Nodular regenerative hyperplasia: The role of the CK7 immunohistochemistry pattern of expression in diagnosis

Abstract

Objectives

Nodular regenerative hyperplasia (NRH) is a rare vascular disorder of the liver. Clinically, patients present with portal hypertension with or without a cholestatic pattern of injury. Histologically, the liver parenchyma is composed of small nodules of hypertrophic hepatocytes surrounded by atrophic hepatocytes without significant fibrosis. Nodular regenerative hyperplasia is a difficult diagnosis on biopsy specimens, but biopsy remains the gold standard for diagnosis. In this retrospective review, cytokeratin 7 (CK7) immunohistochemistry (IHC) was used to aid in the diagnosis and further characterization of NRH and NRH-like changes.

Methods

The H&E-stained slides, reticulin, and CK IHC were reviewed for 22 cases. The percentage of hepatocytes staining for CK7 (0%-100%), the location of staining (centrilobular hepatic progenitor cells vs periportal/bile ductular reaction), and the pattern of staining distribution (patchy or diffuse) were recorded for comparison.

Results

Of the 22 cases, 9 were CK7 positive. Cases of NRH, however, expressed various degrees of CK7 positivity in centrilobular hepatic progenitor cells, unlike NRH-like changes, which were either CK7 negative or CK7 positive in periportal hepatocytes or in areas of bile ductular reaction.

Conclusions

In cases with the appropriate clinical history and histology, CK7 immunohistochemistry can be performed to distinguish nodular regenerative hyperplasia (primary) and NRH-like changes (secondary). In difficult cases, CK7 positivity in centrilobular hepatic progenitor cells can help confirm the diagnosis of NRH. These data support NRH as a true entity with a distinct pathophysiology from NRH-like changes.

Key Points.

• In cases with the appropriate clinical history and histology, CK7 immunohistochemistry can be performed to distinguish nodular regenerative hyperplasia (NRH) and NRH-like changes.

• CK7 positivity in centrilobular hepatic progenitor cells can help confirm the diagnosis of NRH.

• NRH is a true entity with a distinct pathophysiology from NRH-like changes.

Introduction

Nodular regenerative hyperplasia (NRH) is an uncommon vascular disorder of the liver characterized by nodular hepatic parenchyma devoid of fibrous septa. Typically, the nodules are small (1-10 mm) and visualized grossly by a granular-appearing capsule.¹ The entity was first described by Ransom² in 1953 as miliary hepatocellular adenomatosis in a patient with Felty syndrome. Several years later, Steiner³ encountered a similar presentation. He termed the diagnosis NRH and cautioned physicians not to confuse the entity with cirrhosis. He found NRH to be more common among patients with tuberculosis and severe heart failure. From that time until 2000, only about 200 cases had been reported; however, over the past couple of decades, NRH has been recognized more and more frequently.⁴ The prevalence seems to be evenly distributed among men and women and is higher with increasing age.⁵

Nodular regenerative hyperplasia has been consistently associated with underlying systemic diseases, such as rheumatoid arthritis, HIV, systemic lupus erythematosus, inflammatory bowel disease, celiac disease, scleroderma, antiphospholipid syndrome, and sarcoidosis.⁴ The proposed mechanism for many of these conditions is an autoimmune response in which antibodies are generated against the small hepatic vessels in conjunction with hypercoagulation.⁶ In patients with inflammatory bowel disease and posttransplant patients, the causative process has been linked to drugs prescribed for treatment, including azathioprine, a purine analog, and 6-thioguanine, another thiopurine.^4,7 Highly active antiretroviral therapy, particularly the drug didanosine, has also been linked to NRH in patients being treated for HIV.^8,9

In addition to autoimmune diseases and related drugs, NRH has been associated with hematologic disorders, including polycythemia vera, chronic myelogenous leukemia, and essential thrombocythemia, as well as lymphoproliferative conditions, including Hodgkin and non-Hodgkin lymphoma.¹⁰ The associated coagulopathy and platelet dysfunction in these disorders are thought to be causative.

