A Novel Nonobese Diabetic/Severe Combined Immunodeficient Xenograft Model for Chronic Lymphocytic Leukemia Reflects Important Clinical Characteristics of the Disease Free

In the past 10 years, several xenograft models of human chronic lymphocytic leukemia (CLL) using immunocompromised severe combined immunodeficient (SCID) mice have been reported (1–4). Work in this field has generally been hampered by inefficient engraftment (1) and EBV transformation of human cells (2). Recently, some of these pitfalls were successfully addressed by Shimoni et al. (3), who transplanted CLL peripheral blood mononuclear cells (PBMC) into the peritoneal cavity of lethally irradiated mice radioprotected with bone marrow from SCID mice. Importantly, in their initial description of this model, the authors found that PBMC from donors with Rai stage 0 disease led to marked engraftment of human T cells but comparably few leukemic B cells. In contrast, chimeric mice receiving PBMC from patients with advanced CLL (Rai stage III and IV) exhibited a reverse engraftment pattern with a predominance of tumor cells over accessory T cells (3). In a follow-up study, the same authors elegantly showed that autologous T cells can control CLL cell engraftment and suggested that their model could provide a useful tool for the investigation of the pathogenesis and disease progression in CLL (4). However, limitations of this approach have also been noted in that engraftment of CLL cells was followed only over a comparably short period of 2 weeks and the site of engraftment was restricted to the peritoneal cavity.

Following up on the above described studies, we aimed at developing a novel xenograft model for human CLL by transplanting primary CLL PBMC into nonobese diabetic (NOD)/SCID mice and analyzing the engraftment potential of these cells in various murine tissues over a time period of up to 12 weeks. Furthermore, the engraftment capacity of CLL cells was correlated with the disease stage and the molecular risk factor profile of the individual donors.

Patients and isolation of CLL cells. For establishment and evaluation of the model, peripheral blood samples from 39 CLL patients treated at the University Hospital Essen between December 2004 and December 2006 were used (for details, see Table 1). In all patients, the morphologic diagnosis of CLL was confirmed by flow cytometry revealing a typical CD19⁺CD5⁺CD23⁺ immunoglobulin light chain (κ or λ light chain)–restricted immunophenotype. Expression of ZAP-70 and CD38 was determined by flow cytometry as recently described (5). Patient selection was based on a WBC count exceeding 20/nL and a treatment-free interval of ≥6 months before sample acquisition. Heparanized whole-blood cell samples (peripheral blood, 20–50 mL) were obtained during routine follow-up visits to our institutions. Informed consent was obtained according to institutional guidelines. Peripheral blood mononuclear cells (PBMC) were isolated within 3 h of venipuncture by Ficoll-Hypaque (Pharmacia) density centrifugation. The interlayer was collected, washed twice, counted on a hemocytometer, and resuspended in Iscove's modified Dulbecco's medium (IMDM; Life Technologies, Inc.) at the desired cell concentration and transplanted within 2 h. In five experiments, PBMC before transplantation were depleted of CD3⁺ human T cells using the MidiMACS isolation system (Miltenyi Biotech). Purity of CD19⁺ cells after T-cell depletion was >99.5% as analyzed by flow cytometry. T cell–depleted grafts were compared with T cell–enriched grafts, in which 0.5 × 10⁷ to 1.9 × 10⁷ T cells (equivalent to 5–18% of total cells and double the amount of T cells present in the original PBMC sample) were added to the grafts before transplantation.

Transplantation of NOD/SCID mice. NOD/SCID mice of both genders were bred and handled under sterile condition and animals were transplanted at ages 8 to 15 weeks. PBMC were transplanted into NOD/SCID mice at 30 to 60 min after sublethal irradiation (3.0–3.25 Gy) of the animals using i.v. and/or i.p. application. A total of 1.0 × 10⁸ PBMC suspended in 0.2 mL of IMDM were injected per chosen route.

