OUP user menu

Arginase-1, HepPar-1, and Glypican-3 Are the Most Effective Panel of Markers in Distinguishing Hepatocellular Carcinoma From Metastatic Tumor on Fine-Needle Aspiration Specimens

Dana T. Timek MD, Jianhui Shi MD, Haiyan Liu MD, Fan Lin MD
DOI: http://dx.doi.org/10.1309/AJCPK1ZC9WNHCCMU 203-210 First published online: 1 August 2012


Distinction of liver metastases from hepatocellular carcinoma (HCC) may present a diagnostic challenge. Arginase-1 (Arg-1) is a marker for HCC recently described in some literature. Immunohistochemical evaluation of Arg-1, hepatocyte paraffin-1 (HepPar-1), and glypican-3 expression was performed on 1,240 surgical specimens and 62 liver fine-needle aspiration specimens (29 HCCs, 28 metastatic tumors, and 5 benign liver cases). The staining results on tissue microarray sections showed that 2.7% and 3.1% of nonhepatic tumor cases were positive for HepPar-1 and glypican-3, respectively; none was positive for Arg-1. For fine-needle aspiration specimens, 19 HCCs were positive for all 3 markers; 9 were positive for 1 or 2 markers; and only 1 case was negative for all 3 markers. These data demonstrate that Arg-1 is the most specific marker in differentiating a non-HCC from HCC. It is recommended to use 3 markers as a panel in distinguishing HCC from metastatic carcinoma.

Key Words
  • Arginase-1
  • HepPar-1
  • Glypican-3
  • Carcinomas
  • Hepatocellular carcinomas
  • Cytology

Distinction of liver metastatic tumor from hepatocellular carcinoma (HCC) may present a diagnostic challenge, especially in small tissue biopsy or fine-needle aspiration (FNA) biopsy specimens. Hepatocyte paraffin-1 (Hep-Par-1) and glypican-3 are useful diagnostic markers, but the expression of these 2 markers, along with other routinely used markers such as polyclonal carcinoembryonic antigen, CD10, and α-fetoprotein, has also been reported in nonhepatocellular tumors.114 Such markers have been and continue to be thoroughly studied in numerous platforms, including tissue microarray (TMA)–based studies,1518 as well as in cytologic specimens.1923 Arginase-1 (Arg-1), an enzyme involved in the hydrolysis of arginine to ornithine and urea, was recently recognized in some literature as both a sensitive and specific marker for benign and malignant hepatocytes.24,25 This is a useful diagnostic marker in the differential diagnosis of HCC vs metastatic tumor and has only been studied on surgical specimens.24 Studies have shown Arg-1 to be more sensitive and specific than Hep-Par-1 in detecting HCC.24

However, there is no literature that addresses the usefulness of Arg-1 on FNA specimens. This is an important void to address because the diagnostic challenge of determining HCC from other malignancies frequently involves small biopsy or FNA specimens. In this study, we compared the expression of these 3 markers (Arg-1, HepPar-1, and glypican-3) in HCCs and carcinomas from various organs on both surgical and FNA specimens.

Materials and Methods

Construction of TMA blocks

The study was approved by the institutional review board at Geisinger Medical Center (Danville, PA). Carcinomas from various organs (n = 1,240) from 2000 to 2010 were retrieved from the archives of the Department of Laboratory Medicine at Geisinger Medical Center. The number of tumor cases for each specific entity is summarized in Table 1. Multiple TMA blocks with 2 punches of 0.75 mm or 1.0 mm each for each case were constructed as previously described.26,27

Immunohistochemical Stains on Surgical Specimens

Immunohistochemical evaluation of the expression of Arg-1, HepPar-1, and glypican-3 in 1,240 cases of carcinomas from various organs using tissue TMA sections was performed. The detailed antibody information and staining conditions are summarized in Table 2. The staining was done on the DAKO staining system (DAKO, Carpinteria, CA) based on the previously published protocol.28,29 The staining intensity was graded as weak or strong. The distribution was recorded as negative (<5% of tumor cells stained), 1+ (5%–25%), 2+ (26%–50%), 3+ (51%–75%), or 4+ (>75%). Two surgical pathologists (F.L. and H.L.) independently evaluated the immunostained slides.

