Leukemia (2002) 16, 178–185  2002 Nature Publishing Group All rights reserved 0887-6924/02 $25.00 www.nature.com/leu Surface antigen expression in chronic lymphocytic leukemia: clustering analysis, interrelationships and effects of chromosomal abnormalities J Hulkkonen1, L Vilpo2, M Hurme1 and J Vilpo2

1Department of Microbiology and Immunology, University of Tampere Medical School and Laboratory Center of Tampere University Hospital, Tampere, Finland;and 2Department of Clinical Chemistry, University of Tampere Medical School and Laboratory Center of Tampere University Hospital, Tampere, Finland

Chronic lymphocytic leukemia (CLL) is a phenotypically dis- deletion, 13q deletion, 17p deletion and trisomy 12, which tinguishable form of B-lymphoid leukemias. The regularity of are independent predictors of disease progression and sur- surface membrane antigen expression patterns, their inter- 3 relationships as well as the effects of the three frequent vival. The expression variants include, for example, chromosomal aberrations, ie 11q deletion, 13q deletion and tri- CD38 expression, identifying a particular subset of B-CLL cells somy 12, were investigated in 35 classic CLL cases by flow with defined functional properties, including the propensity to cytometry. The two-way cluster analysis of 31 individual anti- undergo apoptosis.8 Another example is significantly higher gens revealed three expression patterns: (1) most cells in most CD23 and CD21 expression in CLL and CLL/PL prolympho- cases positive (CD5, CD19, CD20, CD23, CD27, CD40, CD45, cytes compared with small lymphocytes, ie in the proliferating CD45RA); (2) most cells in most cases negative (CD10, CD14, CD34, CD122, CD154, mIgG); and (3) a mixed pattern with a vari- vs non-proliferating compartment of CLL, suggesting that these able number of positive cases and a variable percentage of two surface antigens may be related to the progression of positive cells in individual cases (CD11c, CD21, CD22, CD25, CLL.9 CD38, CD45RO, CD79b, CD80, CD95, CD124, CD126, CD130, We show here that the expression of several structurally and FMC7, mIgD, mIg␬,mIg␭, mIgM). The expressions of several functionally distinct membrane antigens bears statistically antigens were strongly interdependent, even when antigens highly significant intercorrelations, while some others are belonged to entirely different gene families. Such antigen pairs were: CD11c/CD21; CD19/CD45; CD19/CD79b; CD22/CD45RA; stringently mutually exclusive. Furthermore, new associations CD23/Ig␬; CD25/mIgM; CD27/CD45; CD45/CD79b; CD45RA/Ig␬. of aberrations 11q deletion, 13q deletion and In contrast, the expression of some antigens was mutually trisomy 12 with exceptional membrane antigen expression exclusive, the best examples being CD45RA/CD45RO, will be described. CD38/CD80 and CD45RA/CD80. Deletion of chromosome arm 11q attenuated expression of splicing variant CD45RA, but enhanced CD45RO expression. In contrast, cases of trisomy 12 were associated with enhanced CD45RA and attenuated Materials and methods CD45RO expression. Similarly, trisomy 12 was associated with enhanced CD27 and mIg␬ expression. The variable levels of Patients and samples signaling surface membrane antigens, their interactions and interference by genetic aberrations are likely to affect the clini- Clinical specimens were obtained after informed consent from cal progression and drug response of CLL. 35 consecutive CLL patients referred to the CLL outpatient Leukemia (2002) 16, 178–185. DOI: 10.1038/sj/leu/2402363 Keywords: chronic lymphocytic leukemia; cluster of differentiation; clinic at Tampere University Hospital, Finland. The inclusion × 9 chromosome; cluster analysis criterion was a blood lymphocyte count 30 10 /l or more. Clinical and hematological data of 35 CLL patients is illus- trated in Table 1. Diagnosis and staging were based on stan- Introduction dard clinical, morphological and immunophenotyping criteria (Table 1). In accordance with the B cell phenotype and a com- monly used immunological CLL classification,10 all patients Chronic lymphocytic leukemia (CLL) is the most common + + + form of adult leukemia in western societies. It has attracted had a CD19 /CD5 /CD23 immunophenotype. Typical CLL, CLL/mix or CLL/PL morphology could be assigned to all much basic and clinical interest. Despite recent advances in 4 our understanding of the genetics, biology and clinical patients. Most patients showed chromosomal aberrations behavior of CLL, there is still no established cure for the (Table 1). None of these aberrations indicated chronic lymph- disease and its progression and outcome are highly oid leukemias other than typical CLL. Peripheral blood leu- 1–3 kemic cells were isolated using traditional methods as unpredictable. 11,12 The diagnosis of ‘true’ CLL has become very accurate. The described. The proportion of monocytes plus polyclonal most important methods for making a distinction between CLL T- and B-lymphocytes was 1–13%, indicating that 87–99% of and other chronic lymphoid leukemias are: (1) cell mor- the isolated cells represented the leukemic population (for phology;4 (2) immunophenotyping;5 (3) histology;6 and (4) details of the immunophenotyping, see below). chromosome analyses.7 Using some of these tools, CLL can be subdivided in several categories as regards the cell biology and clinical behavior of the disease. Importantly, distinct dis- Immunophenotyping ease entities may be marked by chromosomal aberrations or by gene expression anomalies. The former include 11q Immunophenotyping was performed by flow cytometry (FACSCalibur, Becton Dickinson Immunocytometry systems, San Jose, CA, USA) using commercial mouse monoclonal antibodies and respective immunoglobulin isotype controls Correspondence: J Vilpo, Department of Clinical Chemistry, Tampere University Hospital, PO Box 2000, FIN-33521 Tampere, Finland; (Table 2). Staining was performed according to the suppliers’ Fax: 358 3 247 4048 instructions. For cell surface immunoglobulin staining, the Received 2 February 2001; accepted 15 October 2001 cells were first incubated in PBS (pH 7.4) at 37°C for 30 min CLL membrane antigens J Hulkkonen et al 179 Table 1 Clinical hematological data at the time of sampling among 35 patients with chronic lymphocytic leukemia

