OUP user menu

An Assessment of the Usefulness of Immunohistochemical Stains in the Diagnosis of Hairy Cell Leukemia

Michael J. Sherman MD, Curtis A. Hanson MD, James D. Hoyer MD
DOI: http://dx.doi.org/10.1309/AJCP5GE1PSBMBZTW 390-399 First published online: 1 September 2011


Annexin-1 and T-bet are recently described immunohistochemical stains that reportedly assist in the diagnosis of hairy cell leukemia (HCL). Our objective was to assess the sensitivity and specificity of a panel of immunohistochemical stains in distinguishing HCL from other B-cell neoplasms, particularly splenic and extranodal marginal zone lymphomas (SMZL and ENMZL, respectively). The study included 234 bone marrow biopsy specimens: 101 HCL, 13 SMZL, and 10 ENMZL cases were assessed with CD20, tartrate-resistant acid phosphatase (TRAP), DBA.44, a-1, T-bet, and cyclin D1, and 110 control cases were assessed with annexin-1 and T-bet. Our study showed that annexin-1 is a specific and sensitive marker for HCL; however, interpretation is limited by positivity within myeloid precursors. T-bet, DBA.44, and TRAP immunohistochemical stains lack sufficient sensitivity and specificity to differentiate HCL from either form of marginal zone lymphoma. However, our data suggest that the addition cyclin D1 to the immunostaining panel will increase the sensitivity and specificity of HCL diagnosis.

Key Words:
  • Hairy cell leukemia
  • Immunohistochemistry
  • Annexin-1
  • T-bet
  • Cyclin-D1
  • Marginal zone lymphoma

Hairy cell leukemia (HCL) is a rare disease of activated mature B cells that requires correlation of morphologic features, cytochemical findings, flow cytometric studies, and immunohistochemical staining as part of the diagnostic evaluation. HCL is usually a straightforward diagnosis owing to the characteristic lymphocyte cytologic features, immunophenotypic pattern, tartrate-resistant acid phosphatase (TRAP) positivity, and the histopathologic pattern of infiltration on the bone marrow (BM) biopsy specimen.13 However, in some cases, it can be difficult to make the diagnosis of HCL, particularly when there is no BM aspirate material available and the number of circulating malignant cells is low. In some cases, overlapping morphologic and immunophenotypic features can make it difficult to separate HCL from other B-cell chronic lymphoproliferative disorders (B-CLPDs), particularly splenic marginal zone lymphoma (SMZL), which can have a similar immunophenotype (including some cases with CD103 expression in flow cytometric studies), cytologic appearance, and varying degrees of TRAP positivity. Immunohistochemical analysis of HCL is helpful in these cases, but it historically has been limited to CD20 (a pan–B-cell marker), anti-TRAP, and DBA.44; the latter 2 antibodies are commonly positive in cases of HCL but will also show expression in other B-CLPDs.1,2,47

Other immunohistochemical stains have become available that have been purported to be useful in the diagnosis of HCL. Annexin-1 is a protein that serves a complex role in the inflammatory response and was described by Falini et al,8 using gene expression profiling, as one of several proteins uniquely up-regulated in HCL.9 T-bet is a T-box transcription factor that was originally described in Th1+ T-helper cells and was subsequently found to be expressed in B cells, possibly related to immunoglobulin class-switching. As described by Dorfman et al,10 T-bet is expressed in several lymphoproliferative disorders, notably including HCL. Cyclin D1, which has been extensively studied in mantle cell lymphomas and plasma cell neoplasms, has also been shown in previous studies to be positive in a variable number of cases of HCL.5,11 Although characterization of HCL by CD20, DBA.44, and TRAP is well established,1,2 the sensitivity and specificity of these newer markers has not been addressed in a large study, and their positivity in a wide variety of other hematologic and nonhematologic disorders is largely unknown.

Our study was designed to investigate the diagnostic usefulness of annexin-1 and T-bet in HCL by determining the sensitivity and specificity of these antibodies in a wide variety of hematologic disorders. We also attempted to find the optimal set of immunohistochemical stains that were needed to differentiate HCL from other B-CLPDs, particularly the marginal zone lymphomas (MZLs).

