MDM2 gene amplification and lack of p53 point mutations in Hodgkin and Reed–Sternberg cells: results from single-cell polymerase chain reaction and molecular cytogenetic studies

Authors


Lorenz Trümper MD, PhD, Department of Internal Medicine I, University of Saarland, D-66421 Homburg/Saar, Germany. E-mail: inltru@med-rz.uni-sb.de

Abstract

Hodgkin's disease (HD) is the most common haematological malignancy after chronic lymphocytic leukaemia, but very little is known about its pathogenesis or the genetic events that contribute to the malignant phenotype of the tumour cells. p53 is assumed to play an important role in the pathogenesis of HD, based on the observation that p53 protein is frequently accumulated in Hodgkin and Reed–Sternberg (H & RS) cells. We investigated single H & RS cells from five different HD patients for point mutations at the genomic level using multiplex polymerase chain reaction amplification and subsequent sequencing. No point mutations were detected in 50 single H & RS cells analysed. Hence, accumulation of p53 protein cannot be explained by mutations within the gene. A genome-wide screening for genomic imbalances using comparative genomic hybridization revealed gain on chromosome 12q14, i.e. the mapping position of the MDM2 gene in several HD cases. Therefore, we assessed the copy number of the MDM2 gene using fluorescence in situ hybridization. In four out of six HD cases analysed, the copy number of the MDM2 gene was found to be increased. As gene amplification is frequently associated with protein overexpression, the observed accumulation of p53 in the nuclei of H & RS cells could be as a result of elevated MDM2 protein levels resulting in stabilization of p53 protein.

Hodgkin's disease (HD) is characterized by the presence of large, usually multinucleated tumour cells, the so-called Hodgkin & Reed–Sternberg (H & RS) cells, that are surrounded by an abundant lymphohistiocytic infiltrate with varying degrees of fibrosis and sclerosis (Aisenberg, 1991). In most cases, H & RS cells represent less than 1% of the cellular infiltrates in involved lymph nodes. The molecular changes contributing to the malignant phenotype of the H & RS cell are still largely unknown. An interesting observation in the majority of HD tumours is an accumulation of the tumour-suppressor protein p53 in H & RS cell nuclei, indicative of a prolonged half-life (Doglioni et al, 1991; Gupta et al, 1992; Said et al, 1992). In normal cells, p53 protein is constitutively expressed at very low, mostly undetectable levels. In response to stimuli, such as DNA damage, levels are drastically raised and p53 finally mediates its functions as a growth suppressor or determines cells for apoptosis (Bates & Vousden, 1996). As p53 seems to be functionally inactive in HD, its accumulation has been interpreted as ‘abnormal protein stabilization’ (Doglioni et al, 1991; Said et al, 1992), possibly playing an important role in the pathogenesis of the tumour.

Two conditions are known that could explain unphysiological p53 accumulation in tumour cells: (i) point mutations within the gene resulting in prolonged half-life and high protein levels in cells (Blagosklonny, 1997), and (ii) binding to other proteins, such as ARF, viral proteins or MDM2 (Momand et al, 1992; Zhang et al, 1998; Prives & Hall, 1999).

In general, information addressing these conditions in H & RS cells is still sparse. Direct evidence for p53 point mutations at the single cell level was only obtained from one HD case at the mRNA level (Trümper et al, 1993). Other studies describing point mutations analysed genomic DNA either from H & RS cells embedded in a broad background of other cell types (Xerri et al, 1995; Chen et al, 1996) or from fractions of flow-sorted nuclei with a high DNA content, i.e. of rather undefined origin (Gupta et al, 1993). Finally, Sánchez-Beato et al (1996) provided indirect evidence for wild-type conformation of p53 in H & RS cells in 32 out of 32 HD cases using an immunohistochemical approach.

The most prominent cellular binding partner of p53 (that could lead to p53 accumulation) is the MDM2 gene product, which regulates p53 activity in an autoregulatory feedback loop (Wu et al, 1993). In many tumours, MDM2 overexpression, which is mostly as a result of gene amplification, was found to inactivate and stabilize p53 at the same time (Oliner et al, 1992; Keleti et al, 1996).

