Potential to Avoid Invasive Procedures for Atypical Focal Nodular Hyperplasia (FNH) and FNH-like Lesions Based on MR Findings

Abstract

Purpose

To identify the imaging characteristics of rare, histopathologically confirmed focal nodular hyperplasia (FNH) and FNH-like lesions (FNH/FNH-like lesions) and determine whether invasive procedures could have been avoided by comprehensive MRI review.

Methods

We retrospectively enrolled patients with pathologically confirmed FNH/FNH-like lesions between January 2011 and December 2023. These cases underwent biopsy or resection due to their atypical hepatobiliary phase appearance. Six patients with hepatocellular carcinoma (HCC) were included from a separate cohort for comparison. Gadoxetic acid-enhanced MRI findings were qualitatively evaluated. Apparent diffusion coefficient (ADC) values and enhancement during the hepatobiliary phase were quantitatively assessed, and the relative enhancement ratio (RER) to liver parenchyma and intrahepatic vessel (RERv) was calculated. Organic anion transporting polypeptides (OATP)1B3 expression was confirmed histopathologically.

Results

Five patients with FNH/FNH-like lesions and 6 patients with HCC were evaluated. The non-contrast MRI findings were non-specific for both groups. All 5 FNH/FNH-like lesions displayed non-rim arterial phase hyperenhancement (APHE) without washout during dynamic contrast study with ADC values of 1.162 ± 0.124 × 10⁻³mm²/sec (mean ± standard deviation). Whereas all 6 HCCs displayed non-rim APHE and washout with ADC values of 0.923 ± 0.138 × 10⁻³mm²/sec. During the hepatobiliary phase, all 5 FNH/FNH-like lesions were hypointense to the liver but hyperintense to intrahepatic vessels, with RER and RERv ≥ 0.90, indicating preserved gadoxetic acid uptake. OATP1B3 membrane expression was confirmed. By contrast, 5 HCCs were hypointense to both the liver and vessels, with RER and RERv < 0.90 and no OATP1B3 expression.

Conclusion

FNH/FNH-like lesions with hepatobiliary phase hypointensity, the characteristic finding of hyperintensity relative to intrahepatic vessels distinguished them from similarly presenting HCCs. Careful evaluation of this specific imaging feature might have potentially avoided the need for invasive treatment or biopsy in these diagnostically challenging cases.

Introduction

Focal nodular hyperplasia (FNH) is the second most common type of benign liver lesion. When FNH presents without all the typical pathological features or occurs in the context of abnormal liver conditions, such nodules are referred to as “FNH-like lesions”.¹ Both FNH and FNH-like lesions (FNH/FNH-like lesions) exhibit similar histological characteristics and imaging findings. Clinically, neither lesion requires a specific treatment, making it essential to accurately diagnose FNH/FNH-like lesions through imaging.

The approval of gadoxetic acid, a liver-specific MR contrast agent, has dramatically improved the accuracy of the diagnosis of focal liver lesions, including FNH/FNH-like lesions.² Gadoxetic acid serves as a dual-function contrast agent that combines the properties of an extracellular contrast agent for dynamic imaging with those of a hepatocyte-specific agent. Approximately 50% of the injected dose is taken up by functioning hepatocytes and excreted in bile, thereby enabling tumor characterization based on the presence or absence of functioning hepatocytes.³

FNH/FNH-like lesions comprise hepatocytes that maintain hepatocyte function based on the pathogenesis of these lesions, which is a local reactive hyperplastic response to abnormal hepatic blood flow.¹ Reflecting their histopathological composition, the majority (94%–97%) of FNH/FNH-like lesions show uptake of gadoxetic acid, displaying hyper- or isointensity during the hepatobiliary phase.² The hyper- or isointensity during the hepatobiliary phase is often a key imaging finding that aids in diagnosis and can reduce unnecessary biopsies for the diagnosis of FNH/FNH-like lesions.² However, a minority of FNH/FNH-like lesions exhibit hypointensity during the hepatobiliary phase,^2,4,5 which can complicate accurate diagnosis and may require histological examination. We believe that examining the imaging characteristics of FNH/FNH-like lesions that have undergone pathological examination, although there is very little research in this direction, is clinically very important.

This study aimed to identify imaging characteristics of rare atypical FNH/FNH-like lesions with hepatobiliary phase hypointensity and determine whether invasive procedures could have been avoided by comprehensive MRI review.

Materials and Methods

This retrospective study was approved by the Research Ethics Committee of Hamamatsu University School of Medicine (Approval no.: 24-157). The requirement for written informed consent was waived due to the retrospective nature of the study.

