Hemangioblastoma is a benign central nervous system

ORIGINAL ARTICLE PAX2( )/PAX8( )/Inhibin A(+) Immunoprofile in Hemangioblastoma: A Helpful Combination in the Differential Diagnosis With Metastatic Clear Cell Renal Cell Carcinoma to the Central Nervous System Erin M. Carney, MD,* Priya Banerjee, MD,* Carla L. Ellis, MD, MS,* Roula Albadine, MD,* Rajni Sharma, PhD,* Alcides M. Chaux, MD,* Peter C. Burger, MD,* and George J. Netto, MD*w z Background: Hemangioblastomas account for up to 2.5% of all intracranial tumors. They may occur sporadically or as a part of the multisystem genetic syndrome of Von Hippel-Lindau syndrome (VHL). Patients with VHL are also at an increased risk of developing clear cell renal cell carcinoma (ccrcc). Distinguishing hemangioblastomas from metastatic ccrcc to the central nervous system (CNS) can be challenging at times when based solely on hematoxylin and eosin-stained sections. We propose an immunohistochemistry (IHC) panel of combination of PAX2, PAX8, and inhibin A as a helpful approach in distinguishing the 2 lesions. Design: Archival tissues from 20 hemangioblastomas and 16 ccrccs metastatic to the CNS were retrieved from our surgical pathology files (2001 to 2010). IHC for PAX2, PAX8, and inhibin A was performed on routine or tissue microarray sections using standard IHC protocol. The intensity of nuclear staining was evaluated for each marker and was assigned an incremental 0, 1+, 2+, and 3+ score. The extent of staining was categorized as focal (<25%), multifocal (25% to 75%), or diffuse (>75%). Result: (1) Hemangioblastoma: The Von Hippel-Lindau syndrome was diagnosed in 4 of 16 (25%) patients, 2 of whom developed multiple hemangioblastomas. All 20 (100%) hemangioblastomas were positive for inhibin A (cytoplasmic). The staining intensity was moderate or strong (2+ or 3+) in 19 cases (95%), all of which were multifocal or diffuse in extent. Nuclear PAX2 staining was present in 1 of 19 evaluable lesions (5%), whereas PAX8 staining was not present in any of the 20 examined lesions. (2) Metastatic ccrcc to the CNS: Fourteen of 16 (88%) examined ccrccs were positive for PAX2, whereas 15 of 16 (94%) lesions showed PAX8 staining. None of 16 (0%) examined ccrccs were positive for inhibin A. From the Departments of *Pathology; wurology; and zoncology, Johns Hopkins Hospital, Baltimore, MD. This study was presented in part at the United States and Canadian Academy of Pathology 2009 meeting in Boston. Correspondence: George J. Netto, MD, 401 N. Broadway/Room 2242, Baltimore, MD 21231-2410 (e-mail: gnetto1@jhmi.edu). Copyright r 2011 by Lippincott Williams & Wilkins Conclusions: We propose the use of the combination of PAX2, PAX8, and inhibin A as a helpful ancillary IHC panel to resolve the differential diagnosis of hemangioblastoma versus metastatic ccrcc. The immunoprofile of PAX2(+) or PAX8(+) and inhibin A( ) supports the diagnosis of metastatic ccrcc with a sensitivity of 94%, specificity of 100%, and positive predictive value of 100%. The PAX2( ), PAX8( ), and inhibin A(+) profile supports the diagnosis of hemangioblastoma with a sensitivity of 95%, specificity of 100%, and positive predictive value of 100%. Key Words: hemangioblastoma, metastatic clear cell renal cell carcinoma, PAX2, PAX8, inhibin A (Am J Surg Pathol 2011;35:262 267) Hemangioblastoma is a benign central nervous system (CNS) neoplasm, which represents 1.5% to 2.5% of all intracranial tumors and 7% to 12% of posterior fossa tumors. Although it most commonly arises in the cerebellum, hemangioblastoma may also arise in the cerebral hemispheres, medulla, and spinal cord with a peak incidence in the third to sixth decades. Approximately 75% of hemangioblastomas are associated with sporadic mutations involving the Von Hippel-Lindau (vhl) gene located at the short arm of chromosome 3 (3p25-26), whereas the remaining 25% occur in the setting of the Von Hippel-Lindau syndrome. The latter is an autosomal dominant genetic syndrome caused by a germline mutation in the vhl tumor suppressor gene. Individuals with the VHL syndrome are at risk of developing clear cell renal cell carcinoma (ccrcc) and hemangioblastoma in addition to pancreatic and renal cysts and cystadenomas of the epididymis, and endolymphatic sac. PAX2 and PAX8 are 2 transcription factors that are normally expressed by cells of the nephric, thyroid, and Mullerian duct lineage. 