Long remissions in hairy cell leukemia with purine analogs

A report of 219 patients with a median follow-up of 12.5 years


  • Monica Else M.A., M.Sc.,

    1. Section of Haemato-Oncology, Royal Marsden NHS Foundation Trust/Institute of Cancer Research (ICR), Sutton, United Kingdom
    Search for more papers by this author
  • Rosa Ruchlemer M.D.,

    1. Section of Haemato-Oncology, Royal Marsden NHS Foundation Trust/Institute of Cancer Research (ICR), Sutton, United Kingdom
    Search for more papers by this author
  • Nnenna Osuji M.B.Ch.B., M.Sc.,

    1. Section of Haemato-Oncology, Royal Marsden NHS Foundation Trust/Institute of Cancer Research (ICR), Sutton, United Kingdom
    Search for more papers by this author
  • Ilaria Del Giudice M.D.,

    1. Section of Haemato-Oncology, Royal Marsden NHS Foundation Trust/Institute of Cancer Research (ICR), Sutton, United Kingdom
    Search for more papers by this author
  • Estella Matutes M.D., Ph.D.,

    1. Section of Haemato-Oncology, Royal Marsden NHS Foundation Trust/Institute of Cancer Research (ICR), Sutton, United Kingdom
    Search for more papers by this author
  • Anthony Woodman M.Sc., Ph.D.,

    1. BioMedical Centre, Cranfield University, Silsoe, United Kingdom
    Search for more papers by this author
  • Andrew Wotherspoon M.B., B.Ch.,

    1. Section of Haemato-Oncology, Royal Marsden NHS Foundation Trust/Institute of Cancer Research (ICR), Sutton, United Kingdom
    Search for more papers by this author
  • John Swansbury,

    1. Section of Haemato-Oncology, Royal Marsden NHS Foundation Trust/Institute of Cancer Research (ICR), Sutton, United Kingdom
    Search for more papers by this author
  • Claire Dearden M.B., M.D.,

    1. Section of Haemato-Oncology, Royal Marsden NHS Foundation Trust/Institute of Cancer Research (ICR), Sutton, United Kingdom
    Search for more papers by this author
  • Daniel Catovsky M.D., D.Sc.

    Corresponding author
    1. Section of Haemato-Oncology, Royal Marsden NHS Foundation Trust/Institute of Cancer Research (ICR), Sutton, United Kingdom
    • Section of Haemato-Oncology, Brookes Lawley Building, Institute of Cancer Research, Cotswold Road, Sutton, Surrey SM2 5NG, UK
    Search for more papers by this author
    • Fax: (011) 44 0 20 8722 4266



Both pentostatin and cladribine have efficacy in hairy cell leukemia (HCL), but it is not known which agent achieves better results.


We reviewed a series of 219 patients with HCL, with median follow-up from diagnosis of 12.5 years (range 1.0 –34.6 yrs), treated with either pentostatin (n = 185) or cladribine (n = 34), to compare these agents and assess the potential for cure.


Overall response to pentostatin was 96% with a complete response (CR) in 81% and a median disease-free survival (DFS) of 15 years. Response to first-line cladribine was 100% with a CR in 82% and DFS of 11+ years. The relapse rates at 5 years and 10 years were 24% and 42%, respectively, with pentostatin, and 33% and 48% with cladribine. Survival at 10 years was respectively 96% and 100%. CR rates decreased with each sequential relapse through 69% to 45% (P ≤ 0.001). Patients achieving CR after first-line treatment had a significantly longer DFS (P = 0.00007) than those achieving a partial response; a similar result was seen after second-line therapy (P = 0.00001). DFS also declined with sequential treatment (P = 0.00005).


We have shown equivalent efficacies for both agents in the treatment of HCL, with DFS showing no plateau. True cure in HCL remains elusive, but the addition of monoclonal antibodies may be beneficial. Our results suggest that achieving CR should remain the main goal of treatment. Cancer 2005. © 2005 American Cancer Society.

The purine analogs 2′-deoxycoformycin (pentostatin, DCF, Nipent, Supergen, Dublin, CA) and 2-chlorodeoxyadenosine (cladribine, 2CDA, Leustatin, Ortho Biotech, Bridgewater, NJ) remain the most significant advances in treatment of hairy cell leukemia (HCL). These agents induce complete and durable responses in up to 98% of cases.1–25 Recent applications of biologic agents such as the monoclonal antibody rituximab offer well tolerated, therapeutic alternatives, particularly for patients with refractory disease,26–28 but complete response (CR) rates with this antibody do not parallel those seen with the two purine analogs.

