Response of hairy cell leukemia to bendamustine

A 47-year-old male was diagnosed with hairy cell leukemia (HCL) in 1998 after presenting with pancytopenia and splenomegaly. Peripheral blood smear, bone marrow, and flow cytometry were diagnostic of HCL, with bright positivity for CD20, CD22, CD11c, CD25, and FMC7, and positive staining for CD19, CD123, CD103, and lambda light chain. He did not respond to a 7-day cycle of cladribine and subsequently underwent splenectomy followed by rituximab, both without response. He achieved a clinical remission with pentostatin lasting 5 years. He relapsed and received multiple courses of pentostatin over several years before progressing and becoming refractory to pentostatin, 10 years after the initial diagnosis. He became transfusion-dependent for both blood and platelets, with neutrophil count 400/mm³, platelet count 13 000/mm³, and hemoglobin 8.7 g/dL, and bone marrow replaced with HCL. Computed tomography (CT) showed no peripheral but extensive retroperitoneal lymphadenopathy. He began eight cycles of bendamustine 90 mg/m² days 1 and 2, every 4–7 weeks. Delays of retreatment were related to infections but not cytopenias worse than baseline. He became transfusion-independent after cycle 3, and achieved a partial response lasting 6 months. Bone marrow biopsy after cycle 7 showed about one-third of the marrow involved. The circulating HCL cell count decreased from 730/mm³ just after cycle 1 to 12/mm³ prior to cycle 3, and was 2/mm³ at 1 month after cycle 8.

Although this patient did not respond to initial cladribine, characterization of his HCL cells, including positivity for B-cell markers, CD103, CD25, and CD123, was consistent with classic HCL based on the World Health Organization (WHO) definition [1]. Using 1:1 conjugates assessed by flow cytometry and control beads as described [2], the numbers of sites/cell on the HCL cells were 70 000 for CD20, 30 000 for CD22, and 4300 for CD25. Molecular characterization of the HCL cells [3] showed them to express the VH3–30-3 immunoglobulin heavy chain variable (IGHV) rearrangement, with 96.28% homology to germline. Analysis in the region of the third complementarity determining region (CDR3) of the heavy chain showed usage of IGHD3–22 and IGHJ4.

To determine whether bendamustine would resolve the patient’s cytopenias or only worsen them by further myelosuppression, blood counts were assessed before each cycle and at multiple time points in between. As shown in Figure 1(B), the platelet count appeared erratic at first due to multiple platelet transfusions required, but platelet transfusions were not needed after cycle 3, and the platelet count rose to a maximum of 193 prior to cycle 7. Similarly, as shown in Figure 1(C), red cell transfusions were not needed after cycle 3 day 2 (C3D2) and the hemoglobin (Hgb) exceeded 10.0 after cycle 7. As shown in Figure 1(A), the neutrophil count did not resolve to the normal range during treatment with bendamustine, but did meet established criteria for partial response (PR) with respect to 50% improvement from baseline independent of transfusions or colony stimulating factors [4,5]. The PR lasted for approximately 6 months, but pancytopenia worsened before and after cycle 8.

Figure 1.

Time course and results of serial measurements of concentrations of (A) neutrophils, (B) platelets, (C) hemoglobin (Hgb), (D) circulating HCL count, (E) soluble CD25 (sCD25), and (F) soluble CD22 (sCD22). Arrows show the dosing time points of bendamustine 90 mg/m²/dose, given on days 1 and 2. (G) Computed tomography (CT) of the abdomen was obtained at the indicated assessment time points, and axial sections at the level of the renal vessels are shown.

To determine the effect of bendamustine on the HCL tumor burden, circulating HCL cells and tumor markers were quantified just after beginning bendamustine and at multiple time points thereafter. As shown in Figure 1(D), the HCL count of 730/mm³ decreased >90% to 66/mm³ in less than 4 weeks. Prior to cycles 3–4, the HCL count was 12 and 6/mm³, respectively, and became undetectable briefly prior to cycle 5. Thereafter, HCL cells became detectable by flow cytometry again but remained low at 1–2 cells/mm³. As shown in Figure 1(E), the soluble CD25 (sCD25) level decreased from 179 ng/mL to 36 ng/mL before cycle 6, but remained elevated. Since sCD25 can be produced by normal T-cells, soluble CD22 (sCD22), recently reported to correlate with tumor burden in HCL [6], was also measured. Figure 1(F) shows that sCD22 decreased from 89 ng/mL just after cycle 1 to 17.9 ng/mL prior to cycle 7. Thus, bendamustine resulted in a significant decrease in HCL tumor burden, which was overestimated by measuring circulating HCL cells.

