OUP user menu

High Serum Thymidine Kinase 1 Level Predicts Poorer Survival in Patients With Chronic Lymphocytic Leukemia

Sergej N. Konoplev MD, PhD, Herbert A. Fritsche PhD, Susan O’Brien MD, William G. Wierda MD, Michael J. Keating MD, Terrie G. Gornet, Susan St Romain MT(ASCP), Xuemei Wang MS, Kedar Inamdar MD, Malisha R. Johnson DO, L. Jeffrey Medeiros MD, Carlos E. Bueso-Ramos MD, PhD
DOI: http://dx.doi.org/10.1309/AJCPHMYT93HUIZKW 472-477 First published online: 1 September 2010

Abstract

Serum thymidine kinase 1 (TK1) levels have been reported to have prognostic significance in patients with chronic lymphocytic leukemia (CLL). Until recently, serum TK1 levels were assessed using inconvenient radioenzyme assays. In this study, we used a novel chemiluminescence assay to assess serum TK1 levels in patients with CLL at the time of first examination. We show that high serum TK1 levels predict poorer overall survival and correlate with unmutated immunoglobulin variable region genes, CD38 and ZAP-70 expression, and subsequent risk of developing large B-cell lymphoma (Richter syndrome). Similar findings were observed in a subset of patients treated with current fludarabine-based chemotherapy regimens. We suggest that serum TK1 levels analyzed using this convenient chemiluminescence assay may be useful in the risk assessment of patients with CLL.

Key Words:
  • Chronic lymphocytic leukemia
  • Serum thymidine kinase 1
  • Prognostic factors

Chronic lymphocytic leukemia (CLL) is the most common hematologic neoplasm of adults in Western countries. In the United States, more than 15,000 new cases are diagnosed annually, and more than 4,000 patients die annually of this disease.1 The clinical course of patients with CLL is variable; most patients have a long survival, but in some patients, early disease progression develops.2 This clinical heterogeneity justifies the need for reliable prognostic markers that allow therapy to be tailored to a particular patient.

In many neoplasms, the clinical course is related to the proliferative activity of the neoplastic cells. Therefore, indicators of proliferation are attractive candidates as prognostic markers. Traditionally, the proliferative component of CLL has been underappreciated, with more focus on apoptosis and other pathogenic mechanisms. However, more recently, the importance of proliferation in CLL pathogenesis has been recognized. One indicator of cell proliferation in CLL is thymidine kinase (TK), a cellular enzyme involved in a salvage pathway for DNA synthesis. There are 2 isoforms of this enzyme: TK1 and TK2. TK1 is found in the cytoplasm of dividing cells and is absent in resting cells. TK2 is located in the mitochondria of resting cells.3,4

In 1984, Kallander and colleagues5 assessed serum TK1 levels in patients with CLL using a radioenzymatic assay (REA) and demonstrated that higher serum TK1 levels correlated with progressive disease and advanced Rai stage. More recently, Hallek and colleagues,6 using similar methods, demonstrated that the serum TK1 levels predicted progression-free survival in patients with CLL independently from other known prognostic factors, such as WBC count and presence of lymphadenopathy. Hallek and others7 confirmed that the serum TK1 level was helpful in identifying a subgroup of patients with CLL at high risk for disease progression. Magnac et al,8 using REA methods, identified a strong correlation between high serum TK1 levels and unmutated immunoglobulin variable heavy chain (IGHV) genes. The last observation raised a question as to whether the serum TK1 level is an independent prognostic marker or simply reflects mutational status of the IGHV gene. In addition, as previous studies analyzed data for patients with CLL not treated with current therapeutic modalities, it is not clear whether the previous observations remain valid in the current era of chemoimmunotherapy for patients with CLL.

In this study, we used a chemiluminescence immunoas-say (CLIA) method to assess serum TK1 levels in untreated patients with CLL. Serum TK1 levels were also correlated with other clinical and laboratory parameters.

Materials and Methods

Patient Study Group

After receiving approval from the institutional review board, we searched the files of our institution from January 2000 through June 2008 for cases of CLL, previously untreated, with serum samples available for TK1 analysis. The original diagnoses were confirmed by review of bone marrow aspirate smears and biopsy specimens, the results of flow cytometry immunophenotypic analysis, and immunohis-tochemical studies performed on bone marrow biopsy and/or clot sections.

Detection of Serum TK1

TK1 was analyzed in baseline serum samples of all patients before therapy. During the initial stage of this project, we analyzed serum samples for TK1 in the first 50 patients in parallel using a standard REA and a novel CLIA.

