients. Our analysis showed an association of FNH with age at transplant 12 years (hazard ratio (HR) 9.10); chronic GVHD (HR 2.99); hormone-replacement therapy (HR 4.02) and abdominal radiotherapy (HR 4.37). MRI proved to be a more accurate diagnostic tool compared with US. Nine out of 12 patients who underwent FibroScan showed hepatic fibrosis. Our study points out that FNH is an emerging complication of HSCT, which requires a lifelong surveillance to follow its course in cancer patients.
Although benign hepatic lesions are rare entities in the general pediatric population, some of them such as hepatic focal nodular hyperplasia (FNH), nodular regenerative hyperplasia, hepatocellular adenoma and hemangioma are frequently detected in children who underwent hematopoietic SCT (HSCT). FNH is a nonmalignant tumor most commonly detected incidentally during ultrasound (US) examination.1, 2, 3, 4, 5, 6, 7, 8 Risk factors are largely unknown and pathogenesis is controversial and not completely understood. Neither complications nor malignant degenerations are described up to now, and the differential diagnosis includes hepatocellular adenoma and fibrolamellar hepatocellular carcinoma.9, 10
In this retrospective study, we analyzed the case series of FNH that occurred in children who underwent HSCT at our institution, with the aim to evaluate incidence, clinical characteristics and instrumental features and to identify the risk factors associated with its occurrence.
Materials and methods
This single-center, retrospective study was realized at the HSCT Pediatric Unit of Padova University Hospital. A total of 587 patients underwent HSCT between June 1983 and December 2012, for either malignant or nonmalignant diseases. Among them, 324 were alive and eligible for the study because of age at HSCT less than 18 years, availability of one abdominal imaging before and after HSCT and no hepatic nodular lesions on imaging evaluation performed before HSCT. To define a lesion consistent with the diagnosis of FNH, the following criteria should be satisfied: an homogeneous, well-circumscribed nodule with variable echogenicity and possibly a fibrotic central scar on US scan; an iso-hypointense signal on T1 and an iso-hyperintense signal on T2-weighted sequences of MRI; a rapid enhancement on the arterial phase and a washout on the late phase of contrast-enhanced MRI. For the purpose of the study, MRI scan was proposed to patients who never underwent it after HSCT. All patients were imaged with a superconducting imager (MAGNETOM Avanto; Siemens, Erlangen, Germany) operating at 1.5 T using a body-array coil. In detail, MRI was performed using T2-weighted turbo spinecho sequences with or without fat saturation, T2-weighted half-Fourier rapid acquisition with relaxation enhancement (RARE; HASTE; Siemens Medical System, Erlangen, Germany), T1-weighted gradient-echo (GRE) in-phase and out-of-phase sequences, diffusion-weighted image series and apparent diffusion coefficient (ADC) maps and T1-weighted volumetric interpolated breath-hold examination (VIBE). Images were acquired before the administration of contrast agent and during the dynamic phase of contrast enhancement (VIBE images only).11
A gradient-echo T1-weighted magnetic resonance sequence was performed during the hepatic arterial and portal venous phases (at 20 and 50 s) after manual administration of 0.1 mmol per kg gadoterate meglumine (Dotarem; Laboratoire Guerbet, Roissy, France), followed by a 20-ml saline flush.11, 12 Diffusion-weighted sequences were useful to roughly discriminate lesions with features suggestive of malignancy, but a more defined radiological diagnosis in cases of benign lesions could be achieved only on the basis of their behavior on sequences exploring the vascular phases. Differential diagnosis between FNH and hepatic angioma or hepatocellular adenoma was based on radiological criteria. A lesion was consistent with the diagnosis of angioma if hypointense on T1 and intensely hyperintense on T2-weighted images, with a peripheral nodular enhancement after contrast injection and a centripetal progression on delayed images.13 As for adenoma, the following criterion was used: nodules hyperintense on T1-weighted sequences, which becomes hypointense after fat suppression, with a rapid contrast enhancement during the arterial phase and a very slow washout in the late phase (with focal areas heterogeneously hypointense).13, 14
In addition, liver transient elastography (FibroScan, Echosens, Paris, France) was performed in all patients with FNH to evaluate noninvasive liver stiffness (LS) and assess the severity of associated liver disease, by obtaining 10 valid measurements, within an interquartile range of 3 and a success rate of 60%.15, 16 When a definite diagnosis could not be achieved, a biopsy was mandatory. Clinical features, laboratory findings and imaging characteristics of the patients with FNH in accordance with the criteria mentioned above were collected and analyzed. In particular, an accurate description of nodule characteristics was performed comparing US and MRI.