Collection of cells from NOD/SCID mice. Peripheral blood was obtained every 2 weeks from the tail veins of transplanted animals. For harvest of cells from the peritoneal cavity, bone marrow, and spleen, mice were sacrificed by cervical dislocation 4, 8, or 12 weeks after transplantation or earlier when they became listless or unwell. Peritoneal cells were obtained by peritoneal lavage with 5 mL of 0.9% NaCl whereas bone marrow cells were obtained by flushing the femoral, tibial, and hip bones with IMDM and afterward filtering cells through 70-μm nylon sieves (Becton Dickinson). For analysis of engraftment within the murine spleen, one half of the organ was fixed immediately in buffered 3.7% formalin and afterward embedded in paraffin for further histologic and immunohistochemical analyses. The other half was mechanically homogenized, pressed through 70-μm nylon sieves (BD Falcon), and suspended in PBS for further analysis by flow cytometry.

Characterization of human cell engraftment in NOD/SCID mice using flow cytometry. For detection of cell-surface antigens, single-cell suspensions obtained from the different organs were incubated for 30 min on ice with a mixture of the indicated appropriate fluorescently labeled monoclonal antibodies. In some experiments, cell-surface antigens were first stained and fixed, followed by permeabilization of the plasma membrane (Cytofix/Cytoperm, Becton Dickinson) and staining of intracellular antigens. After washing, four-color flow cytometric analysis of human antigens was done on a Coulter Epics XL analyzer (Coulter Electronics). Erythrocyte lysis (Pharmlyse, Becton Dickinson) of the cells was done before staining. The following directly labeled antibodies were used for the characterization of human cells: CD45-cyanine dye (Cy5), CD19-phycoerythrin, CD5-FITC and CD23-energy coupled dye, λ light chain (FITC, Kallestad), κ light chain (FITC, Kallestad), CD38 phycoerythrin (Becton Dickinson), CD100 (FITC, Serotec), and survivin (FITC, R&D Systems) for CLL cells; CD45-Cy5, CD4-phycoerythrin, CD8-FITC, and CD3-energy coupled dye for human T cells. Antibodies were purchased from Beckman Coulter Immunotech except indicated otherwise. For the determination of ZAP-70, we used an indirect staining approach as previously described (5) using anti-ZAP-70 (clone 2F3.2, Upstate Biotechnology) as a primary antibody and goat-anti-mouse immunoglobulin-FITC (DAKO) as secondary antibody. Phycoerythrin-, phycoerythrin-cyanin 5.1–, energy coupled dye–, and FITC-conjugated mouse isotypic monoclonal antibodies served as controls and were included in all experiments.

Bromodeoxyuridine incorporation experiments. Analysis of 5-bromodeoxyuridine (BrdUrd) incorporation was done using the FITC BrdUrd Flow kit (PharMingen) after a single i.p. injection of BrdUrd (PharMingen; 1 mg/6 g of mouse weight within 5 days after transplantation) and admixture of 1 mg/mL BrdUrd (Sigma) to drinking water for 5 days within week 1 and week 3 each, after transplantation (6). Analysis for BrdUrd incorporation in recovered CLL cells was only done when splenic engraftment >1% CLL cells of total spleen cells was evident.

Fluorescence in situ hybridization analysis. To detect specific anomalies of the leukemic clone in chromosomal regions 11q, 13q, and 17p in cells recovered after transplantations, the following fluorescence-labeled DNA probes were used for interphase cytogenetic analyses: locus-specific identifier (LSI) ATM (11q23), LSI D13S319 (13q14), LSI p53 (17p13.1), and chromosome enumeration probe (CEP) 12 (centromere 12). All probes were purchased from Abbot Vysis. Sample preparation, fluorescence in situ hybridization (FISH), and counterstaining with 4,6-diamidino-2-phenylindole dihydrochloride were done as previously reported (5).