REA is based on the conversion of iodine 125 (125I)-labeled deoxyuridine to 125I deoxyuridine mono-phosphate and was performed on the Prolifigen TK REA, DiaSorin, Stillwater, MN, as described previously.9 Briefly, 500 μL of radioactive substrate was added to 20 μL of patient serum sample and incubated for 4 hours at 37°C in a water bath. A TK-REA separator tablet was added to each sample tube and incubated at room temperature for 1 hour. Tube contents were washed for a total of 4 washes and then counted on a gamma counter. A series of standards and controls were set up simultaneously with patient samples to allow the construction of a standard curve and monitor for interbatch and intrabatch variation.

The CLIA, which is easy to perform, is based on TK1 converting AZT (3′-azido–3′deoxythymidine) to AZTMP (3′-azido–3′-deoxythymidine monophosphate) and is commercially available (DiaSorin). The correlation between these methods was excellent (R = 0.9764; y = 1.0158x + 3.2691). For this reason, for the remainder of this study, we used only the CLIA method to detect serum TK1 levels. There was no significant difference between clinical and laboratory parameters for the first and second groups of patients, which allowed us to combine data for both groups of patients in the analysis.

CD38 Expression

CD38 expression was evaluated in all patients by flow cytometric immunophenotypic analysis of bone marrow aspirate specimens at the time of initial diagnosis. The percentage of CD38+ B cells was determined by the fraction of CD19+ cells that coexpressed CD38. A 30% cutoff was used for positive CD38 expression.

ZAP-70 Expression by Immunohistochemical Analysis

ZAP-70 expression was evaluated by immunohistochemical analysis using paraffin-embedded bone marrow tissue sections and a monoclonal murine anti–ZAP-70 antibody (Upstate Cell Signaling Solutions, Charlottesville, VA), as previously described.10 A positive result was determined by observing ZAP-70 expression by B cells. ZAP-70 is expressed strongly by T cells alone and served as an internal control.

Somatic Mutation Status of the IGHV Gene

Sequence analysis of the IGHV genes was performed using total RNA extracted from bone marrow aspirates or peripheral blood specimens as described previously. To determine the level of somatic mutation, sequences were aligned with the germline sequences in the V-BASE 2 database.11 The IGHV mutation status was designated as unmutated if there were fewer than 2% mutations (≥98% homology to germline sequences) or as mutated if there were 2% or more mutations (<98% homology to germline sequences) compared with the germline sequence.12

Interphase and Conventional Cytogenetics

Conventional karyotype analysis using GTG banding was performed on bone marrow aspirate specimens according to methods described previously.13 Interphase cytogenetic analysis by fluorescence in situ hybridization (FISH) was performed on a subset of patient samples by using a commercially available set of DNA probes to detect the most frequent genomic aberrations in CLL as defined by conventional cytogenetics studies (Vysis, Downers Grove, IL).1417 The probe set allowed detection of the following genomic aberrations in CLL: deletion of 11q22.3 at the ATM locus, deletion of 17p13.1 at the p53 locus, deletion of 13q14.3 at the D13S319 locus, deletion of 13q34 at the LAMP1 locus, and trisomy 12.

Statistical Analysis

Serum TK1 levels determined by the REA and CLIA methods were correlated by using the Spearman correlation coefficient. Similar analyses were performed to assess the association of serum TK1 levels and other clinical and laboratory data.

The Kaplan-Meier method was used to estimate overall survival (OS) and event-free survival (EFS). The OS was defined from the time of CLL diagnosis to death or last follow-up date. The EFS was defined from the time of complete remission (CR) to disease relapse, development of Richter syndrome, or death. Only data for patients who achieved CR were included in the EFS analysis. Patients who were alive and without disease relapse were censored at the last follow-up date.

Univariate Cox proportional hazards models were fitted to assess the effect on OS and EFS of serum TK1 level and other clinical and laboratory characteristics, including age, sex, race, Rai stage, presence or absence of Richter transformation, performance status, leukocytosis, hemoglobin level, platelet count, serum lactate dehydrogenase level, serum β2-microglobulin level, IGHV mutational status, CD38 positivity, ZAP-70 positivity, conventional karyotype, and cytogenetic markers detected by FISH (del17p13, del11q22–23, trisomy 12, and del13q14). For multivariate analysis, a full Cox proportional hazards model on OS or EFS, including all patient characteristics listed above, was fitted initially. A reduced Cox model was then built using stepwise model selection such that all variables remaining in the final model were statistically significant at the .05 level. All analyses were performed using SAS software (SAS Institute, Cary, NC) and S-plus software (Insightful, Seattle, WA).