In order to determine the role of mutated p53 and MDM2 in HD, we performed detailed analyses on single H & RS cells. CD30-positive H & RS cells were subjected to a newly developed multiplex polymerase chain reaction (PCR) assay that allows the simultaneous amplification of exon 5 through to 9, in which 98% of all known p53 mutations reside (Hollstein et al, 1991). MDM2 copy number was assessed by comparative genomic hybridization (CGH) with genomic DNA universally amplified from H & RS cells, and combined analysis of CD30 immunostaining and interphase cytogenetics.

PATIENTS and METHODS

Patients HD patient data are summarized in Table I. Small parts of lymph node biopsies from seven patients (five men, two women) performed for diagnostic reasons were used for our analyses with permission and according to the guidelines of the local ethics review committee (‘Ethikkommission der Ärztekammer des Saarlandes’). According to the availability of material, we used five of these for mutation analysis of the p53 gene. A set of six patients was assessed using interphase cytogenetics. Histological diagnosis was made or confirmed by a member of the reference panel of the German Hodgkin Study Group.

Table I.  Clinical data of the cases included in this study.

Case
Number


Sex

Age
(years)

Histology
subtype
Immunohistology
EBV
status
CD 30CD 20CD 3
  1. Patient numbers were assigned consecutively when cells are processed in the laboratory and correspond to the ones described in Roth et al (1994), Daus et al (1995) and Trümper et al (1997). All cases are part of on ongoing study. CD-30/20/3-staining: positivity of H & RS cells for the respective antigens (Daus et al, 1995); EBV status: presence of EBV mRNA in single H & RS cells as determined using EBNA-1 PCR (Trümper et al, 1997) (cases 2, 9, 11, 12, 16) or using immunohistochemistry with an antibody against LMP-1 (case 17). cHL-MC, classic Hodgkin's lymphoma mixed-cellularity subtype; cHL-NS, classic Hodgkin's lymphoma nodular sclerosis subtype; NLPHL, nodular lymphocyte-predominant Hodgkin's lymphoma; NA, not assessed.

2Female32cHL-NS++
7Male63cHL-MC+NANA
9Male25cHL-MC+NA
11Male38NLPHL+++
12Male49cHL-MC++
16Female21cHL-MC+
17Male26cHL-NS+

Isolation of H & RS cells Cell suspension from fresh lymph nodes of patients with HD were prepared as described elsewhere (Küpper et al, 1996). For analysis, single H & RS cells were either aspirated from suspension with a microcapillary or picked from cytospin preparations with a micromanipulator (Eppendorf; Hamburg, Germany). H & RS cells were identified by positive staining for CD30 antigen, as well as cellular size and presence of multiple nuclei.

Amplification and sequencing of p53 Single H & RS cells from five male patients with HD, three with mixed cellularity (cHL-MC) (cases 7, 9, 12), one patient with nodular sclerosing (cHL-NS) subtype (case 17) and one case of nodular lymphocyte-predominant Hodgkin's lymphoma (NLPHL-n) (case 11) were subjected to PCR amplification of exons 5 through to 9, which was carried out as a two-step multiplex PCR assay using the primers indicated in Table II. The first amplification round was performed in 50 μl of reaction buffer [10 mmol/l Tris pH 8·8, 50 mmol/l KCl, 0·001% (w/v) gelatin], containing 200 μmol/l each nucleotide, 30 nmol/l each primer, 2·5 mmol/l MgCl2 and 1 U of Taq DNA-Polymerase (Taq2000; Stratagene Europe, Amsterdam, The Netherlands). Temperature cycles were 3 min denaturation at 95°C, 4 min annealing at 54°C and 80 s extension at 72°C, followed by 35 cycles of 1 min denaturation, 30 s annealing and 80 s extension (Robocycler, Stratagene Europe). After the last cycle, the 72°C step was extended to 8 min. In the second amplification round, 1 μl of the primary product was transferred into each of four test tubes containing fresh buffer (1·5 mmol/l MgCl2, 200 μmol/l per nucleotide, 0·5 U of Taq Polymerase) and only one nested primer pair specific for one exon (i.e. exons 5, 6, 7, as well as 8/9; see Table II) at a final concentration of 125 nmol/l each. Cycles were carried out as before, but the annealing temperature was raised to 56°C. As a negative control, we used a sample of the buffer of the cell suspension.

Table II.  Primer sequences used in the p53 multiplex PCR.
ExonPrimerSequenceProduct size
  1. Primers in the first amplification round were used together in one reaction tube. In the second amplification, the sample was divided into four parallel assays that contained only one primer pair.

