Leukemic CD52 Negative Subclones Due to Defective Glycophosphatidyl-Innositol Anchoring Are Common in Acute Precursor B Lymphoblastic Leukemia, Escape Alemtuzumab Therapy, but Display Increased Sensitivity to Rituximab Mediated Complement Dependent Cytotoxicity: a Mechanistic Rationale for Antibody Combination Therapy.

Abstract 835

The prognosis of B-lineage acute lymphoblastic leukemia (B-ALL) in adults is poor despite intensive polychemotherapy. Treatment intensification is limited by treatment related toxicity. Alemtuzumab (ALM) is a humanized monoclonal antibody directed against the glycophosphatidyl-innositol (GPI-) anchored membrane protein CD52. Because B-ALL frequently express CD52, ALM may be of therapeutic value in this disease. However, loss of CD52 due to acquired defects in GPI biosynthesis has been described in various hematological settings. Presence of GPI-defective leukemic subclones in ALL will have implications for the response of ALL to ALM. We analyzed leukemic cells from 24 randomly selected cases for expression of CD52 by flow cytometry using FITC-conjugated ALM. To identify GPI-defective cells counter staining was performed using PE-conjugated antibodies against the GPI-anchored surface markers CD55 and CD66c. Of the 24 cases, 19 expressed CD52 (median MFI: 220, range 34.0-617). In 16 of these 19 cases CD52⁻ events were detected. In 11 of these 16 cases these events represented a discernable CD52⁻CD55⁻CD66c⁻ GPI-defective subpopulation (median 0.02% of all cells, range 0.01-25.8%). We subsequently evaluated the in vivo activity of ALM against ALL, and the relevance of the GPI-defective subpopulations, in a preclinical model of human ALL. For this, we selected case BV (4.0% CD19⁺CD52⁻CD55⁻ cells, overall MFI 137) and case CM (no detectable CD19⁺CD52⁻CD55⁻ cells, overall MFI 616). NOD/scid mice were inoculated i.v. with primary BV or CM cells. Starting 14 days after inoculation, animals received daily i.p. injections of 250μg ALM (n=8) or saline (n=4). After 3 weeks of treatment animals were sacrificed and bone marrow (BM) was analyzed for the presence of leukemic cells by flow cytometry. BM of control treated animals engrafted with BV cells contained 80±20% leukemic cells, 95% of which were CD52⁺. BM of control treated animals engrafted with CM cells contained 69%±14% leukemic cells, 99% of which were CD52⁺. BM of ALM treated animals engrafted with BV or CM cells contained 75% (±5.9%) and 21% (±5.4%) leukemic cells, respectively. However, these cells were exclusively CD52⁻CD55⁻CD66c⁻, demonstrating that ALM had efficiently eradicated all CD52⁺cells but in both cases clonogenic GPI-defective subpopulations had been selected. The cells remained CD52⁻CD55⁻CD66c⁻ when transplanted into secondary untreated recipients, demonstrating that the loss of expression was irreversible. As known from paroxysmal nocturnal hemoglobinuremia, GPI-defective cells are particularly susceptible to complement mediated cytotoxicity (CDC) because they lack the protective function of CD55. Because the cells recovered from ALM treated animals and control treated mice expressed similar levels of CD20, we compared their susceptibility to rituximab (RTX) mediated CDC. GPI-defective and wild type CM cells were coated with 0-10μg/mL RTX, exposed to dilutions of complement, and assayed for viability by flow cytometry. In the presence of 10μg/mL RTX, 50% lysis of wild type cells required a 1:40 complement dilution while 50% lysis of GPI-defective cells required a 1:200 dilution. Likewise, in the presence of a 1:100 complement dilution, 2.5μg/mL RTX mediated 10% lysis in wild type cells but 75% lysis in GPI-defective cells. To determine whether the GPI-defective cells that escape ALM treatment could therefore be targeted by RTX in vivo, we engrafted NOD/scid mice with CM or BV cells and administered RTX (250μg daily) or RTX plus ALM (250μg each daily) in the set-up described above. Administration of RTX alone resulted in partial reduction of engraftment levels only, most likely due to low expression levels of CD20 on the leukemic cells. However, combined administration of ALM and RTX resulted in complete disappearance of leukemic cells in 8 out of 8 CM-engrafted animals, 6 of which were in molecular remission as demonstrated by RT-PCR, and in 4 out of 5 BV-engrafted animals. We conclude that ALM is highly effective against CD52⁺ ALL cells but CD52⁻ GPI-defective subclones may pre-exist in the majority of primary ALL and escape therapy. Because GPI-defective cells are particularly susceptible to complement mediated lysis, combination of ALM with antibodies directed against non-GPI anchored proteins, such as RTX, may be highly synergistic.