A Triple Stain of Reticulin, Glypican-3,.pptx

triple stain of liver

1. A Triple Stain of Reticulin, Glypican-3, and Glutamine Synthetase A Useful Aid in the Diagnosis of Liver Lesions

2. PUBLICATION SITE: PUB MED IMPACT FACTOR: INDEXES:

3. AUTHORS Benjamin J. Swanson, MD, PhD; Martha M. Yearsley, MD; William Marsh, MD; Wendy L. Frankel, MD

4. ABSTRACT Context. — The correct histologic diagnosis of mass lesions of the liver can be difficult, especially in biopsy samples. Reticulin, glypican-3, and glutamine synthetase are stains that can help distinguish hepatocellular carcinoma, hepatic adenoma, and focal nodular hyperplasia. Objective. — To evaluate the utility of a triple stain of reticulin, glypican-3, and glutamine synthetase in distinguishing hepatocellular carcinoma, hepatic adenoma, and focal nodular hyperplasia.

5. Results — The combination of intact reticulin with either glypican-3 negativity or negative glutamine synthetase was 92% sensitive and 95% specific in the distinction of tissue microarray hepatic adenoma from hepatocellular carcinoma. For the distinction of tissue microarray focal nodular hyperplasia and hepatic adenoma, map like glutamine synthetase was most useful and was 85% sensitive and 100% specific. Conclusions — The triple stain of reticulin, glypican-3,and glutamine synthetase is useful in the differentiation of hepatocellular carcinoma, hepatic adenoma, and focal nodular hyperplasia on biopsy specimens. Furthermore, this triple stain is advantageous to single stains and can help when aberrant staining patterns are observed.

6. INTRODUCTION T he classification of liver lesions can be a challenging endeavor. With advances in imaging and biopsy techniques, pathologists are increasingly required to render diagnoses when there is scant diagnostic tissue or when the prebiopsy interpretation of radiologic images is equivocal.

7. Hepatocellular carcinoma (HCC) is the most common primary malignancy of the liver and is one of the most common tumors worldwide.1,2 Common risk factors include cirrhosis, viral hepatitis C infection, and viral hepatitis B infection.3 However, up to 40% of HCCs arise in noncirrhotic livers.4 Ultrasound is used to screen patients with cirrhosis.5

8. Hepatic adenomas (HAs) are benign tumors that most commonly occur in reproductive-age women.6 Transcriptional and protein microarray profiling have demonstrated distinct subgroups among these tumors, some of which demonstrate unique genetic and histologic alterations.7 Hepatic adenomas harboring mutations in HNF1 a often have steatosis and lack cytologic dysplasia.

9. When mutations in b -catenin are present within HA, there is often small cell cytologic dysplasia as well as a significantly increased risk of progression to HCC. Those lesions can be confused with HCC. Inflammatory HAs demonstrate telangiectatic blood vessels and increased inflammatory infiltrates and were initially classified as telangiectatic focal nodular hyperplasia (FNH), which can be difficult to distinguish from FNH on slides stained with hematoxylin-eosin. However, genetic and proteomic data suggest that these tumors are best classified as HAs.8,9

10. The distinction of a well-differentiated HCC from a HA can be difficult on needle biopsy, especially in a patient without cirrhosis. Histologic features that favor HCC include increased nuclear/cytoplasmic ratios and nuclear pleomorphism. Furthermore, although many HCCs and HAs have classic radiologic findings, some have variant imaging features.5,1 Likewise, distinguishing FNH from HA can be challenging, especially when characteristic radiologic features , such as a central scar, are not present.17 Histologic features that favor FNH include fibrous bands without portal tracts, reactive ductules , thick-walled blood vessels, cholestasis, and nodularity.20,21

11. To aid in the differential diagnosis of these various entities, authors22 have developed panels of immunostains that can be used primarily for resection specimens. In addition, others23,24 have developed immunostain panels for biopsy specimens. In our laboratory, we find that reticulin, glypican-3 (Gly3), and glutamine synthetase (GS) are the most reliable stains and the easiest to interpret. Thus, we developed a triple stain (reticulin, GS, Gly3) and applied it to tissue microarrays (TMAs) to simulate a needle biopsy and compared the findings in whole tissue sections of HCC, HA, and FNH to determine the utility of this triple stain, particularly with limited material available.