View this table:
Table 1

Immunohistochemical Stains on Cytology Cell Block Specimens

Sixty-two FNA biopsy specimens of the liver (29 HCCs, 28 metastatic tumors, and 5 benign liver cases) with adequate cellularity on cell blocks were included. The 29 HCCs were divided into 2 groups: well to moderately differentiated HCC (n = 22) and moderately to poorly differentiated HCC (n = 7). The microscopic grading system for HCC on both surgical and FNA specimens was based on the modified criteria in the Atlas of Tumor Pathology.30 For grade 1 HCC, the nuclear features were similar to those of hepatocellular adenoma. For grade 2 to 4 HCC, the tumor grade was increased, with increased nuclear/cytoplasmic ratio, prominent nucleoli, nuclear membrane irregularities, and nuclear pleomorphism. To simplify this grading system on an FNA sample, grade 1 and grade 2 HCC cases were classified as well to moderately differentiated, and grade 3 and grade 4 HCC cases were included in the moderately to poorly differentiated group. The cell block was prepared as previously described.31 Immunostains for Arg-1, HepPar-1, and glypican-3 were performed on these cell blocks with the same protocol described before. Adequate cellularity for malignant cases was tentatively defined as at least 3 groups of atypical epithelial cells (more than 10 cells in each group) and single atypical cells. For benign liver tissue, a minimum of 3 groups of hepatocytes in the cell block was considered adequate. Most cell blocks contained more than 5 groups of atypical epithelial cells or benign hepatocytes. The staining intensity (weak or strong) and distribution (negative [<5% of tumor cells stained], focal staining [5%–50%], and diffuse [>50%]) were recorded. Two surgical pathologists (D.T.T. and F.L.) independently evaluated the immunostained slides.

View this table:
Table 2


Arg-1, HepPar-1 and Glypican-3 Expression in Tumors on TMA Sections

Seventeen (94%) of 18 HCC cases were positive for both Arg-1 and HepPar-1, with diffuse staining (4+) in 15 cases for Arg-1 and 13 cases for HepPar-1. In contrast, only 14 (78%) of 18 cases were positive for glypican-3, with diffuse staining in 9 cases. For the nonhepatocellular tumors (n = 1,222), 33 (2.7%) and 38 (3.1%) cases were positive for HepPar-1 and glypican-3, respectively; in contrast, none was positive for Arg-1. The results are summarized in Table 1. Representative cases with diffuse and focal positive staining for Arg-1 are shown in Image 1. Representative cases of nonhepatic tumors positive for HepPar-1 and glypican-3 are shown in Image 2 and Image 3.

Arg-1, HepPar-1, and Glypican-3 Expression on FNA Cell Block Specimens

Of 29 HCC cases, 19 were positive for all 3 markers, and 6 were positive for 2 markers (2 of these cases were in the moderately to poorly differentiated HCC group, and 4 were in the well to moderately differentiated group). Three cases were positive for only 1 marker: 2 cases were in the moderate to poorly differentiated group and were focally weakly positive for glypican-3; 1 case was in the well to moderately differentiated group and stained strongly and diffusely positive for HepPar-1. Only 1 case was negative for all 3 markers and was in the moderately to poorly differentiated group. The staining results are summarized in Table 3 and Table 4. Representative HCC cases focally and diffusely positive for Arg-1 are shown in Image 4A and Image 4B. Representative HCC cases with HepPar-1 and glypican-3 positivity are shown in Image 4C and Image 4D.

Image 1

Diffuse (A) and focal (B) Arg-1 staining (both cytoplasmic and nuclear pattern) on tissue microarray sections of hepatocellular carcinoma (×400).

Image 2

Cytoplasmic HepPar-1 staining on samples of esophageal adenocarcinoma (A) and lung adenocarcinoma (B) (×400).

Image 3

Cytoplasmic staining of glypican-3 on lung squamous cell carcinoma (A) and endocervical adenocarcinoma (B) (×400).

All 22 (100%) well to moderately differentiated HCC cases were positive for 1 or more of the immunomarkers studied. For glypican-3, the majority (41%) of these cases stained strongly and diffusely positive. Another 27% and 23% stained focally strong and focally weak, respectively. Hep-Par-1 and Arg-1 showed similar staining patterns, with similar percentages, staining either diffusely strong (45% for HepPar-1 and 41% for Arg-1) or focally strong (41% for HepPar-1 and 36% for Arg-1), and fewer staining focally weak (5% for Hep-Par-1 and 14% for Arg-1).