Patient Sex/Age Stage Blood FAB Progressionb Previous or Cytogenetic findingsd No. (Binet) lymphocytes diagnosisa ongoing therapyc (×109/l)

1 M/53 A 63 CLL S None 11q del (I,II); 13q del (IV,VI) 2 M/63 A 46 CLL T LP 7q del (I,VI); 11q del (I,II,VI) 3 M/72 B 84 CLL F None Trisomy 12 (I,III,VI) 11q del (II) 4 M/69 B 92 CLL F L None 5 F/61 A 86 CLL S None 6q del (VI) 6 F/49 A 130 CLL S None 13 q del (IV) 8 M/58 C 310 CLL/PL T LP, 6×COP,7×CH Complex (I, VI) OP 9 M/68 B 67 CLL S None 6q del (I,VI); 13q del (I,IV,VI) 10 M/68 A 51 CLL I None 13q del (IV,VI) 11 F/68 C 88 CLL T L 11q del (I,II,VI); 13q del (I,IV,VI) 12 M/64 C 132 CLL T LP 13q del (VI) 13 M/68 C 114 CLL F None 11q del (I,II,VII); 13q del (IV) 14 F/66 C 188 CLL/PL T LP, Complex (VI) 7×COP,10×CHOP 15 M/57 A 63 CLL F None 13q del (IV) 16 M/68 A 81 CLL I None 13q del (IV,VI) 17 M/57 A 120 CLL F None 13q del (IV) 18 F/73 C 79 CLL/mix F None Trisomy 12 (I,III,VI) 19 M/55 A 36 CLL/PL F None Trisomy 6 (I,VI); Trisomy 12 (I,III,VI); 13q del (IV) 20 M/53 B 69 CLL/PL I None None 21 M/71 B 89 CLL/mix S None 11q del (I,II,VI) 22 M/55 B 59 CLL/PL I None 6q del (VI); 13q del (IV) 23 M/62 B 59 CLL S None 11q del (I,II,VII) 24 F/79 B 97 CLL/mix T 4×CHOP,LP None 25 M/54 B 216 CLL/mix T None Trisomy 12 (I,III,VI); 13q del (IV) 26 M/48 A 93 CLL S None None 27 M/67 A 67 CLL I None 13q del (IV) 28 M/66 C 178 CLL F/T LP 11q del (I,II) 29 M/77 A 94 CLL F LP Other 30 F/69 A 69 CLL F None Complex (V) 32 M/70 A 134 CLL F None None 33 M/57 B 68 CLL S None None 34 F/75 C 173 CLL I None None 35 F/76 A 206 CLL F None 11q del (I) 36 F/78 C 155 CLL S None Trisomy 16 (VII) 37 M/63 C 340 CLL/PL F None Trisomy 12 (III,VII) aBennett et al.4 bAbbreviations: S (slow) = blood lymphocytes increased less than 20% within a year; F (fast) = lymphocyte doubling time 1 year or less; I (intermediate) between S and F; T (therapy) = Chemotherapy given, natural disease progression non-evaluable. cAbbreviations: L, Leukeran; LP, L + prednisolone; COP, cyclophosphamide-Oncovin-prednisolone; CHOP, COP + hydroxydaunorubicine. dThe classification of Do¨hner et al was applied.3 The cytogenetic methods and most of the findings have been published previously. The abbreviations are: I = CGH (comparative genomic hybridization, applied to all patients);33 II = 11q del/FISH (11q del was assessed with a 7.8 Mb contig of probes spanning region 11q23.1-q23.3. applied to patients 1–31);35 III = 12 trisomy/FISH (trisomy 12 determination was performed with a chromosome 12 pericentromeric alphoid probe, pA12H8, and applied to all patients);40 IV = Rb1/FISH (LOH at the RB locus was determined with a locus-specific P1 probe for Rb1, RMC13-P001, applied to all patients;40 V = traditional G-banding (applied to all patients);33,34 VI = G-banding after optimized mitogen stimulation (patients 1–25);34 VII = Multicolor-FISH after optimized mitogen stimulation (Karhu et al, unpublished). to remove heterophilic antibodies. The buffers used in all the The percentages of cells were calculated on the basis of steps did not contain serum. data obtained from two-color immunofluorescence dot blots. The forward scatter channel was checked prior to every Analyses were carried out using CellQUEST Software (Becton analysis using unlabeled calibration particles (SPHERO cali- Dickinson). Quadrant markers were set relative to negative bration particles, blank 6.5–8 ␮m, Pharmingen, San Diego, immunoglobulin isotype controls in a such way that 99% CA, USA). Instrument linearity was checked using commercial (99.05 ± 0.04) of the cells labeled with the control antibodies standard reagents (Immuno-Brite Standards Kit; Coulter, Hia- were located in the left lower quadrant. When estimating the leah, FL, USA) prior to study. Fluorescence compensations reactivity of known antigens on the surface of malignant were set prior to every analysis using calibrate beads B cells, the proportion of CD2 cells was taken into account. (caliBRITE; Becton Dickinson) and FACSComp software Furthermore, an aberrant expression of T cell antigens CD4 (Becton Dickinson). Instrument calibration was additionally and CD8 in B cells was investigated. As only one case with checked using Immunobrite level II standard beads (Coulter). aberrant CD8 expression was found (patient No. 30 with 32% On the basis of the data thus obtained, the interassay varia- of CD8-positive B cells), these antigens were not taken into bility in this study was 6.5% for FL1 channel and 7.6% for account in cluster analysis. The intensity of antigen expression FL2 channel. in the cell membrane was analyzed in a histogram data dis-

