International Journal of Hematology

, Volume 93, Issue 1, pp 74–82

Telomeres and prognosis in patients with chronic lymphocytic leukaemia

Authors

    • Department of HaematologyUniversity of Duisburg Essen
    • Institute of Cell Biology (Cancer Research)University of Duisburg Essen
    • Department of HaematologyUniversity Hospital
  • Dirk de Beer
    • Experimental Haematology, Department of Clinical ResearchUniversity of Bern
  • Marius Bartels
    • Department of HaematologyUniversity of Duisburg Essen
  • Bertram Opalka
    • Department of HaematologyUniversity of Duisburg Essen
  • Holger Nückel
    • Department of HaematologyUniversity of Duisburg Essen
  • Ulrich Dührsen
    • Department of HaematologyUniversity of Duisburg Essen
  • Jan Dürig
    • Department of HaematologyUniversity of Duisburg Essen
  • Marc Seifert
    • Institute of Cell Biology (Cancer Research)University of Duisburg Essen
  • Dörte Siemer
    • Institute of Cell Biology (Cancer Research)University of Duisburg Essen
  • Ralf Küppers
    • Institute of Cell Biology (Cancer Research)University of Duisburg Essen
  • Gabriela M. Baerlocher
    • Experimental Haematology, Department of Clinical ResearchUniversity of Bern
    • Department of HaematologyUniversity Hospital Bern
  • Alexander Röth
    • Department of HaematologyUniversity of Duisburg Essen
Original Article

DOI: 10.1007/s12185-010-0750-2

Cite this article as:
Sellmann, L., de Beer, D., Bartels, M. et al. Int J Hematol (2011) 93: 74. doi:10.1007/s12185-010-0750-2
  • 163 Views

Abstract

In the present study, telomere length, telomerase activity, the mutation load of immunoglobulin variable heavy chain (IGHV) genes, and established prognostic factors were investigated in 78 patients with chronic lymphocytic leukaemia (CLL) to determine the impact of telomere biology on the pathogenesis of CLL. Telomere length was measured by an automated multi-colour flow-FISH, and an age-independent delta telomere length (ΔTL) was calculated. CLL with unmutated IGHV genes was associated with shorter telomeres (p = 0.002). Furthermore, we observed a linear correlation between the frequency of IGHV gene mutations and elongation of telomeres (r = 0.509, p < 0.001). With respect to prognosis, a threshold ΔTL of −4.2 kb was the best predictor for progression-free and overall survival. ΔTL was not significantly altered over time or with therapy. The correlation between the mutational load in IGHV genes and the ΔTL in CLL might reflect the initial telomere length of the putative cell of origin (pre- versus post-germinal center B cells). In conclusion, the ΔTL is a reliable prognostic marker for patients with CLL. Short telomeres and high telomerase activity as occurs in some patients with CLL with a worse prognosis might be an ideal target for treatment with telomerase inhibitors.

Keywords

CLLPrognostic factorsIGHV mutationsTelomeresTelomerase

1 Introduction

B cell chronic lymphocytic leukaemia (CLL) is the most common type of leukaemia in adults in Western countries, and mainly affects elderly individuals [1]. CLL is characterized by a clonal expansion of mature B lymphocytes and is clinically heterogeneous. Whereas, some patients with CLL have an unaltered life expectancy, others die within a few years of diagnosis [2]. A number of markers with prognostic relevance for CLL have been identified [3]. The established prognostic markers for CLL include the disease stage, as defined by the system of Binet et al. [4] and Rai et al. [5], serum parameters, including beta 2-microglobulin (β2M) levels and thymidine kinase (TK) activity, CD38 and ZAP70 expression, the mutation status of the immunoglobulin variable heavy chain (IGHV) genes, and specific chromosomal aberrations, which can be identified by FISH in approximately 80% of cases [69]. The most common chromosomal aberrations in patients with CLL are deletions in human chromosome bands 13q14, 11q22-23, 17p13, and 6q21, and trisomy 12 [10].

