Usefulness of Dual Immunohistochemistry Staining in Detection of Hairy Cell Leukemia in Bone Marrow

Abstract

Objectives

We evaluated efficacy of two dual immunohistochemistry (IHC) staining assays in assessing hairy cell leukemia (HCL) involvement in core biopsies and compared the results with concurrently collected flow cytometric data.

Methods

Overall, 148 patients with HCL (123 male, 25 female; mean age: 59.8 years; range: 25-81 years) had multiparameter flow cytometry performed using CD19, CD20, CD22, CD11c, CD25, CD103, CD123, surface light chains, CD5, and CD23. In parallel, bone marrow IHC was done using PAX5/CD103 and PAX5/tartrate-resistant alkaline phosphatase (TRAP) dual IHC stains.

Results

Overall sensitivity of dual IHC stains was 81.4%, positive predictive value was 100%, and negative predictive value was 81.7%. All IHC-positive cases concurred with flow cytometry data, even when HCL burden was extremely low in the flow cytometry specimens (as low as 0.02% of all lymphoid cells).

Conclusions

Dual IHC stain is a sensitive tool in detecting HCL, even in cases with minimal disease involvement.

Hairy cell leukemia (HCL) is a B-cell lymphoproliferative disorder characterized by distinct immunophenotype (positive for CD19, CD20, PAX5, CD22, CD11c, CD25, CD103, CD123, and CD200). Immunophenotypic analysis by flow cytometry (FC) is considered the gold standard for diagnosis of HCL. However, both FC and immunohistochemistry (IHC) can be used to determine these markers. Although both trephine bone marrow biopsy and aspirate are vital for assessment of the extent of bone marrow infiltration, in some cases a cellular aspirate cannot be obtained because of extensive fibrosis (ie, “dry tap”). In such cases, IHC stains are crucial for assessment of bone marrow leukemic involvement. To date, IHC detection of HCL in bone marrow sections has been limited to overt disease and could not be used reliably in patients with very low-level HCL involvement. Monoclonal anti-CD103 antibody has been previously established as a specific HCL marker that reliably distinguishes HCL from most other B-cell lymphomas.¹ However, CD103 is also expressed in dendritic cells and in background T cells in the bone marrow, which makes CD103 difficult to interpret by light microscopy. Tartrate-resistant alkaline phosphatase (TRAP) is another IHC marker for HCL that lacks adequate specificity.² In trephine biopsies, TRAP stains nonneoplastic cells such as macrophages and osteoclasts.³ Annexin A1, BRAF V600E, and DBA.44 are other IHC stains used in the evaluation of HCL. However, annexin A1 is strongly positive in myeloid cells; this makes morphological interpretation challenging, especially in the setting of low disease burden. BRAF V600E stain sometimes has low reliability in decalcified bone marrow core biopsy specimens. Although neoplastic cells can be differentiated using all of these single IHC stains based on morphologic features, interpretation can be challenging in the setting of low disease burden. We used dual IHC staining with a specific B-cell marker (PAX5; nuclear stain) along with either CD103 or TRAP (membranous stain) to help overcome these limitations, especially in patients with low-level HCL involvement. We evaluated the efficacy of these dual IHC staining assays (PAX5/CD103 and PAX5/TRAP) in a large cohort (148 patients) evaluated for HCL involvement in marrow core biopsies and compared the results with concurrently collected FC data.

Materials and Methods

Case Selection

The data were analyzed on 148 bone marrow biopsy specimens (123 male and 25 female patients; mean age, 59.8 years; range, 25-81 years) collected from patients evaluated for HCL between 2016 and 2017 at our institution. All specimens were submitted to the Flow Cytometry Unit, Laboratory of Pathology, National Cancer Institute, and the Hematology Section, Department of Laboratory Medicine, Clinical Center, National Institutes of Health (NIH), as part of routine screening and diagnostic evaluation for multiple intramural NIH clinical protocols. Institutional review board–approved informed consent was signed by all patients.