Leukemia CLL membrane antigens J Hulkkonen et al 180 Table 2 Monoclonal antibodies used order to fit the program, the percent of positive cells values were divided by 25. After log2 transformation of these data, Antigen and Label Isotype Clone hierarchical clustering was applied to both axes using average a source linkage clustering, as implemented in the program Cluster 2.11 (M Eisen; http://rana.lbl.gov/EisenSoftware.htm).13 The ␥ CD2 (B) FITC 2a leu-5b results were analyzed with Tree View 1.50 (M Eisen; CD4/CD8 (B) FITC/PE ␥ /␥ leu-3a/leu-2a 1 2a http://rana.lbl.gov/EisenSoftware.htm).13 CD20 (B) FITC ␥1 leu16 ␥ CD45RA (B) FITC 1 leu-18 ␥ CD45RO (B) PE 2a leu-45RO ␥ CD122 (B) PE 1 anti-IL-2R-p75 Results CD19 (B) PE ␥1 SJ25C1 ␥ CD5 (B) FITC 2a Leu-1 ␥ Surface antigen expression patterns CD23 (B) PE 1 Leu-20 FMC7 (I) FITC ␮ FMC7 ␥ CD22 (B) FITC 2b Leu-14 Two-way cluster analysis of 31 different membrane antigens ␥ CD34 (B) PE 1 8G12 of 35 CLL patients is illustrated in Figure 1. Three different CD38 (B) PE ␥ Leu-17 1 types of clusters could be demonstrated. Antigens CD45, CD10 (B) FITC ␥2a Anti-CALLA ␥ CD5, CD45RA, CD20, CD19, CD23, CD27 and CD40 formed CD124 (I) PE 1 S4–56C9 ␥ CD126 (I) PE 1 M91 a red cluster of predominantly positive antigens. In contrast, ␥ CD11c (D) FITC 1 KB90 the antigens CD14, CD10, CD154 (CD40 ligand), CD34, ␥ CD14 (D) FITC 2a TUK4 mIgG and CD122 (IL-2 receptor, beta chain) were predomi- ␥ CD21 (D) FITC 1 1F8 ␥ nantly negative (green). The most variation was observed in CD79b (D) FITC 1 SN8 mIgG (P) FITC ␥ G18–145 the third cluster, containing antigens: CD11c, CD25, CD21, 1 CD79b,CD22, CD124 (IL-4 receptor) (less variation), and mIgD (P) FITC ␥2a IA6–2 ␥ ␭ ␬ mIgM (P) PE 1 G20–127 CD45RO, mIgM, mIgD, CD38, mIg ,mIg, FMC7, CD130 ␥ CD95 (P) FITC 1 DX2 (gp130), CD80 (B7), CD95 (APO-1/Fas) and CD126 (IL-6 ␥ CD40 (P) FITC 1 5C3 receptor) (more variation). In this third cluster, there was a CD27 (P) FITC ␥1 M-T271 ␥ tight correlation between the percentage of positive cells and CD130(P) PE 1 VC041 CD25 (B) FITC ␥ Anti-IL-2R the geometric mean fluorescence of the corresponding anti- 1 gens (Figure 1 and Table 3). Although statistically significant CD154 (B) PE ␥1 Anti-gp39 ␥ Ͻ CD80 (B) PE 1 Anti-B7 (P 0.05), poorer correlations were demonstrated for CD40, ␥ ␥ CD45/CD14 (B) FITC/PE 1/ 2b Anti-Hle-1/Leu- CD130 and mIgG. 3M The percentages of positive cells in the red cluster cases ␬ ␭ ␥ ␥ / (B) FITC/PE 1/ 1 TB28–2/1–155–2 approached 90–100%. However, the geometric mean fluor- escence (GMF) values were more variable. Hence, GMF a = = = = B Becton Dickinson; D Dako; I Immunotech; P Pharmingen. values instead of percentages of positive cells were used for For more accurate description of antigen , see, for example, http://www.ncbi.nlm.nih.gov/PROW. further analyses of the association between different patients and the expression of different membrane antigens.