12. MATERIALS AND METHODS The current study was approved by the Ohio State Biomedical Sciences’ institutional review board. Cases were retrieved from archived specimens, and the slides were reviewed. Of 65 TMAs with cores in duplicate, there were 13 HAs, 13 FNHs, 19 HCCs, and 20 nonmalignant liver parenchyma adjacent to hepatocellular carcinoma. Unstained, formalin-fixed slides of both whole tissue sections (WTS; 44 of 109; 40%) and TMA (65 of 109; 60%) underwent sequential staining. A triple stain of reticulin ( Dako , Carpinteria, California) was followed by antigen retrieval using alow -pH buffer. This was followed by an immunostain for glutamine synthetase as per manufacturer’s protocol.

13. Next, slides were immunostained with glypican-3 (1:900, 15-minute incubation; clone 1G12, BioMosaics , Burlington, Vermont), which was developed with the Leica Bond Polymer Refine Red Detection kit (alkaline phosphatase red chromogen). Reticulin staining was scored as intact or lost. Reticulin loss was defined as greater than 3 to 4 hepatocytes between reticulin fibers. Glypican-3 staining was scored as negative (0%–5% of cells) or positive ( . 5% of cells). Glutamine synthetase staining was scored as negative (0%–5%), perivenular, diffuse or map like (solid areas with a heterogenous pattern).25

14. RESULTS

15. Clinical Characteristics of Liver Lesions The WTS of HA were from 15 patients, with an average age of 34.9 years (range, 14–51 years), and all patients were women. Four (27%) had significant steatosis, 3 (20%) had telangiectasia, and 3 (20%) met the diagnostic criteria for adenomatosis. The TMA cases of HA were from 13 patients,10 (77%) of which were the same as the WTS cases. The average age was 34.4 years (range, 23–51 years), and all cases were from women. Three cases (23%) showed significant steatosis, 3 (23%) showed telangiectasia, 2 (15%) met the diagnostic criteria for adenomatosis, and 1 (8%) showed large cell change.

16. The WTS and TMA of FNH were derived from the same 13 patients. The average age of these patients was 36.5 years (range, 21–61 years) and comprised 10 women (77%) and 3 men (23%).

17. The WTS of HCC were from 16 patients, with an average age of 59.5 years (range, 28–83 years). Twelve cases (75%) were from men, and 4 cases (25%) were from women. Three cases (19%) were well-differentiated, 10 (63%) were moderately differentiated, and 2 (12%) were poorly differentiated. One case (6%) was a fibrolamellar variant of HCC. The TMA of HCC was from 19 different patients with an average age of 65.2 years (range, 49–80 years). Ten cases (53%) were from women, and 9 (47%) were from men. Nine cases (47%) were well-differentiated, 6 (32%) were moderately differentiated, and 4 (21%) were poorly differentiated.

18. Evaluation of Reticulin, Gly3, and GS in WTS and TMA Liver Lesions The staining results of reticulin, Gly3, and GS on WTS and TMA from HA, FNH, and HCC cases are presented in Table 1. The WTS HA showed intact reticulin with negative Gly3 in 100% (15 of 15) of the cases. The GS was negative in 26% (4 of 15) of the cases and perivenular in 67% (10 of 15) of the cases. One of the 15 cases (7%) was diffusely positive for GS. The TMA HA reticulin was intact in 92% (12 of 13) of the cases.

19. One of the 13 cases (8%) showed significant reticulin widening in an area of steatosis. The Gly3 was negative in 100% (13 of 13) of the TMA HA cases, whereas GS was negative in 84% (11 of 13) of the cases. One case(8%) showed perivenular GS staining, whereas another case (8%) showed diffuse GS staining. The diffuse GS staining of HA TMA and WTS was from the same patient. No significant steatosis, telangiectasia, inflammatory infiltrates, or small or large cell changes were present in that case.