Histologic staining of spleen sections. Four-micrometer-thick sections of paraffin-embedded spleen sections were obtained and stained with H&E according to standard procedures. Additional serial sections were prepared for immunohistochemical investigation. Immunohistochemistry was done using the EnVision+ System peroxidase method (DAKO) with monoclonal mouse antibodies (immunoglobulin G) against human leukocyte common antigen (CD45, clones 2B11+PD7/26, DakoCytomation; dilution, 1:1,000), CD3 (clone F7.2.38, DakoCytomation; dilution, 1:100), CD20 (clone L26, DakoCytomation; dilution, 1:4,000), CD21 (clone 1F8), and Ki67 (clone BGX-297, ready to use, BioGenex).

For double immunohistochemistry, the same CD3 and CD20 antibodies and an additional Ki67 antibody were used (Ki67, clone SP6, DCS; dilution, 1:1,200). Antigen retrieval was carried out at 98°C for 10 min in a water bath (Target retrieval buffer, DAKO). The CD3/Ki67 and CD20/Ki67 double labelings were done in two steps. First, Ki67 labeling with an immunohistochemical staining technique based on a horseradish peroxidase (HRP)–labeled polymer that is conjugated to secondary antibodies (ZytoChemPlus HRP Polymer Kit, Zytomed Systems) was done. Staining was completed by an incubation with 3,3′-diaminobenzidine + substrate-chromogen (Zytomed Systems), which results in a brown-colored precipitate at the antigen site. Second, CD3 or CD20 labeling was done with an alkaline phosphatase–labeled polymer [ZytoChemPlus (AP) Polymer Kit, Zytomed Systems] and developed with a Permanent Red chromogenic substrate system (Zytomed Systems). At the end of the procedure, nuclei were counterstained with hematoxylin for 5 min.

In situ hybridization. For the detection of a latent EBV infection by in situ hybridization, sections from paraffin-embedded tissue were analyzed. In brief, slides were dewaxed, rehydraded, and dried at 45°C. Afterward, they were pretreated with proteinase K for 25 min at 37°C. Two drops of EBV-specific FITC-labeled oligonucleotide probe [DAKO, Epstein-Barr Virus (EBER) PNA Probe/Fluorescein] were added to the tissue section and slides were covered with coverslips. The hybridization was carried out for 90 min at 55°C. After washing with diethyl pyrocarbonate water, oligonucleotide binding was indirectly detected with an anti-FITC antibody conjugated to alkaline phosphatase (DAKO, PNA ISH detection kit).

Statistical analysis. Data are presented as the mean ± SE unless otherwise specified. Statistical analysis was done using SPSS 13.0 for Windows (SPSS, Inc.). The significance of differences between groups was determined with the Friedman test (differences between the application routes and differences in CLL or T-cell engraftment within the different tissue compartments), the Kruskal-Wallis test (differences in engraftment according to stage), the Wilcoxon test (differences between the mean fluorescence intensities of survivin and CD100 expression of transplanted and engrafted CLL cells and differences in CLL or T-cell engraftment within the same compartment), or the Mann-Whitney U test [differences in engraftment according to lactate dehydrogenase (LDH), lymphocyte doubling time, or prognostic parameters], with P ≤ 0.05 considered to be significant.

All in all, the transplantation of 43 primary human peripheral blood-derived CLL samples from 39 different CLL patients was investigated in a total of 167 NOD/SCID mice. The clinical characteristics of the patients are shown in Table 1.

In initial experiments using 5 different CLL samples and 35 NOD/SCID mice, the most suitable route for transplantation was determined by transplanting 1 × 10⁸ CLL cells from the same sample either i.p. (n = 12 mice), i.v. (n = 12 mice), or via both routes (i.p. + i.v.; n = 11 mice). Figure 1A shows representative flow cytometry plots of CLL cells before transplantation and of human cells recovered from the bone marrow, spleen, and peritoneal cavity of a NOD/SCID animal showing that the recovered cells from all three compartments exhibited the CD19⁺CD5⁺CD23⁺ immunophenotype characteristic of CLL cells. On quantitative analysis, the number of recovered CLL cells was significantly higher in the spleen and peritoneal cavity as compared with the bone marrow (Fig. 1B). Whereas CLL recovery in the murine spleen was higher following i.v. versus i.p. transplantation, cell recovery in the peritoneal cavity was mainly dependent on the i.p. route. Combination of both routes, however, yielded the best results (Fig. 1B).