Several retrospective and prospective series have evaluated treatment of HCL with either pentostatin or cladribine. Only one report, an earlier analysis of this series,6 attempted to compare these drugs directly and suggested some advantage of pentostatin over cladribine for disease-free survival (DFS) at a median follow-up of 71 months and 45 months, respectively. The present study was, therefore, undertaken to assess the relative efficacies of these two agents and the feasibility of achieving long-term remissions.


A clinical database of patients with HCL (n = 226) was retrospectively reviewed. Of these, 219 patients were included in the current study. Seven patients had received no treatment. There was some crossover of patients at second and third-line treatment (Table 1).

Table 1. Summary of Number of Patients Treated with Pentostatin and Cladribine at Each Successive Course of Therapy
Course of therapyPentostatinCladribineTotal
  1. Numbers in brackets indicate patients retreated with the same agent.

First line18534219
Second line20 (17)53 (9)73
Third line7 (6)13 (11)20

Data were extracted on baseline patient characteristics at the start of treatment, including gender, prior therapy with interferon-alpha (IFN-α) and/or splenectomy, age, blood counts, presence of splenomegaly, and interval from diagnosis to purine analog treatment (Table 2).

Table 2. Demographics of HCL Patients Receiving First-Line Therapy
 Initial treatment
  • a

    excluding interferon given for 2-3 months before purine analog treatment.

Total patients185 34 
Prior therapy    
Prior interferona
Splenectomy and prior interferon351939
No prior therapy76411853
Age in yrs, at first purine analog treatment
 Median52 52 
 Range24-85 30-70 
White blood count (× 109/L)    
 Median3.7 3.3 
 Range0.0-90.0 0.6-20.5 
Hemoglobin (g/dL)
 Median12.0 12.7 
 Range4.1-16.3 8.1-15.7 
Platelets (× 109/L)
 Median120 73 
 Range5-695 22-278 
Splenomegaly if spleen not removed
Interval from diagnosis to purine analog treatment
 Median in yrs1.7 0.6 
 Range in yrs0-18 0.02-29 
Follow-up in yrs since diagnosis
 Median13.6 8.8 
 Range1.5-24.4 1.0-34.6 

The diagnosis of HCL was established by morphology, flow cytometry, and immunohistochemical analysis of peripheral blood, bone marrow, and/or spleen specimens; HCL-variant was excluded by central pathology review.


The first 40 patients treated with pentostatin received 4 mg/m2 intravenously (i.v.) weekly for the first 4 weeks, every 2 weeks thereafter until maximum response, and then one additional injection. All other patients received pentostatin 4 mg/m2 i.v. every 2 weeks until maximal response, consolidated with two further doses. Cladribine was given by continuous i.v. infusion at a dose of 0.1 mg/kg/day over 7 days. A further course of cladribine was given, within 7 months, in 8 patients where CR was not initially attained. Such sequential treatment was considered a single course for the purposes of this analysis. Prophylaxis was given with cotrimoxazole or nebulized pentamidine (against pneumocystis carinii pneumonia) and aciclovir.

Response Evaluation

Response was evaluated on completion of pentostatin treatment and at 3 and 6 months after cladribine therapy, by morphology and immunophenotyping of peripheral blood and bone marrow trephine biopsy. The criteria for response were those of the Consensus Resolution 1987.29 Briefly, CR required morphologic absence of hairy cells in blood and bone marrow and normalization of any organomegaly and cytopenias. Partial response (PR) required normalization of peripheral counts, together with reduction of at least 50% in organomegaly and bone marrow hairy cells, and < 5% circulating hairy cells. All other responses were considered nonresponses (null response). Immunostaining with CD20 and DBA44 was included in the pathology evaluation of more recent cases.

Outcomes and Statistical Analysis

Comparisons were made based on overall response rate, quality of response, relapse rate, DFS, and overall survival (OS). DFS was defined as time from start of treatment to first relapse or death from HCL. Observations for DFS and OS were censored at the date of death from causes unrelated to HCL or the date of most recent follow-up for patients who were alive and (DFS only) with no relapse. Of 34 deaths, only 7 (21%) were related to HCL. In 11 (32%), the cause of death was a second malignancy, in 9 (26%) cardiovascular, 1 (3%) accident, and 6 (18%) unknown. Survival curves were drawn according to the method of Kaplan and Meier,30 and differences in response were estimated by the log-rank test.31 Regression analysis to determine which variables were predictive of quality of response, DFS, and OS was carried out, and those variables attaining a significance of P ≤ 0.2 were included in a Cox proportional hazards model.32 Analysis was performed using the Statistica (StatSoft, Tulsa, OK) software package.


