Neutropenic fever following cladribine therapy for symptomatic hairy-cell leukemia: Predictive factors and effects of granulocyte-macrophage colony-stimulating factor

Annals ofOncology 6: 371-375, 1995. O 1995 Kluwer Academic Publishers. Printed in the Netherlands.

Original article Neutropenic fever following cladribine therapy for symptomatic hairy-cell leukemia: Predictive factors and effects of granulocyte-macrophage colony-stimulating factor G. Juliusson,1-3 R. Lenkei,5 G. Tjonnfjord,2'6 D. Heldal2 & J. Liliemark4 departments of Medicine, Huddinge Hospital, Huddinge Sweden; 2The National Hospital, Oslo, Norway; 3 Department ofHematology, Linkoping Hospital; 4Departments of Oncology and Clinical Pharmacology, Karolinska Hospital, Karoiinska Institute; !SBL Calab Diagnostics, Stockholm, Sweden; 6 The Institute of Transplantation Immunology, Oslo, Norway

a low percentage of myelopoietic cells, low albumin, and high C-reactive protein predicted neutropenic fever. The addition Background: Neutropenic fever is the commonest complica- of GM-CSF did not improve the kinetics of recovery for neution of cladribine therapy for hairy-cell leukemia (HCL), trophils, hemoglobin or platelets, as compared to matched leading to a 3% mortality rate. Our aim was to identify pre- control patients. However, GM-CSF significantly reduced dictive factors and evaluate the effects of concomitant granu- cladribine-induced lymphopenia, but not the incidence of neutropenic fever. locyte-macrophage colony-stimulating factor (GM-CSF). Patients and methods: We studied 102 patients with active Conclusion: Factors predicting febrile neutropenia were HCL given cladribine for 7 days. Pretreatment parameters identified. GM-CSF protected from cladribine lymphotoxicpredicting neutropenic fever were analysed. Twelve patients ity but did not improve neutropenia or febrile episodes. at high risk for febrile complications also received 400 (ig GM-CSF per day on days 1 through 21. Results: Pretreatment anemia, hypocholesterolemia, bone Key words: 2-chlorodeoxyadenosine, CdA, cladribine, GMmarrow differential with a high percentage of hairy cells and CSF, infections, T-lymphocyte subsets, hairy-cell leukemia Summary

Introduction

Hairy-cell leukemia (HCL) [1] is a malignant disease of differentiated B-cells with typical morphology, phenotype [2], and genotype [3]. The clinical characteristics are cytopenia, splenomegaly, and impaired immunity. Opportunistic infections are common and may be fatal [4]. The diagnosis of HCL is made on bone marrow analyses, using morphology, cytochemistry [5], and flow cytometry. Recently cladribine (2-chlorodeoxyadenosine, CdA) [6, 7], was shown to induce durable complete remissions when given as a single continuous intravenous infusion of 0.085 to 0.1 mg/kg per day over seven days, and it has become the first-line choice of treatment for HCL. The response rate and duration with the current cladribine regime are excellent, but little effort has been made to improve feasibility and to reduce toxicity. We have recently shown that cladribine may be given by subcutaneous injection, with the same response rate and less toxicity due to the amelioration of phlebitis, and at a much lower cost [8]. The only significant short-term toxic effect is neutropenic fever [9-11] that develops during the first month after the start of treatment in about one-third to onehalf of the patients. No clinical long-term toxicity has been reported, but the long-lasting lymphopenia has been a concern [2, 12, 13]. Most episodes with neutro-

penic fever present during or shortly following therapy, are brief, with a limited deterioration of general wellbeing, and with no documented etiologic agent or infectious focus. It has thus been suggested that the fever is not caused by infection, but by cytokine release [11], associated with rapid lysis of circulating hairy cells. However, documented infections, mainly opportunistic [9], occur in a significant proportion of the cases [9-11, 14-18]. A cytomegalovirus (CMV) viremia complicating the first cladribine course of a HCL patient delayed a scheduled second course, and enabled the observation that a single course may be sufficient to achieve long-term complete remission (E. Beutler, public lectures). Significantly, such infections may be fatal. In our first HCL study, 5 of 16 patients had opportunistic infections caused by CMV, Candida and aspergillus, two of which were fatal [9]. In the combined study of 196 patients [18) at the Scripp's clinic [11] and the MD Anderson Cancer Center [14], 69% developed fever >100°F, 47% had neutropenic fever, and infectious signs were seen in 28%. Of 54 episodes of documented infections, 23 were caused by bacteria, 11 by viruses, and 11 by fungi, and 6 patients (3%) died of infection [18]. Fatal septicemias are also reported [10, 14, 17]. In our studies 4 of 102 patients have died of infections [8,9]. The prophylactic use of oral antibiotics has not been

372

shown to prevent fever, and virtually all HCL patients with neutropenic fever are hospitalized and given intravenous antibiotics [11,18]. This complication thus has a significant influence of the quality of their lives and the total cost of treatment. The objectives of the present study were to identify risk factors for subsequent neutropenic fever in a large patient population, and to evaluate the addition of granulocyte-macrophage colony-stimulating factor (GM-CSF) and cladribine dose reduction to reduce toxicity.