Sensitivity for all 3 markers was decreased in the moderately to poorly differentiated HCC group (n = 7): 71% of the total HCC cases in this group were positive for glypican-3, 57% were positive for HepPar-1, and 43% were positive for Arg-1. In the majority of positive cases, glypican-3 demonstrated focal weak staining (57%) with some focal strong staining (14%). HepPar-1 demonstrated no focal weak staining in this group. Arg-1 was focally strong in 29% and focally weak in 14%; there was no strong diffuse positivity for Arg-1 in this group. These results are summarized in Table 5.

View this table:
Table 3
View this table:
Table 4

Among the 28 metastatic carcinoma cases, no cases were positive for HepPar-1 or Arg-1. Conversely, 3 cases were positive for glypican-3 (11%). The glypican-3-positive cases were described in their original cytologic reports as metastatic urothelial carcinoma, metastatic adenocarcinoma compatible with esophageal primary, and metastatic adenocarcinoma of unknown primary.

Diagnostic Sensitivity and Specificity of Arg-1, HepPar-1, and Glypican-3

The diagnostic sensitivity and specificity of Arg-1, Hep-Par-1, and glypican-3 were calculated based on the combined numbers of TMA and FNA specimens, with a total of 47 HCC cases and 1,250 cases of nonhepatocellular tumors. The results are summarized in Table 6.

Image 4

Staining results on cell block sections with diffuse Arg-1 staining (A) and focal Arg-1 staining (B), HepPar-1 staining (C), and glypican-3 staining (D) (×600).

View this table:
Table 5


Arg-1 has been described in recent literature as the most sensitive and specific marker for identifying an HCC when working on a tumor of unknown primary, especially compared with HepPar-1, glypican-3, and other markers.24 The aims of our study were to (1) confirm Arg-1 as the most specific marker for identifying HCC compared with HepPar-1 and glypican-3; (2) test the diagnostic usefulness of Arg-1 on FNA samples of the liver on cell blocks because most HCCs were diagnosed on FNA or small biopsy specimens; and (3) identify the most effective panel of markers in distinguishing a metastasis from an HCC on FNA specimens on cell blocks.

View this table:
Table 6

Our data support the conclusion that Arg-1 is a highly specific marker for HCC on both TMA and FNA specimens.24 TMA data demonstrated Arg-1 with a 100% diagnostic specificity in a large series of surgical cases (n = 1,222), which was similar to the report by Yan et al.24 In that study, only 1 prostatic adenocarcinoma case was positive for Arg-1. The current study showed all nonhepatic tumors, including 136 cases of prostatic adenocarcinoma, were negative for Arg-1. In contrast, 33 (2.7%) and 38 (3.1%) cases of nonhepatic tumors were positive for HepPar-1 and glypican-3, respectively. Similarly, all nonhepatocellular tumors (n = 28) were negative for both Arg-1 and HepPar-1 on FNA specimens. In contrast, 3 (11%) nonhepatocellular cases were positive for glypican-3. The expression of HepPar-1 in nonhepatocellular tumors is well documented in the literature. For example, Kakar and coworkers32 reported that the expression of Hep-Par-1 was observed in 3 of 11 lung adenocarcinomas, 6 of 8 esophageal carcinomas, and 7 of 10 gastric carcinomas. Therefore, caution should be taken when using HepPar-1 to confirm a diagnosis of HCC.

Arg-1 also has been shown to be a very sensitive marker for hepatocytes,24 and our study supports this conclusion as well. In our TMA study, 17 (94%) of the 18 HCC cases were positive for Arg-1, which is comparable to the 96% finding of Yan et al.24 In our FNA specimen study, we found the sensitivity to be somewhat lower than this; 23 (79%) of 29 HCC cases were positive for Arg-1. The lower diagnostic sensitivity on FNA samples may be because of the small series of cases in this study (n = 29) and the limited number of tumor cells available compared with the surgical cases. In addition, 1 of the HCC cases in this study was nonreactive for all 3 markers; in retrospect, this case may not have been a classic HCC. However, the sensitivity for HepPar-1 (83%) in our cytologic samples was similar to that of Yan et al (84.1%), and our sensitivity for glypican-3 (86%) was also similar to that in other reported literature.7,8,13,15,19,3335 Therefore, until larger studies of Arg-1 on FNA samples are completed, the possibility that the Arg-1 antibody-antigen interaction may be altered in some fashion (such as with inadequate fixation) in cytologic preparations must at least be entertained.