Data regarding all the eligible population were collected from the local database. The following risk factors for FNH were analyzed: gender; age at HSCT; type of underlying disease (neoplastic vs non-neoplastic); disease stage at HSCT (standard risk vs high risk); type of transplant (Auto vs Allo); conditioning regimen (myeloablative vs nonmyeloablative); use of TBI, melphalan and CY; abdominal radiotherapy (RT) before and/or after HSCT; occurrence of acute or chronic GVHD (cGVDH) and hepatic veno-occlusive disease; and use of hormone replacement therapy (HrT). A logistic regression model was used for univariate and multivariate analysis.17 Variables with a P-value <0.05 were considered statistically significant. In multivariate Cox regression analysis, factors with a P-value <0.2 were included. The risk of developing FNH was expressed as the hazard ratio (HR) with 95% confidence interval. Statistical analysis was performed using SAS software (SAS-PC, version 9.3, SAS Institute, Inc., Cary, NC, USA).
Seventeen out of 324 eligible patients (5.2%) were diagnosed with FNH. Characteristics of FNH patients are summarized in Tables 1 and 2. All patients were children at the time of transplantation, with a median age at HSCT of 8.9 years (range 2.9–16.3), including 16 (94.2%) younger than 12 years. The median age at diagnosis of nodules was 14.5 years (range 8.9–21.7); the median interval between HSCT and FNH diagnosis was 5.7 years (range 3.1–11.4). Eleven patients (64.7%) were females, with a female-to-male ratio of 1.8:1. Ten (58.8%) received an Allo-graft and all were given a myeloablative conditioning regimen. Four out of 17 patients (23.5%) received abdominal RT before and/or after HSCT. Nine (52.9%) developed acute GVHD; seven (41.2%) developed cGVHD (never involving the liver). There were no cases of active cGVHD at the time of FNH diagnosis. All the 11 females had a history of gonadal insufficiency and received estrogens (before diagnosis of FNH n=9, after diagnosis of FNH n=2); 2 out of 6 males received testosterone. The median interval between the begining of HrT and diagnosis of FNH was 2.6 years (range 0.7–6.4). HrT was not modified in males and in 6 out of 11 females, whereas it was stopped in 5 females for medical reasons other than the diagnosis of FNH. The size of nodules remained unchanged in patients who stopped HrT, as well as in the six patients who started and continued HrT after the diagnosis of FNH. In 12 patients, FNH was an incidental finding during follow-up post HSCT. None of the patients had palpable abdominal masses. Only three patients presented symptoms: right hypochondrium pain (n=1) and asthenia and abdominal pain (n=2). Eight out of 17 (47.1%) patients presented with altered liver function tests (glutammate pyruvate transaminase (GPT), glutammate oxalacetate transaminase (GOT), gamma glutamyl transferase (GCT), alkaline phosphatase (ALP) outside the normal limits according to the local laboratory reference values). One patient was affected by hepatitis A (patient #14); the others had negative virus serology results (CMV, adenovirus, EBV, HAV, HBV, HCV, HDV, HEV). No patient had abnormal plasma neoplastic markers (CEA, AFP, CA 19-9). None of the 17 patients presented signs of underlying disease progression at the time of FNH diagnosis.
Nodules were first found on US examination in 15 cases, on MRI in 1 patient and on computed tomography scan in 1 case. Abdominal US, performed in all patients, showed a single lesion in 6 patients and multiple nodules in both hepatic lobes in 11 cases. The median number of nodules was 2 (range 1–7), with a median size of 20 mm (range 8–60 mm). FNH nodules were iso-hypoechoic in 14 cases and hyperechoic in 3. Central stellate scar was detected in five cases (29.4%). Color Doppler US demonstrated intratumoral or peripheral hypervascular areas. Contrast-enhanced US (CEUS) was performed in two cases showing a contrast enhancement during the arterial phase and an enhancement equal to the liver parenchyma during the portal venous and late phases. Abdominal MRI, performed in 14 of 17 cases, demonstrated a single lesion in one patient and multiple nodules in 13 cases. The median number of nodules was 4 (range 1–9), with a median size of 14 mm (range 5–60 mm) and localization in both hepatic lobes. MRI showed typical features in 10 patients (Figure 1, patient #9). On the contrary, in four patients, nodules were iso-hyperintense on T1-weighted images and hypointense on T2-weighted images. The central area, corresponding to the fibrotic scar, was noted in six cases (42.9%). Concomitant hepatic hemangiomas were detected on MRI in three patients and hepatocellular adenoma in one case.