First amplification
 55–5-1AAGCTCCTGA GGTGTAGACG C548 bp
 66–3CCAGTTGCAA ACCAGACCTC AGG 
 77–5-3CTTGCCACAG GTCTCCCCAA G211 bp
 77–3GTGGCAAGTG GCTCCTGACC TGG 
 88–5-1GGACCTGATT TCCTTACTGC C368 bp
 99–3CCCAAGACTT AGTACCTGAA G 
Second amplification
 55–5-1AAGCTCCTGA GGTGTAGACG C345 bp
 55–3CTGCTCACCA TCGCTATCTG AGC 
 66–5-1TGGAGAGACG ACAGGGCTGG TTGC195 bp
 66–3CCAGTTGCAA ACCAGACCTC AGG 
 77–5-2CCTGTGTTAT CTCCTAGGTT GG145 bp
 77–3GTGGCAAGTG GCTCCTGACC TGG 
 88–5-2CCTCTTGCTT CTCTTTTCCT ATCC348 bp
 99–3CCCAAGACTT AGTACCTGAA G 

Product bands were excised from 2% agarose gel (Seakem Agarose; Biozym, Eggenstein, Germany) and the DNA was subsequently recovered using a DNA extraction kit (Qiagen; Hilden, Germany). Purified PCR products were either sequenced directly (Cycle Sequencing Excel; Epicentre c/o Biozym) or first cloned into pCR 2·1 vector (Topo TA Cloning; Invitrogen, deSchelp, The Netherlands) and then sequenced (T7 Sequenase Quick Denature; Amersham, Braunschweig, Germany). For sequencing, 5′ and 3′ PCR primers from the second amplification round were used in independent reactions.

Analysis of pools of H & RS cells using degenerate oligo-primed (DOP) PCR and CGH Pools of 20–30 H & RS cells were collected and analysed as described elsewhere (Joos et al, 2000). Briefly, cells were proteinase K-digested and the genomic DNA universally amplified according to the protocol of Telenius et al (1992). CGH analysis including preparation of metaphase chromosomes, probe labelling, hybridization and image acquisition were performed as described previously (Lichter et al, 1995; Joos et al, 1996).

Analysis of MDM2 gene copy number in H & RS cells using fluorescent in situ hybridization (FISH) FISH analysis of H & RS cells required identification of these cells after the hybridization procedure using immunohistochemistry. For this purpose, we developed a new protocol that enabled the combination of a cell surface staining with an efficient hybridization. H & RS cells on cytospin preparations were immunostained for 30 min with anti-CD30 monoclonal antibody HRS4 (Engert et al, 1990) (1:50 diluted in Dako S 3022; Dako, Hamburg, Germany), followed by 30 min incubation with Dako Envision AP-conjugate. CD30-positive cells were then visualized using Fast Red (Sigma-Aldrich, Deisenhofen, Germany) solution. For the FISH analysis, cell membranes were permeabilized with 1·5% Triton X-100 (Sigma-Aldrich) in 0·01 mol/l HCl (20 min). Denaturation of nuclear DNA was achieved by 5 min incubation in 70% formamide (FA), 2× saline sodium citrate (SSC) and 0·05 mol/l sodium phosphate buffer (pH 7·2) at 76°C, followed by dehydration in ice-cold 70% ethanol/0·01 mol/l HCl, 90% ethanol/0·01 mol/l HCl and 100% ethanol for 5 min each and, finally, air drying. For the hybridization, 200 ng of a digoxigenin-labelled YAC clone covering the MDM2 gene on chromosome 12q13–14 (YAC clone 571A4; CEPH, Paris, France) and 200 ng of a biotin-labelled centromere-specific α-satellite clone (Oncor; Gaithersburg, MD, USA) were precipitated separately in the presence of 20 μg and 10 μg of Cot1 DNA (Gibco BRL; Eggenstein, Germany), respectively, and resuspended in a hybridization mix (50% FA, 10% dextran sulphate, 2× SSC). After denaturation at 80°C for 5 min, probes were chilled on ice, preannealed for 10 min at 37°C and transferred to prewarmed slides. Hybridization was allowed for 36 h at 37°C in a wet chamber. For detection, the slides were gently rocked three times for 5 min in 50% FA at 42°C, followed by three washing steps in 0·1× SSC at 60°C for 5 min. Unspecific binding was blocked [3% bovine serum albumin (BSA), 4× SSC, 0·1% Tween 20] at 37°C for 30 min. Slides then were incubated for 30 min at 37°C with Cy5-conjugated avidin (1:200 in 1% BSA, 4× SSC, 0·1% Tween 20), washed three times for 5 min in 4× SSC and 0·1% Tween 20, and incubated again with biotin-conjugated anti-avidin antibody for 30 min at 37°C. After a repeated washing step, Cy5-conjugated avidin and fluoroscein isothiocyanate (FITC)-conjugated anti-DIG antibody (1:200) were again incubated for 30 min at 37°C. Finally, nuclear DNA was counterstained with 4′6′-diamideno-2-phenylindole (DAPI) and specimens were embedded in Vecta Shield anti-fade solution (Vector Laboratory, Burlingame, CA, USA). Signals were detected by fluorescence microscopy using an Axioplan microscope (Carl Zeiss; Jena, Germany) and documented with a cooled charge-coupled device (CCD) camera (Photometrics; Tucson, AZ, USA) using dedicated software (IPLab Spectrum 3·1.1; Signal Analysis, Vienna, VA, USA). For each tumour case, at least nine H & RS cells were evaluated. In addition, signal numbers from 200 normal lymphocytes of each case were counted. In less than 10% of those, signal numbers for either probe were different from the assumed normal value of 2 (data not shown).