play using logarithmic scale. The number of CD45-positive cells was always Ͼ98% (99.79 ± 0.35). The mean proportion Correlations in the expressions of individual of CD14-positive cells in an analyzed gate was 0.77% (s.d. membrane antigens 1.11, range 0.05–6.82). The intensity of mIgD and mIgM expression in the cell Membrane immunoglobulins: Light chain expression was membrane was estimated in a histogram data display, sel- easily demonstrable in all cases. Fifty-four percent of the cases ␬ ␭ ecting the log scale in the fluorescence axis and arithmetic were -positive and 46% -positive. The individual variation ␬ statistics. The criteria were as follows: weak expression (+/−) of positive cells among different patients was: -positive from ␭ when the mean fluorescence value was within the first logar- 44 to 99% (mean 78.4, s.d. 19.4%) and -positive from 91 to ithmic percentile (ie Ͻ10); moderate (+), when the peak was 99% (mean 95.0, s.d. 2.7%). within the second percentile (10–100), and strong (++) or very Patient No. 18 showed strong mIgG expression and patient strong (+++) when the peak was within the third percentile Nos 27 and 34 showed moderate mIgG expression. None of (100–1000) or more (Ͼ1000). the cases showed strong or very strong mIgM or mIgD expression. The classification of the remaining 32 cases which could be evaluated according to weak vs moderate Statistical treatment mIgD and mIgM expression resulted in four categories: (1) mIgMweak/mIgDweak (12 cases); (2) mIgMmod/mIgDmod (11 mod weak Comparison of two mean values was performed by using cases); (3) mIgM /mIgD (6 cases); and (4) weak mod Mann–Whitney U-test and more than two means by using mIgM /mIgD (3 cases). Kruskal–Wallis test (ANOVA). Correlations were calculated using Spearman’s order correlation test. These statistical calculations were carried out using Statistica software (Win Other surface membrane antigens: Pairwise correlations 5.1F; StatSoft Inc, Tulsa, OK, USA). of expression levels of individual antigens were determined by Two-way joining cluster analysis was based on joining or using Spearman’s rank order regression analysis. This analysis tree clustering methods which makes use of dissimilarities or revealed a statistically significant (P Ͻ 0.05) correlation of distance between the objects when forming the clusters. In expression in 22 individual pairs amongst a total of 19 differ-

Leukemia CLL membrane antigens J Hulkkonen et al 181

Figure 1 The left panel demonstrates a two-way joining cluster analysis (M Eisen: Cluster and Tree View) of expressions of 31 surface membrane antigens in peripheral blood leukemic cells of 35 CLL patients. The right panel demonstrates the distribution of geometric mean fluorescence values amongst 35 CLL patients. Median (black point), 25–75 percentile (red column) and 10–90 percentile (whiskers) are illustrated.

Table 3 Spearman’s rank order correlations between percentage Table 4 Interrelationships of expressions of various membrane of positive cells vs geometric mean fluorescence of peripheral blood antigens among 35 CLL patients leukemic cells of 35 CLL patients. Antigens R P Marker R P CD11c CD21 0.60 0.0001 CD11c 0.88 Ͻ0.0001 CD19 CD45 0.57 0.0003 CD21 0.78 Ͻ0.0001 CD19 CD79b 0.57 0.0004 CD22 0.86 Ͻ0.0001 CD22 CD45RA 0.61 Ͻ0.0001 CD25 0.91 Ͻ0.0001 CD23 Ig-kappa 0.69 Ͻ0.0001 CD38 0.90 Ͻ0.0001 CD25 IgM 0.55 0.0008 CD40 0.51 0.0020 CD27 CD45 0.61 0.0001 CD45RA 0.75 Ͻ0.0001 CD45 CD79b 0.77 Ͻ0.0001 CD45RO 0.97 Ͻ0.0001 CD45RA Ig-kappa 0.59 0.0002 CD79b 0.85 Ͻ0.0001 Ͻ CD80 0.87 0.0001 The antigens were selected on the basis of Spearman’s rank order CD95 0.63 0.0001 regression analysis: all antigen pairs with a highly significant corre- Ͻ CD124 0.81 0.0001 lation (R у 0.55, n = 35) are included provided that the background Ͻ CD126 0.84 0.0001 geometric mean fluorescence +2 s.d. (95% confidence limit of CD130 0.40 0.0186 assay sensitivity; 5 for FITC- and 4 for PE-labeled antibodies) was Ͻ CD154 0.70 0.0001 smaller than the median fluorescence of each antigen to be com- Ͻ FMC7 0.74 0.0001 pared. mIgD 0.90 Ͻ0.0001 mIgG 0.40 0.0188 mIg␬ 0.86 Ͻ0.0001 mIg␭ 0.89 Ͻ0.0001 not shown, but the patterns can be discerned by referring to mIgM 0.96 Ͻ0.0001 Figure 1).