Materials and Methods

A total of 234 BM studies were selected from the Mayo Clinic files from a 113-month period. All HCL cases collected during this period with material available for review were included, all SMZL and extranodal MZL (ENMZL) cases collected during the latter 64-month period were included, and control cases were chosen as random representatives of various diseases during this period. All BM studies included peripheral blood and BM aspirate smears stained by Wright-Giemsa per standard methods. BM biopsies were decalcified using a 20% formic acid solution, fixed using B-5, embedded in paraffin, and stained with H&E per standard methods. All immunohistochemical stains on BM biopsy specimens were performed on a DAKO Autostainer (DAKO, Carpinteria, CA) on 4-μm sections. Antibodies used in this study included CD20 (clone L26, DAKO), TRAP (clone 26E5, Novocastra, Newcastle upon Tyne, England), DBA.44 (clone DBA.44, DAKO), annexin-1 (clone 29, BD Transduction Laboratories, San Diego, CA), T-bet (clone 4B10, Santa Cruz Biotechnologies, Santa Cruz, CA), and cyclin D1 (clone SP4, Thermo Fisher Scientific, Fremont, CA).

Immunohistochemical stains for CD20, TRAP, DBA-44, annexin-1, T-bet, and cyclin D1 were evaluated in 101 HCLs, 13 SMZLs, and 10 ENMZLs. CD20 was used to determine the extent of BM involvement by the HCL, SMZL, and ENMZL. The remaining immunohistochemical stains were compared with CD20 to assess the range of antigen expression within the neoplastic process. All HCLs were diagnosed using a combination of blood/aspirate smear cytologic studies, BM biopsy histologic studies, flow cytometric immunophenotype, TRAP studies, and cytochemical and immunohistochemical stains. All SMZL and ENMZL cases were diagnosed and classified on non-BM tissue. In addition, a series of 110 control cases was evaluated that included a wide variety of neoplastic and nonneoplastic hematologic diseases and metastatic malignancies. These latter diagnoses are listed in Table 1 and were evaluated only with annexin-1 and T-bet antibodies.

Staining for all immunohistochemical antibodies was graded as positive (strong staining involving a majority of cells), weakly positive (pale or diminished staining involving a majority of cells), focally positive (strong staining involving a minority of cells), weakly and focally positive (pale or diminished staining involving a minority of cells), and negative. Cases considered weakly, focally, or weakly and focally positive were considered with the positive cases for the purposes of data analysis.


The results of the immunohistochemical studies for annexin-1 and T-bet in the control cases are summarized in Table 1. Annexin-1 positivity was typically seen as a cytoplasmic and nuclear staining pattern. Annexin-1 stained the normal or malignant granulocytic component in all cases, including normal BM samples and all cases of acute myeloid leukemia, myelodysplastic syndromes, and myeloproliferative neoplasms. Of the lymphoproliferative disorders, annexin-1 was positive in all cases of precursor T-lymphoblastic leukemia and in 2 cases of lymphoplasmacytic lymphoma. In addition, annexin-1 positivity was seen in 1 case of metastatic adenocarcinoma and 1 case of metastatic melanoma Image 1.

View this table:
Table 1
Image 1

A and B, A representative case of acute myeloid leukemia. C and D, A representative case of metastatic adenocarcinoma. E and F, A case of metastatic melanoma. These cases show cytoplasmic and nuclear staining that defines positivity for annexin-1. Background staining of normal myeloid cells and connective tissue is also noted (A, C, and E, H&E; B, D, and F, annexin-1; AF, oil immersion ×40).

As expected, T-bet positivity was present only within the nucleus. T-bet was positive in all mature T-cell lymphoproliferative disorders but was absent in the precursor T-lymphoblastic leukemias. T-bet was also positive in all 3 chronic lymphocytic leukemias, 1 large B-cell lymphoma, 2 Hodgkin lymphomas, and in 2 BM samples with metastatic adenocarcinoma.

The comparison of antigen expression by immunohistochemical staining in HCL, SMZL, and ENMZL is summarized in Table 2. CD20 was positive in all cases and was used primarily to determine extent of BM involvement. Anti-TRAP immunohistochemical staining was positive in all cases of HCL, but was also identified in all cases of SMZL and in a majority of ENMZL cases. The strength of cytoplasmic TRAP immunohistochemical positivity in SMZL was equivalent to that seen in the HCL specimens. As previously described, DBA.44 stained a subpopulation of lymphocytes in 96.0% of cases of HCL, but also stained a lymphoid population in approximately half of the SMZL specimens.

Annexin-1 stained 96% of cases of HCL (95 positive/99 tested) and was absent in all cases of SMZL and ENMZL. However, assessment of annexin-1 positivity was difficult when the tumor burden was low because it was extremely difficult to distinguish the positivity in the HCL infiltrate from the reactivity that was present in the background normal granulocytic component. T-bet showed immunohistochemical positivity in a significant number of HCL, SMZL, and ENMZL cases.