FC Immunophenotyping

Specimens were stained within 24 hours of collection with a panel of antibodies. Erythrocytes were lysed by incubating with lysing solution (150 mmol/L NH₄Cl, 10 mmol/L KHCO₃, and 0.1 mmol/L EDTA) for 10 minutes at room temperature at a ratio of 1:9 (volume of sample:volume of lysing solution). Specimens were subsequently washed with phosphate-buffered saline and stained for 30 minutes at room temperature with antibody combinations in eight-color cocktails Table 1, according to recommendations of Clinical Laboratory Standards Institute document H43-A2.⁴ All cells were fixed in 1.0% paraformaldehyde after staining and stored at 4°C for up to 12 hours before acquisition. Specimens were acquired on FACSCanto II (BD Biosciences). One million events were acquired per tube. Data collected in list mode were analyzed with FCS Express v6 (De Novo Software). For analysis, cell populations were gated by forward and side light-scatter properties consistent with mononuclear cells, in conjunction with antigen backgating to ensure that relevant cell populations were included. FC analysis and diagnostic interpretation were performed as described previously.⁵ The diagnosis of HCL was made in accordance with the current World Health Organization classification guidelines.⁶

Table 1.

Antibody Panel Used for Flow Cytometric Evaluation of Hairy Cell Leukemia Cases

FITC	PE	PerCP	PC7	APC	AH7	V450	V500
CD103	CD25	CD123	CD19	CD23	CD20	CD11c	CD45
κ-m	λ-m	CD20	CD19	CD10	CD14	CD11c	CD45
λ-p	CD22	CD5	CD19	κ-p	CD20	CD11c	CD45
κ-m	λ-m	CD19	CD200	CD5	—	CD11c	CD45

AH7, APC-cyanine tandem conjugated;

APC, allophycocyanin; FITC, fluorescein isothiocyanate; PC7, phycoerythrin-cyanine 7; PE: phycoerythrin; PerCP: peridinin-chlorophyll-protein.

Morphology and IHC

The bone marrow biopsies were fixed in B-Plus fixative and decalcified in Rapid Cal Immuno (BBC Biochemical) and paraffin embedded using Tissue Tek processor (Sakura Finetek). Four μm sections were cut and subjected to H&E and IHC stains. The antibody panel included CD20 (L26 clone; predilute), PAX5 (SP34 clone; predilute), CD103 (EPR4166[2] clone; 1:100), and TRAP (9C5 clone; predilute). The IHC staining was performed with an automated immunostainer BenchMark ULTRA (Ventana Medical Systems), as follows. CD20 staining was performed as per the manufacturer's instructions. For PAX5/CD103 dual IHC staining, we selected PAX5 to stain first with 3,3'-diaminobenzidine (DAB), and CD103 was selected second to stain with red chromogen. Similarly, for PAX5/TRAP dual IHC staining, PAX5 was stained first with DAB and TRAP was selected second to stain with red chromogen. The expression of immunostains was evaluated by two hematopathologists (I.M. and G.K.G.) blinded for FC results. All concurrent positive and negative (isotype-matched antibody) controls stained appropriately. PAX5/CD103 and PAX5/TRAP dual IHC positivity was determined by nuclear PAX5 staining and membranous staining for CD103 or TRAP, and cases were categorized as follows: negative (no staining), rare (<5% of all cells), or positive (≥5% of all cells).