The cases were selected from the mixed cluster (see Results), when there were at least four cases with у20% positive cells. Membrane phenotypes in 11q deletion, 13q deletion and trisomy 12 ent antigens. Nine pairs with the strongest associations are The expression patterns of leukocyte common antigen (CD45) illustrated in Table 4. splicing variants CD45RA and CD45RO were strongly affec- The expression patterns of mIg␭ and mIg␬ were mutually ted by the presence of 11q deletion, as well as by trisomy 12: exclusive, as expected (Figure 2). Interestingly, the same tend- the presence of 11q deletion was associated with significantly ency also concerned several other antigen pairs. The clearest suppressed expression of CD45RA and enhanced expression examples were CD45RA/CD45RO, CD38/CD80 and of CD45RO, while the opposite situation was demonstrated CD45RA/CD80 (Figure 2), and it was also seen in other anti- in case of trisomy 12 (Figure 3). Furthermore, 11q deletion gen pairs including CD21/CD80 and CD80/CD126 (results was associated with enhancement of CD20 expression and

Leukemia CLL membrane antigens J Hulkkonen et al 182 Clinical parameters and surface antigen expression

Generally, no marked correlations were noted between the clinical stage (Binet) and expression of CD19, CD22 and CD23. Stage C patients (n = 9) tended to have smaller GMF- CD22 values than Binet A and B patients (n = 26; P = 0.0379, Mann–Whitney U test). The only surface membrane antigen expression, which sig- nificantly changed according to progression status (see Table 1), was GMF-CD40 (P = 0.0123, Kruskal–Wallis ANOVA). In particular, stronger expression was noted in fast- progressing cases with chemotherapy (mean 37.1) vs fast-pro- gressing cases not undergoing chemotherapy (mean 23.2, P = 0.0450, Mann–Whitney U test). Expression of none of the membrane antigens correlated significantly to chemotherapy, if patients were dichotomized to two groups, one having received or receiving chemotherapy (n = 8) and the other without previous or ongoing treatment (n = 27).

Antigen expression in CD38 subgroups

It has been shown that CLL can be divided into two functional Figure 2 The mutually exclusive tendency of expression of surface groups depending on CD38 antigen expression.8,13,14 Further- membrane antigens in 35 CLL patients: (a) chains, (b) CD45 iso- more, according to Damle and coworkers,15 the CD38 positiv- forms (splicing variants) RA and RO, (c) CD38 and CD80, and (d) у CD45RA and CD80. The line represents curve fitting using a non- ity ( 30%) was suggested to correlate with the clinical + linear least-squares method (Statistica). Assay sensitivity (isotype outcome of CLL patients. Hence, we analyzed whether CD38 control mean fluorescence +2 s.d.) was 5 for FITC- and 4 for PE- cases (30% or more positive cells as used by Damle and labeled antibodies). coworkers; n = 14) vs CD38− cases (n = 21) were different regarding the expression of other surface membrane antigens. Only one marginal difference could be demonstrated among suppression of CD11c, CD21, CD22 and mIg␬ expression. the 30 surface markers investigated. The percentage of Trisomy 12 cases, on the other hand, were associated with CD124+ cells was 86% in the CD38− group and 70% in the enhancement of mIg␬ expression, as well as in CD27 CD38+ group (P = 0.0071, Mann–Whitney U test). A similar expression. 13q deletion was only associated with an difference was noticed if cases with 20% or more of CD38+ increased CD20 expression (Figure 3). cells were regarded as CD38-positive.

Figure 3 Surface membrane antigen expression in 11q-deletion (n = 9) vs non-deletion (n = 26) groups, in trisomy 12 (n = 5) vs non-trisomy groups (n = 30), and in 13q-deletion (n = 21) vs non-deletion groups (n = 14). Mean (point), s.e.m. (column) and 1.96 × s.e.m. (whiskers) is illustrated. The P-values given above each panel were calculated by using Mann–Whitney U test.