Results

Clinical and Pathologic Features

Patient characteristics are given in Table 1 and Table 2. There were 80 men and 37 women with a median age of 63 years (range, 35–87 years). In this group, 103 were white, 9 were African American, and 5 were Hispanic. At the time of diagnosis, Rai stage was 0 in 38 patients, I in 35 patients, II in 23 patients, III in 10 patients, and IV in 11 patients. Thirty-four of the patients received no treatment, and 83 received chemotherapy according to the FCR protocol (fludarabine, cyclophosphamide, rituximab),18 including 4 patients who received stem cell transplantation for refractory disease later in the disease course (grafts from sibling donors, 2 patients; from matched unrelated donors, 2 patients). Among patients who received chemotherapy, 56 (69%) patients achieved CR. Sixteen patients died. The median survival time was not reached at the time of writing the manuscript, and the median follow-up time was 64 months. In 4 patients, large B-cell lymphoma (Richter syndrome) developed later in their disease course.

View this table:
Table 1

The median WBC count at diagnosis was 63,000/μL (63 × 109/L; range, 35,000–87,000/μL [35–87 × 109/L]; reference range, 4,000–11,000/μL [4–11 × 109/L]). IGHV mutational status was analyzed in 94 cases; 52 patients had unmutated and 42 patients had mutated IGHV. CD38 expression was analyzed in 109 cases: in 37 patients (33.9%), the CLL cells expressed CD38. ZAP-70 expression was analyzed in 69 cases and was positive in 39 (57%). Conventional cytogenetic studies were performed using bone marrow aspirate material for 93 patients. In 56 patients, the karyotype was normal, and in 37 patients, there were karyotypic abnormalities.

FISH studies were performed on interphase cells in 101 cases. In 29 cases, there were normal signals with all probes tested. Deletion 11q22–23 was detected in 25 cases, including 10 cases in which it was the sole abnormality. Trisomy 12 was detected in 18 cases and was the sole abnormality in 12. Deletion 13q14.3 was detected in 45 cases and was the sole abnormality in 25. Deletion 13q34 was detected in 4 cases and was associated with del11q22–23 in 1 patient and with del13q14.3 in 3 patients. Deletion 17p13 was detected in 4 cases, in 2 as the sole abnormality.

View this table:
Table 2

Correlation Between Serum TK1 Levels and Clinical and Other Laboratory Data

Increased serum TK1 levels significantly correlated with male sex (r 2 = 0.644), poor performance status (r 2 = 0.285), increased serum β2-microglobulin level (r 2 = 0.470), unmutated IGHV genes (r 2 = 0.316), advanced Rai stage (r 2 = 0.552), lack of likelihood of achieving CR (r 2 = 0.400), CD38 expression (r 2 = 0.202), ZAP-70 expression (r 2 = 0.283), lack of del13q14.3 as assessed by FISH (r 2 = 0.247), and development of Richter syndrome later in the disease course (r 2 = 0.200). Serum TK1 levels were not associated with age, normal vs abnormal karyotype, and FISH evidence of del11q22–23, trisomy 12, del13q34, or del 17p13.

Prognostic Factors Influencing OS and EFS

Univariate Cox proportional hazards models for the entire patient population were fit for OS, which suggested that only development of Richter syndrome and high serum TK1 levels were significantly associated with OS Table 3. The risk of death was higher in patients in whom Richter syndrome developed (hazard ratio, 5.03; P = .01) and in patients with a higher serum TK1 level (hazard ratio, 1.13 for each increment of 10 U/L in TK1 level; P = .02). The fitted multivariate Cox model confirmed the independent adverse effect of higher serum TK1 level and presence of Richter syndrome on OS. It was found that patients tended to have a higher risk of death if they had Richter syndrome (hazard ratio, 5.45; P = .01) and a higher serum TK1 level (hazard ratio, 1.15 for each increment of 10 U/L in TK1 level; P = .01).

Univariate Cox proportional hazards models for the entire patient population were fit for EFS, which suggested that only a high serum β2-microglobulin level was significantly associated with a worse EFS (hazard ratio, 1.34; P = .03). Because only 1 covariate was statistically significant, no further multivariate analysis was conducted for EFS.