Response to First-Line Treatment, DFS, and OS

Of 219 patients, 185 received initial pentostatin, and 34 received cladribine. Baseline characteristics were similar, apart from an increased male predominance in the pentostatin group (Table 2). Response rates were also similar, with overall response rates of 96% for pentostatin and 100% for cladribine; responses were durable and comparable for both agents, as were the relapse rates at 5 and 10 years (Table 3). With non-HCL–related deaths censored, estimated median DFS was 15 years for pentostatin and 11+ years for cladribine (Table 3, Fig. 1); OS remained at 96% at median follow-up 10.8 years (pentostatin) (range 0.3–17.9 yrs) and 100% at median follow-up 7.2 years (cladribine) (range 0.5–11.5 yrs) (Table 3, Fig. 2). Median OS has not been reached for either agent. No statistical differences were found between the two agents.

Table 3. Response, Disease-Free Survival, Relapse Rate, and Overall Survival after First-Line Therapy
 Pentostatin (n = 185)aCladribine (n = 34)a
  • a

    Of these, 2 pentostatin patients and 1 cladribine patient were not evaluable for response.

 Complete (CR)81%82%
 Partial (PR)15%18%
 Overall rate (CR + PR)96%100%
Median disease-free survival (DFS) in yrs1511+
Relapse rate  
 at 5 yrs24%33%
 at 10 yrs42%48%
Overall survival (OS) at 10 yrs96%100%
Figure 1.

disease-free survival (DFS) after first-line therapy, comparing pentostatin with cladribine. The seven nonresponders (to pentostatin) are excluded. Log-rank test P = 0.55.

Figure 2.

Overall survival, comparing first-line treatment with pentostatin and cladribine. Log-rank test P = 0.34.

Patients achieving a CR, regardless of the agent used, showed significantly longer DFS than those attaining only a PR (P = 0.00007) (Fig. 3).

Figure 3.

DFS after first-line pentostatin and cladribine, analyzed by response. Complete responders had a significantly better chance of prolonged remission. Log-rank test P = 0.00007.

Treatment after Relapse

Patients who relapsed after initial therapy were retreated with one of the purine analogs (Table 1). Of these, 61 were evaluable for quality of response and 69 for DFS. Twenty patients went on to receive third-line therapy, of whom 11 were evaluable for quality of response and 20 for DFS. The responses to second and third-line therapies are summarized in Table 4 and illustrated in Figure 4. The ability to attain CR diminished with each course of therapy (P ≤ 0.001). The probability of achieving CR was not affected by whether the patient was retreated with the same agent (CR = 71%) or switched to the alternative (CR = 68%).

Table 4. Response to Second-and Third-Line Therapy
 Percentage second line P ≤ 1Percentage third line P ≤ 0.2
  • CR: complete response; PR: partial response.

  • a

    patients not evaluable = 12.

  • b

    patients not evaluable = 9.

ResponsePentostatin (n = 16)aCladribine (n = 45)aPentostatin (n = 4)bCladribine (n = 7)b
Overall Rate (CR + PR)94100100100
Figure 4.

Response to first, second and third-line therapy. The likelihood of achieving complete response decreased with each course. Chi-square test P ≤ 0.001.

Although relapse rates remained relatively stable after each course of treatment (38% after first-line, 35 % after second-line, 28 % after third-line), there was a significant reduction in DFS with successive courses of treatment (P = 0.00005) (Fig. 5). DFS after second-line did not correlate with quality of first-line response.

Figure 5.

DFS after successive courses of pentostatin and/or cladribine. Remissions were progressively shorter with each course of therapy. Cox–Mantel test P = 0.00005.

DFS was comparable for pentostatin and cladribine after second-line therapy (P = 0.46). However, as with first-line treatment, patients achieving CR after second line had a better chance of prolonged remission (P = 0.00001) than those achieving PR.

Factors Predicting Quality of Response and DFS

No study variable was found to be significantly associated with quality of response (CR or partial response) after first-line therapy. However, at second-line, no prior therapy with splenectomy and/or IFN-α predicted for CR (P = 0.01).