Telomeres, which are specialized protective structures at the end of eukaryotic chromosomes, shorten with each round of cell division due to the inability of DNA polymerase to completely replicate the 3′ end of chromosomes, among other causes [11]. As a consequence, the average telomere length in most human cells decreases with age in vivo and with cell culture in vitro [12]. Telomerase is a ribonucleoprotein complex that is able to maintain telomere length by adding hexameric TTAGGG repeats to chromosomal ends, thus compensating for the continued replicative loss of telomeric material [13]. Telomerase activity is regulated by the expression of the human telomerase reverse transcriptase (hTERT) gene, which encodes the catalytic subunit of telomerase [14], and by multiple factors that interact with telomerase or that may mediate the formation of alternative telomere structures [15].

Whereas most human somatic tissues lack telomerase expression [16], telomerase is specifically re-activated in the germinal centers (GCs) of B cells, which extends the telomeres by several kilobase pairs [17, 18]. Consequently, GC-derived memory B cells have longer telomeres than naive B cells [17]. Longer telomeres provide memory B cells with a telomere reservoir for future divisions, ranging from approximately 100 base pairs up to 1 kb [17], and may be a determining factor for the individual replicative lifespan of B cells.

IGHV gene mutations occur during the GC reaction. Naive B cells carry unmutated IGHV genes, whereas memory B cells carry mutated IGHV genes. Therefore, IGHV gene mutations in patients with CLL are not only important as a prognostic marker, but may also reflect the putative cell of origin, i.e., pre-GC (presumably antigen-activated) B cells as cells of origin for CLL cells with unmutated IGHV genes versus post-GC memory B cells as cells of origin for CLL cells with mutated IGHV genes. However, this remains a controversial topic because both mutated and unmutated CLL cells have been reported to show a gene expression profile more similar to memory cells than to naive or CD5+ B cells [19]. In addition, CLL cells usually express CD5+ on the cell surface, like a distinct subset of normal B cells. The fact that the vast majority of normal CD5+ B cells carry unmutated IGHV genes, whereas approximately one-half of CLL cells have somatically mutated IGHV genes, together with the results of a gene expression study provides a strong argument against a CD5+ B cell derivation of CLL [20, 21].

Telomeres also have a biological role in CLL. Poncet et al. [22] conducted a transcriptomic analysis of telomerase components, shelterin proteins, and a set of multifunctional proteins involved in telomere maintenance in CLL. They identified a distinct expression pattern in CLL samples relative to controls, suggesting that both telomerase down-regulation and changes in telomeric protein composition may be involved in the pathogenesis of CLL [22]. Several studies have shown that short telomeres are associated with unmutated IGHV genes, and a shorter median survival in patients with CLL compared to patients with mutated IGHV genes [2326]. Furthermore, telomere length, as measured by a quantitative real-time polymerase chain reaction (qRT-PCR), has been shown to correlate with progression-free survival (PFS) and overall survival (OS) in patients with CLL; specifically, patients with CLL with shorter telomeres have a worse outcome [27].

The aim of the present study was to correlate telomere length, as measured by automated multi-colour flow-FISH, and telomerase activity with the IGHV gene mutation load to gain insight into the cell of origin in patients with CLL. In addition, telomere length was compared to common prognostic factors in patients with CLL to determine the prognostic impact.

2 Materials and methods

2.1 Patients and samples

Seventy-eight patients with CLL in the Department of Haematology of the University of Duisburg-Essen who were diagnosed according to the World Health Organization (WHO) criteria were selected for this retrospective, single centre analysis. The study was approved by the Ethics Committee of the Faculty of Medicine at the University of Duisburg-Essen, and is in accordance with the Helsinki Declaration of 1997. The indication for treatment was based on standard criteria [28]. The PFS was calculated from the date of first diagnosis until the first progression-related therapy. The OS was calculated from the date of first diagnosis until death. Comprehensive clinical information, including treatment history, was available for all patients. Leucocytes were obtained from peripheral venous blood samples after informed consent and according to institutional guidelines. Peripheral blood mononuclear cells (PBMCs) were separated by density gradient centrifugation using Ficoll-Hypaque (Pharmacia, Freiburg, Germany) and cryopreserved until further analysis.