Statistical analysis was carried out using the Wilcoxon signed-rank test (Lehmann, 1975); computations were performed using the software packages splus, version 3·4 (MathSoft; Cambridge, MA, USA), and StatXact4 for Windows (Cytel Software; Cambridge, MA, USA). A result was always judged as statistically significant if the P-value of its respective test statistic was less than or equal to 1% (P ≤ 0·01).

Results

Genomic sequencing of the p53 gene locus

Between 10 and 20 individual H & RS cells from each of five HD patients were isolated and subjected to PCR analysis of exons 5 through to 9. Up to 10 PCR products from every exon were sequenced in two independent reactions from either end of the amplification fragment, either after cloning or by cycle sequencing. All sequences that were obtained in this way represented the wild-type configuration, i.e. no mutations within exons 5–9 of p53 were detected compared with the germline sequence (Chumakov, P.M., Genbank Acc. No. X54156) (results not shown).

Assessment of the MDM2 gene copy number by CGH and interphase FISH

As there was no experimental evidence for p53 deregulation as a result of genomic point mutations, we focused on the assessment of the MDM2 gene copy number in order to provide evidence for overexpression of p53 binding MDM2 protein. In another study, pools of 30 H & RS cells from 12 patients with HD had been subjected to comprehensive analysis of chromosomal imbalances applying CGH (Joos et al, 2000). Gain on chromosome 12 was observed to affect the entire chromosome 12 (case 7) or only a distinct chromosomal band, i.e. 12q14 representing the mapping position of MDM2 (see case 2) (Fig 1). CGH does not enable the assessment of the absolute copy number of chromosomal arms or subregions. Therefore, in the present study, we investigated the absolute gene copy number using dual-colour FISH experiments on CD30-positive H & RS cells with a differentially labelled 300 kb YAC clone covering the MDM2 locus, as well as a chromosome-specific α-satellite probe.

Figure 1.

CGH analysis of five cases of HD. The average ratio profiles of chromosomes 12 and 17 from pooled H & RS cells are described elsewhere (Joos et al, 2000). Profiles exceeding the green cut-off level of 1·25 (first green vertical line) indicate overrepresentations, a cut-off level of 0·75 (first red vertical line) indicates underrepresentation of chromosomal material. For reasons described elsewhere (Lichter et al, 1995), the centromeric regions indicated by grey bars are not evaluated by CGH experiments. In case 12, no gain of chromosome 12 material was detected. In case 7, the short and the long arm of chromosome 12 were overrepresented, while in case 16, gain of only the long arm was observed. Distinct overrepresented chromosomal sub-bands were observed in case 11 on 12p13 and in case 2 on 12q14, which represents the mapping position of the MDM2 gene.

The FISH results are presented in Fig 2 and Fig 3. Signal numbers in the H & RS cells ranged from 2 to 16 for the MDM2 probe and from 2 to 11 for the centromeric probe in individual cells (data not shown). In case 12, the lowest average signal numbers of both MDM2 and the centromere, i.e. 2·8 and 3·0, were observed. Increased copy numbers (> 2) of the centromere as well as of the MDM2 region were detected in case 7 (MDM2 = 5, centromere = 4·2), case 9 (5·4 and 4·6) and case 11 (5 and 4·5). A relatively higher copy number of MDM2 vs. centromere signals was observed in cases 2 (5·5 and 4·1) and 16 (4·9 and 3·6). In normal lymphocytes, at least 90% showed two signals for MDM2 and the centromere region.