Baseline anemia (P = 0.0004) and thrombocytopenia (P = 0.003) in patients attaining CR after first-line treatment were found to correlate with shorter DFS. In a Cox proportional hazards analysis, anemia remained a significant independent predictor of shorter DFS (P = 0.02). In the first-line pentostatin-treated group, patients with both anemia and thrombocytopenia at the start of treatment had a 64% chance of relapse compared with a 22% chance for patients with neither (chi-square P ≤ 0.001).


Pentostatin and cladribine have become the standard treatment of HCL. Both agents show high response rates with durable remissions. Randomized controlled trials have demonstrated superiority of pentostatin over IFN-α14, 33 for this condition. In some reports,22, 34 patients refractory to pentostatin have responded to cladribine, suggesting a lack of cross-resistance between these two drugs. But which, if either, of these agents is superior for treatment of HCL remains unknown. The feasibility of a randomized controlled trial to compare pentostatin and cladribine is restricted by high response rates and durable remissions, which mandate large sample sizes and markedly prolonged follow-up to detect significant differences. An earlier retrospective analysis performed at our institution,6 suggested possible benefits of pentostatin over cladribine in terms of DFS. Median follow-up in published series for pentostatin ranges from 1.1 to 9.8 years, and for cladribine from 0.75 to 10.2 years. We assessed retrospectively the efficacies of these agents in a large sample cohort with prolonged follow-up (median 10.8 yrs and 7.2 yrs, respectively) and demonstrated no difference in terms of response rates, relapse rates, DFS, and OS between them. This equivalency of outcome persisted after second- and third-line therapy.

Despite best treatments, a subset of patients may remain refractory. Although inadvertent inclusion of HCL-variant may explain this finding in some studies, this differential diagnosis was actively excluded in our series. Catovsky et al.16 and Mercieca et al.35, 36 identified the presence of lymphadenopathy as a feature associated with poor response. Other studies have evaluated a number of factors that impact on response, DFS and/or OS, with varying results. Age,1, 5, 14, 23, 25 hemoglobin,3, 5, 14, 25 splenectomy,13 splenomegaly,14, 23 adenopathy,16, 25 leukopenia,25 leukocytosis,3, 5, 23 number of previous treatments (including splenectomy),18 disease duration and performance status5, 13 have emerged as possible prognostic factors in published data. Although there is some overlap of these factors between studies, there remains no consensus on prognostic factors. Our analysis points to previous splenectomy and/or IFN-α as a significant predictor for failure to attain CR after second-line purine analog therapy. In patients attaining CR at first-line, baseline anemia emerges as the most significant predictor of shorter DFS. Relapse rates in this study were comparable to those of other published studies.

Residual hairy cells in bone marrow persist after both pentostatin and cladribine and may be detected using immunohistochemical or molecular methods.37–42 Such residual disease is possibly more common after cladribine,41 but this finding may reflect differences in endpoints used for therapy. The fate of this residual cell population is not clear, with some authors describing a steady increase,39 whereas others report a stable, unchanging population.38, 40, 41 The presence of residual cells, in conjunction with disease relapse even after 16 years, suggests that a true cure remains elusive. It is not clear how far the presence and size of this residual cell population affects outcome. In this series and others, residual disease has previously been associated with shorter event-free survival after both pentostatin and cladribine.6, 43 With extended follow-up and increased accrual, we confirm a significant difference in outcome between patients achieving CR versus PR, with DFS curves diverging after approximately 2 years. Attainment of CR leads to superior DFS after both first- and second-line therapy with purine analogs. But, with successive courses of treatment, the ability to reattain CR is significantly reduced.

Hoffman et al.20 reported shorter remissions after retreatment with cladribine, and Tallman et al.19 suggested that prior treatment with IFN-α may be predictive of relapse in patients receiving cladribine. We have expanded on this point by showing significant reduction in DFS with successive courses of therapy.


Our findings highlight the need for accurate assessment of response after treatment, and the need to maximize primary responses, in order to prolong remissions. The use of rituximab or other antibodies to improve the quality of responses to pentostatin or cladribine44–46 is an attractive approach in light of these findings. Assessment of bone marrow response remains critical to ensuring that CR has been achieved to secure a favorable long-term outlook.


The authors thank Anton Planting, data manager and to all the clinicians who provided patient data. This study was approved by the Royal Marsden/ICR Committee for Clinical Research and by the Royal Marsden Local Research Ethics Committee.