The underlying etiology in NRH seems to be related to abnormalities in portal hepatic blood flow, which results in atrophy of the ischemic portion of hepatic parenchyma and hypertrophy in the well-supplied areas.^4,11 Nodular regenerative hyperplasia is seen in obliterative portal venopathy, which is characterized by eradication of the small portal veins, leading to scarring and fibrosis in the affected area, while the larger and medium-sized portal vein branches show enlargement with subendothelial thickening.¹²

As a result of abnormal blood flow, portal hypertension is the most common clinical presentation. With portal hypertension, patients usually present with splenomegaly, varices, or ascites of varying degrees.¹⁰ Nodular regenerative hyperplasia comprises an estimated 27% of noncirrhotic portal hypertension cases in Europe and 14% of noncirrhotic portal hypertension cases in Japan.⁴ In the United States, such statistics have yet to be elucidated. The portal hypertension is usually slow to progress compared with portal hypertension associated with cirrhosis and is considered pre-sinusoidal.¹³ The hepatic venous pressure gradient (HVPG), if measured, is usually increased only slightly. In a study of 24 patients with NRH and symptomatic portal hypertension, the mean HPVG was 8.9 mm Hg (normal HPVG, 1-5 mm Hg). A significant difference exists, however, between the HPVG and the portal venous pressure gradient (when calculated), helping confirm the pre-sinusoidal nature of the condition. In the majority of patients, liver function test results are normal.¹⁰ Typically, the prothrombin time, international normalized ratio, and albumin levels are within the standard reference range. Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) may be slightly elevated. If liver enzymes are abnormal, levels are usually consistent with a cholestatic pattern of injury.¹⁴ Alkaline phosphatase has been elevated in approximately 25% of cases reported,¹⁵ but current data may be biased toward patients who are symptomatic. An autopsy study of 2500 patients found a prevalence of NRH of 2.6%.⁵ Of the 64 patients with a diagnosis, only 1 carried the diagnosis before death. In addition, portal hypertension was evident in only 4.7% of cases.

Imaging is often pursued in patients with unexplained portal hypertension or elevated liver enzymes; however, the sensitivity and specificity for NRH in most modalities is poor.⁴ On ultrasonographic images, NRH may appear similar to metastases. The lesions are well delineated and either hypoechoic or isoechoic, with a sonolucent or dark rim. The regenerative nodules themselves may not be visualized because of their small size. On computed tomographic imaging, NRH may present as a heterogenous hepatic parenchyma with an enlarged portal vein, indicating portal hypertension.¹⁵ Such findings on imaging may be confused with cirrhosis. On contrast-enhanced computed tomographic imaging, the regenerative nodules are either hypodense or isodense on both arterial and portal venous phases, differentiating NRH from focal nodular hyperplasia and adenomas, both of which show arterial enhancement.¹⁵ On magnetic resonance imaging, NRH appears hyperintense on T1-weighted imaging and hypointense on T2-weighted imaging.

Because of the low sensitivity and specificity of imaging, a liver biopsy remains the gold standard for diagnosis of NRH.¹⁶ Biopsy is typically performed in the setting of portal hypertension of unknown etiology. Cirrhosis is first clinically ruled out before an invasive procedure is implemented. Bleeding esophageal varices may be another indication. In some patients, the clinical decision to perform a biopsy is solely based on abnormal liver function tests found either through routine monitoring or annual wellness screenings.¹⁴ Biopsy can be performed either percutaneously or using a transjugular approach. The latter is becoming more common because the procedure decreases bleeding risk and allows for measurement of hepatic vein pressures concomitantly.¹⁰