Leukemia CLL membrane antigens J Hulkkonen et al 183 Discussion mIgM+/mIgD+ cells.32 We found only six mIgM+/mIgD− cases (17.6%) and 10 mIgM+/mIgD+ cases (29.4%). Most of the Relatively little is known about the regulation of gene cases (35.2%) of the present cohort showed only very weak expression in CLL. However, it is well known that CD5, or absent expression of mIgD and mIgM and were difficult to CD19, CD20, CD23, CD40 and CD45, as well as its isoform put into the above categories. Hence, the expression of mIgM CD45RA are moderately or strongly expressed in CLL.16 The and mIgD in these CLL cases did not follow the normal B cell results with CD27 have been conflicting. While two previous maturation pattern. studies have indicated surface membrane expression of CD27 It has recently been established that genomic aberrations in in most or all CLL cases,17,18 a more recent and larger investi- CLL are important independent predictors of disease pro- gation regarded the expression as negative in CLL.19 Our gression and survival.3 Sembries et al19 have shown that B- results confirm the unequivocal expression of CD27 in typical CLL cells with or without 11q deletion differentially express CLL. The percentage of positive cells ranged from 59 to 98, functionally relevant cell surface molecules. All of these mark- median 94% (n = 35). ers (CD6, CD11a, CD18, CD31, CD35, CD39, CD45, CD48 The expressions of most of the comparable antigens were and CD58) were found at lower levels on cells with 11q very similar to those reported recently from another labora- deletion than on leukemic cells without this chromosomal tory,19 these being CD11c, CD21, CD22, CD38, CD79b and aberration. In the present investigation, the inferior expression CD95. This high rate of inter-laboratory concordance reflects of CD11c, CD22 and mIg␬ in cases with 11q deletion pointed the good accuracy and precision of the flow cytometric deter- in the same direction as in the study by Sembries et al. In the minations in these two laboratories. On the other hand, in current study, the difference reached a statistical significance. contrast to two previous studies19,20 and in accordance with In contrast, slightly stronger expression of CD20 was noted one study,21 we found five of 35 cases with more than 50% here in 11q deletion cases. The most striking novel difference of CD80-positive cells. Expression of CD154 was negative or was demonstrated with CD45 isoforms CD45RA and low in all cases of the present study. In an earlier report,22 CD45RO. Deletion of 11q was associated with a remarkable elevated expression was noted in some CLL cases. attenuation of CD45RA expression, while enhancement of There is very little information concerning the expression of CD45RO expression was noted. Experimental evidence indi- CD124, CD126 or CD130 in CLL. The regulatory functions cates that there may be negatively acting trans-factors that of CD124 on human normal and CLL B cells are gradually allow the skipping of alternate exons, while the full-length iso- emerging,23,24 but very little quantitative data about the form of CD45 represents the default pattern of splicing.16 Itis expression are available. The moderate but variable tempting to think that this kind of differential splicing existed expression (37–98% positive cells) of CD124 in CLL, as noted in our cases, where 11q23 deletion has been carefully charac- here, deserves further investigation regarding the pathogenesis terized.33–35 Although chromosome 11q22–23 is a gene rich and minimal residual disease diagnostics of CLL. The area, we are not aware of any gene which might change the expression patterns of CD126 and CD130 have been the sub- CD45 splicing strategy. ject of only a few studies. In a murine hybridoma model a It is of interest that trisomy 12 had the opposite effect on negative correlation has been noted between IL-6R alpha and CD45RA/CD45RO expression compared with 11q deletion. IL-6 expression, suggesting negative feedback regulation.25 The mechanisms of this on CD45 splicing remain to be Allelic exclusion ensures that functional cells never contain determined. Similarly, contrasting ratios of mIg␬ expression 26 more than one VLJL unit. In other words, the expressions of were noted when 11q deletion and trisomy 12 cases were mIg␬ and mIg␭ are mutually exclusive. In contrast, it is not compared. These observations supported the idea that CLL understood why two splicing variants of the CD45 molecule, cases with 11q deletion and those with trisomy 12 may rep- ie CD45RA and CD45RO, are differentially expressed in CLL. resent independent disease entities. It is interesting to note that The heterogeneous expression of these two variants has pre- 11q23 deletion is also frequent in B cell prolymphocytic viously been noted in CLL, but the interpretations of its back- leukemia, which ususally has a fast progression.36 ground have been conflicting. One possibility is that CLL A recent cDNA mRNA profiling analysis from this same expressing different isoforms may originate from neoplastic patient cohort revealed some interesting gene expression pat- expansion of either CD45RO-positive or CD45RO-negative B terns concerning CD14, CD25, CD79b.37 CD25 mRNA was cells.27 One study group suggested that CLL cells expressing upregulated in 11q deletion patients. The same tendency, CD45RO may be more mature cells than those expressing albeit statistically not significant, was observed at the only CD45RA,28 while another group concluded that the level in this study. Another similarity was a higher expression expression of CD45RO in CLL may reflect relative immaturity of CD14 in more advanced disease stage, ie Binet A vs B + C of these cells.16 The distinctive association of 11q deletion and (results not shown). Furthermore, CD79b mean expression at trisomy 12 with CD45RA vs CD45RO expression supports the the mRNA level was slightly higher in the faster-progressing hypothesis that these splicing variants are selected early in the cases.37 This was not observed at the protein level in the malignant process in CLL. We revealed two other antigen current work. The explanation may be that the extracellular pairs where expression tended to be mutually exclusive: epitope of CD79b is not necessarily expressed,38 despite the CD80 vs CD38 and CD45RA vs CD80. The present investi- presence of the pertinent mRNA species.39 gation did not reveal any remarkable antigen expression In conclusion, our results revealed a large group of CLL sur- differences in CD38-positive vs CD38-negative cells. face membrane antigens with a heterogeneous expression pat- An interesting question from the clinical and cell biological tern. Some of them are previously unexplored. Many of these point of view is whether CLL could be subdivided to similar antigen expression patterns are case-specific and differ from maturational stages as in, for instance, acute lymphoblastic normal polyclonal B cells. Hence, if determined in the presen- leukemia. If CD27 is accepted as a memory cell marker, as tation of a case of CLL, they should be very useful in the diag- proposed,28–31 all cases of the current CLL cohort belonged to nostics of minimal residual disease after intensive chemo- that category (CD27+ cells: 59–98%, median 94%). Usually, therapy and stem cell transplantation. Moreover, the mIgM+ cells are regarded as more immature than the interrelationships of the expression of seemingly independent