20. The WTS FNH had intact reticulin in 100% (13 of 13),Gly3 negativity in 100% (13 of 13), and map like GS in 100% (13 of 13) of the cases. The TMA FNH cases had intact reticulin in 100% (13 of 13), Gly3 negativity in 100% (13 of 13), and map like GS in 85% (11 of 13) of the cases. Two of the 13 TMA FNH cases (15%) were interpreted as diffusely positive for GS.

21. The WTS HCC had reticulin loss in 94% (15 of 16), Gly3 positivity in 50% (8 of 16), and diffuse GS in 56% (9 of 16) of the cases. The TMA HCC had reticulin loss in 89% (17 of 19), Gly3 positivity in 42% (8 of 19), and diffuse GS in 79% (15 of 19) of the cases.

22. Sensitivity and Specificity of Stains in the Diagnosis of Liver Lesions Intact reticulin was 92% sensitive and 89% specific. The Gly3 negativity was 100% sensitive and 58% specific. The GS negativity (not diffuse) was 92% sensitive and 79% specific. The combination of intact reticulin with either Gly3 negativity or negative GS was 92% sensitive and 95% specific. For the distinction of TMA FNH and HA, map like GS was most useful, being 85% sensitive and 100% specific.

23. DISCUSSION The distinction between HCC, HA, and FNH on biopsy material can be difficult. Several stains have been developed to assist in these differential diagnoses. Here, we developed a triple stain of reticulin, Gly3, and GS and evaluated its utility on both TMA and WTS of liver lesions. The advantages of a triple stain, rather than using a panel of 3 individual stains, is both in the ability to compare stains directly in the same cells and in cases with very small amounts of tumor available to stain

24. Reticulin fibers are composed primarily of type III collagen bundles, which in HCC, often have greater than 4 malignant hepatocytes between fibers.26,27 The reticulin network is widened or lost in most HCC, although it is known that well-differentiated cases of HCC can have an intact reticulin framework.28,29 In both HA and FNH, the reticulin framework is intact; however, cases of steatotic liver that are not HCC are known to have focal reticulin widening.30 These areas can be challenging to distinguish with the reticulin stain, and additional stains may be helpful

25. Glypican-3, a heparan sulfate proteoglycan, is expressed in HCC in 50% to 80% of HCC, depending on the degree of differentiation.31–34 In general, poorly differentiated HCC is more commonly positive for Gly3 than is well-differentiated HCC. Hepatic adenoma and FNH are not known to express Gly3; however, malignant tumors from other organs can express Gly3.35,36

26. Glutamine synthetase is a target gene of the Wnt / b -catenin pathway, which in normal liver, stains perivenular hepatocytes.37 In FNH, GS is found in a maplike pattern38 in virtually all patients. When b -catenin is mutated and, thus, the Wnt pathway is activated, GS is upregulated.39 In HCC, approximately 80% of cases stain diffusely positive.23,40 Furthermore, in cases of HA with activation of the Wnt / b -catenin pathway, GS will diffusely stain hepatocytes.22,37

27. In the distinction of HCC from HA, reticulin was the most useful single stain. However, variant staining of reticulin was encountered. Three cases of well-differentiated HCC (2 TMA, 1 WTS) showed an intact framework, a known phenomena.29 Importantly, all 3 cases had either positive Gly3 or diffuse GS, which would support the diagnosis of HCC. Conversely, one case of TMA HA showed focal widening/loss of reticulin in an area of steatosis, which has also been described.30 Likewise, Gly3 and diffuse GS were negative, supporting the benign nature of the lesion.