2.2 IGHV gene characterization

Immunoglobulin variable heavy chain gene characterization was performed, as described previously [29]. The 78 CLL patients investigated in the present study were part of the previously studied cohort.

2.3 FISH analysis

Chromosomal alterations were identified using FISH, as described previously [30]. The following gene probes were used: LSI 13q14, LSI 13q34, CEP 12, LSI 17p13, and LSI 11q22-23 (Vysis, Downers Grove, PA, USA). Döhner’s hierarchical model was used to categorize patients with ≥2 FISH abnormalities [31].

2.4 CD38 and ZAP70 expression analysis

Flow cytometry was performed for quantification of CD38 and ZAP70 expression, as previously described [30]. Chronic lymphocytic leukaemic cells were considered CD38-positive when ≥30% expressed the membrane antigen. A CLL population was considered ZAP70-positive when ≥20% of the gated CD19-positive cells expressed the marker [30].

2.5 Telomere length

The average length of telomere repeats at chromosome ends in individual cells from the mononuclear cell fraction of CLL samples was measured by automated FISH and flow cytometry (flow-FISH), as previously described [32, 33]. Briefly, mononuclear cells were isolated by Ficoll-Hypaque separation and nucleic acids were denatured by heat and formamide. For FISH, mononuclear cells were hybridized with a fluorescein-conjugated (C3TA2)3 peptide nucleic acid probe, stained with cell type-specific antibodies (CD20 and CD45RA), and counterstained with LDS 751 DNA dye. Acquisition and analysis of the telomere-specific fluorescence signals in the different subsets of lymphocytes were then performed on a FACSCalibur instrument. Reference telomere length values and percentiles established from B cells of >400 healthy individuals (between 0 and 102 years of age) were used for comparison and calculation of the delta telomere length (ΔTL; median telomere length of healthy population − telomere length of CLL sample = ΔTL of the CLL sample). The age-dependent percentiles for telomere length of healthy individuals have been previously described [34].

2.6 Telomerase activity measurement

Telomerase activity was measured in 51 CLL cell samples using the telomeric repeat amplification protocol enzyme linked immunosorbent assay (TeloTAGGG Telomerase PCR Elisa; Roche Diagnostics, Penzberg, Germany), according to the manufacturer’s instructions. Frozen aliquots of the human erythroid-myeloid cell line, K562, were analysed with each test as a positive control, and heat-treated cellular lysates (85°C for 10 min) and samples without cell extract were used as negative controls. Telomerase activity was determined relative to the positive control K562 cell line, which was considered to have 100% activity. Telomerase activities of 0 to <1, ≥1 to <10, and ≥10 to 100% were semi-quantitatively considered negative/weak, moderate, and strong, respectively.

2.7 Statistical analysis

The PFS and OS were measured from the time of primary diagnosis until the first progression-related therapy for CLL and from the time of primary diagnosis until death, respectively. The results were analysed using SPSS for Windows 15.0 software (SPSS, Inc., Chicago, IL, USA) and GraphPad Prism 3.0 (GraphPad Software, San Diego, CA, USA). For survival analyses, a Kaplan–Meier curve was depicted. Receiver-operating characteristic (ROC) analysis was used to define the best threshold for the investigated parameters. Multivariate analyses revealed hierarchical clustering of prognostic factors. A comparison of clinical or laboratory parameters between patient subgroups was performed using the non-parametric Mann–Whitney U test for comparison of two categories and the Kruskal–Wallis test for the comparison of more than two categories. The log-rank test was used to compare strata defined by the respective factors. A log-rank test for trend was used to test a linear relationship in the median survival times of these strata. Pearson correlation was used for bivariate correlations. Differences were scored as statistically significant when the p < 0.05.