Patients

We retrospectively enrolled patients with pathologically confirmed FNH/FNH-like lesions between January 2011 and December 2023 based on pathological records. These cases represented rare variants that underwent biopsy or resection due to their atypical hepatobiliary phase appearance. The inclusion criteria were as follows: (1) patients aged ≥ 20 years; (2) those with focal liver lesions histopathologically diagnosed as FNH/FNH-like lesions; and (3) those who underwent gadoxetic acid-enhanced MRI. For comparison, patients with hepatocellular carcinoma (HCC) were included in the study between January 2018 and December 2023 based on the following criteria: (1) aged ≥ 20 years; (2) histopathologically confirmed HCC with a maximum diameter of ≤ 20 mm; and (3) low HCC signal intensity (SI) in the hepatobiliary phase of gadoxetic acid-enhanced MRI. These HCCs were purposefully collected for comparison with the FNH/FNH-like lesions described above, characterized by similar lesion sizes, typical arterial phase hyperenhancement (APHE), and clear hypointensity in the hepatobiliary phase. A flowchart summarizing the study population is shown in Fig. 1.

Fig. 1.

Patient flowchart of the study. (a) Inclusion criteria of the patients with FNH/FNH-like lesions. (b) Inclusion criteria of the patients with HCC.FNH, focal nodular hyperplasia; FNH/FNH-like leisons, FNH and FNH-like leisons; HCC, hepatocellular carcinoma.

Gadoxetic acid-enhanced MRI protocol

Gadoxetic acid-enhanced MRI was performed using a 3T system (Discovery 750 DV 25.1; GE Healthcare, Waukesha, WI, USA) with an 8-channel body phased-array coil. All patients included in the study underwent scans using a standard liver MRI protocol that included the following sequences: chemical shift imaging and 3D T1-weighted fat-suppressed spoiled gradient-recalled echo sequences (liver acquisition with volume acceleration [LAVA; GE Medical Systems]) as pre-contrast sequences. After administering a dose of 0.1 mL/kg of gadoxetic acid (Primovist, 0.25 mmol/mL; Bayer Schering Pharma, Berlin, Germany) at a rate of 1 mL/s, followed by a 20 mL saline flush using a power injector and a bolus tracking technique, late arterial, portal venous, and transitional-phase images were acquired using LAVA. After obtaining fat-suppressed fast spin-echo T2- and diffusion-weighted (b = 0 s/mm², b = 800 s/mm²) images, hepatobiliary phase images were acquired at least 20 mins after contrast administration. Details of the MRI protocols are presented in Table 1.

Table 1.

Details of the MRI acquisition parameters

Parameter	Chemical shift image	T2-weighted image	Diffusion-weighted image	Dynamic study*	Hepatobiliary phase image
Sequence	2D gradient echo	Fast spin echo	Echo planar imaging	Liver acquisition with volume acceleration	Liver acquisition with volume acceleration
Fat suppression	—	Yes	Yes	Yes	Yes
Acquisition plane	Transverse	Transverse	Transverse	Transverse	Transverse
Respiratory compensation	Breath-hold	Respiration trigger	Respiration trigger	Breath-hold	Breath-hold
Repetition time (ms)	170–270	2000–15000	8000–12000	6	6
Echo time (ms)	2.2/1.1	82	90	1.6	1.6
Flip angle (degrees)	50–80	90	90	15	30
Section thickness (mm)	4	5	5	4	4
Intersection gap (mm)	1	1	1	1	1
FOV (mm2)	40 × 40	40 × 40	40 × 40	42 × 42	42 × 42
Matrix	320 × 224	320 × 320	128 × 128	320 × 192	320 × 224
Acquisition time	20	230	210	20 × 4	20

Note: *Dynamic study consists of pre-contrast, late arterial, portal venous, and transitional phases.

Qualitative MRI analyses

Images were analyzed by 2 board-certified radiologists, with 2 and 25 years of experience in liver imaging, respectively. In case of disagreement between the 2 evaluators, a 3rd radiologist (with 23 years of experience) reanalyzed the data. The qualitative assessment of gadoxetic acid-enhanced MRI included the following criteria: (a) SI of the lesions on fat-suppressed T1-, T2-, and diffusion-weighted images was classified as hypo-, iso-, or hyperintensity compared to the background liver parenchyma; (b) central scar on any images; (c) hemorrhage, defined as hyperintensity on fat-suppressed T1-weighted imaging without enhancement during dynamic contrast study; (d) fat deposition on chemical shift imaging; (e) an enhancement pattern during the arterial phase, categorized as hypo-, iso-, and hyperenhancement; arterial phase enhancement was further classified as either with or without a rim pattern; (f) washout during the portal venous phase; (g) SI on hepatobiliary phase image (compared to the background liver parenchyma); (h) SI on hepatobiliary phase image (compared to the first branch of the portal vein), and the enhancement was classified as homogeneous or heterogeneous; and (i) liver imaging reporting and data system v2018 category.