28. Glypican-3 was expressed in 58% of TMA HCC, 0% of HA, and 0% of FNH, thus being 100% specific but not sensitive. Furthermore, Gly3 was expressed in 50% of WTS HCC. Our findings show greater sensitivity of Gly3 for HCC than a study performed by Lagana and colleagues23 on TMA HCC, which found 42% Gly3 positivity. The discrepancy could be related to the presence of moderately and poorly differentiated HCC cases within our TMA, which are known to have higher expression of Gly3.31

29. Glutamine synthetase was expressed diffusely in 15 of 19 TMA HCCs (79%) and 1 of 13 TMA HAs (8%), similar to other published results.22 The one HA case with diffuse GS raises the possibility of a b -catenin mutation being present. After more than 15 years of follow-up, this patient has not developed HCC, and there were no atypical features identified on the hematoxylin-eosin stain.

30. Because no one stain was entirely sensitive or specific at distinguishing HA from HCC on TMA, we evaluated combinations of intact reticulin with either Gly3 negativity or GS negativity (see Table 2). The combination of intact reticulin with either Gly3 negativity or GS negativity had a sensitivity of 92% and a specificity of 95%. Although there was a trend toward better specificity with the combination of these stains, these results were not statistically significant.

31. Usual staining patterns of liver lesions. A, Hepatic adenoma with intact reticulin and negative for glutamine synthetase and glypican-3 on tissue microarray, and on whole tissue section (B).

32. C, Focal nodular hyperplasia with intact reticulin, maplike glutamine synthetase (brown), and negative for glypican-3 on tissue microarray, and on whole tissue section (D).

33. Variant staining patterns in liver lesions. A, Hepatic adenoma with steatosis and focal reticulin widening. B, Focal nodular hyperplasia with diffuse glutamine synthetase staining

34. C, Hepatocellular carcinoma with only glutamine synthetase staining (brown). D, Hepatocellular carcinoma with only glypican-3 staining (red) (original magnification x 200)

35. A map like GS pattern was the most useful pattern to distinguish FNH from HA on TMA, as expected.25 Two FNH TMA cases showed diffuse GS staining (see Figure 2, B). The WTS of those cases showed a maplike GS pattern. This finding is likely a sampling artifact but should be considered a diagnostic pitfall. The differential diagnosis of diffuse GS staining would include b -catenin–mutated HA and HCC. Thus, diffuse GS within a liver biopsy in the absence of features suspicious for malignancy should be interpreted cautiously.

36. Overall, the expression patterns of reticulin, Gly3, and GS were similar between the TMA and the WTS of all cases supporting the utility of these stains even in biopsies. Moreover, this triple stains offers several advantages. Our triple stain allows the colocalization of reticulin loss with either Gly3 or GS positivity, which is similar to the use of P504S (racemase) and basal markers in prostate cancer.41 Furthermore, this triple stain allows the preservation of limited tissue, which could be used for potential molecular tests.

37. One potential disadvantage of this triple stain is that both Gly3 and GS are expressed in the cytoplasm. Some cases of HCC have coexpression of both proteins, potentially masking the detection of one of the proteins, which can make interpretation difficult. Therefore, caution must be used when analyzing this triple stain. Furthermore, familiarity with both the single-positive staining patterns as well as the double-positive pattern is helpful in interpretation.

38. The aberrant staining patterns seen in Figure 2, A through D, in combination with the findings of Table 2, highlight the main advantage of performing the triple stain as described here. The cases of TMA HCC with intact reticulin were positive for Gly3 (1 case) and/or GS (both cases). Conversely, the HA cases with reticulin widening were negative for both Gly 3 and diffuse GS. Thus, in all cases, careful consideration of the hematoxylin-eosin slides together with the combination of stains should help the pathologist arrive at the correct diagnosis.

39. In summary, we have developed a triple stain of reticulin, GS, and Gly3 that is useful in the distinction of HCC, HA, and FNH on biopsy specimens. The combination of reticulin loss with either Gly3 or GS positivity was useful at discriminating HCC from HA. A maplike GS pattern was useful at discriminating HA from FNH.

40. THANK YOU