Soluble CD22 as a tumor marker for hairy cell leukemia

Abstract

CD22 is an important immunotherapeutic target on B-cell malignancies, particularly hairy cell leukemia (HCL), but its soluble extracellular domain, sCD22, has not yet been reported in the blood. By immunoaffinity and enzyme-linked immunosorbent assay techniques using anti-CD22 monoclonal antibodies, we identified the 100-kDa extracellular domain of CD22 and an 80-kDa processed form in serum of patients with HCL. The median sCD22 level measured by enzyme-linked immunosorbent assay was 18 ng/mL for 93 patients with HCL. sCD22 levels varied from 2.1 to 163 ng/mL and were higher (P < .001) than 23 normal donors (median, 0.6 ng/mL). More than 95% of normal donors had sCD22 levels less than 1.9 ng/mL. sCD22 levels were proportional to concentrations of circulating HCL cells (P = .002), and HCL spleen size (P < .001). sCD22 levels normalized with complete but not partial response to treatment. sCD22 levels up to 300 ng/mL had less than a 2-fold effect on the cytotoxicity of the anti-CD22 recombinant immunotoxin BL22. sCD22 levels may be useful to follow in patients with HCL and may be more specific than sCD25 in patients with CD22(+)/CD25(-) disease. Trials are listed on www.cancer.gov as NCT00002765, NCT00021983, NCT00074048, NCT00085085, NCT00337311, and NCT00462189.

Figures

Figure 1

Isolation of sCD22. Shown are Western blots of recombinant purified sCD22-Fc (lane 1), sCD22 (lane 2), serum from a healthy donor (lane 3), serum from an HCL patient (lane 4), culture medium (land 5), and culture medium from the HCL Eskol cell line (lane 6).

Figure 2

sCD22 standard curve. Known concentrations of purified sCD22-Fc were tested and the curve solved by 4-parameter logistic regression. Each point is the mean of 3 triplicate experiments, and SD values are shown when larger than the point markers. The inset curve magnifies the data at low concentrations.

Figure 3

sCD22 levels in HCL and CD22-negative malignancies. Serum samples were tested from the indicated number of patients with HCL, normal donors (NL), T-NHL, and HD. Median results of each set are shown.

Figure 4

Accuracy and precision. Serum samples from 2 normal donors (represented by gray and black), diluted 10-fold with PBS, contained the indicated known final concentrations of purified sCD22-Fc. ELISA-determined sCD22 concentrations in the samples were compared with known concentrations to determine accuracy (A) and precision (B). COV indicates coefficient of variance. Shown are the means of 2 or 3 independent experiments, and error bars represent SD.

Figure 5

sCD22 vs tumor burden. sCD22 levels were obtained in 82 HCL patients with the indicated concentrations of circulating leukemic cells, determined by flow cytometry (A). The products of malignant cell counts and CD22 sites/cell are shown in panel B as cell-bound CD22 sites/mm3. sCD22 vs spleen height was assessed for 22 HCL patients with spleens present (C). The vertical dashed line indicates the spleen size (160 mm) used in comparing sCD22 levels in patients with large and small spleens by rank-order (Wilcoxon) analysis.

Figure 6

Tumor burden before and after response to BL22. Matched sCD22 levels before (PRE) and after achieving PR (A) or CR (B) in patients with HCL. HCL cell concentrations were determined by flow cytometry before (PRE) and after achieving PR (C) and CR (D) to BL22.

Figure 7

Competition of BL22 cytotoxicity by sCD22-Fc. Raji cells were incubated with the indicated concentrations of BL22 either alone (●) or combined with 30 (○), 300 (▾), or 3000 (▿) ng/mL of sCD22-Fc. Error bars as in Figure 2.