Quantitative MRI analyses

The maximum lesion diameter and SI were assessed by 2 radiologists independently. The maximum lesion diameter was measured during the hepatobiliary phase. The SI of the lesion was measured by placing ROIs in the maximum oval area at the level of the largest diameter on the diffusion-weighted, pre-contrast, and hepatobiliary phase images. On the pre-contrast and hepatobiliary phase images, a similarly sized ROI was placed on the adjacent liver parenchyma while avoiding large blood vessels, and an additional ROI was placed on the vessels (Fig. 2). Three ROIs were placed on the areas of the nodule and background liver parenchyma, ensuring that the positions were consistent across all patients. The mean of the 3 ROI measurements was recorded as the SI of each lesion. All the ROIs were checked by a 3rd radiologist with 23 years of experience to ensure accurate placement. The average SI values measured by the 2 evaluators were used in subsequent calculations:

$$ADC(apparentdiffusioncoefficient)=\frac{log\left(\frac{S{I}_1}{S{I}_2}\right)}{b_2-b_1}$$

Fig. 2.

Examples of ROI placements. ROIs were placed in the maximum oval area at the level of the largest tumor diameter on pre-contrast T1-weighted images (a: arrowhead). A similarly sized ROI was placed on the adjacent liver parenchyma to avoid the large blood vessels (a: black arrow). An ROI was also placed on the intrahepatic vessel (a: arrow). The same ROIs (3 green circles) were copied and pasted onto the corresponding hepatobiliary phase images for each patient (b).

where SI₁ and SI₂ are the SI of the lesion in the image with b-values of 0 and 800 s/mm², respectively. b₁ is the b-value of 0 s/mm², and b₂ is the b-value of 800 s/mm².

$$RER=\frac{RI{R}_{HBP}\left(\frac{S{I}_{nod}}{S{I}_{liver}}\right)}{RI{R}_{pre}\left(\frac{S{I}_{nod}}{S{I}_{liver}}\right)}$$

The relative enhancement ratio (RER)⁶ was calculated as RIR_HBP/RIR_pre, where SI_nod and SI_liver are the SI of the lesion and liver parenchyma, respectively, and RIR_HBP is the relative intensity ratio (RIR) on the hepatobiliary phase images, and RIR_pre is the RIR before the contrast enhancement. In this study, to enable comparison with intrahepatic vessels, RERv is defined as RER applied to intrahepatic vessel modified from the original RER. RERv was also calculated by substituting the SI of the lesion (SI_nod) with that of the intrahepatic vessels (SI_vessel):

$$RERv=\frac{RI{R}_{HBP}\left(\frac{S{I}_{nod}}{S{I}_{vessel}}\right)}{RI{R}_{pre}\left(\frac{S{I}_{nod}}{S{I}_{vessel}}\right)}$$

Histological and immunohistochemical analysis

For comparisons with the imaging results, a board-certified pathologist confirmed that the recorded pathological diagnosis was correct from hematoxylin and eosin staining and verified the presence or absence of membrane expression with organic anion transporting polypeptides (OATP)1B3 immunostaining for all lesions.

Statistical analysis

The inter- and intraobserver agreements were assessed using the intra-class correlation coefficient (ICC). According to Koo et al., the ICCs were interpreted as follows: < 0.50, poor agreement; 0.50–0.75, moderate agreement; 0.75–0.90, good agreement; and > 0.90, excellent agreement.⁷ Continuous variables are expressed as mean ± standard deviation. The values of ADC, RER, and RERv were compared between the 2 groups using the Mann–Whitney U test. Based on a previous study, the optimum RER cutoff point for the determination of Wnt/β-catenin activation was set as 0.90.⁴ For all statistical analyses, P-values were two-sided, with statistical significance set at P < 0.05. All statistical analyses were performed using the SPSS Statistics (version 27.0; IBM, Armonk, NY, USA).

Results

Patients and clinical data

Five patients with FNH/FNH-like lesions (3 males and 2 females, mean age 61.4 years) and 6 patients with HCC (5 males and 1 female, mean age 72.6 years) were included. Clinical data are summarized in Table 2.