The diagnosis of NRH is definitively made under the microscope. Histology shows a nodular architecture with little to no fibrosis. The nodules are typically small—in general, less than 3 mm in diameter.⁴ Two populations of hepatocytes are present. The hepatocytes within the nodules are hypertrophied and often arranged in double-cell plates.¹⁶ The hepatocytes surrounding the nodules are small, with decreased amounts of cytoplasm. These hepatocytes are atrophic, presumably because of variable ischemia. The formal grading system proposed by Wanless⁵ in 1990 is traditionally used. For the Wanless criteria, the degree of nodularity and fibrosis are quantitatively assessed on a scale from 0 to 3. Diffuse nodularity is graded as 3, while indistinct, vague nodules are given a score of 1. No fibrosis is given a score of 0, whereas definitive fibrosis surrounding nodules is given a score of 3. A score of 3 for nodularity and 0 for fibrosis correlates with NRH. In addition, sinusoidal dilation, portal vein dilation, and atretic portal tracts may be seen.¹⁶ The diagnosis, however, may be difficult to make because of inadequate sampling or small sample size lacking diagnostic histopathologic features. Nodular regenerative hyperplasia may also be overlooked if the biopsy shows evidence of another entity, such as viral hepatitis or alcoholic liver disease, simultaneously⁴

In this retrospective study, we assessed the utility of cytokeratin 7 (CK7) immunohistochemistry (IHC) to aid in the pathologic diagnosis and characterization of NRH (de novo/primary) and NRH-like changes (secondary changes), particularly in regard to clinical settings.

Methods

We searched the pathology database within the electronic health record (EHR) system for cases diagnosed as NRH or suggestive of the diagnosis. The search term nodular regenerative hyperplasia was used to acquire cases from July 2010 through August 2022. A total of 25 possible cases were identified, 3 of which were subsequently eliminated due to lack of slides or paraffin-embedded tissue samples. Each case was subsequently evaluated, with a thorough EHR review. Information obtained included patient age, sex, race, presenting symptoms, imaging reports, indication for biopsy or resection, laboratory values at the time of the procedure, endoscopy findings, concurrent medications, coexisting systemic diseases (HIV, hepatitis C virus [HCV], and hepatitis B virus [HBV]), autoimmune markers, and the patient’s condition as of August 2022. The 22 cases included 20 liver core biopsies, 1 partial hepatectomy, and 1 transplanted liver. The H&E-stained slides at the time of original diagnosis were reviewed. Each was classified according to the criteria originally proposed by Wanless.⁵ Reticulin special stain was acquired for all cases and used to identify fibrosis and atrophic or hypertrophic hepatocytes. The extent of hepatocellular nodules less than 3 mm in diameter were noted, along with the presence or absence of fibrous septa. Immunohistochemistry for CK7 was performed using SP52 CK7 clone from Roche. Any staining of CK7 in hepatocytes was considered positive. CK7 staining was graded on a scale from 0 to 3, with 3 being the strongest intensity compared with the control and 0 being no staining. The percentage staining for CK7 (0%-100%), location (centrilobular hepatic progenitor cells [CHPCs] vs periportal/bile ductular reaction [PP/BDR]), and pattern of distribution (patchy or diffuse) were recorded for comparison.

Results

All 22 cases identified were diagnosed as NRH or NRH like, or NRH was listed within the differential diagnosis. The patients ages ranged from 24 years to 78 years, with the average being approximately 53 years. Of these 22 possible cases of NRH, 7 patients were female and 15 were male; 15 identified as white, 4 as African American, 2 as Hispanic, and 1 as Asian. For the liver transplant, hepatocellular carcinoma was suspected; for the partial hepatectomy, metastatic adenocarcinoma was speculated. The indications for biopsy included metastatic carcinoma (n = 1), suspected infection (n = 1), elevated liver enzymes (n = 5), signs and symptoms of portal hypertension (n = 5), possible cirrhosis (n = 7), and transplant rejection (n = 1).

Clinical Presentation

Eleven patients (50%) had an elevated alkaline phosphatase or γ-glutamyl transpeptidase value, indicating a cholestatic pattern of injury. Only 2 patients (9%) had elevated AST or ALT value, with no elevation in alkaline phosphatase or γ-glutamyl transpeptidase level. At the time liver biopsy or resection, 12 patients (55%) had thrombocytopenia and 6 (27%) had evidence of splenomegaly on imaging. Of the 11 patients with endoscopic findings, only 3 had esophageal varices. Similar to previous studies, many of the patients had a coexisting systemic illnesses (n = 13 [59%]), such as HCV (n = 3 [14%]), hepatitis B (n = 1 [5%]), hematologic malignancy (n = 3 [14%]), metastatic carcinoma (n = 2 [9%]), autoimmune hepatitis (n = 1 [5%]), focal segmental glomerulosclerosis (n = 1 [5%]), Ehlers-Danlos syndrome (n = 1 [5%]), and diabetes (n = 1 [5%]).