We then proceeded to validate this model in a cohort of 39 CLL patients (Table 1). For these experiments, only the combined i.v. + i.p. injection route was used and a total of 126 mice were evaluated. In line with our previous results, analysis at week 4 revealed consistent engraftment of human CD45⁺ cells in 37 of 39 (94.9%) experiments. The highest number of CD45⁺ cells was detected in the murine spleens whereas cell recovery from bone marrow and peritoneal cavity was significantly lower (Fig. 1C; P = 0.00000001). Human T cells engrafted more efficiently than their leukemic counterparts, accounting for the majority of CD45⁺ cells in the murine spleen and bone marrow (Fig. 1C). Of note, a reverse engraftment pattern was noted in the murine peritoneal cavity, where the number of CLL cells recovered after 4 weeks exceeded that of the accessory T cells (P = 0.038; Fig. 1C). In the peripheral blood, comparably low numbers of CLL cells were detected in 12 of 18 and 8 of 22 samples 2 and 4 weeks after transplantation, accounting for 1.1 ± 0.5% and 0.7 ± 0.3% of the total peripheral blood leukocytes at these time points. Again, higher engraftment rate and absolute number of human T cells were detected in the peripheral blood of the transplanted animals 2 weeks (17 of 17) and 4 weeks (14 of 22) after transplantation, accounting for 10.2 ± 4.0% and 6.6 ± 3.5% of the total peripheral blood leukocytes, respectively.

Taken together, these results suggest that murine splenic tissue offers a more supportive microenvironment for the engraftment of CLL PBMC than peritoneal cavity, bone marrow, and peripheral blood.

Individual reproducibility. To evaluate the reproducibility of our model in four individual patients showing high, moderate, low, or no splenic CLL cell recovery, transplantation was repeated at a later time point. Comparative analysis of human CLL cells recovered from the spleens in these four paired experiments showed reproducible recovery of CLL cells for individual patients, and 29.9 × 10⁵ versus 19.6 × 10⁵, 15.8 × 10⁵ versus 15.4 × 10⁵, and 2.2 × 10⁵ versus 11.2 × 10⁵ recovered cells, and no engraftment versus no engraftment were observed. To unequivocally confirm that the recovered CD45⁺CD19⁺CD5⁺CD23⁺ cells originated from the transplanted leukemic cell clones, we carried out comparative FISH and immunoglobulin light chain restriction analyses in a series of four cases with informative cytogenetic aberrations. Reassuringly, concordant results with both assays were observed in all of the four cases (Supplementary Fig. S1).

Time course of CLL cell engraftment. In a total of five experiments, the recovery of human CLL cells in the spleen as well as the peritoneal cavity of NOD/SCID mice was evaluated over a more extended time period. Whereas absolute numbers of recovered CLL cells remained relatively stable between weeks 4 and 8 (spleen, 5.5 ± 3.2 × 10⁵ versus 2.4 ± 1.4 × 10⁵; peritoneal cavity, 0.6 ± 0.3 × 10⁵ versus 3.2 ± 2.9 × 10⁵); at week 12, a substantial decline of cells was noted in the spleen (0.6 ± 0.3 × 10⁵) as well as the peritoneal cavity (0.2 ± 0.1 × 10⁵).