Fourteen patients underwent both imaging investigations: MRI detected a higher number of nodules than US (median=4 vs median=2), with a smaller diameter (median=14 mm vs median=20 mm). Two patients were further examined on computed tomography scan; during the arterial phase, the lesions showed high enhancement and acquired iso- and low attenuation on delayed scans (typical features). Positron emission tomography was performed in two cases and scintigraphy in one patient: these imaging techniques were not helpful for diagnosis because of absence of uptake. Liver biopsies were requested in six patients to rule out malignant masses: in four cases for the presence of atypical imaging findings, whereas in the remaining two cases for the appearance of new nodules or the growth in size of previous lesions. Histology of bioptic samples revealed well-preserved hepatocyte trabeculae often with a fibrous area containing abnormal vessels and proliferation of bile ductules. No post-procedural complications were observed. Only the patient affected by HAV and secondary hemochromatosis (patient #14), with MRI findings suggestive of concomitant FNH and hepatocellular adenoma, underwent a complete surgical excision (left hepatectomy). FibroScan was performed in 12 patients, showing no fibrosis (LS<5.3 kPa) in 3 cases, mild fibrosis (LS=5.3–7.0 kPa) in 3 cases, moderate fibrosis (LS=7.0–9.5 kPa) in 5 patients and cirrhosis (LS>12.5 kPa) in 1 patient.
Up to January 2014, all patients with FNH were alive. Size, number and radiologic features of FNH did not vary in 3 of 17 cases; in 9 cases, one or more additional lesions with the same or increased diameter were observed at the follow-up. For the remaining five patients, the diagnosis of FNH is recent and, at the present time, it is too early to assess its course.
By univariate model, the use of HrT and the age at HSCT 12 years were significantly associated with a higher probability of FNH (P=0.02 and P=0.049, respectively). Other factors adjusted for multivariate analysis, although not statistically significant, were: gender (P=0.054), cGVHD (P=0.064), hemochromatosis (P=0.09) and abdominal RT (P=0.11). In multivariate Cox regression analysis, factors significantly associated with a higher probability to develop FNH were: age at transplant 12 years (HR 9.1), cGVHD (HR 2.99), HrT (HR 4.02) and abdominal RT (HR 4.37; Table 3).
FNH is a rare lesion, but it represents the second commonest nonmalignant hepatic tumor, outnumbered only by hemangioma.8, 18, 19 It is detected in both sexes and at all ages, but it is predominant in females aged between 20 and 50.20 It is unusual in children, representing 2% of all pediatric hepatic tumors, with an incidence of ~0.02% in the general pediatric population, which seems to be higher in children previously treated for cancer.2, 5, 21, 22 These data are probably underestimates as nodules are incidentally detected during imaging examinations performed for other reasons. There are few reports about FNH in pediatric oncology: the largest series, all dealing with HSCT patients, were described by Bouyn,2 Masetti3, 4, 7 and Sudour.5 Smith et al.6 reported 46 cases of FNH (incidence 5.1%), focusing on the specific subset of solid tumor pediatric patients (regardless of HSCT status), without a description of the previous treatments’ details nor a risk factor analysis.
To our knowledge, this is one of the largest cohorts describing FNH in patients who underwent HSCT performed at a pediatric age either for malignant or nonmalignant disease. Our data showed an incidence of FNH, among all transplanted patients, of 5.2% (260 times higher than incidence reported in general pediatric population and 11 times higher than the one reported in children with cancer) and a median age at diagnosis of FNH of 14.5 years (range: 8.9–21.7), intermediate between the ones reported by other authors.2, 4, 5
In our series, there were three cases of hemangioma (17%), all females who were receiving oral contraceptive therapy, confirming what is known from literature.4, 23 FNH is associated with hepatic hemangioma in 20% of cases23 and the only variables associated with its development are female sex and the use of oral contraceptives.4 In our study, one female was diagnosed with hepatic adenoma, which is known to be associated with FNH.9 As malignant transformation from adenoma to hepatocellular carcinoma is rare but well documented,24 she underwent surgical resection of the mass.
FNH is most often solitary (up to 95%) and usually less than 5 cm in diameter.2, 3, 5, 25 In our transplanted patients’ group, lesions involved the liver in multiple sites showing smaller average diameters. A standardized nomenclature was first proposed in 1994, placing FNH in the group of regenerative nodules, as opposed to dysplastic or neoplastic nodules.26 The hypothesis of the benign nature of FNH nodules is supported by the polyclonal origin of the hepatocytes.18, 19 Nowadays, the most widely accepted theory is that FNH is a hyperplastic response to hyperperfusion by the characteristic anomalous arteries found in the center of nodules.9 The association of FNH with some congenital vascular diseases such as hereditary hemorrhagic telangiectasia27 and hepatic hemangiomas26 strengthens this hypothesis. FNH could also be an acquired condition, for instance, after a localized vascular injury primed by chemotherapy and RT.3, 4 This would explain the higher incidence of FNH among children treated for cancer (with or without HSCT).