Leukemia CLL membrane antigens J Hulkkonen et al 184 surface membrane antigens, as noted here, provide a new 16 Yu Y, Rabinowitz R, Polliac A, Ben-Bassat H, Schlesinger M. B- basis for more extensive exploration of their role in CLL patho- lymphocytes in CLL and NHL differ in the mRNA splicing pattern genesis. Several previously unknown correlations between of the CD45 molecule. Eur J Haematol 2000; 64: 376–384. 17 Ranheim EA, Cantwell MJ, Kipps TJ. Expression of CD27 and its antigen expression and genomic aberrations were found. It is ligand, CD70, on chronic lymphocytic leukemia B cells. Blood conceivable that some antigens may have a prognostic role 1995; 85: 3556–3565. and some could possibly be used as surrogate markers for 18 van Oers MH, Pals ST, Evers LM, van der Schoot CE, Koopman clonal chromosomal abnormalities. These possibilities remain G, Bonfrer JM, Hintzen RQ, von dem Borne AE, van Lier RA. to be investigated in a considerably larger patient group. Expression and release of CD27 in human B-cell malignancies. Blood 1993; 82: 3430–3436. 19 Sembries S, Pahl H, Stilgenbauer S, Do¨hner H, Schriever F. Reduced expression of adhesion molecules and cell signaling Acknowledgements receptors by chronic lymphocytic leukemia cells with 11q deletion. Blood 1999; 93: 624–631. This work was supported by grants from the Medical Research 20 van den Hove LE, van Gool SW, Vandenberghe P, Bakkus M, Thi- Fund of Tampere University Hospital and the Finnish Foun- elemans K, Boogaerts MA, Ceuppens JL. CD40 triggering of dation for Cancer Research. We thank Leena Pankko and chronic lymphocytic leukemia B cells results in efficient alloan- Merja Suoranta for their skilful technical assistance. tigen presentation and cytotoxic T lymphocyte induction by up- regulation of CD80 and CD86 costimulatory molecules. Leukemia 1997; 11: 572–580. 21 Tsukada N, Aoki S, Maruyama S, Kishi K, Takahashi M, Aizawa Y. References The heterogenous expression of CD80, CD86 and other adhesion molecules on leukemia and lymphoma cells and their induction 1 Kipps T. Chronic lymphocytic leukemia. Curr Opion Hematol by interferon. J Exp Clin Cancer Res 1997; 16: 171–176. 2000; 7: 223–232. 22 Schattner EJ, Mascarenhas J, Reyfman I, Koshy M, Woo C, Fried- 2 Meinhardt G, Wendtner C, Hallek M. Molecular pathogenesis of man SM, Crow MK. Chronic lymphocytic leukemia B cells can chronic lymphocytic leukemia: factors and signaling pathways express CD40 ligand and demonstrate T-cell type costimulatory regulating cell growth and survival. J Mol Med 1999; 77: 282–293. capacity. Blood 1998; 91: 2689–2697. 3 Dohner H, Stilgenbauer S, Benner A, Leupolt E, Krober A, Bul- 23 McKay C, Hewitt E, Ozanne B, Cushley W. A functional role for linger L, Dohner K, Bentz M, Lichter P. Genomic aberrations and interleukin (IL-)4-driven cyclic AMP accumulation in human B survival in chronic lymphocytic leukemia. N Engl J Med 2000; lymphocytes. Cytokine 2000; 12: 731. 343: 1910–1916. 24 Fluckiger AC, Briere F, Zurawski G, Bridon JM, Banchereau J. IL- 4 Bennett J, Catovsky D, Daniel M, Flandrin G, Galton D, Gralnick 13 has only a subset of IL-4-like activities on B chronic lympho- H, Sultan C. Proposals for the classification of chronic (mature) B cytic leukaemia cells. Immunology 1994; 83: 397–403. and T lymphoid leukaemias. J Clin Pathol 1989; 42: 567–584. 25 Iwasaki T, Hamano T, Fujimoto J, Kakishita E. Regulation of 5 Matutes E, Owusu-Ankomah K, Morilla R, Garcia Marco J, Houli- interleukin-6 and interleukin-6R alpha (gp80) expression by han A, Que TH, Catovsky D. The immunological profile of B-cell disorders and proposal of a scoring system for the diagnosis of murine immunoglobulin-secreting B cell hybridomas. Immu- CLL. Leukemia 1994; 8: 1640–1645. nology 1998; 93: 498–504. 6 Rozman C, Montserrat E, Rodriguez-Fernandez J et al. Bone mar- 26 Yancopoulos G, Alt F. Regulation of the assembly and expression row histologic pattern – the best single prognostic parameter in of variable-region . Ann Rev Immunol 1986; 4: 339–348. chronic lymphocytic leukemia: a multivariate survival analysis of 27 Zola H, Melo J, Zowtyj H, Nikoloutsopoulos A, Skinner J. The 329 cases. Blood 1984; 64: 642–648. leukocyte-common antigen (CD45) complex and B-lymphocyte 7 Dohner H, Stilgenbauer S, Do¨hner K, Bentz M, Lichter P. Chromo- activation. Hum Immunol 1990; 27: 368–377. some aberrations in B-cell chronic lymphocytic leukemia: reas- 28 Maddy AH, Sanderson A, Mackie MJ, Smith SK. The role of cell sessment based on molecular cytogenetic analyses. J Mol Med maturation in the generation of phenotypic heterogeneity in B-cell 1999; 77: 266–281. chronic lymphocytic leukaemia. Immunology 1989; 68: 346–352. 