In view of the fact that the highest absolute number of human cells was consistently observed in the murine spleens, we further investigated the histomorphologic appearance of these cells using H&E as well as immunohistologic staining of spleen sections. All in all, 46 mice transplanted with 15 different CLL samples were analyzed (Supplementary Table S1). Strikingly, a focal infiltration pattern of human lymphoid CD45⁺ cells situated along the splenic arterioles was observed in the majority of animals (37 of 46; representing 14 of 15 CLL samples; Fig. 2A). On immunohistochemistry, the majority of cells in these focal aggregates exhibited a CD20⁺ CLL phenotype (88.3 ± 3.6%; Fig. 2B), whereas a smaller proportion represented accessory CD3⁺ T cells (11.0 ± 3.4%; Fig. 2C). In some cases, these focal aggregates consisted almost exclusively (>99%) of CD20⁺ CLL cells (Supplementary Table S1). The majority of the H&E-stained lymphoid CD20⁺ cells morphologically exhibited typical features of prolymphocytes: cells were larger than a normal lymphocyte and their nuclei contained finely distributed chromatin as well as small nucleoli (Fig. 2D,, arrows). Besides this population, a fraction of paraimmunoblasts characterized by a nucleus containing a solid centrally located nucleolus was observed (Fig. 2D , cones). When the potential presence of human follicular dendritic cells within the splenic aggregates was evaluated by immunostaining of CD21, no follicular dendritic cells were detected.

A minority of animals (9 of 46; 6 of 15 CLL samples; Supplementary Table S1) showed a more diffuse infiltration pattern of human cells consisting mostly of CD3⁺ T cells (72.5 ± 8.4%) and less frequently of CD20⁺ CLL cells (27.5 ± 8.4%). Of note, histomorphologic analysis of spleen sections 12 weeks after transplantation frequently revealed a considerable amount of fibrosis with a consecutive loss of structural integrity of the murine splenic tissue, which was not detectable 4 weeks after transplantation. To rule out EBV-transformed lymphoma cells as the source of the recovered human cells in several experiments, additional in situ hybridization analysis for EBV-encoded RNAs was done (5 CLL samples, n = 10 mice), revealing no EBV-RNA–positive cells within the focal aggregates (data not shown).

To further characterize the human CLL cells recovered from the NOD/SCID spleens, a panel of surface and intracellular marker molecules was used. Given the histologic similarities between the focal aggregates observed in our NOD/SCID model and the pseudofollicular proliferation centers described in human CLL (7), this analysis focused on marker molecules described as characteristic of CLL cells located in proliferation centers, survivin, and CD100 (8–10). In 23 mice presenting with focal aggregates in the spleen, consistently a substantial percentage of human CD45⁺ cells (19.5 ± 2.7%) stained positive for Ki67. Analysis of focal aggregates composed almost exclusively (>99%) of CD20⁺ CLL with only sporadic interspersed T cells indirectly confirmed that Ki67 was indeed expressed in CLL cells (data not shown). In addition, Ki67 expression in CLL cells and absence of Ki67 expression in human T cells could be shown directly by double immunostaining of Ki67 and CD20 as well as CD3 (Fig. 3A–C). Here, hardly any Ki67 staining was detected in human CD3⁺ T cells (Fig. 3B), but coexpression of Ki67 and CD20 was evident in a substantial fraction of CLL cells recovered from the murine spleens (Fig. 3C). Of note, high-power magnification of splenic focal CLL aggregates revealed that the Ki67⁺ cells exhibited morphologic features of either paraimmunoblasts or prolymphocytes (Fig. 3D).

Activation of CLL cells within proliferation centers has also been shown to result in increased expression of survivin, one of the inhibitor of apoptosis proteins, as well as up-regulation of the semaphorin protein family member CD100 (semaphorin 4D; refs. 8–10). As depicted in Supplementary Fig. S2, the same pattern of activation was observed in our CLL model. Of note, significant up-regulation of survivin (P = 0.028; Supplementary Fig. S2B) as well as CD100 (P = 0.043; Supplementary Fig. S2D) was detected in leukemic cells recovered from NOD/SCID spleens when compared with the initially transplanted CLL cells.

To further investigate the proliferative potential of human CLL cells recovered from NOD/SCID spleens, 5-bromodeoxyuridine (BrdUrd) incorporation during a 1- to 3-week in vivo exposure period was analyzed in five samples showing typical focal aggregates in the corresponding histomorphologic analyses. In these experiments, 6.7% to 25.0% of the recovered splenic CLL cells stained positive for BrdUrd (Supplementary Fig. S3), indicating that a significant proportion of splenic CLL cells located within splenic aggregates had entered the cell cycle during the BrdUrd exposure period.