3 Results

3.1 Patient characteristics

Seventy-eight CLL patients, including 50 males and 28 females, were treated at the University Hospital of Essen in Germany. The clinical characteristics of all 78 patients are summarized in Table 1. For patients who received therapy, the individual treatment regimens are described in the “Supplementary Results.” In addition to the FISH results shown in Table 1, FISH analyses detected 7 cases with 2 concurrent abnormalities, as follows: deletion of 17p13 and trisomy 12 in 1 case; deletion of 13q14 and trisomy 12 in 1 case; deletion of 11q22 and 13p14 in 3 cases; and deletion of 17p13 and 13q14 in 2 cases. An analysis of established prognostic factors revealed the expected distributions (data not shown). Taken together, the cohort described an unselected population of patients with CLL, which reflects the normal clinical course of the disease.
Table 1

Clinical characteristics and prognostic factors of patients with CLL

 

n (%)

Median (range)

Total

78

 

Male

50 (64)

 

Female

28 (36)

 

Binet at primary diagnosis

 A

55 (71)

 

 B

15 (19)

 

 C

8 (10)

 

Binet at last presentation

 A

33 (42)

 

 B

15 (19)

 

 C

30 (39)

 

Therapy needed

44 (56)

 

PFS (months)

 

34 (0–297)

Deaths

17 (22)

 

OS (months)

 

Not reached

Beta 2-Microglobulin (mg/l)

57

2.5 (1.2–10)

Thymidinkinase (U/ml)

49

10.4 (2.1–134)

CD38 positive

37 (47)

 

CD38 negative

41 (53)

 

ZAP70 positive

31 (47)

 

ZAP70 negative

35 (53)

 

Mutated IGHV

34 (60)

 

Unmutated IGHV

23 (40)

 

FISH inconspicuous

11 (14)

 

FISH del 13q14

50 (64)

 

FISH trisomy 12

9 (12)

 

FISH del 11q22

7 (9)

 

FISH del 17p13

8 (10)

 

PFS progression-free survival, OS overall survival, FISH fluorescence in situ hybridization karyotype

CD38 status: ≥30% positive cells, ZAP70 status: ≥20% positive cells

3.2 IGHV mutation status and ΔTL

In the present study, the mutation load in IGHV genes had a positive correlation with the ΔTL, i.e., telomeres of unmutated CLL cells were considerably shorter (median −5.2 kb, range 1.8 to −6.7; n = 23) and highly mutated CLL cells had only marginally shorter telomeres (median −3.0 kb, range 2.8 to −6.2; n = 34) when compared to telomeres from B cells of healthy age-related controls (Fig. 1a). The difference in the telomere length between unmutated and mutated CLL cells was statistically significant (p = 0.002). Furthermore, we found a linear correlation between the IGHV gene mutation frequency and the telomere length (r = 0.509, p < 0.001) and the IGHV gene mutation percentage and the telomere length (r = 0.534, p < 0.001). There was no significant difference in the follow-up period between patients with IGHV mutated and unmutated CLL (p = 0.99), suggesting that the clinical course cannot account for the differences in telomere length between these groups. Our data support the hypothesis that the telomere length is dependent on the putative cell of origin. Patients with CLL with somewhat lower levels of IGHV mutations had a reduction in telomere length that was intermediate between unmutated and highly mutated CLL (Fig. 1b). This may reflect the relatively shorter time period of GC participation compared to patients with a higher mutation load.
https://static-content.springer.com/image/art%3A10.1007%2Fs12185-010-0750-2/MediaObjects/12185_2010_750_Fig1_HTML.gif
Fig. 1

Delta telomere length (ΔTL) and IGHV mutation status. a The mutation load in IGHV correlates with the ΔTL, i.e., telomere lengths in unmutated CLL cases were significantly shorter compared to mutated cases. b There was a linear correlation between the number of IGHV mutations and the ΔTL

3.3 ΔTL and established prognostic markers

Because previous studies have shown that telomere length has some prognostic value in patients with CLL [26], the present study investigated telomere length using the flow-FISH technique, which allows the precise determination of telomere length in individual cells [24]. Telomeres are known to shorten with age, thus the telomere length measured in CLL cells was age-adjusted to the telomere length of control cells representing the given age-percentile, and a ΔTL value was calculated by subtracting the telomere length of the CLL sample from the median telomere length for healthy individuals of the same age, as described in “Materials and methods”. In general, the telomere length was shorter in CLL cells when compared to healthy control cells. When patients were grouped based on known prognostic factors, the telomere length in the CLL samples was significantly shorter in the groups with the worst prognosis. With respect to karyotypic aberrations, as determined by FISH, CLL cells with a deletion of 11q22 or 17p13 had shorter telomeres compared to CLL cells with a normal karyotype, deletion of 13q14, or trisomy 12. A comparison of the 5 subgroups (normal karyotype, trisomy 12, and deletions of 13q14, 11q22, or 17p13) did not show any significant differences (Table 2). These data confirm earlier observations that telomere length in CLL correlates with established prognostic factors.
Table 2