Pathologic Findings

All 22 cases were adequate for evaluation and had vague nodularity on H&E-stained or reticulin-stained sections, giving the impression of NRH. In addition to the possible NRH, several other findings were observed. Of the 22 cases, 5 had mild periportal or portal fibrosis. There were 9 cases with prominent sinusoidal dilation, while 4 cases had portal inflammation. Mild to moderate steatosis was identified in 2 cases, and iron overload was recognized in 2 separate cases. The patient with partial hepatectomy had metastatic carcinoma, and the patient who had undergone liver transplant had multiple hepatic adenomas. CK7 immunohistochemistry (IHC) was positive in 9 of 21 cases (43%) (CHPC and PP/BDR) and negative in 12 of 21 cases (57%). CK7 staining could not be evaluated in 1 case due to exhaustion of tissue. The 12 patients with negative CK7 IHC results had a variety of systemic diseases and underlying etiologies, with the most common being congestive heart failure (n = 6). The others included focal segmental glomerulosclerosis (n = 1), type 1 diabetes (n = 1), Ehlers-Danlos syndrome (n = 1), chronic HCV (n = 1), and chronic HBV (n = 1).

In 4 of the 9 CK7-positive cases, CK7-positive hepatocytes were noted in periportal areas or areas of bile ductular reaction (PP/BDR) TABLE 1. Clinically, these 4 patients had additional underlying etiologies contributing to the NRH-like changes (secondary NRH), including HCV, autoimmune hepatitis, inferior vena cava stenosis, and congestive hepatopathy. In the remaining 5 cases, CK7-positive cells were located around the central veins in zone 3 (CHPCs; primary NRH/de novo disease). The percentage of positive CHPCs ranged from 30% to 75%. Three cases were diffusely positive around all central veins, while 2 cases had selective positivity. The reticulin in all 5 cases clearly demonstrated areas of hypertrophic hepatocytes around the central veins surrounded by atrophic curvilinear hepatocytes. Periportal or portal fibrosis was absent.

TABLE 1.

Findings in CK7-Positive Cases

Case No.	Age (sex)	CK7 location, pattern of distribution	CK7 positive, %	CK7 intensity	Portal hypertension	Elevated liver enzymes	Systemic diseases	Treatment before biopsy	Final diagnosis
1	60 (M)	PP/BDR, diffuse	90	Moderate	Present	AST, ALT, alkaline phosphatase, γ-glutamyl transpeptidase	HCV, post–liver transplant	—	NRH-like changes: active hepatitis
2	54 (F)	CHPCs, patchy	70	Moderate	Present	None	Metastatic carcinoma	Carboplatin/etoposide Platinum/irinotecan Rovalpituzumab tesirine	NRH
3	55 (M)	CHPCs, patchy	40	Moderate	Present	Alkaline phosphatase	Hematologic malignancy	Fludarabine, cyclophosphamide, and rituximab Azacytidine	NRH
4	42 (F)	PP/BDR, patchy	30	Moderate	Present	None	Autoimmune hepatitis	—	NRH-like changes: active autoimmune hepatitis
5	24 (F)	PP/BDR, patchy	70	Moderate	Absent	AST, γ-glutamyl transpeptidase	Focal segmental glomerular sclerosis, inferior vena cava stenosis	—	NRH-like changes: congestive hepatopathy
6	67 (M)	CHPCs, diffuse	75	Strong	Present	None	Hematologic malignancy	No treatment	NRH
7	59 (F)	CHPCs, diffuse	30	Moderate	Present	Alkaline phosphatase	Metastatic carcinoma	Leucovorin calcium, fluorouracil, and oxaliplatin Right hepatic lobe ablation	NRH
8	45 (M)	CHPCs, diffuse	75	Moderate	Present	AST, ALT, alkaline phosphatase, γ-glutamyl transpeptidase (post transplant)	Liver transplant	—	NRH
9	30 (M)	PP/BDR, patchy	20	Moderate	Present	None	Type 1 diabetes	—	NRH-like changes: congestive hepatopathy