Importantly, Yan and coworkers24 demonstrated that Arg-1 has a better sensitivity in identifying higher-grade HCC (moderately differentiated and poorly differentiated HCC; n = 81) than does HepPar-1. This finding is very useful because one of the most frequent diagnostic challenges is to differentiate a high-grade HCC from a metastasis, especially on an FNA or small biopsy specimen. Our cytologic data on HCC were divided into 2 categories based on degree of differentiation (well to moderately differentiated and moderately to poorly differentiated) to reduce any intra- and interobserver discrepancies in grading that a 3-tiered level may incur. Among the well to moderately differentiated HCC cases, 22 (100%) were positive for 1 or more of any combination of the 3 markers (glypican-3, HepPar-1, and Arg-1), while approximately 86% of cases in the moderately to poorly differentiated HCC group were positive for any combination of those markers. In general, the well to moderately differentiated HCC group stained more diffusely strong and focally strong than focally weak compared with the moderately to poorly differentiated HCC group, which, among the positive cases, stained more focally weak with glypican-3, more focally strong with HepPar-1, and focally strong to focally weak with Arg-1. However, among negative cases, Arg-1 (57%) had the highest negative rate compared with HepPar-1 (43%) and especially with glypi-can-3 (29%). These findings demonstrate that well to moderately differentiated HCCs stain positively more often for these markers than do higher grade HCCs. However, our data fail to demonstrate a better sensitivity of Arg-1 for higher-grade HCC compared with the other markers. This may certainly be because of the small sampling of the cytologic specimens in the moderately to poorly differentiated HCC category (n = 7), limited amount of sample for each case, and patchy/focal staining for Arg-1 in higher-grade HCC.

When combining both TMA and FNA samples, the diagnostic sensitivities for Arg-1, HepPar-1 and glypican-3 were 85%, 87%, and 83%, respectively. The diagnostic specificities for Arg-1, HepPar-1, and glypican-3 were 100%, 97.4%, and 96.7%, respectively.

In addition, our data showed 3 cases of HCC that were positive for HepPar-1 but negative for Arg-1. Conversely, 2 cases were positive for Arg-1 but negative for HepPar-1. Furthermore, 6 HCC cases were positive for 2 of the 3 markers used in our study, and 3 HCC cases were positive for only 1 marker; these cases were from both the well to moderately differentiated and the moderately to poorly differentiated groups, and the markers for which they were positive also varied. These findings reinforce the concept of using a diagnostic panel of these 3 markers to best capture and classify HCC vs other carcinomas, especially in limited biopsy or FNA samples, where scant amounts of cells may be available for staining.

It should be pointed out that similar to HepPar-1, Arg-1 plays no role in distinguishing between normal and adenoma/malignant hepatocytes, and this was reflected in our FNA specimen study, in which 100% of the benign liver cases were positive for Arg-1 and HepPar-1. Glypican-3 has been shown to be less sensitive as well as less specific than HepPar-1 and Arg-1, with positive staining on various tumors,24,33,34 and can occasionally be positive in regenerative nodules associated with cirrhosis.35

Another interesting point that needs further investigation is that Arg-1 typically shows cytoplasmic staining with some nuclear staining.24 Only cytoplasmic staining was regarded as positive both in this study and in the study of Yan et al,24 but strong and diffuse nuclear staining can be seen in some of our cases (both normal liver and HCC) along with the cytoplasmic staining. The significance of Arg-1 nuclear reactivity still remains unknown.

In summary, these data demonstrate that Arg-1 has a similar sensitivity and higher specificity in differentiating a non-HCC from HCC when compared with HepPar-1 and glypican-3. These 3 markers are recommended as the most effective panel for small tissue biopsy or FNA specimens in the distinction of HCC from metastatic carcinoma.


We thank Tina Brosious and Erin Powell for construction of TMA blocks and cutting TMA sections, Angie Bitting for her assistance with immunostains, and Kathy Fenstermacher for editing the manuscript.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
View Abstract