Telomere length in patients with CLL grouped by established prognostic factors in CLL

Prognostic factor

Median (range)

p

Binet at primary presentation

 A

−4.2 (4.1 to −6.7)

0.402

 B

−4.0 (1.0 to −6.2)

 

 C

−5.3 (−1.9 to −5.8)

 

Binet at last presentation

 A

−2.9 (4.1 to −6.7)

0.022

 B

−4.2 (3.5 to −6.6)

 

 C

−4.9 (1.0 to −6.2)

 

CD38

 Positive

−5.2 (3.5 to −6.7)

<0.001

 Negative

−2.8 (4.1 to −6.6)

 

ZAP70

 Positive

−5.4 (−1.8 to −6.7)

<0.001

 Negative

−2.7 (4.1 to −5.9)

 

IGHV

 Mutated

−3.0 (2.8 to −6.2)

0.002

 Unmutated

−5.2 (4.1 to 6.7)

 

Beta 2-Microglobulin

 <3.9 mg/l

−3.2 (4.1 to −6.7)

0.135

 ≥3.9 mg/l

−4.7 (0.1 to −6.2)

 

Thymidinkinase

 <10 U/ml

−2.6 (2.2 to −5.8)

0.018

 ≥10 U/ml

−4.8 (3.5 to −6.7)

 

FISH

 Inconspicuous

−4.7 (2.2 to −6.6)

0.163

 del13q14

−3.1 (4.1 to −6.7)

 

 Trisomy 12

−3.6 (−1.7 to −5.4)

 

 del11q22

−5.2 (3.5 to −6.2)

 

 del17p13

−5.3 (−4.2 to −6.0)

 

 Good prognosis

−4.0 (4.1 to −6.7)

0.002

 Bad prognosis

−5.2 (3.5 to −6.2)

 

p Values were calculated with Mann–Whitney U test for 2 or Kruskal–Wallis test for more than 2 variables

CD38 status: ≥30% positive cells, ZAP70 status: ≥20% positive cells

Good prognosis includes inconspicuous, deletions 13q14, trisomies 12; bad prognosis includes deletions 11q22, deletions 17p13

Bold values indicate significant values (p < 0.05)

FISH fluorescence in situ hybridization karyotype

3.4 A novel threshold of the ΔTL for prediction of PFS and OS

ROC analyses showed that a threshold ΔTL of −4.2 kb resulted in the best separation of CLL patients with respect to the PFS (Fig. 2a). Also, the OS was significantly different when using the −4.2 kb ΔTL threshold (Fig. 2b). To determine whether or not prognostic factors were associated with the ΔTL, multivariate analyses for the PFS were performed for the following variables: gender; Binet A, B, or C stage at the time of primary diagnosis; FISH karyotypes with deletions of 11q22 and 17p13 compared to inconspicuous deletions of 13q14 and trisomy 12; ZAP70 positivity or negativity; and ΔTL> or <−4.2 kb. Data were available for all variables for all 78 cases, with the exception of ZAP70 data, which was available for 66 cases. Data for IGHV mutation status and β2M levels were not available in a sufficient number of cases for multivariate analysis. The Binet score at the time of primary diagnosis and ZAP70 status were the best predictors of the PFS (Table 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs12185-010-0750-2/MediaObjects/12185_2010_750_Fig2_HTML.gif
Fig. 2