Table 2.

Clinical data of patients and qualitative and quantitative assessment of FNH/FNH-like lesion and HCC

Characteristics	FNH/FNH-like lesion					Hepatocellular carcinoma
	Case 1	Case 2	Case 3	Case 4	Case 5	Case 1	Case 2	Case 3	Case 4	Case 5	Case 6
Clinical characteristics
Age, years	65	78	73	68	23	73	75	74	62	74	78
Sex; male/female	Female	Female	Male	Male	Male	Male	Male	Male	Male	Female	Male
Background liver	Normal	Normal	SVR	Normal	Alcoholic	Unknown	HCV	Unknown	Normal	HCV	Alcoholic
ALBI grade	—	—	1	—	1	1	1	1	—	2	2
Clinical symptom	NS	NS	NS	NS	NS	NS	NS	NS	NS	NS	NS
Tumor makers
AFP (< 13.4 ng/mL)	8	10	14.7	12	11	31.2	19.5	15.8	22.2	5	9
PIVKA-II (< 40 mAU/mL)	5	5	17	9	8	10	7	45	23	18	32
CEA (< 5.0 ng/mL)	3.0	1.7	3.3	2.1	1.9	4.1	—	—	3.6	—	—
CA19-9 (< 37 U/mL)	10.2	13.2	25.9	8	11	—	—	—	—	—	—
Resection or biopsy	Resection	Resection	Biopsy	Biopsy	Biopsy	Resection	Resection	Resection	Resection	Resection	Resection
Imaging characteristics of the lesion
Size (mm)	18	16	13	14	18	19	18	20	9	11	10
SI on T1-weighted image	Hypo	Hypo	Hypo	Hypo	Hyper	Hyper	Hypo	Hypo	Hypo	Hyper	Hypo
SI on T2-weighted image	Hyper	Hyper	Hyper	Hyper	Hyper	Hyper	Hyper	Hyper	Hyper	Hyper	Hyper
SI on diffusion weighted image	Hyper	Hyper	Hyper	Iso	Hyper	Hyper	Hyper	Hyper	Hyper	Hyper	Hyper
ADC values (×10−3mm2/sec)	1.02	1.20	1.15	1.35	1.09	0.99	1.09	0.95	0.85	0.97	0.69
Central scar at any images	Absent	Absent	Absent	Absent	Absent	Absent	Absent	Absent	Absent	Absent	Absent
Hemorrhage at T1-weighted image	Absent	Absent	Absent	Absent	Absent	Absent	Absent	Absent	Absent	Absent	Absent
Fat deposition on chemical shift image	Absent	Absent	Absent	Absent	Absent	Absent	Absent	Absent	Absent	Absent	Absent
Enhancement on arterial phase; Hypo-, iso, or hyper-	Hyper	Hyper	Hyper	Hyper	Hyper	Hype	Hype	Hype	Hype	Hype	Hype
Rim or non-rim on arterial phase	Non-rim	Non-rim	Non-rim	Non-rim	Non-rim	Non-rim	Non-rim	Non-rim	Non-rim	Non-rim	Non-rim
Washout on portal phase	Absent	Absent	Absent	Absent	Absent	Present	Present	Present	Present	Present	Present
SI on HBP image (compared to the background liver parenchyma)	Hypo	Hypo	Hypo	Hypo	Hypo	Hypo	Hypo	Hypo	Hypo	Hypo	Hypo
SI on HBP image (compared to the intrahepatic vessels)	Hyper	Hyper	Hyper	Hyper	Hyper	Hypo	Hypo	Hypo	Hypo	Hypo	Hypo
Homogeneous (Homo) or heterogeneous (Hetero) on HBP image	Hetero	Hetero	Hetero	Homo	Homo	Homo	Homo	Homo	Homo	Homo	Homo
RER at HBP image	0.99	1.07	1.21	1.03	0.95	0.71	0.83	0.59	0.65	0.60	0.69
RERv at HBP image	0.93	0.91	1.41	1.13	0.99	0.71	0.83	0.59	0.65	0.60	0.69
LI-RADS category	3	3	3	3	3	5	5	5	4	5	5
Histological evidence related with HBP imaging findings
Membrane expression of OATP1B3	Present	Present	Present	Present	Present	Absent	Absent	Absent	Absent	Absent	Absent

Note: The numbers in parentheses in tumor makers represent the upper limits of normal values.