Figure 2.

Morphology of a CD30-positive H & RS cell (case 16). (A) DAPI counterstaining of nuclear DNA. (B) Detection of the anti-CD30 monoclonal antibody using Fast Red, which shows fluorescence in the Rhodamine channel; signals of the centromere 12 probe are detected in FITC. Note that the MDM2 signals are not depicted here, as they were detected by Cy5, which emits in the infrared range (magnification 600×).

Figure 3.

Amplification of the MDM2 gene in a H & RS cell from one case. (A) CD30 staining using Fast Red. (B) Overlay-image of the DAPI counterstaining of nuclear DNA, the centromere 12 probe (FITC; green) and the MDM2 probe (Cy5; red). The picture is a false colour image, acquired with a CCD camera mounted on a fluorescence microscope (magnification 600×).

According to statistical analysis, the MDM2 signal number was significantly increased vs. the centromere signal numbers in cases 2 (P-value = 1·27e-5), 16 (P-value = 1·12e-4), 11 (P-value = 2e-3) and 7 (P-value = 2e-3). No significant results were obtained for cases 9 (P-value = 3·9e-2) and 12 (P-value = 2·5e-1) (P ≤ 0·01) (Table III).

Table III.  FISH analysis of MDM2 and centromere 12 signal numbers in H & RS cells.

Case
number


N
MDM2 signalsCen 12 signals
MDM2 gene
imbalances


P-value
total numberaverage numbertotal numberaverage number  
  1. The table indicates the total signal number as the sum of all signals in the investigated cells (column ‘N’). The column ‘MDM2 gene imbalances’ indicates whether MDM2 signal numbers were elevated vs. Centromere 12 signal numbers according to statistical analysis. Cen 12, centromere of chromosome 12.

2281545,51144,1overrepresented1·27e-5
16301464,91093,6overrepresented1·12e-4
7231155974,2overrepresented2·0e-3
9341835,41554,6normal3·9e-2
113316551464,5overrepresented2·0e-3
1219583542,8normal2·5e-1

Discussion

The molecular changes underlying the pathogenesis of H & RS cells in HD are still largely unknown. As mutant p53 shows a prolonged half-life time and therefore accumulates within cells, the common finding of immunohistochemical staining for p53 in the H & RS cells of HD was hitherto understood to be an indicator for the presence of mutations within the p53 gene which, in turn, were assumed to play an important role in the pathogenesis of HD. However, discrepant results concerning mutations were obtained that may be best explained by methodological differences (Gupta et al, 1993; Trümper et al, 1993; Xerri et al, 1995; Chen et al, 1996; Elenitoba-Johnson et al, 1996; Sánchez-Beato et al, 1996).

With respect to the scarcity of malignant cells in affected lymph nodes, we addressed genomic p53 mutations on single H & RS cells. We have previously demonstrated the reliability of the cell selection by assessing gene expression patterns of Epstein-Barr virus EBNA-1 and -2 (Trümper et al (1997) genes, as well as other genes such as restin (Delabie et al, 1992). Furthermore, chromosomal aberrations were detected in another study in various H & RS cell populations isolated in the same manner (Joos et al, 2000).

p53 point mutations were studied using a novel single-cell multiplex PCR assay that was established for amplification and sequence analysis of exons 5 through to 9, in which 98% of all mutations in different tumour systems were described (Hollstein et al, 1991). From five different HD tumours, we isolated 10–20 individual cells. In all the malignant cells, the wild-type configuration of exons 5 through to 9 was found. We conclude that p53 mutations are not commonly present in H & RS cells and, therefore, cannot account for the observed p53 protein accumulation. Although mutations in small subpopulations cannot be excluded, our data do not argue for a pivotal role of mutated p53 in the pathogenesis of HD (see also Elenitoba-Johnson et al, 1996). This finding is in accordance with the recently published results of Montesinos-Rongen et al (1999).