Delta telomere length (ΔTL) alone, and in combination with IGHV mutation status as a predictor for PFS and OS in patients with CLL. Kaplan–Meier curves of PFS and OS dependent on the ΔTL are shown for 78 CLL patients. a, b The best threshold ΔTL for predicting PFS in patients suffering from CLL was −4.2 kb, as determined by ROC analysis. CLL samples with ΔTLs >−4.2 kb (long-ΔTL) had improved PFS and OS. c, d Inclusion of IGHV mutation status resulted in 3 groups of CLL samples. The first CLL group had a presumably good prognosis with mutated IGHV genes and ΔTLs >−4.2 kb (long-Δtl, mutated); the second CLL group had a presumably poor prognosis with unmutated IGHV genes and ΔTLs <−4.2 kb (short-Δtl, unmutated); and the third group had discordant values concerning IGHV mutation status and ΔTL (discordant). The group with a presumably good prognosis had the best PFS and OS compared to discordant cases and the presumably poor prognosis group, respectively

Table 3

Multivariate Cox regression analysis model

Variable

Hazard ratio

CI (95%)

p

Gender

1.07

0.5–2.2

0.861

Disease stage at primary diagnosis

2.12

1.3–3.5

0.003

Genomic aberrations

1.21

0.5–3.1

0.688

CD38 status

0.96

0.4–2.5

0.925

ZAP70 status

0.76

0.3–2.0

0.575

Delta telomere length

0.22

0.1–0.6

0.002

The cut-off levels used in the analysis were: disease stage at primary diagnosis Binet stage A versus B versus C; genomic aberrations: normal risk—normal karyotype, deletion 13q14 and trisomy 12, poor risk—deletion 11q22 and deletion 17p13; CD38 status: ≥30% positive cells; ZAP70 status: ≥20% positive cells; delta telomere length: good prognosis delta telomere ≥−4.2 kb, poor prognosis <−4.2 kb

Bold values indicate significant values (p < 0.05)

CI confidence interval

Despite the fact that IGHV mutation status could not be used in the multivariate analysis in the present study, IGHV mutation status is known to have a high predictive value in patients with CLL. Therefore, IGHV mutation status was combined with ΔTL using a threshold of −4.2 kb for OS analyses. Patients with CLL were divided into the following 3 groups: (1) a presumably good prognosis, consisting of CLL with mutated IGHV genes and a ΔTL > −4.2 kb (n = 23); (2) a presumably poor prognosis, consisting of CLL with unmutated IGHV genes and a ΔTL < −4.2 kb (n = 20); and (3) discordant values concerning these 2 variables (n = 14). The clinical characteristics of the 3 subgroups are shown in Table 4. As expected, the groups had significant differences in several established prognostic factors (Table 4). The group with a good prognosis had the best PFS and OS (Fig. 2c, d). The 3 groups showed a significant difference in the PFS (p for trend = 0.0023) and OS (p for trend = 0.0001), as determined by the log-rank test. The group with a presumably good prognosis had a better PFS and OS compared to discordant cases (p = 0.09, p < 0.001) and patients with CLL with a presumably poor prognosis (p < 0.001, p < 0.001), respectively. There were no significant differences in the PFS (p = 0.820) and OS (p = 0.496) between the discordant cases and CLL cases with a presumably poor prognosis (Fig. 2c, d). Interestingly, of the CLL cases with mutated IGHV genes, the CLL cases with a short ΔTL (n = 11) had a significantly worse PFS compared to the CLL cases with a long ΔTL (n = 23, p = 0.004). Of the CLL cases with unmutated IGHV genes, cases with a short ΔTL (n = 20) had a worse PFS compared to CLL cases with a long ΔTL (n = 3, p = 0.067). However, no significant differences existed between CLL cases with mutated IGHV genes (n = 11, n = 23) and unmutated IGHV genes (n = 20, n = 3) in the group of CLL patients with a short ΔTL and in the group of CLL patients with long ΔTL, respectively (p = 0.233, p = 0.775).
Table 4

Clinical characteristics of patients with CLL with different prognostic scores based on telomere length and IGVH mutation

 

Good (n)

Discordant (n)

Bad (n)

p

Male

13

11

16

0.179

Female

10

3

4

 

Binet at primary diagnosis

 A

16

11

13

0.564

 B

6

1

5

 

 C

1

2

2

 

Binet at last presentation

 A

14

5

2

0.003

 B

6

2

5

 

 C

3

7

13

 