ALT, alanine aminotransferase; AST, aspartate aminotransferase; CHPC, centrilobular hepatic progenitor cells; CK7, cytokeratin 7; F, female; HCV, hepatitis C virus; M, male; NRH, nodular regenerative hyperplasia; PP/BDR, periportal/bile ductular reaction.

FIGURE 1 shows an easily identifiable example of NRH. Histology shows a nodular architecture with hypertrophied hepatocytes within the nodule and atrophic hepatocytes surrounding it. The CK7 IHC is clearly staining the CHPCs around the central veins. FIGURE 2 is another example of NRH. The nodular architecture is less apparent but better visualized with the reticulin stain. The CK7 IHC is positive within the CHPCs. FIGURE 3 is the third example of NRH in which the nodularity is vague. CK7 IHC staining, however, correlates with glutamine synthetase, confirming the centrizonal location. FIGURE 4 is a case of NRH-like changes due to congestive hepatopathy. Sinusoidal dilation gives the appearance of nodularity, but the CK7 positivity is only within the bile ducts.

FIGURE 1.

Nodular regenerative hyperplasia. A, Nodular regenerative hyperplasia with distinct nodularity (H&E, ×4). B, A higher-power image shows hypertrophied hepatocytes within the nodule and atrophic hepatocytes surrounding it (H&E, ×10). C, Cytokeratin 7 immunohistochemistry (×8). Arrows point to the positive cells surrounding the central veins in zone 3, the centrilobular hepatic progenitor cells. Ovals highlight the bile ducts for comparison.

FIGURE 2.

Nodular regenerative hyperplasia. A, Nodular regenerative hyperplasia with vague nodularity (H&E, ×4). B, Reticulin stain demonstrates areas of hypertrophic hepatocytes around the central veins, surrounded by atrophic curvilinear hepatocytes (×4). Periportal or portal fibrosis is absent. C, Cytokeratin 7 immunohistochemistry (×8). Arrows point to the positive centrilobular hepatic progenitor cells, and the ovals highlight the bile ducts/ductular reaction.

FIGURE 3.

Nodular regenerative hyperplasia. A, Nodular regenerative hyperplasia with slight nodularity (H&E, ×4). B, Reticulin stain better defines the nodularity (×4). C, Glutamine synthetase highlights strongly zone 3 of the liver around the central veins (×4). D, Cytokeratin 7 immunohistochemistry gives a similar staining pattern in a centrizonal location, highlighting the centrilobular hepatic progenitor cells (×4).

FIGURE 4.

Nodular regenerative hyperplasia (NRH)–like changes (secondary NRH). A, NRH with nodularity at low power, with associated sinusoidal dilatation (H&E, ×4). B, Reticulin stain fails to show distinct hypertrophic or atrophic hepatocytes pericentral and periportal, respectively (×4). C, Cytokeratin 7 immunohistochemistry is negative (×8). The ovals highlight the bile ducts with portal tracts.

The patients in all 5 cases with positive centrizonal CK7 IHC had evidence of portal hypertension confirmed by evaluation of transjugular pressures, imaging, or symptoms of ascites. Two patients had normal liver enzymes; 2 had elevated alkaline phosphatase; and 1 had elevated alkaline phosphatase, AST, and ALT. The patient with multiple elevated liver enzymes was 4 months post–liver transplant. Hematologic malignancies were diagnosed in 2 of the other patients, and 2 had metastatic carcinoma. Three patients were male and 2 were female; the average age was 56 years.