We then proceeded to correlate splenic CLL PBMC engraftment with the disease characteristics observed in individual patients at the time of transplantation (Fig. 4). Here, we found that parameters reflecting disease activity (e.g., Binet stage, lymphocyte doubling time, and LDH serum activity) exhibited a significant positive correlation with CLL cell engraftment (Fig. 4A). Disease stage also affected the histologic appearance of the transplanted cells in the murine spleens in that samples from Binet stage C patients exclusively engrafted in a follicular pattern. By contrast, all of the nine cases with a diffuse infiltration pattern were observed in samples obtained from Binet stage A and B patients (Supplementary Table S1). In line with these findings, samples obtained from poor-prognosis patients, as defined by expression of CD38 (8, 11–13), ZAP-70 (5, 14–17), unmutated IgVH genes (11, 18, 19), or high-risk cytogenetics (20), generally showed a higher leukemic cell engraftment than CLL cells isolated from patients with good prognosis (Fig. 4B). However, the observed differences failed to reach statistical significance, suggesting that the model may better reflect disease activity at the time of venipuncture rather than the long-term clinical course of individual cases.

Interestingly, increased leukemic B-cell engraftment in advanced disease stages inversely correlated with the number of recovered accessory T cells (Fig. 5), suggesting that the biological properties of CLL T cells may change in the course of the disease. To further define the role of accessory human T cells in this model, comparative transplantation experiments using T cell–depleted and T cell–enriched grafts were done. In line with the results shown in Fig. 5, we found that T cells isolated from patients with advanced/active disease seemed to increase leukemic cell recovery as compared with T cells obtained from patients with early/inactive disease (Supplementary Table S2). However, these potentially interesting findings are limited by the small number of patients analyzed and thus need to be confirmed in a larger series. Furthermore, these experiments did not reveal a consistent effect of cotransplanted T cells on either histomorphologic appearance (focal aggregate formation) or survivin and CD100 expression by the leukemic cells (data not shown).

We here describe a novel and robust in vivo model of human CLL using NOD/SCID mice as recipients of primary peripheral blood-derived CLL cells. In this model, we show reproducible and stable recovery of PBMC from CLL patients over a time span of 8 weeks after transplantation. Comparing different routes of transplantation, the highest number of human CD45⁺ cells was detected in the murine spleen followed by the peritoneal cavity, bone marrow, and peripheral blood, suggesting that splenic tissue may offer the most suitable microenvironment for the transplanted cells.

Given that the highest number of engrafted human cells was observed in the murine spleen, we aimed to further investigate the morphologic and functional characteristics of these cells in this tissue compartment. Histomorphologic evaluation of spleen sections revealed focal aggregates of CD45⁺ human cells consisting of both CLL and accessory T cells, which were clustered around splenic arterioles. At higher magnification, these focal aggregates were found to contain a mixture of paraimmunoblasts, prolymphocytes, and interspersed small lymphocytes. Further immunohistochemical evaluation showed that the prolymphocytic and paraimmunoblastic cells exhibit a CD20⁺CD23⁺CD5⁺ phenotype and that a substantial proportion of these cells stain positive for the proliferation marker Ki67. The proliferative potential of these cells was confirmed by BrdUrd DNA incorporation experiments, which unequivocally showed that a substantial proportion of the transplanted CLL cells entered the cell cycle during the 4-week posttransplant period. Consistent with their proliferative potential and their “activated morphology,” these cells were found to express the B-cell activation markers survivin and CD100, as detected by multiparameter flow cytometry. Both of these molecules have previously been shown to exert prosurvival effects in CLL cells and may play an important functional role in the molecular pathogenesis of the disease (8–10). Whereas, to our knowledge, this is the first detailed report on the occurrence of splenic focal aggregates following the transplantation of CLL cells into mice, this is probably not a unique characteristic of CLL cells as previous work by Burakova et al. (21) has shown that transplanted normal human PBMC form follicles in the spleen and lymph nodes of SCID mice.