ADC, apparent diffusion coefficient; AFP, alpha fetoprotein; ALBI grade, albumin-bilirubin grade; Alcoholic, alcoholic liver disease without cirrhosis; CA19-9, carbohydrate antigen 19-9; CEA, carcinoembryonic antigen; FNH, focal nodular hyperplasia; HBP, hepatobiliary phase; HCC, hepatocellular carcinoma; HCV, chronic liver disease related with hepatitis C virus; Hyper, hyperintensity; Hypo, hypointensity; Iso, isointensity; LI-RADS, liver imaging reporting and data system; Normal, patients without chronic liver disease; NS, No significant findings were observed; OATP1B3, organic anion transporting polypeptides (OATP)1B3; PIVKA-II, protein induced by vitamin K absence or antagonist-II; RER, relative enhancement ratio; RERv, relative enhancement ratio to intrahepatic vessels; SI, signal intensity; SVR, sustained virological response (hepatitis C virus is not detected in the blood 24 weeks or more after completing treatment); T1- and T2-weighted image, fat-suppressed T1- and T2-weighted image; Unknown, chronic liver disease with unknown etiology.

Inter- and intraobserver agreement of quantitative assessment

There was an excellent interobserver agreement between the 2 radiologists regarding the quantitative assessments of size, ADC values, RER, and RERv during the hepatobiliary phase, with ICC values of 0.94, 0.95, 0.95, and 0.92, respectively. The intraobserver agreement for readers 1 and 2 was excellent for size, ADC values, RER, and RERv during the hepatobiliary phase, with ICC of 0.96 and 0.97, 0.95 and 0.97, 0.95 and 0.96, and 0.95 and 0.95, respectively (Table 3).

Table 3.

Interobserver and intraobserver agreements

Measurements	Interobserver agreement	Intraobserver agreement reader 1	Intraobserver agreement reader 2
Size (mm)	0.94	0.96	0.97
ADC (×10−3mm2/sec)	0.95	0.95	0.97
RER at HBP image	0.95	0.95	0.96
RERv at HBP image	0.92	0.95	0.95

Note: There was excellent interobserver agreement between the quantitative assessments by the 2 radiologists. The intraobserver agreement for readers 1 and 2 was excellent.

ADC, apparent diffusion coefficient; HBP, hepatobiliary phase; RER, relative enhancement ratio; RERv, relative enhancement ratio to intrahepatic vessels.

Qualitative and quantitative MRI analyses

FNH/FNH-like lesions displayed hypointensity (n = 4) and isointensity (n = 1) on fat-suppressed T1-weighted images and hyperintensity (n = 5) on fat-suppressed T2-weighted images relative to the background liver parenchyma. Lesions displayed hyperintensity (n = 4) and isointensity (n = 1) on diffusion-weighted images with ADC values of 1.162 ± 0.124 ×10⁻³mm²/sec (mean ± standard deviation). No cases exhibited central scar, hemorrhage, or fat deposition. All lesions showed non-rim arterial-phase hyperenhancement and no washout in the portal venous phase. Moreover, all lesions appeared hypointense relative to the background liver parenchyma but demonstrated hyperintensity compared to the intrahepatic vessels during the hepatobiliary phase. All RER and RERv values (1.050 ± 0.100 and 1.074 ± 0.207, respectively) were ≥ 0.90, and OATP1B3 membrane expression was confirmed by histological examination.

All 6 cases of HCC exhibited hypointensity (n = 4) and hyperintensity (n = 2) on fat-suppressed T1-weighted images and hyperintensity (n = 6) on fat-suppressed T2-weighted images relative to the background liver parenchyma. Lesions displayed hyperintensity (n = 6) on diffusion-weighted images with ADC values of 0.923 ± 0.137 ×10⁻³mm²/sec. No cases exhibited central scar, hemorrhage, or fat deposition. All lesions showed non-rim arterial-phase hyperenhancement and washout during the portal venous phase. Moreover, all lesions appeared hypointense compared both to the background liver parenchyma and intrahepatic vessels during the hepatobiliary phase. All RER and RERv values (0.678 ± 0.882 and 0.678 ± 0.882, respectively) were < 0.90, and no OATP1B3 membrane expression was confirmed by histological examination. The radiological data are summarized in Table 2, and the key images are shown in Figs. 3, 4, 5, 6.

Fig. 3.