In addition, we analyzed prognostic factors influencing OS and EFS in patients treated with chemotherapy (n = 83). Univariate Cox proportional hazards models for treated patients were fit for OS, which suggested that only Richter syndrome, serum TK1 level, and age were significantly associated with OS Table 4. Patients tended to have a higher risk of death if they were older (hazard ratio, 1.06; P = .03), developed Richter syndrome (hazard ratio, 4.09; P = .03), or had a higher serum TK1 level (hazard ratio, 1.10; P = .02). The fitted multivariate Cox model confirmed the independent influence of those factors on OS and showed that patients tended to have a higher risk of death when they were older (hazard ratio, 1.09; P = .02), had Richter syndrome (hazard ratio, 4.08; P = .04), and had a higher level of TK1 (hazard ratio, 1.18 for each increment of 10 U/L in TK1 level; P = .02).

For treated patients, univariate Cox proportional hazards models were fit for EFS and demonstrated that only a high serum β2-microglobulin level was significantly associated with worse EFS (hazard ratio, 1.32; P = .04). When we tried to fit a multivariate Cox model with serum β2-microglobulin and TK1 levels, the effect of TK1 level became insignificant (P = .18).

Discussion

Initial studies investigating the prognostic importance of serum TK1 levels in patients with CLL were performed using an REA.58 Although results were promising, technical difficulties and the use of radioactivity resulted in serum TK1 analysis not being widely accepted. Eventually, serum TK1 testing was virtually abandoned. In this study, we demonstrated an excellent correlation between serum TK1 levels detected by standard REA technology and by a novel CLIA technique. This new method does not require radioactivity and is convenient and rapid. Therefore, serum TK1 analysis by this CLIA method has the potential to become a useful test for a routine clinical practice.

View this table:
Table 3
View this table:
Table 4

A milestone in predicting the prognosis of patients with CLL was achieved with the observations that somatic mutations of the IGHV genes occur in a subset of CLL cases and that IGHV gene mutation status correlates with clinical course.17,19 Patients with CLL in which IGHV genes show fewer than 2% somatic mutations (unmutated CLL) have a poorer outcome. In contrast, patients with CLL in whom IGHV genes show 2% or more mutations (mutated CLL) have a clinically indolent course.17,19 Although the prognostic importance of IGHV gene mutational status has been confirmed in numerous studies and has become a “gold standard” in the prognostic assessment of patients with CLL, it remains elusive as to why patients with unmutated IGHV genes have a more aggressive clinical course. One plausible explanation is that CLL cells with unmutated IGHV genes receive continuous antiapoptotic and/or proliferating stimuli via the B-cell receptor pathway, leading to more aggressive disease. If this presumption is correct, CLL cells with unmutated IGHV genes should be more proliferative than CLL cells with mutated IGHV genes. The results of this study serve as indirect evidence to support this theory because serum TK1 levels were higher in patients with unmutated CLL. Our study confirms similar observations by Magnac and colleagues,8 who reported in a small series of patients with CLL a correlation between higher serum TK1 level and unmutated IGHV genes. These authors further suggested using the serum TK1 level as a surrogate for assessing IGHV gene mutation status.

In addition to an association between serum TK1 levels and unmutated IGHV genes, we showed an association between serum TK1 level and expression of CD38 and ZAP-70 by CLL cells. To our knowledge, this observation has not been reported in the literature, but it was expected because CD38 and ZAP-70 have been proposed as surrogate markers for unmutated IGHV genes.17,20 Similarly, the observed correlation between the serum TK1 level and a high serum β2-microglobulin level has a biologic explanation because both markers reflect increased cell proliferation.

Previous studies have shown that increased serum TK1 levels adversely effect OS.58 This study confirmed the independent adverse effect of increased serum TK1 levels on OS. Because all previous studies were conducted before fludarabine-based regimens entered clinical practice, our results address the important question as to whether increased serum TK1 levels retain prognostic importance in the current era of chemoimmunotherapy. To our knowledge, this is the first report of an adverse effect of increased serum TK1 levels in patients with CLL subsequently treated with a fludarabine-based regimen.

In summary, we used a novel CLIA to determine serum TK1 levels in patients with CLL. This chemiluminescence assay obviates the need for radioactivity and is more convenient and easier than older radioimmunoassay; the results of both methods are highly concordant. By using this chemiluminescence assay, we showed that serum TK1 levels predict the prognosis of patients with CLL treated with current fludarabine-based chemotherapy regimens. A high serum TK1 level also correlates with unmutated IGHV genes and CD38 and ZAP-70 expression.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
View Abstract