Therapy needed

10/23

9/14

16/20

0.048

Deaths

1/23

5/14

9/20

0.007

2-Microglobulin, <3.9 mg/l

11

6

1

0.014

2-Microglobulin, ≥3.9 mg/l

3

6

18

 

Thymidinkinase, <10 U/ml

10

4

2

0.023

Thymidinkinase, ≥10 U/ml

5

6

11

 

CD38 positive

3

11

16

0.001

CD38 negative

20

3

4

 

ZAP70 positive

3

9

16

0.001

ZAP70 negative

18

4

4

 

Mutated IGHV

23

11

0

0.001

Unmutated IGHV

0

4

20

 

FISH inconspicuous

4

3

4

0.140

FISH del 13q14

15

5

7

 

FISH trisomy 12

4

1

3

 

FISH del 11q22

0

2

4

 

FISH del 17p13

0

3

2

 

p Values were calculated with Pearson Chi-quadrate-test

CD38 status: ≥30% positive cells; ZAP70 status: ≥20% positive cells

Good mutated IGHV genes and delta telomere lengths greater than −4.2 kb, discordant discordant values concerning IGHV mutation status and delta telomere length, bad unmutated IGHV genes and delta telomere lengths less than −4.2 kb

Bold values indicate significant values (p < 0.05)

FISH fluorescence in situ hybridization karyotype

CLL expressing a somatically mutated VH3-21 gene is associated with a poor clinical outcome, which differs from the improved clinical outcome associated with CLL with other mutated IGHV genes. However, in the present study, exclusion of 3 cases with mutated VH3-21 genes did not alter the statistical data regarding the PFS or OS.

3.5 ΔTL and disease progression

To evaluate the influence on disease progression, the ΔTL was evaluated longitudinally in 47 cases during the clinical course of the disease. The median follow-up period was 22 months. The mean ΔTL was −4.0 kb at the first evaluation and −3.7 kb at the last evaluation (p = 0.393). In 16 of 47 cases, the CLL progressed and therapy was eventually needed. The median telomere length in this group was −4.6 kb at the first evaluation and −4.8 kb at the last evaluation (p = 0.794). The remaining 31 CLL patients had stable disease during the observation period. The median ΔTL in this group was −3.6 kb at the first evaluation and −3.4 kb at the last evaluation (p = 0.815).

3.6 Telomerase activity, telomere length, and IGHV gene mutations

Telomerase activity plays an important role in tumourigenesis. Telomerase activity was measured in 51 of the CLL patient samples in the present study. Seventeen samples had negative/weak telomerase activity, 27 samples had moderate telomerase activity, and 7 samples had strong telomerase activity. The median values for the ΔTL were −2.5, −4.5, and −5.6 kb in the groups with negative/weak, moderate, and strong telomerase activity, respectively, indicating a tendency towards shorter telomeres in CLL cells with stronger telomerase activity (Fig. 3a). The CLL samples with unmutated IGHV genes had lower telomerase activity compared to CLL cells with mutated IGHV genes, although these differences were not statistically significant (p = 0.179; Fig. 3b).
https://static-content.springer.com/image/art%3A10.1007%2Fs12185-010-0750-2/MediaObjects/12185_2010_750_Fig3_HTML.gif
Fig. 3

Telomerase activity compared to PFS, telomere length, and IGHV mutation status. a The median values for the ΔTL were −2.5, −4.5, and −5.6 kb in the groups with negative/weak, moderate, and strong telomerase activity, respectively, indicating a tendency towards shorter telomeres in CLL with higher telomerase activity. b CLL patients with unmutated IGHV genes had higher telomerase activity compared to CLL patients with mutated IGHV genes, although these data did not reach statistical significance

4 Discussion

Previous studies have shown that telomeres have a biological role in patients with CLL [22]. Furthermore, telomere length and telomerase activity have been shown to predict progression and survival in patients with CLL [26]. The present study determined the influence of telomere length and telomerase activity in CLL cells on the biology of CLL and CLL progression. We found that CLL cells with mutated IGHV genes had longer telomeres than CLL cells with unmutated IGHV genes, which is in agreement with the hypothesis that the former are derived from post-GC B cells and the latter are derived from pre-GC B cells, as post-GC B cells have elongated telomeres. However, a caveat of this interpretation is that most proliferation of CLL cells appears to take place in proliferation centers in the lymph nodes, and we do not know if the telomerase activity in these proliferating CLL cells might influence the degree of telomere shortening or preservation during tumour clone expansion.