To determine whether bendamustine would result in the regression of large abdominal adenopathy, serial CT scans were obtained. As shown in Figure 1(G), serial axial sections at the level of the renal vessels showed significant regression of periaortic adenopathy. After cycle 7, abdominal adenopathy remained decreased, without significant growth more than 2 months after starting cycle 8. Thus, bulky adenopathy in HCL, which is not uncommon in patients many years after splenectomy, is capable of undergoing at least partial regression with bendamustine.

To determine whether bendamustine would reduce the number of normal B- and T-lymphocytes, as expected for purine analogs, B- and T-cell subsets were measured. As shown in Table I, the patient had very few normal B-cells during the entire course. Natural killer (NK) cells peaked prior to cycle 4, and decreased below baseline after cycle 8. Although a pre-bendamustine sample was not available, the total T-cell counts increased with response to bendamustine, and decreased after cycle 6. Levels of CD4+ and CD8+ T-cells correlated approximately with the total T-cell number, as did the CD56+ or CD57+ cytotoxic T-cells. Thus, bendamustine was not at least initially toxic to normal NK and T-cells. The decrease in T-cells and T-cell subsets in the latter half of the treatment course may represent cumulative toxicity to normal lymphocytes. Toxicity to normal B-cells could not be assessed due to their rarity, but normal B-cells could be detected in low numbers after cycle 5, when the response to bendamustine was evident. Thus, bendamustine, after enough cycles, might have caused cumulative toxicity to normal lymphocytes in this patient with HCL, although suppression of normal lymphocytes due to disease might also have occurred, and improvements in normal lymphocytes could be due to response.

Table I.

Effect of bendamustine on normal B- and T-cells^*.

				T-cells
Time point	WBC	Normal B-cells	NK cells	Total	CD4+	CD8+	CD56+	CD57+
C1D7	1400	0	73	118	64	50	8	51
C2D9	1200	0	41	171	108	50	ND	ND
C3D1	1200	1	69	193	108	69	3	66
C4D0	1400	1	450	160	92	74	6	74
C5D1	1100	0	131	243	ND	ND	ND	ND
C6D1	2000	3	113	421	205	227	5	200
C7D1	1500	1	116	215	142	83	2	122
C8D28	1040	2	44	213	166	70	2	135

^*All cell counts /mm³.

C1D7, cycle 1, day 7; WBC, white blood cells; NK, natural killer; ND, not determined.

Hairy cell leukemia is a B-cell malignancy first reported by Bouroncle et al. in 1958, constituting about 2% of all leukemias [7], and the median survival without effective treatment is about 4 years [8]. Treatment with a purine analog, cladribine or pentostatin, has dramatically improved the outlook of HCL, with 75–90% of patients attaining complete remission (CR), most lasting >10–15 years [4,9,10]. However, many patients become resistant to purine analogs and other therapies, and will die of disease-related complications. Some of these patients have responded to rituximab [11] or recombinant immunotoxins [5,12], but the optimal treatment for purine analog-refractory HCL is unknown.

Bendamustine has structural features of both an alkylating drug and a purine analog [13]. Bendamustine interacts with DNA in a way that is non-cross-resistant with other alkylators including cyclophosphamide, carmustine, and melphalan. The lack of cross-resistance between bendamustine and other cytotoxic drugs is consistent with its multiple mechanisms of action, including: (1) activation of DNA-damage stress responses and apoptosis, (2) inhibition of mitotic checkpoints, (3) induction of mitotic catastrophe, (4) activation of a DNA repair pathway involving base excisions, (5) p53-dependent stress pathway initiation leading to apoptosis, and (6) down-regulation of genes needed for mitotic checkpoint regulation [13]. Several recent trials have documented the activity of bendamustine in B-cell hematologic malignancies. Bendamustine has demonstrated efficacy in chronic lymphocytic leukemia (CLL), follicular lymphoma, multiple myeloma, and aggressive B-cell non-Hodgkin lymphomas (NHLs) [14]. We believe this is the first report of a patient with refractory HCL who had a clinically valuable partial response despite multiply resistant and refractory transfusion-dependent HCL.

Using lymphoma cell lines, bendamustine was synergistic with either cladribine or rituximab [13]. Data are relatively limited regarding the use of rituximab combined with chemotherapy for HCL. Small series have shown >90% CR rates with rituximab combined with either cladribine or pentostatin, most without minimal residual disease [10,15], and the cladribine + rituximab trial is ongoing at M. D. Anderson. A randomized trial exploring the role and timing of rituximab with cladribine is under way at the National Institutes of Health (NIH) for newly diagnosed or once-relapsed patients with HCL. To determine the potential role of bendamustine + rituximab in the treatment of multiply relapsed HCL, a clinical trial is under way at NIH in which patients are randomized to either pentostatin + rituximab or bendamustine + rituximab, with crossover if needed from one regimen to the other. The goal of this trial will be to prospectively define the activity of each combination in multiply relapsed HCL, and to compare both the efficacy and the toxicity of these combinations.