FNH-like lesion in a 65 year-old woman (Case 1). A nodule in the segment 8 was incidentally detected in a 65 year-old woman with pancreatic adenocarcinoma. The nodule shows hyperintensity on fat-suppressed T2- and diffusion-weighted images (a, b: arrowheads) and hypointensity on pre-contrast images (c: arrowhead). In the dynamic contrast study, arterial phase hyperenhancement is observed (d: arrowhead), with no washout during the portal venous phase (e: arrowhead). The nodule shows hypointensity during the hepatobiliary phase compared to the background liver parenchyma (f: arrowhead), whereas it shows hyperintensity compared to the intrahepatic vessel (f: arrow). Neither a central scar nor a hemorrhage was observed. The nodule was pathologically diagnosed as FNH-like lesion by resection. Hematoxylin and eosin staining revealed that the nodule consists of cell proliferation without nuclear atypia, resembling normal hepatocytes, with marked sinusoidal dilatation (g, original magnification, ×100). Immunostaining for OATP1B3 demonstrated a similar membrane expression to that of normal hepatocytes (h, original magnification, ×100). FNH, focal nodular hyperplasia; OATP, organic anion transporting polypeptide.

Fig. 4.

FNH-like lesion in a 78 year-old woman (Case 2). A nodule was incidentally detected in a 78 year-old woman without any chronic liver disease. The nodule in segment 8 shows hyperintensity on fat-suppressed T2-weighted image (a: arrowhead) and diffusion-weighted image (b: arrowhead) and shows hypointensity on pre-contrast images (c: arrowhead). In the dynamic contrast study, arterial phase hyperenhancement is observed (d: arrowhead), whereas washout is not observed during the portal venous phase (e: arrowhead). The nodule shows hypointensity during the hepatobiliary phase compared to the background liver parenchyma (f: arrowhead), whereas it shows hyperintensity compared to the intrahepatic vessel (not shown). Neither a central scar nor a hemorrhage was observed. The nodule was pathologically diagnosed as FNH-like lesion by resection. FNH, focal nodular hyperplasia.

Fig. 5.

FNH in a 68 year-old man without chronic liver disease (Case 4). Focal liver lesions were detected incidentally in a 68 year-old man without chronic liver disease. The nodule in segment 7 displayed hyperintensity on fat-suppressed T2-weighted images (a: arrowhead), iso-hyperintensity on diffusion-weighted images (b: arrowhead), and hypointensity on fat-suppressed T1-weighted images (c: arrowhead). The nodule demonstrated arterial phase hyperenhancement (d: arrowhead) and showed no washout during the portal venous phase (e: arrowhead). The nodule shows hypointensity during the hepatobiliary phase compared to the background liver parenchyma (f: arrowhead) but was hyperintense relative to the intrahepatic vessel (f: arrow). The nodule was pathologically diagnosed as FNH-like lesion on biopsy. FNH, focal nodular hyperplasia.

Fig. 6.

HCC in a 74 year-old man with chronic liver disease of unknown etiology (Case 3). The nodule in segment 8 displayed hyperintensity on fat-suppressed T2-weighted images (a: arrowhead) and diffusion-weighted images (b: arrowhead). The nodule demonstrated hypointensity on pre-contrast images (c: arrowhead) and arterial phase hyperenhancement (d: arrowhead) and showed washout during the portal venous phase (e: arrowhead). The nodule shows hypointensity during the hepatobiliary phase compared to the background liver parenchyma (f: arrowhead) and to the intrahepatic vessel (f: arrow). The nodule was pathologically diagnosed as HCC following resection. Hematoxylin and eosin staining revealed characteristic HCC cell features (g: asterisk, original magnification, × 100) compared to the backgroud liver parenchyma (g: double asterisk). HCC cells demonstrated the membrane expression absence of OATP1B3 (h: asterisk) compared to the background liver parenchyma (h: double asterisks) (original magnification, ×100). HCC, hepatocellular carcinoma; OATP, organic anion transporting polypeptide.

ADC values of FNH/FNH-like lesions were significantly higher than that of HCCs (P < 0.009). RER and RERv values of FNH/FNH-like lesions were significantly higher than that of HCCs (P < 0.004 and P < 0.004, respectively).

Discussion

The present study examined extremely rare and atypical cases of FNH/FNH-like lesions that underwent biopsy or resection specifically due to their atypical hepatobiliary phase appearance on gadoxetic acid-enhanced MRI. These cases represented a challenging diagnostic scenario where conventional imaging criteria for FNH/FNH-like lesions were inconclusive, necessitating histological confirmation. Despite exhibiting hypointensity relative to background liver parenchyma during the hepatobiliary phase—an appearance typically associated with malignant lesions—all histologically confirmed FNH/FNH-like lesions in this study demonstrated hyperintensity compared to the intrahepatic vessels during the hepatobiliary phase and preserved gadoxetic acid uptake, as evidenced by RER values ≥ 0.90. This finding was corroborated by histological confirmation of OATP1B3 membrane expression, indicating functional hepatocyte-like uptake mechanisms despite the atypical imaging appearance.