Our data showed a consistently higher incidence in females and in patients with an underlying malignant disease. None of the patients who developed FNH experienced hepatic veno-occlusive disease in contrast to what has been reported by Bouyn2 (10 out of 14 cases of hepatic veno-occlusive disease) and Sudour5 (3 out of 17 cases). Comparable to the data reported by Masetti4 and Sudour,5 we found a significant association between FNH and abdominal RT and younger age (12 years). We also found a significant association with occurrence of cGVHD, abdominal RT and HrT. As for the association between HrT and FNH, there is no consensus in literature, and the most recent studies do not support this hypothesis.28, 29, 30, 31 According to our data, HrT was a significant risk factor for the development of FNH, even if the therapy withdrawal did not modify the nodules’ size and characteristics. Unlike hepatic adenomas, FNH rarely presents with symptoms due to complications.2, 3, 4, 5, 6, 7, 8, 9, 32 Our study confirmed this trend. Liver tests as well as viral and neoplastic markers are usually normal in FNH patients, although slightly elevated GOT, GPT, ALP and GGT have been reported,2, 3, 4, 5, 6, 7, 8, 9, 32 as it happened in 47.1% of our patients.
Imaging has 82.6% sensitivity and 97.4% specificity for the FNH diagnosis.33 There are no universally accepted radiological criteria to distinguish FNH from other liver lesions, although characteristic imaging features have been described.34, 35 US is generally the first technique that allows to detect nodules.34, 36 Color Doppler US and CEUS can be used to increase US diagnostic accuracy.37 MRI has been shown to be the most accurate imaging modality (sensitivity 81.3%).33 Both US and MRI avoid exposure to radiation, and this is a significant advantage in such a cohort of patients at risk of secondary malignancies. If diagnosis remains unclear, a liver biopsy may be helpful. Surgery should be reserved for rare, symptomatic FNH nodules and for highly suspicious lesions, to rule out disease recurrence or a second tumor.8 Our experience confirms the leading role of US and MRI as diagnostic tools for FNH and shows a superior accuracy of MRI (a higher number of nodules with smaller size diagnosed with MRI). The use of hepato-specific contrast agents (that is Eovist, Primovist, Bayer Shering Pharma Ag, Leverkusen, Germany) has been proposed by some authors to further increase MRI diagnostic specificity, but their availability is limited because of the lack of experience in pediatric population and their cost. In addition, we evaluated the degree of hepatic fibrosis in FNH patients with a quite recently introduced, noninvasive imaging technique, liver FibroScan.15, 16, 38 We also tried to identify any relationship between fibrosis and nodules’ size and number. We found out that the majority of FNH patients who underwent FibroScan (75%) had a moderate degree of LS not correlated with nodules’ characteristics. This data should be interpreted with caution, as the presence of moderate–severe steatosis, often present in patients who underwent HSCT and suffer of metabolic syndrome, may overestimate LS measurements. Moreover, we lack a control group, as the exam was proposed only to patients with an established diagnosis of FNH. So far, FNH seems to have a benign and uncomplicated course: all the existing studies report that nodules generally do not change their size and characteristics with time.2, 3, 4, 5, 6, 7, 8, 9, 32 In our case series, in 53% of patients, one or more additional lesions with the same or increased diameter were discovered on follow-up, without evidence of malignant transformation.
In conclusion, FNH is an emerging complication of cancer treatment, especially after HSCT during childhood. Although it seems to have an indolent course, its pathogenesis is not yet fully understood and its evolution, in this specific subset of patients (at higher risk of second malignancies compared with the general population), should be carefully monitored. It may be postulated that HSCT patients receive an amount of MSC that may be able to migrate into the liver, engraft and enhance the hyperplasic regenerative response to a preexisting injury. This hypothesis is under investigation in animal models. By now, the donor origin of neoplastic cells has been demonstrated only in one patient with hepatic tumors after HSCT.39, 40 For all these reasons, it is important to be aware of its existence and features to avoid unnecessary diagnostic interventions and excessive concerns in patients, parents and treating physicians. As there is no consensus about optimal follow-up protocols for pediatric cancer survivors, it is important to share single institutions’ experience on this subject and to include abdominal US in lifelong surveillance of these patients.
About this article
Focal Nodular Hyperplasia After Treatment With Oxaliplatin: A Multiinstitutional Series of Cases Diagnosed at MRI
American Journal of Roentgenology (2018)