Results

PAX5/CD103 dual IHC staining generated brown nuclear staining for PAX5 and red membranous and cytoplasmic staining for CD103. PAX5/TRAP dual IHC staining showed similar results for PAX5 and red membranous and cytoplasmic staining for TRAP. PAX5 also highlighted background nonneoplastic B cells that were negative for CD103 and TRAP. CD103 also highlighted few scattered background T cells. TRAP also highlighted background macrophages and osteoclasts. The cases were divided into three groups based on PAX5/CD103 or PAX5/TRAP dual IHC staining results; the positive group Image 1, the rare dual IHC staining positive group Image 2, and the negative group Image 3. The results of PAX5/CD103 dual staining revealed that 82 of 148 cases were negative (55.4%), 21 of 148 cases showed rare dual IHC staining positive cells (14.2%), and 45 of 148 cases were dual IHC positive (30.4%). The dual IHC staining data were compared with concurrent FC results Table 2. FC data concurred with IHC results in all IHC-positive cases (HCL cells out of all lymphoid cells by FC: median, 7.05%; range, 0.05%-91.7%) and all rare dual IHC staining–positive cases (HCL cells out of all lymphoid cells by FC: median, 0.98%; range, 0.02%-19.04%). The sensitivity of PAX5/CD103 dual IHC staining was 81.4%, positive predictive value was 100%, and negative predictive value was 81.7%. Results revealed that 18.5% (15/81) cases were positive by FC analysis (HCL cells out of all lymphoid cells by FC: median, 0.13%; range, 0.01%-9.21%) but negative for PAX5/CD103 dual IHC staining. The data on PAX5/TRAP dual IHC staining were analyzed similarly (Table 2). PAX5/TRAP dual IHC staining was slightly less sensitive compared with PAX5/CD103 dual IHC staining. Results showed that 27.1% (22/81) of FC-positive cases (HCL cells: median, 0.27%; range, 0.01%-29.5%) were negative for PAX5/TRAP dual IHC staining.

Image 1.

Representative case of extensive bone marrow biopsy involvement by hairy cell leukemia (HCL). A, H&E. B, CD20 stain showing sheets of B cells. C, PAX5/CD103 dual immunohistochemistry (IHC) staining showing sheets of dual-positive cells (PAX5, brown nuclear stain; CD103, red membranous and cytoplasmic stain). D, PAX5/tartrate-resistant alkaline phosphatase (TRAP) dual IHC staining showing sheets of dual-positive cells (PAX5, brown nuclear stain; TRAP, red membranous and cytoplasmic stain). Bottom row shows concurrent flow cytometry on bone marrow aspirate showing atypical B cells (red) with immunophenotype diagnostic of HCL.

Image 2.

Representative case of detection of low-level involvement by hairy cell leukemia (HCL) in bone marrow biopsy by dual immunohistochemistry (IHC) staining. A, H&E. B, CD20 stain showing focally increased predominantly scattered B cells. C, PAX5/CD103 dual IHC staining showing rare dual-positive cells (arrows; PAX5, brown nuclear stain; CD103, red membranous and cytoplasmic stain). D, PAX5/tartrate-resistant alkaline phosphatase (TRAP) dual IHC staining showing rare dual-positive cells (arrows; PAX5, brown nuclear stain; TRAP, red membranous and cytoplasmic stain). Bottom row shows concurrent flow cytometry on bone marrow aspirate, with gating on B cells (purple) showing few (0.4% of marrow lymphoid cells) atypical B cells (red) with immunophenotype diagnostic of HCL.

Image 3.

Representative case of hairy cell leukemia (HCL) negative by dual immunohistochemistry (IHC) staining but positive by flow cytometry (FC). A, H&E. B, CD20 stain is negative (after rituximab). C, PAX5/CD103 dual IHC staining showing no definite dual-positive cells, PAX5 stain showing brown nuclear staining in nonneoplastic B cells (arrowhead), and CD103 showing membranous and cytoplasmic staining in a subset of T cells (arrow). D, PAX5/tartrate-resistant alkaline phosphatase dual stain showing no definite dual-positive cells. Arrow points at B cells. Bottom row shows concurrent FC on bone marrow aspirate, with gating on B cells (purple) showing rare (0.03% of marrow lymphoid cells) atypical B cells (red) with immunophenotype diagnostic of HCL.

Table 2.

Summary of Flow Cytometry vs Dual Immunohistochemistry Staining Results

Flow Cytometry	PAX5/CD103 Dual Stain, No. (%)			PAX5/TRAP Dual Stain, No. (%)
	Positive	Rare	Negative	Positive	Rare	Negative
Positive (n = 81)	45 (55.6)	21 (25.9)	15 (18.3)	45 (55.6)	14 (17.2)	22 (27.1)
Negative (n = 67)	0	0	67 (100)	0	0	67 (100)

TRAP, tartrate-resistant alkaline phosphatase.