Only a limited number of FNH/FNH-like lesions cases show hypointensity in the hepatobiliary phase.^4,5 In the report by van Kessel et al.,⁴ hypointensity was noted in 1/26 cases (approximately 4%), and in the large-scale study by Bilreiro et al., ⁵ FNH cases showing hypointensity were rare (0.6%). Therefore, the details of FNH/FNH-like lesions showing hypointensity during the hepatobiliary phase have not yet been described. In the present study, all nodules exhibited hypointensity relative to the surrounding liver parenchyma during the hepatobiliary phase. However, upon closer observation, all FNH/FNH-like lesions showed hyperintensity than intrahepatic vessels during the hepatobiliary phase, with an RER of ≥ 0.90. By comparing hepatobiliary phase findings to vascular structures rather than background liver parenchyma, it is possible to accurately assess gadoxetic acid uptake. This approach allows for correct diagnosis of FNH/FNH-like lesions that may appear hypointense at first glance. The evaluation of RERv values also supported this approach.

The SI of FNH/FNH-like lesions during the hepatobiliary phase significantly correlates with OATP1B3 expression, as previously demonstrated in HCC.⁸ In HCC, OATP1B3 expression, induced by the activation of the Wnt/β-catenin signaling pathway, was detected using a cutoff point of 0.90 for the RER.⁶ Referring to this established cutoff value, all lesions in the present study exhibited RER ≥ 0.90, confirming gadoxetic acid uptake despite their hypointense appearance compared to the background liver parenchyma.

FNH/FNH-like lesions are among the hepatic lesions that present with non-rim APHE. Given that HCC and hepatocellular adenoma (HCA) also exhibit non-rim APHE, differentiation between these entities is clinically important. In the hepatobiliary phase, FNH/FNH-like lesions commonly demonstrate areas with iso- to hyperintensity relative to the background liver parenchyma.^2,4,9 This imaging characteristic overlaps with HCC and HCA, which also typically show hyperintensity in the hepatobiliary phase.^10–14 Consequently, the majority of previous studies have focused on establishing imaging criteria to differentiate FNH/FNH-like lesions from these hepatobiliary phase hyperintense HCC and HCA with non-rim APHE. The differences in MRI findings when HCC showing high signal intensity in the hepatobiliary phase exhibits a pattern similar to FNH/FNH-like lesions. While 95.9% of HCCs showed washout in portal venous phases following APHE during the arterial phase, 77.8% of FNH/FNH-like lesions did not exhibit washout. Additionally, 88.2% of HCCs showed enhancing capsule and 72.5% demonstrated mosaic-like structure, whereas these features were observed in only 22.2% and 11.1% of FNH/FNH-like lesions, respectively. By contrast, 55.6% of FNH/FNH-like lesions showed central scar, and no central scar was observed in HCCs.¹⁰ Based on these findings, FNH/FNH-like lesions are characterized by central scar and the absence of washout, moderate diffusion signal, while HCC is characterized by APHE, high frequency of washout, capsule, mosaic pattern, and low ADC.¹⁰ Therefore, regardless of findings in the hepatobiliary phase, it is often possible to accurately diagnose FNH/FNH-like lesions based on other findings. However, conversely, if one is not aware that FNH/FNH-like lesions may not exhibit typical hemodynamics and can appear hypointense on the hepatobiliary phase, making a correct diagnosis can be challenging. Additionally, HCC is a tumor with biological diversity, and thus exhibits a wide variety of imaging findings. It has not been confirmed in this study whether HCC presenting with signal intensity between the parenchyma and vessels, as seen in this case, exists.