9,12,16 Inhibin A is a 32-kd dimeric glycoprotein, which is normally secreted by ovarian granulosa cells and testicular Sertoli and Leydig cells inhibiting the pituitary follicle-stimulating hormone. 262 www.ajsp.com Am J Surg Pathol Volume 35, Number 2, February 2011

Am J Surg Pathol Volume 35, Number 2, February 2011 PAX2, PAX8, and Inhibin A Combination in Hemangioblastoma Inhibin A is used as a marker of ovarian and testicular sex cord differentiation 1 and has been previously shown to be expressed by hemangioblastomas. 5 We examined the use of the combination of all 3 markers in distinguishing hemangioblastoma from CNS metastasis of clear cell renal cell carcinoma. MATERIALS AND METHODS The study was approved by the Johns Hopkins Institutional Review Board. Patient Cohort Electronic surgical pathology records at our hospital were retrospectively searched for all cases with the diagnosis of hemangioblastoma treated at our hospital in the period between 2001 and 2010. A total of 23 hemangioblastoma cases were identified. Archival sections were available in 20 cases that were included in this study. In addition, 17 ccrccs metastatic to the CNS were identified in the same period. Archival blocks were available for the study in 16 of the ccrcc cases. All sections were reviewed by 2 pathologists (G.J.N. and E.M.C.) on the study confirming the original diagnosis. All pertinent clinical information was obtained from the hospital electronic medical records. Immunohistochemical Analysis Representative formalin-fixed, paraffin-embedded sections were selected for immunohistochemical analysis. Routine 5-mm unstained sections were used for immunohistochemical analysis on all cases with the exception of 3 of 16 metastatic ccrccs that were represented on an earlier constructed tissue microarray (TMA) for metastatic RCC. TMA construction was performed using a Beecher Instrument (Silver Spring, MD) as detailed earlier. 4 Each RCC tumor was spotted in triplicate as 1.6-mm diameter cores. Four-micrometer TMA sections were used for immunostains. Antibody clones and the commercial sources are summarized in Table 1. PAX2 and PAX8 Immunohistochemical studies were carried out using an automated Bond max-leica autostainer from Leica Microsystems (Bannockburn, IL). The sections were deparaffinized, hydrated, and subjected to heatinduced antigen retrieval for 20 minutes with a high ph ER2 solution. For PAX2, Rabbit polyclonal anti-pax2 (Invitrogen, Camarillo, CA) was applied for 45 minutes at room temperature followed by the application of biotinfree bond polymer refine detection. For PAX8, Rabbit polyclonal anti-pax8 (ProteinTech Group, Chicago, IL) was applied for 15 minutes at room temperature followed by the application of biotin-free bond polymer refine detection. 3-3 0 -Diaminobenzidine chromogen was used to visualize the reactions (all reagents from Leica Microsystems) and then counterstained with hematoxylin. Inhibin A Immunohistochemical studies were also carried out using the same Bond-Leica autostainer. The tissue sections were baked for 20 minutes at 761C, deparaffinized, and hydrated. Heat-induced antigen retrieval was performed for 20 minutes with a high ph ethylenediaminetetraacetic acid retrieval buffer. Mouse antihuman inhibin a (AbD Serotech, Martinsreid, Germany) was applied for 15 minutes followed by the application of the bond polymer detection system (Leica Microsystems). Reaction visualization was achieved similarly to that of PAX2 and PAX8. Appropriate positive and negative controls were used for all 3 markers. Acceptable positive expression patterns were cytoplasmic for inhibin A and nuclear for PAX2 and PAX8. The intensity of nuclear staining was evaluated for each marker and was assigned an incremental 0, 1+, 2+, and 3+ score. The extent of staining was categorized as focal (<25%), multifocal (25% to 75%), or diffuse (>75%).The sensitivity and specificity of each marker and the diagnostic positive predictive values of their combination were subsequently determined in both types of tumors. RESULTS Cohort Characteristics Hemangioblastoma In our cohort of 16 patients with 20 lesions, the mean patient age at the time of surgery was 40.1 years (range, 16 to 68 y) and the mean tumor size was 2.8 cm (range, 0.8 to 5.9 cm). The male to female patient ratio was 4:1. The ethnic heritage of the patients was 12 white, 3 black, and 1 patient was of middle-eastern origin. The majority (12 of 20) of the lesions were sporadic. Four (25%) patients had a documented diagnosis of the VHL syndrome, and 2 of these patients had 2 and 4 lesions each. Eight lesions were located within the spinal cord, 6 lesions were located in the cerebellum, 3 lesions were located in the medulla, and 1 lesion each was located in the cerebellopontine angle, vertebra, and the fourth ventricle/posterior fossa. TABLE 1. Commercial Sources and Characteristics of Inhibin A, PAX2, and PAX8 Antibodies Name Clone/Animal Species Dilution Retrieval AB Incubation Time (min) Source Anti-inhibin a Clone R1/mouse monoclonal 1:25 ER2 (high ph) 15 AbD Serotech Martinsried, Germany Anti-PAX2 Polyclonal/rabbit 1:100 ER2 (high ph) 45 Invitrogen, Camarillo, CA Anti-PAX8 Polyclonal/rabbit 1:100 ER2 (high ph) 15 Protein tech group, Chicago, IL r 2011 Lippincott Williams & Wilkins www.ajsp.com 263

Carney et al Am J Surg Pathol Volume 35, Number 2, February 2011 TABLE 2. Immunohistochemical Expression of Inhibin A, PAX2, and PAX8 in Hemangioblastoma and Metastatic Clear Cell Renal Cell Carcinoma to the Central Nervous System PAX2 PAX8 Inhibin A Tumor Type Positive (%) Negative (%) Positive (%) Negative (%) Positive (%) Negative (%) Hemangioblastoma 1/19 (5) 18/19 (95) 0/20 (0) 20/20 (100) 20/20 (100) 0/20 (0) ccrcc 14/16 (88) 2/16 (12) 15/16 (94) 1/16 (6) 0/16 (0) 16/16 (100) ccrcc indicates metastatic clear cell renal cell carcinoma to the CNS. Metastatic ccrcc At the time of surgery, the mean patient age in our group of 16 metastatic ccrccs was 58.7 years (range, 41 to 80 y). The mean tumor size was 3.1 cm (range, 0.9 to 6.5 cm). The male to female patient ratio was 2:1. Twelve of the patients with ccrcc were white, and 3 patients were black. All the lesions were sporadic in origin, with none of the patients having a history of the VHL syndrome. One patient experienced 2 separate metastases to the brain at a 4-year interval. Thirteen lesions were located in the cerebral lobes, and 3 lesions were located in the cerebellum. Immunohistochemical Findings The immunohistochemical results are summarized in Tables 2, 3, and 4 and are illustrated in Figures 1 and 2. PAX2/PAX8 Nuclear PAX2 expression was present in 1 of 19 (5%) evaluable hemangioblastomas and absent in all remaining 18 cases. Nuclear PAX8 expression was absent in all 20 (0%) examined cases including the 1 hemangioblastoma that was positive for PAX2. Overall, 14 of 16 (88%) metastatic ccrccs were positive for PAX2 and 15 of 16 (94%) metastatic ccrccs expressed PAX8. Fourteen ccrccs that were positive for PAX2 were also positive for PAX8. One ccrcc was negative for both markers, whereas an additional ccrcc was negative for PAX2 but positive for PAX8. PAX2 expression in ccrcc was diffuse in 8 cases, multifocal in 3 cases, and focal in 3 cases. The intensity of staining ranged from 1+ to 3+. PAX8 