Discussion

To our knowledge, this study is the first to explore utilization of PAX5/CD103 and PAX5/TRAP dual IHC stains in evaluation of bone marrow involvement by HCL, including cases with low-level involvement, and then compare results with concurrent FC studies. Results from this study showed that dual IHC could reliably detect HCL cells even when the disease burden was as low as 0.02% of cellular elements by FC. Morphologically, it was possible to detect dual IHC-stained cells at very low levels, even when only rare dual-positive cells were present.

The advancement of HCL therapy with utilization of purine nucleoside analogs has significantly improved the prognosis of HCL.^7,8 Unfortunately, the disease relapses frequently. Published data suggest that persistence of minimal residual disease (MRD) could predict disease recurrence.⁹ Recent consensus guidelines do not recommend MRD eradication as a well-established goal of therapy, so consistent detection of minimal disease involvement is important given that treatment decisions are sometimes based on these data.¹⁰ In one study, Bengio and colleagues¹¹ reported comparative analysis of FC (CD20, CD22, CD11c, CD25, CD103, SIg) vs single IHC staining (CD20 and DBA.44 antibodies) for MRD detection status after treatment with 2-chlorodeoxyadenosine. The criteria for MRD detection by IHC was 1% to 10% CD20- and DBA.44-positive cells occurring singly or in clusters with hairy cell morphology. MRD by FC was defined as less than 0.3% cells with high expression of CD11c, CD25, or CD103. The results showed a significantly higher MRD detection rate (64%) by FC vs IHC (46%), despite the fact that disease burden by FC may not always be adequately assessed because of marrow hemodilution and/or fibrosis.

CD103 is a specific marker for HCL and has been in use for HCL detection by FC and by frozen-section IHC^12-14 for more than a decade. Recently, Morgan et al¹ demonstrated successful use of CD103 rabbit monoclonal antibody in formalin-fixed paraffin-embedded sections of HCL and HCL-variant cases, providing decisive aid in HCL detection. However, CD103 is also expressed in a subset of T cells, making interpretation difficult, especially in the setting of minimal disease. In a recent study, Rozenvald et al¹⁵ reported improved sensitivity and specificity in HCL detection among various B-cell lymphoma cases by PAX5/CD103 dual IHC staining. We observed similar results from PAX5/CD103 dual IHC staining in our cohort. We applied additional dual IHC staining (PAX5/TRAP) and went on to compare combined dual IHC staining results with FC evaluation. PAX5/CD103 dual IHC staining showed slightly higher sensitivity compared with PAX5/TRAP dual IHC staining. These findings were in accordance with the previous study by Shao et al⁵ showing that TRAP is positive in only 95% of HCL cases.

The focus of our study was on detection of HCL in bone marrow specimens regardless of treatment status. However, the morphologic assessment of hairy cells can be particularly challenging in the setting of minimal disease involvement. The following criteria for MRD detection by IHC were defined by Tallman et al¹⁶: (1) morphologic absence of HCL in peripheral blood, bone marrow aspirate, and bone marrow core (H&E); (2) presence of CD20- or DBA.44-positive cells equal to or greater than CD45RO-positive cells; and (3) presence of more than 50% of CD20- or DBA.44-positive cells showing morphology of hairy cells. In our study, the cases with rare dual IHC staining–positive cells were either inconclusive or negative for morphologic evidence of disease by CD20 and H&E stains. Therefore, IHC-MRD detection by dual IHC stains might be a more sensitive tool than single IHC stains, meriting possible future studies. However, approximately 20% of cases with extremely low disease burden were missed by dual IHC staining, suggesting that FC remains the gold standard for detecting minimal disease when a robust protocol with collection of at least 1 million events is used.

In summary, dual IHC staining is a sensitive tool for detecting HCL, even in cases with minimal disease involvement. All IHC-positive cases concurred with FC data, even when HCL burden was extremely low (as low as 0.02% of all lymphoid cells by FC analysis). Only 18.3% of dual IHC–negative cases were positive for low-level involvement by FC analysis. PAX5/CD103 dual IHC staining was slightly more sensitive than PAX5/TRAP dual IHC staining. Therefore, dual IHC staining is a sensitive new tool for evaluation of marrow involvement by HCL and could be used as a reliable surrogate in the event of nonavailability of aspirate specimens for FC analysis.