In differentiating between HCA and FNH/FNH-like lesion, previous reports have shown that the signal in the hepatobiliary phase provides the highest diagnostic performance.¹¹⁻¹⁴ This is based on the fundamental premise that the majority of FNH/FNH-like lesion exhibits hyper- to iso-intensity. In addition, HCA with β-catenin-mutation, which exhibits hyperintensity in the hepatobiliary phase, is not particularly common.¹⁵ However, in this study, it is necessary to differentiate rare hypointense FNH/FNH-like lesion from HCA. HCAs frequently show intratumoral hemorrhage, fat, and glycogen accumulation, while FNH/FNH-like lesions are significantly more likely to have central scars,^13,14 although characteristic findings vary depending on the HCA subtype. The presence of a central scar is more frequent in FNH than in HCA.¹⁶ Regarding diffusion-weighted imaging (DWI), as mentioned previously, there are reports stating that HCC shows significantly lower ADC compared to FNH/FNH-like lesions,¹⁰ while other reports indicate that ADC values overlap between FNH/FNH-like lesions or HCA and HCC, making differentiation difficult with DWI and ADC values alone.^17,18 In this study, statistical verification was difficult due to the small number of cases examined. Overlap in ADC values was observed, although the mean ADC value was higher in FNH/FNH-like lesions.

Detailed reports on FNH/FNH-like lesions showing hypointensity in the hepatobiliary phase are scarce and, to date, there are no studies addressing the differentiation between FNH/FNH-like lesions showing hypointensity in the hepatobiliary phase and HCC or HCA. However, in this study, FNH/FNH-like lesions showing hypointensity in the hepatobiliary phase demonstrated signal characteristics on T1, T2, and diffusion-weighted images that were consistent with previous reports,^18,19 suggesting that the differentiating points used for typical FNH/FNH-like lesions showing high signal intensity can be directly applied. Even when encountering nodules showing rim APHE and hypointensity in the hepatobiliary phase, we believe that differential diagnosis is possible by comparing the signal intensity in the hepatobiliary phase with intrahepatic vessels and combining this with findings from sequences other than the arterial phase.

FNH/FNH-like lesions display various imaging appearances on hepatobiliary phase images, despite exhibiting uniform enhancement in the arterial phase. Four major patterns have been described for FNH during the hepatobiliary phase: homogeneous hyperintensity, heterogeneous hyperintensity, homogeneous isointensity, and peripheral ring-like hyperintensity.^4,9 Additionally, although very few in number, FNH presenting with hypointensity also exists.^4,5 The variation in findings cannot be solely attributed to the presence of a central scar, even though the central scar appears hypointense due to the absence of hepatocyte function. The different patterns observed in the hepatobiliary phase may be related to the zonation of liver lobules.²⁰ In normal liver parenchyma, hepatocyte function differs between the periportal (zone 1) and centrilobular areas (zone 3) in the lobules,^20,21 and OATP1B3 expression is predominantly observed in zone 3.²¹ Although the reason for this remains unclear, hyperplastic hepatocytes within FNH/FNH-like lesions can show that different zonal functions depend on the area, and their distribution varies. Therefore, the different zonal functions of hepatocytes result in varying intensities during the hepatobiliary phase. Based on this speculation and assuming that FNH/FNH-like lesions are mainly composed of hepatocytes with zone 1 characteristics due to unclear underlying causes, they may exhibit relatively mild uptake during the hepatobiliary phase, potentially appearing as a lower SI than the background liver parenchyma. However, van Kessel et al.⁴ correlated the pathological findings of hypointense FNH observed in 1 out of 26 cases, reporting that increased fibrous tissue and absence of bile duct structures were the causes in the hypointense case in the hepatobiliary phase. Therefore, there may be multiple histopathological backgrounds for hypointense area in the hepatobiliary phase.

This study has some limitations. First, this was a retrospective study. Second, the patient population was quite small, as most FNH/FNH-like lesions with typical findings were not subjected to resection or biopsy. In addition, all 5 cases of FNH/FNH-like lesions that underwent histological examination exhibited hypointensity during the hepatobiliary phase compared to the background liver parenchyma, suggesting the presence of selection bias. The same limitations of small sample size and pronounced selection bias are equally applicable to the HCC control group, which was specifically assembled to serve as a comparator for these rare and atypical FNH/FNH-like lesions cases. Third, this study focused solely on FNH/FNH-like lesions and HCC that appeared hypointense on hepatobiliary phase images; we did not compare other hypervascular tumors, such as non-hypointense HCC and HCA, and the discrimination from these is not definite. Future studies should aim to include a larger number of cases to enhance the generalizability of the findings. It would also be valuable to perform comparative analyses with other hypervascular hepatic lesions, such as HCC and HCA, in order to better distinguish FNH/FNH-like lesions from these differential diagnoses.

Conclusion

Rare and atypical FNH/FNH-like lesions presenting with hypointensity during the hepatobiliary phase relative to background liver demonstrated consistent hyperintensity compared to intrahepatic vessels, unlike small typical hypointense HCCs. It is possible that invasive surgery or biopsy could have been avoided if these hepatobiliary phase imaging findings had been carefully considered.