29 Agematsu K, Nagumo H, Yang F et al. B cell subpopulation separ- 8 Zupo S, Isnardi L, Megna M, Massara R, Malavasi F, Dono M, + Cosulich E, Ferrarini M. CD38 expression distinguishes two groups ated by CD27 and crucial collaborations of CD27 B cells and of B-cell chronic lymphocytic leukemias with different responses helper T cells in immunoglobulin production. Eur J Immunol to anti-IgM antibodies and propensity to apoptosis. Blood 1996; 1997; 27: 2073–2079. 88: 1365–1374. 30 Klein U, Rajewsky K, Kuppers R. Human immunoglobulin + + 9 Lopez-Matas M, Rodriguez-Justo M, Morilla R, Catovsky D, (Ig)M IgD peripheral blood cells expressing the CD27 cell sur- Matutes E. Quantitative expression of CD23 and its ligand CD21 face antigen carry somatically mutated variable region genes: in chronic lymphocytic leukemia. Haematologica 2000; 85: CD27 as a general marker for somatically mutated (memory) B 1140–1145. cells. J Exptl Med 1998; 188: 1679–1689. 31 Tangye SG, Weston KM, Raison RL. Interleukin-10 inhibits the in 10 Matutes E, Catovsky D. The value of scoring systems for the diag- + nosis of biphenotypic leukemia and mature B cell disorders. Leuk vitro proliferation of human activated leukemic CD5 B cells. Leuk Lymphoma 1994; 13: 11–14. Lymphoma 1998; 31: 121–130. 11 Vilpo J, Vilpo L, Hulkkonen J, Lankinen M, Kuusela P, Hurme 32 Goldsby R, Kindt T, Osborne B. Kuby Immunology (4th edn). WH M. Non-specific binding compromises the purification yields of Freeman: New York, 2000. leukemic B cells in chronic lymphocytic leukemia: prevention by 33 Karhu R, Knuutila S, Kallioniemi OP, Siitonen S, Aine R, Vilpo L, collagen coating. Eur J Haematol 1998; 60: 65–67. Vilpo J. Frequent loss of the 11q14–24 region in chronic lympho- 12 Vilpo J, Koski T, Vilpo L. Selective toxicity of vincristine against cytic leukemia: a study by comparative genomic hybridization. chronic lymphocytic leukemia cells in vitro. Eur J Haematol 2000; Tampere CLL Group. Gene Chrosome Cancer 1997; 19: 286–290. 65: 1–9. 34 Larramendy M, Siitonen S, Zhu Y, Hurme M, Vilpo L, Vilpo J, 13 Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis Knuutila S. Optimized mitogen stimulation induces proliferation and display of genome-wide expression patterns. Proc Natl Acad of neoplastic B cells in chronic lymphocytic leukemia: signifi- Sci USA 1998; 95: 14863–14868. cance for cytogenetic analysis. Cytogenet Cell Genet 1998; 82: 14 Zupo S, Massara R, Dono M, Rossi E, Malavasi F, Cosulich M, 215–221. Ferrarini M. Apoptosis or plasma cell differentiation of CD38-posi- 35 Zhu Y, Monni O, El-Rifai, W, Siitonen S, Vilpo L, Vilpo J, Knuutila, tive B-chronic lymphocytic leukemia cells by cross-linking of sur- S. Discontinuous deletions at 11q23 in B cell chronic lymphocytic face IgM or IgD. Blood 2000; 95: 1199–1206. leukemia. Leukemia 1999; 13: 708–712. 15 Damle RN, Wasil T, Fais F et al. Ig V gene mutation status and 36 Lens D, Matutes E, Catovsky D, Coignet LJ. Frequent deletions CD38 expression as novel prognostic indicators in chronic lym- at 11q23 and 13q14 in B cell prolymphocytic leukemia (B-PLL). phocytic leukemia. Blood 1999; 94: 1840–1847. Leukemia 2000; 14: 427–430.

Leukemia CLL membrane antigens J Hulkkonen et al 185 37 Aalto Y, El-Rifai W, Vilpo L, Ollila J, Nagy B, Vihinen M, Vilpo 39 Alfarano A, Indraccolo S, Circosta P et al. An alternatively spliced J, Knuutila S. Distinct gene expression profiling in chronic lympho- form of CD79b gene may account for altered B cell receptor cytic leukemia with 11q23 deletion. Leukemia 2001; 15: 1721– expression in B-chronic lymphocytic leukemia. Blood 1999; 93: 1728. 2327–2335. 38 Garcia Vela J, Delgado I, Benito L, Monteserin M, Garcia Alonso 40 Koski T, Karhu R, Visakorpi T, Vilpo L, Knuutila S, Vilpo J. Com- L, Somolinos N, Andreu M, Ona F. CD79b expression in B cell plex chromosomal aberrations in chronic lymphocytic leukemia chronic lymphocytic leukemia: its implication for minimal are associated with cellular drug and irradiation resistance. Eur J residual disease detection. Leukemia 1999; 13: 1501–1505. Haematol 2000; 65: 32–39.

Leukemia