Clearly, the most important finding of this study is that the engraftment capacity of leukemic cells closely correlated with the disease characteristics observed in individual patients at the time of transplantation. In line with data published by the Reisner group (3, 4), we found that transplantation of PBMC obtained from patients in stage Binet A resulted in marked i.p. engraftment of human T cells, whereas only few leukemic cells were recovered from the peritoneal cavity of the recipient animals. Conversely, infusion of samples from donors with Binet C disease led to the predominant engraftment of tumor cells compared with a low number of accessory T cells. Importantly, using the combined i.v./i.p. transplantation route, which has previously not been used in this experimental setting, we observed a similar disease stage–dependent engraftment profile in the murine spleen and peritoneal cavity. However, absolute human cell recovery was higher in the murine spleen as compared with peritoneal cavity, raising the possibility that using splenic CLL cell engraftment as readout may be advantageous due to a potentially higher sensitivity of the assay system. Of note, disease stage and cellular composition of the human graft also affected the histologic appearance of the transplanted cells in the murine spleens in that samples from Binet stage C patients exclusively engrafted in a follicular pattern (Supplementary Table S 1). By contrast, all of the nine cases with a diffuse infiltration pattern were observed in samples obtained from Binet stage A and B patients. Furthermore, we found that parameters reflecting disease activity other than Binet stage [ref. 22; e.g., lymphocyte doubling time (23–25) and LDH serum activity] also exhibited a significant positive correlation with CLL cell engraftment. Similarly, when samples were grouped according to the expression/presence of the molecular risk factors (19) CD38 (13), ZAP-70 (5, 26), IgVH mutation status (11, 18), and karyotype (20), the prognostically unfavorable group generally showed a higher leukemic cell engraftment than CLL cells isolated from patients with good prognosis. However, the observed differences failed to reach statistical significance, suggesting that the model may better reflect disease activity at the time of venipuncture rather than the long-term clinical course of individual cases.

Taken together, leukemic cells from advanced-stage patients have a higher engraftment potential than early-stage CLL cells, whereas T cells derived from patients with advanced disease engraft less efficiently than their counterparts obtained from Binet stage A patients. This mutually exclusive pattern of engraftment first described by Shimoni et al. (4) raises the possibility that T cells may at least partly control the expansion of the tumor cell clone in vivo. Our data are compatible with the concept that T cells isolated from early-stage patients may be capable of suppressing tumor cell growth (27, 28) whereas T cells from Binet stage C may have lost their antineoplastic immunosurveillance function (29, 30). In support of this hypothesis, we found that addition of autologous T cells purified from Binet stage A patients reduces splenic leukemic cell engraftment whereas addition of T cells from patients with progressive/advanced disease increases leukemic cell recovery from murine spleens. In this context, it is worth noticing that T cells of CLL patients with indolent disease have been shown to exhibit a lower production of leukemic cell survival–promoting cytokines as compared with T cells from patient with advanced disease (31, 32).

In conclusion, our data show that PBMC from CLL patients can be effectively engrafted into NOD/SCID mice. Using combined i.v./i.p. transplantation, the highest number of engrafted cells is detected in the murine spleen, where these cells can be conveniently quantified and subjected to further functional analysis. Our finding that CLL cells obtained from advanced-stage patients with elevated LDH serum activity and short lymphocyte doubling time exhibit a higher engraftment potential as compared with samples derived from donors with indolent disease suggests that our model reflects the heterogeneity and important clinical features of the disease. Thus, we suggest to further evaluate this model as a tool for preclinical drug testing and the investigation of the pathophysiology of CLL.

Grant support: Förderverein Innere Klinik und Poliklinik (Tumorforschung) e.V. and Deutsche José Carreras Leukämie-Stiftung grant DJCLS R 06/24 (J. Dürig and T. Moritz).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank A. Feldmann, W. Stellberg, A. Führer, and B. Friedmann for excellent technical help, K. Metz for histopathologic advice, and J. Cancelas and D.A. Williams for helpful discussions.

This work is dedicated to Prof. G. Brittinger on the occasion of his 75th birthday.