expression in ccrcc was diffuse in 13 lesions, multifocal in 2 lesions, and none of the lesions showed focal expression. The intensity of staining ranged from 2+ to 3+. Inhibin A All tested hemangioblastomas (20 of 20; 100%) expressed inhibin A. None of the 16 tested ccrccs were positive for inhibin A (0%). Inhibin A expression was diffuse in 18 lesions, multifocal in 1 lesion, and focal in 1 lesion. The intensity of staining ranged from 1+ to 3+. Combination of PAX2/PAX8/Inhibin A With the exception of 1, all 19 hemangioblastomas that were evaluable for all 3 markers were negative for PAX8, negative for PAX2, and positive for inhibin A. The 1 exception to the latter profile was a case that was negative for PAX8 and positive for PAX2 and inhibin A. Therefore, the immunoprofile of PAX2 and PAX8 negative and inhibin A positive supported the diagnosis of hemangioblastoma with a sensitivity of 95%, a specificity of 100%, and a positive predictive value of 100%. Of the 16 cases of metastatic ccrcc that were evaluable for all 3 markers, 14 cases were positive for PAX2 or PAX8 and negative for inhibin A giving the PAX2 or PAX8 negative and inhibin A positive immunoprofile a sensitivity of 94%, specificity of 100%, and a positive predictive value of 100% for metastatic ccrcc. DISCUSSION Retinal angiomas and CNS hemangioblastomas are some of the earliest manifestations of the VHL syndrome. TABLE 3. Intensity and Extent of Immunohistochemical Staining of Inhibin A, PAX2, and PAX8 in Hemangioblastoma and Metastatic Clear Cell Renal Cell Carcinoma Hemangioblastoma ccrcc Antibody Category Intensity (Extent) Intensity Extent Intensity Extent PAX2 Negative 18 (95) 18 (95) 2 (12) 2 (12) 1 (focal) 0 (0) 0 (0) 1 (7) 3 (19) 2 (multifocal) 0 (0) 1 (5) 5 (31) 3 (19) 3 (diffuse) 1 (5) 0 (0) 8 (50) 8 (50) PAX8 Negative 20 (100) 20 (100) 1 (7) 1 (6) 1 (focal) 0 (0) 0 (0) 0 (0) 0 (0) 2 (multifocal) 0 (0) 0 (0) 5 (31) 2 (13) 3 (diffuse) 0 (0) 0 (0) 10 (62) 13 (81) Inhibin A Negative 0 (0) 0 (0) 16 (100) 16 (100) 1 (focal) 1 (5) 1 (5) 0 (0) 0 (0) 2 (multifocal) 2 (10) 1 (5) 0 (0) 0 (0) 3 (diffuse) 17 (85) 18 (90) 0 (0) 0 (0) 264 www.ajsp.com r 2011 Lippincott Williams & Wilkins

Am J Surg Pathol Volume 35, Number 2, February 2011 PAX2, PAX8, and Inhibin A Combination in Hemangioblastoma TABLE 4. Sensitivity, Specificity, and Positive Predictive Value of PAX8, PAX2, Inhibin A, and PAX2/PAX8/Inhibin A Immunostaining in the Diagnosis of Hemangioblastoma Versus Metastatic Clear Cell Renal Cell Carcinoma to the Central Nervous System Test Sensitivity (%) Specificity (%) PPV (%) PAX2 for ccrcc 88 95 94 PAX8 for ccrcc 94 100 100 PAX2(+) or PAX8(+), and inhibin A( ) for ccrcc 94 100 100 Inhibin A(+) for hemangioblastoma 100 100 100 PAX2( ) and PAX8( ) and inhibin A(+) for hemangioblastoma 95 100 100 PPV indicates positive predictive value. The diagnosis of VHL can be made clinically when either type of these lesions is found in the presence of an extracranial tumor or in the context of a family history of VHL. Given the significantly increased risk of developing multiple, often bilateral ccrcc in patients with VHL, the question of whether a posterior fossa tumor found simultaneously with a renal mass represents a metastatic renal cell carcinoma is not completely uncommon in this setting. Similarly, finding a CNS mass during the clinical workup of a sporadic renal mass could bring the same differential diagnosis. Although subtle morphologic differences, such as the differing cytoplasmic characteristics of the neoplastic cells, amount of reticulin, and the MIB-1 index can help distinguish hemangioblastoma from metastatic ccrcc, 2 such distinction can be challenging at times due to the fact that both lesions are richly FIGURE 1. Hemangioblastoma showing characteristic richly vascularized proliferation of polygonal cells with clear-to-foamy cytoplasm with ill-defined cytoplasmic borders (A; hematoxylin and eosin stains; magnification, 200). Negative immunoreactivity for PAX2 (B; magnification, 200) and PAX8 (C; magnification, 200) is illustrated. Diffuse, 3+ positive cytoplasmic expression of inhibin A is depicted (D; magnification, 200). r 2011 Lippincott Williams & Wilkins www.ajsp.com 265