The present study demonstrated that telomere length can be used to predict disease progression in patients with CLL. The novel defined threshold of a ΔTL of −4.2 kb, as measured by flow-FISH, was able to predict the PFS and OS in patients with CLL. With a follow-up of 81 months, this study is of the longest duration to date regarding the prognostic value of telomere length [35]. Unmutated IGHV genes are associated with shorter telomeres in CLL cells [25, 36]. When IGHV gene mutation status and telomere length were discordant, and outcomes were better predicted by telomere length than by IGHV mutation, confirming the findings of Ricca et al. [37]. Based on these results, we hypothesized that the clinical course of patients with CLL could be predicted by a combination of the ΔTL and IGHV mutation status. Indeed, only 1 of 23 CLL patients (4%) with mutated IGHV genes and ΔTL values >−4.2 kb died during the observation period, and this death was due to urothelial carcinoma rather than CLL.

Different methodologic approaches, including Southern blot analysis, real-time PCR, dideoxy-primed in situ (PRINS) labelling, and flow-FISH can be used to determine telomere length [24, 25, 37, 38]. However, Southern blot analysis also measures the length of the subtelomeric regions. Furthermore, in multi-colour flow-FISH experiments, telomere length can be measured in specific tumour cells using selected surface markers, e.g., a B cell marker for CLL cells. Thus, telomere length can be determined more precisely with the flow-FISH method used in the present study.

A previous report showed that telomerase activity, independent of telomere length, predicted survival in patients with CLL [26]. However, CLL cells with short telomeres and high telomerase activity defined a small group of patients with a worse prognosis. Similarly, T-prolymphocytic leukaemia (T-PLL) cells have short telomeres and high telomerase activity, and T-PLL is associated with a poor prognosis [39]. Currently, telomerase inhibitors are being explored as a possible therapy for patients with T-PLL. In like manner, telomerase inhibitors might be a promising novel therapy for patients with CLL characterized by CLL cells with short telomeres and high telomerase activity [39]. Indeed, the telomerase inhibitor, imetelstat, is under investigation in ongoing phase I/II studies for its role in cancer therapy [40].

Another advantage of telomere length as a prognostic marker is the marginal decline of telomere length during the clinical course of CLL. This may reflect the biological behaviour of CLL cells, which is mainly characterized by a defect in apoptosis rather than a high proliferative potential. In the present study, neither time nor therapy changed the ΔTL in patients with CLL. Ricca et al. [37] reported preliminary findings which indicated that telomere length is not always stable over time in CLL patients. Damle et al. [25] reported decreasing telomere length in CLL cells at follow-up; however, when detected, the declines in telomere length over time were marginal.

Taken together, the present study showed that slowly expanding CLL cells have longer telomeres, whereas more aggressive CLL cells have shorter telomeres. This finding not only reflects the telomere load of the putative cell of origin, i.e., pre- or post-GC B cells, but may also result from a differential mechanism for tumour cell maintenance. Using multi-colour flow-FISH, the most accurate method to determine telomere length available to date, we showed that telomere length is a reliable prognostic marker for patients with CLL. A threshold ΔTL of −4.2 kb was the best predictor for the PFS and OS. The ΔTL was not significantly altered over time or with therapy. Thus, drugs which inhibit telomerase activity could be promising novel therapeutic strategies for CLL with short telomeres and high telomerase activity.

Acknowledgments

We thank Anja Führer, Songül Basoglu, Barbara Friedmann, Andrea Kopplin, and Ute Schmücker for their expert technical assistance. This work was supported by a grant from the Bernese Cancer League (G.M.B.) and from the IFORES program of the University Duisburg Essen (L.S.).

Conflict of interest

The other authors declare no competing financial interests.

Supplementary material

12185_2010_750_MOESM1_ESM.doc (50 kb)
Supplementary material 1 (DOC 49 kb)

Copyright information

© The Japanese Society of Hematology 2010