Discussion

Nodular regenerative hyperplasia is difficult to diagnose. It is a rare entity with many nonspecific and overlapping histologic features. Studies have shown poor interobserver agreement among pathologists on the diagnosis.⁶ It is often described as a clinicopathologic diagnosis in which histology alone is not sufficient. Although clinical correlation is necessary, ancillary studies are invariably applied. A reticulin stain is performed routinely to evaluate for the presence of alternating regenerative and atrophic hepatic plates in the absence of fibrosis. In normal hepatic parenchyma, reticulin fibers made of type III collagen surround the hepatic plates in a regular and radial pattern, forming the overall acinar structure.¹⁷ In NRH, reticulin staining proves that nodularity is caused by compression of the hepatic plates due to hypertrophy of cells, not fibrosis.¹⁸ More recently, CK7 IHC has also been used.¹⁶ Normal hepatocytes do not express CK7. In the liver, under normal conditions, CK7 is expressed only by interlobular bile ducts and hepatocyte progenitor cells. In a cholestatic pattern of liver injury, CK7 positivity has been found in centrilobular hepatocytes in association with centrilobular scarring.¹⁹ The CK7 expression is thought to either be an adaptive response to ischemia by hepatocytes or activation of hepatic progenitor cells.¹⁶

Using CK7 IHC in our study, we were able to differentiate NRH (primary/de novo) from NRH-like changes (secondary) FIGURE 4. In half the patients with negative CK7 IHC, congestive heart failure was well documented. In such cases, a prominent histologic finding was sinusoidal dilation with vague nodularity. On a reticulin stain, sinusoidal dilation can give the impression of areas of atrophic hepatocytes due to compression. Although sinusoidal dilation can be evident in cases of NRH, if such findings are pronounced, a more thorough look into the clinical history is advised.¹⁶ In most of our cases in which NRH was listed as a differential diagnosis, only early signs of congestive hepatopathy were present; thus, detailed electronic health record review was required.

In patients with the typical presentation of portal hypertension with unknown etiology, CK7 positivity in the centrizonal hepatocytes (CHPCs) was indicative of the diagnosis, similar to the results shown by Bakshi et al.¹⁶ The percentage of positive centrizonal areas, however, varied. This, in turn, could be a potential pitfall. Biopsy specimens are limited, and at times, few central veins are observed. Nonrepresentative sampling could render CK7 IHC impractical and nondiagnostic. Conversely, if CK7 IHC is positive, careful observation of the location is crucial. In entities with cholestatic liver pattern of injury, PP/BDR, CK7 IHC will be positive. CK7 positivity in portal areas has been described in cases such as HCV, as we witnessed.¹⁹ Without cautious evaluation, such staining could lead to an inappropriate conclusion.

In the 5 cases of confirmed NRH, 2 patients had metastatic carcinoma, small cell carcinoma, and colorectal carcinoma, and 2 patients had hematologic disorders, B-cell lymphoma, and myelodysplastic syndrome. We did not see a correlation with autoimmune disorders, as usually described, but this could be due to the limited sample size.⁴ The last patient had recently undergone a liver transplant for alcoholic cirrhosis. Although the mechanism is unknown, NRH has been documented in approximately 1.3% of liver transplant patients.²⁰

This study has several limitations. It was single-institution retrospective review. The number of cases was limited, and only 5 positive cases were confirmed as primary NRH (de novo). Some cases had limited clinical history, and follow-up was at times infrequent. To our knowledge, however, it is the first study to describe IHC differences between NRH primary (de novo NRH) and NRH-like changes (secondary).

In patients with a clinical history of portal hypertension without an alternative cause, CK7 IHC findings can be used to justify the diagnosis of NRH. The IHC and clinical history, however, must be thoroughly evaluated to reach that conclusion. In cases with NRH-like changes due to etiologies such as congestive hepatopathy and portal vein thrombosis, CK7 is negative. In NRH, variable CK7 positivity in centrizonal areas is observed. These data support NRH as a true entity with a distinct pathophysiology from NRH-like changes, but larger studies are required for validation.

Conflict of interest disclosure: The authors have nothing to disclose.