Carney et al Am J Surg Pathol Volume 35, Number 2, February 2011 FIGURE 2. Metastatic clear cell renal cell carcinoma to the cerebellar region (A; hematoxylin and eosin stain; magnification, 200). Diffuse, 3+ positive nuclear expression of PAX2 (B; magnification, 200) and diffuse 3+ positive nuclear expression of PAX8 (C; magnification, 200) is shown. The tumor was negative for inhibin A (D; magnification, 200). vascularized and are composed of clear cells. The clinical implications of making the proper distinction are crucial in terms of management, staging, and prognostication. Inhibin A, a glycoprotein produced by ovarian granulosa cells and testicular Sertoli and Leydig cells, 1 has been previously shown to be expressed by hemangioblastomas as well, and has been shown to be of some use in differentiating such tumors from RCC when combined with renal markers, such as RCC and CD10. Ingold et al 6 found RCC monoclonal antibodies to have variable sensitivity for detecting ccrcc at 77% and 66%, with only about 50% of ccrcc metastases to the brain showing RCC expression, whereas none of the hemangioblastoma cases in this series was labeled with RCC. In their study, Jung and Kuo 8 showed CD10 expression in all 5 examined cases (100%) of ccrcc metastases to the CNS, but none of their 22 cases of hemangioblastoma showed CD10 expression. In the same study, however, inhibin A was found to be expressed in 91% of hemangioblastomas and in 60% of ccrcc metastases to the CNS, calling into question the added diagnostic value of inhibin A immunohistochemistry in this setting. Finally, Weinbreck et al 17 using a panel of aquaporin 1, inhibin A, D2-40, cytokeratin AE1/AE3, epithelial membrane antigen, and CD10 also found inhibin A to be a helpful positive marker for hemangioblastoma albeit with slightly lower sensitivity than aquaporin 1 (88% vs 97%, respectively). Our group and other researchers have recently shown the superior sensitivity and specificity of newer markers of the renal lineage, such as PAX2 and PAX8, compared with traditional markers, such as CD10 and RCC. 3,11,13,16 In this study, we showed a panel of PAX2/PAX8/ inhibin A markers to be of great use in the differential diagnosis of ccrcc and hemangioblastoma. An immunoprofile of PAX2( ), PAX8( ), and inhibin A(+) was highly sensitive (95%) and specific (100%) for hemangioblastoma in our group of 20 lesions. Our findings are in line with a recent study by Rivera et al 14 pointing at PAX2 and inhibin A as the best 2 markers among a panel of 6 markers that also included D2-40, CD10, FLI-1, and 266 www.ajsp.com r 2011 Lippincott Williams & Wilkins

Am J Surg Pathol Volume 35, Number 2, February 2011 PAX2, PAX8, and Inhibin A Combination in Hemangioblastoma RCC in differentiating ccrcc from hemangioblastomas. It is interesting to note the slightly better performance of inhibin A in this study compared with the study by Rivera et al. 14 The latter group found inhibin A to be positive in 87% of their hemangioblastoma cases compared with 100% of our cases, but also found it to be expressed in 12% of their metastatic ccrcc, whereas none of our ccrcc was positive for inhibin A. The difference in performance can be potentially ascribed to the different sources of the antibodies used in both studies. In this study, the addition of PAX8 as a second renal cell carcinoma marker offers an increase in the sensitivity for metastatic ccrcc from 88% for PAX2(+) and inhibin A( ) profile (results not shown) to 94% for a PAX2(+) or PAX8(+) and inhibin A( ) profile without sacrificing specificity or positive predictive value with both remaining at 100%. Finally, it is important to keep in mind that rare metastatic tumors to the CNS other than RCC can express PAX8 and/or PAX2 transcription factors and at times show focal cytoplasmic clearing. These primarily include thyroid carcinoma, serous and clear cell ovarian carcinoma, fallopian tube carcinoma, and more recently well-differentiated neuroendocrine tumors, including pancreatic islet cell tumors. 