It is interesting that almost all cases of HCL demonstrated varying degrees of positivity with cyclin D1. The strength of nuclear positivity with cyclin D1 was variable; the majority of cases showed weak or focal positivity with only a minority of cases showing uniform and strong positivity. Only 1 case of SMZL showed cyclin D1 positivity, and none of the ENMZL cases were positive for cyclin D1 Image 2, Image 3, and Image 4.


The diagnosis of HCL ideally requires a multifaceted and correlative approach looking at blood and BM aspirate smear cytologic features, BM biopsy histologic features, TRAP cytochemical results, immunophenotyping by flow cytometry, and immunohistochemical staining. However, it is not uncommon for the BM biopsy to be the primary diagnostic material as the number of circulating cells can be very minimal and BM fibrosis prevents both adequate cytologic review and flow cytometric immunophenotyping. In these latter situations, the evaluation of the BM biopsy specimen relies heavily on immunohistochemical stains, especially in cases with minimal or focal BM involvement. The historic combination of CD20, TRAP, and DBA.44 has been shown to have diagnostic usefulness, although it is not entirely specific for HCL,1,5,6,12 with TRAP and DBA.44 positivity being present in some cases of SMZL and ENMZL. Thus, there has clearly been a need to expand this basic panel so as to improve the sensitivity and specificity of immunohistochemical staining in adequately separating HCL cases from non-HCL disorders.

Separating HCL from these diseases has more than simple academic interest. As medical therapies develop and evolve, it is critical to isolate diseases as pure entities to ensure that patients are treated correctly and that the efficacy of the treatments can be appropriately assessed. In the case of HCL, the advent of purine analog medications (cladribine and pentostatin) has provided a particularly effective therapy. Because the usefulness of these therapies has not been fully assessed in MZLs, it remains a diagnostic and therapeutic necessity to separate MZLs from HCL to appropriately treat patients.3,13

View this table:
Table 2
Image 2

A representative case of hairy cell leukemia (HCL) showing classic H&E-stained morphologic features (A), strong membranous CD20 staining (B), cytoplasmic tartrate-resistant acid phosphatase (TRAP) immunohistochemical staining (C), variably strong nuclear T-bet staining (D), strong nuclear and cytoplasmic annexin-1 staining (E), and prominent nuclear cyclin D1 staining (F). Other cases of HCL showed variable staining patterns for TRAP immunohistochemical staining, T-bet, annexin-1, and cyclin D1 with the lower end of positivity being focal and weak (AF, oil immersion ×40).

Image 3

Two cases of hairy cell leukemia (HCL). H&E (A), CD20 (B), and annexin-1 (C) stains from 1 representative case of HCL that was negative for annexin-1. H&E (D), CD20 (E), and T-bet (F) stains from 1 representative case of HCL that was negative for T-bet (AF, oil immersion ×40).

Annexin-1 and T-bet have been promulgated as worthwhile additions to the immunohistochemical repertoire in the diagnosis of HCL. Studies by Jöhrens et al12 and Falini et al8 that initially addressed the usefulness of these stains showed promising results with the suggestion that annexin-1 and T-bet greatly enhanced the diagnostic sensitivity and specificity of an immunohistochemical panel for HCL. These initial studies suggested that annexin-1 achieved 100% specificity and sensitivity and that T-bet achieved 100% sensitivity for HCL. As these immunohistochemical stains have become more prominently used in clinical practice, we believed it was necessary to do a large study comparing the results for HCL with the results for B-CLPDs that can have morphologic or immunophenotypic overlap with HCL and to understand their expression in a wide variety of BM diseases not previously pursued. Our current study attempted to comprehensively analyze annexin-1 and T-bet within a larger panel of immunohistochemical stains to determine their usefulness across a wide spectrum of BM diseases.

Our study reveals, consistent with immunohistochemical studies for most disease processes, that no single immunohistochemical stain achieves 100% sensitivity and specificity in HCL. However, practicing hematopathologists have a large cadre of immunohistochemical stains available that can be used in the evaluation of a possible HCL. These stains, which include TRAP, T-bet, DBA.44, annexin-1, and cyclin D1, all demonstrate a high degree of sensitivity with more than 90% of HCL cases expressing antigens recognized by these antibodies. Even though TRAP and DBA.44 have long been considered critical stains for the diagnosis of HCL, our data clearly show that both lack absolute specificity and sensitivity.

Image 4

A representative case of splenic marginal zone lymphoma showing classic H&E-stained morphologic features (A), strong membranous CD20 staining (B), cytoplasmic tartrate-resistant acid phosphatase immunohistochemical staining (C), variably strong nuclear T-bet staining (D), absent annexin-1 staining in the neoplastic cells (E), and absent cyclin D1 staining (F). G, CD20 staining of tumor cells revealing an intrasinusoidal growth pattern (A–F, oil immersion ×40; and G, oil immersion ×100).