An alternative mechanism for p53 protein stabilization is complex formation with viral proteins, such as SV 40 large T antigen (Linzer & Levine, 1979) or overexpressed MDM2 (Keleti et al, 1996). Evidence for the involvement of MDM2 in HD comes from immunohistochemical studies demonstrating high levels of this protein in H & RS cells (Chilosi et al, 1994; Sánchez-Beato et al, 1996). MDM2 overexpression has frequently been found in the context of amplification of the gene in various tumours (Oliner et al, 1992; Leach et al, 1993; Reifenberger et al, 1993). Since a recent CGH analysis also indicated additional chromosomal material around the MDM2 locus in H & RS cells, we were interested in the precise determination of the copy number of this gene.

FISH analysis with simultaneous hybridization of a centromere-specific and a MDM2-specific probe revealed that H & RS cells are quite heterogeneous with regard to the centromere 12 and MDM2 signal number. Heterogeneity was described in previous cytogenetic studies and appears to be a typical feature of the multinucleated H & RS cells (Weber-Matthiesen et al, 1995). Our data indicate that not only numerical but also structural chromosome 12 aberrations are present to various degrees. Numerical aberrations were indicated by elevated but equal centromere- and MDM2-signal numbers, e.g. in case 9. Structural aberrations were characterized by different numbers of MDM2 and centromere 12 respectively. As approved using the Wilcoxon signed-rank test, a significantly increased number of MDM2 vs. centromere 12 signals were found in four out of six HD cases. This indicates that an amplification of the MDM2 locus has occurred. It is important to note that both numerical and structural chromosome 12 aberrations resulted in increased copy numbers of the MDM2 gene.

Comparing the FISH data with previous CGH results (Fig 1), similar results were found in four out of six tumours (2, 7, 8, 12). The apparently discrepent results regarding tumours 11 and 16 were probably as a result of the much lower resolution of CGH (approx. 10 MB) compared with FISH, which is determined by the length of the probe, i.e. 300 kb with the corresponding MDM2 clone.

For the FISH analysis described, a fast and simple protocol combining immunohistochemistry and interphase FISH was developed. Based on conventional alkaline phosphatase anti-alkaline phosphatase (APAAP) staining, CD30 is detected by Fast Red precipitation using a one-step amplification system. As Fast Red exhibits strong fluorescence in the Rhodamin/Cy 3 channel, it can be visualized together with different-coloured FISH signals using a fluorescent microscope. This protocol circumvents the drawback of the previously described ‘FICTION’ method (Weber-Matthiesen et al, 1993) in which a complex cascade of various dye-labelled antibodies is used for detection, which is prone to the risk of high background staining.

The finding of wild-type p53, accumulation of p53 protein and MDM2 gene amplification in tumour cells described for H & RS cells in this study may seem to be contradictory, as wild-type p53 activates the expression of MDM2 protein (Barak et al, 1993) that not only inhibits p53 functions, but also leads to its rapid ubiquitin-mediated degradation (Haupt et al, 1997; Lane & Hall, 1997; Kubbutat et al, 1998a). However, the same constitution has been described in human sarcomas (Leach et al, 1993; Cordon-Cardo et al, 1994). A possible explanation for this apparent discrepancy could be the recent finding that the C-terminal domain of MDM2 is essential for p53 degradation but not for functional inhibition of p53 (Kubbutat et al, 1998b, 1999; Dobbelstein et al, 1999). In addition, alternatively spliced variants of MDM2 that lack the p53 binding domain have been described (Sigalas et al, 1996). It is therefore conceivable that overexpressed MDM2 protein in H & RS cells stabilizes wild-type p53 but lacks the ability to promote proteasomal p53 protein degradation. Further detailed molecular studies in HD are necessary in order to test this hypothesis.

The small number of cases examined here is not sufficient for statistical correlation between histological subtypes or clinical outcome in patients and the results of MDM2–FISH analyses. It is therefore not possible at present to assess the prognostic significance of MDM2 amplifications or overexpression in the H & RS cells of HD. This should be part of a future study carried out in a prospective fashion within a clinical trial setting.

Acknowledgments

We thank Prof. P. Möller, University of Ulm, for providing HD-lymph node biopsies and Axel Benner, Biostatistics German Cancer Research Centre, for statistical analysis. This work was supported by grants from the Deutsche Forschungsgemeinschaft to L.T. and M.P. (SFB 399/Project A8) and to S.J. (Grants Be 1454/5–2 and Li 406/4–1), and from the Tumorzentrum Heidelberg/Mannheim to S.J. (Grant 1/1.1).

Ancillary