7,10,15 Attention to other distinguishing morphologic and immunohistochemical features of such tumors should facilitate their distinction from metastatic ccrcc and primary hemangioblastoma. REFERENCES 1. Anderson RA, Cambray N, Hartley PS, et al. Expression and localization of inhibin alpha, inhibin/activin betaa and betab and the activin type II and inhibin beta-glycan receptors in the developing human testis. Reproduction. 2002;123:779 788. 2. Burger PC, Scheithauer BW. Armed Forces Institute of Pathology Atlas of Tumor Pathology Series 4, Fascicle 7: Tumors of the Central Nervous System. Washington, D.C.: The American Registry of Pathology in collaboration with AFIP; 2007. 3. Daniel L, Lechevallier E, Giorgi R, et al. Pax-2 expression in adult renal tumors. Hum Pathol. 2001;32:282 287. 4. Fedor HL, De Marzo AM. Practical methods for tissue microarray construction. Methods Mol Med. 2005;103:89 101. 5. Hoang MP, Amirkhan RH. Inhibin alpha distinguishes hemangioblastoma from clear cell renal cell carcinoma. Am J Surg Pathol. 2003;27:1152 1156. 6. Ingold B, Wild PJ, Nocito A, et al. Renal cell carcinoma marker reliably discriminates central nervous system hemangioblastoma from brain metastases of renal cell carcinoma. Histopathology. 2008;52:674 681. 7. Jadallah S, Albadine R, Sharma R, et al. PAX8 expression in clear cell, papillary, and chromophobe RCC and urothelial carcinoma of the renal pelvis. Mod Pathol. 2009;23:174A. 8. Jung SM, Kuo TT. Immunoreactivity of CD10 and inhibin alpha in differentiating hemangioblastoma of central nervous system from metastatic clear cell renal cell carcinoma. Mod Pathol. 2005;18: 788 794. 9. Lang D, Powell SK, Plummer RS, et al. PAX genes: roles in development, pathophysiology, and cancer. Biochem Pharmacol. 2007; 73:1 14. 10. Long KB, Srivastava A, Hirsch MS, et al. PAX8 expression in welldifferentiated pancreatic endocrine tumors: correlation with clinicopathologic features and comparison with gastrointestinal and pulmonary carcinoid tumors. Am J Surg Pathol. 2010;34:723 729. 11. Mazal PR, Stichenwirth M, Koller A, et al. Expression of aquaporins and PAX-2 compared to CD10 and cytokeratin 7 in renal neoplasms: a tissue microarray study. Mod Pathol. 2005;18: 535 540. 12. Narlis M, Grote D, Gaitan Y, et al. PAX2 and PAX8 regulate branching morphogenesis and nephron differentiation in the developing kidney. J Am Soc Nephrol. 2007;18:1121 1129. 13. Ozcan A, Zhai Q, Javed R, et al. PAX-2 is a helpful marker for diagnosing metastatic renal cell carcinoma: comparison with the renal cell carcinoma marker antigen and kidney-specific cadherin. Arch Pathol Lab Med. 2010;134:1121 1129. 14. Rivera AL, Takei H, Zhai J, et al. Useful immunohistochemical markers in differentiating hemangioblastoma versus metastatic renal cell carcinoma. Neuropathology. 2010;30:580 585. 15. Robson EJD, He SJ, Eccles MR. A PANorama of PAX genes in cancer and development. Nat Rev Cancer. 2006;6:52 62. 16. Tong GX, Yu WM, Beaubier NT, et al. Expression of PAX8 in normal and neoplastic renal tissues: an immunohistochemical study. Mod Pathol. 2009;22:1218 1227. 17. Weinbreck N, Marie B, Bressenot A, et al. Immunohistochemical markers to distinguish between hemangioblastoma and metastatic clear-cell renal cell carcinoma in the brain: utility of aquaporin1 combined with cytokeratin AE1/AE3 immunostaining. Am J Surg Pathol. 2008;32:1051 1059. r 2011 Lippincott Williams & Wilkins www.ajsp.com 267