Annexin-1 has been reported to have high sensitivity for HCL and up to 100% specificity.8 We also found that annexin-1 showed high sensitivity (96%) for HCL and was the only antibody that did not show staining in any SMZL or ENMZL cases, thus displaying 100% specificity for HCL when only these 3 entities were considered. However, annexin-1 has limited usefulness owing to the high level of background staining involving normal granulocytic elements. We could not interpret 2% of de novo HCL cases, 23% of SMZL cases, and 20% of ENMZL cases for annexin-1 owing to low tumor burden relative to the high background staining seen in the normal BM precursors. Annexin-1 also showed positive staining in many of the control cases, which confirmed the limited usefulness of this antibody when a broad differential diagnosis is being considered. Thus, annexin-1 must be used hesitantly and carefully when the lineage of the neoplasm is uncertain or when the disease process only minimally involves the BM. Clearly, annexin-1 has a limited or no role in evaluating follow-up HCL cases with minimal BM involvement.

T-bet, as a relatively newer antibody, has been suggested to have a 100% sensitivity but low specificity for HCL.12 The results from our study showed that T-bet failed to stain a few cases of HCL (93.0% of cases were positive) and confirmed its previously described low specificity. Previous studies have shown that T-bet is expressed in several low-grade B-cell lymphomas and a variety of T-cell malignancies. Thus, the usefulness of T-bet in immunohistochemical staining is limited to the cases in which the differential diagnosis has been narrowed to a mature B-cell neoplasm and has excluded chronic lymphocytic leukemia/small lymphocytic lymphoma.

In our series, cyclin D1 showed similar sensitivity (96%) to annexin-1 and aberrantly stained only 1 case of MZL (cytogenetic analysis of the splenectomy specimen in this case was negative for t(11;14) and lacked all of the morphologic and genetic characteristics of mantle cell lymphoma). The high sensitivity of cyclin D1 for HCL was unexpected because the literature has suggested a much lower proportion of HCL cases expressing cyclin D1. However, Miranda et al11 showed a similar proportion (100%) of HCLs staining with cyclin D1 in their series.

Our study promotes the use of a panel of immunohistochemical stains in lieu of relying on 1 or 2 antibodies in the diagnostic evaluation of a possible HCL. TRAP and DBA.44 retain their usefulness in identifying a hairy cell infiltrate in BM biopsy specimens, but sensitivity and specificity can be improved if annexin-1, T-bet, and cyclin D1 are included in an HCL immunohistochemical panel. A thorough understanding of the limitations of these antibodies is required for accurate interpretation of immunohistochemical stains in difficult cases.

The limitations of these stains include high background staining of marrow elements with annexin-1, which may mask a sparse infiltrate of HCL; imperfect sensitivity and low specificity of T-bet for HCL, even in cases in which the differential diagnosis has been narrowed; and the inconsistent staining quality of cyclin D1 in HCL and its reactivity with mantle cell lymphoma and plasma cell disorders. The use of a wide panel of immunohistochemical stains interpreted in conjunction with cytologic and histopathologic evaluation and flow cytometric immunophenotyping, all taken within the context of the clinical history and findings, will greatly facilitate the diagnosis of HCL in difficult cases and even cases with minimal BM involvement. Figure 1 is a summary of a proposed algorithm as an aid to the workup of HCL and its differentiation from SMZL and ENMZL.

Figure 1

A proposed algorithm to aid in the work-up of hairy cell leukemia (HCL), and the differentiation from marginal zone lymphoma (MZL), including splenic MZL (SMZL) and extranodal MZL (ENMZL). IHC, immunohistochemical stain; TRAP, tartrate-resistant acid phosphatase.


Upon completion of this activity you will be able to:

  • identify the morphologic mimics of hairy cell leukemia (HCL) and the clinical importance of definitive diagnosis.

  • list 4 immunohistochemical stains that, as a panel, will assist in diagnosing HCL.

  • describe the limitations of current diagnostic techniques when HCL is considered.

The ASCP is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The ASCP designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 Credit ™ per article. Physicians should claim only the credit commensurate with the extent of their participation in the activity. This activity qualifies as an American Board of Pathology Maintenance of Certification Part II Self-Assessment Module.

The authors of this article and the planning committee members and staff have no relevant financial relationships with commercial interests to disclose.

Questions appear on p 478. Exam is located at www.ascp.org/ajcpcme.


  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.
View Abstract