Clinical intervention Twelve consecutive patients presumed to be at high risk for infection, i.e., with hemoglobin <120 g/1, neutrophils < 1 x lOTliter, and platelets <100 x lOTliter, were given GM-CSF (Leucomax*, Schering-Plough Int., Kenilworth, NJ; provided for the study by Schering-Plough, Stockholm, Sweden), 400 |ig total dose per day in one or two daily doses by subcutaneous injection from day 1 through day 21, in conjunction with cladribine given on days 1 through 7. For the statistical analysis, patients fulfilling these criterias who were treated with the same cladribine dose but who did not have GM-CSF were used as controls. Another eight patients received low-dose cladribine (2 mg/sqm/day on days 1-7), as previously described [21].

Patients and methods Patients One hundred two patients with active hairy-cell leukemia were included from March 1990 through November 1993. There were 20 females and 82 males, and their mean (± SD) age was 56 ± 12 years. Forty-six patients were previously untreated, 30 had prior interferon only, 5 were splenectomized, and 10 had had both treatments. In addition, 11 patients had prior alkylating agents or anthracyclines with or without interferon and splenectomy. Active disease was defined by hemoglobin < 110 g/1, neutrophil count < l x 109/l, platelets <100 x lOTliter, repeated infections, and/or general symptoms. Diagnosis was made through morphology with cytochemistry [5] and flow cytometry [2] on repeated bone marrow aspirations and biopsy specimens. Flow cytometry on pre-cladribine bone marrow aspirates were successfully analysed in 87/102 patients. Pretreatment evaluation consisted of extensive blood tests including C-reactive protein, albumin, p2-microglobulin, lactic dehydrogenase, bone marrow aspiration and biopsy, isotope or ultrasound scan of the liver and spleen in non-splenectomized patients, and flow cytometry of blood and bone marrow cells with enumeration of hairy cells, CD20+, CD4+, CD8+, NK-cells, and myeloid cells, as previously described [2]. Mean (± SD) pretreatment blood counts were as follows: hemoglobin 111 ± 21 g/1, lymphocyte count 4.8 ± 7.8 x 109/l, neutrophil count 0.99 ± 0.85 x 1071, platelets 102 ± 72 x 1071. Low-density lipoprotein (LDL)-bound cholesterol were either measured directly or calculated by FriedewaJd's formula (LDLcholesterol — total cholesterol - high-density lipoprotein (HDL)bound cholesterol — (0.45 x triglyceride value). Bone marrow cellularity was assessed morphologically on bone marrow aspirate clots and bone biopsies. The percentages of hairy cells and myelopoietic cells in the bone marrow were analysed by flow cytometry. The mean (± SD) bone marrow cellularity was 69% ± 25%, with hairy cells 38% ± 24%, and myelopoietic cells 27% ± 19%.

Cladribine treatment Cladribine was gratefully received from Dr. Ernest Beutler (March 1990 through April 1992) or purchased from Dr. Zygmunt Kazimierczuk, The Foundation for the Development of Diagnostics and Therapy, Warsaw, Poland, and a solution of 2 mg/ml in saline was prepared in the Huddinge Hospital pharmacy. Seventeen patients received 0.085 mg cladribine/kg/day through continuous intravenous infusion for 7 days [9|. From December 1991 subcutaneous injections were used instead, according to a new protocol [8] with the corresponding dose of 3.4 mg cladribine per square meter once daily for seven consecutive days. This schedule yields the same drug plasma levels [8, 19, 20], clinical effect [8| and toxicity to lymphoid subsets [2] as those of the conventional infusion regimen. The protocol was approved by the Medical Products Agency, and local ethics committees, and informed consent was obtained.

Patient monitoring Peripheral blood counts were in most instances evaluated once or twice weekly during the first months following cladribine treatment. Blood and bone marrow sampling for morphology and flow cytometry was repeated at about 3 months (n - 78), 6 months (n — 70), and 1 year (n — 49) following treatment [2]. Additional blood for flow cytometry was sampled 1 week (n — 29) and 1 months (n - 11) after start of treatment in some patients [2]. Definition of infection Patients who were hospitalized with intravenous antibiotics for fever with or without associated infectious symptoms were regarded as having a major infection. Our prestudy recommendation was that patients with neutrophil counts of less than lOTliter and fever of 38.5 "C or more for at least 4 hours be given intravenous antibiotics following appropriate investigations, also considering opportunistic agents. Patients who were given oral antibiotics or no therapy for infectious symptoms with or without fever were classified as having minor infection. Moderate chills with temperature less than 38 "C did not qualify for the designation of a minor infection.

Criteria for response Response was assessed according to the consensus resolution [22]. Complete remission (CR) was defined as a normalization of organomegaly, hemoglobin > 120 g/liter, platelets > 100 x lOTliter, neutrophils >1.5 x lOTliter, and no hairy cells in blood or bone marrow aspirate and biopsy by morphology. In addition, less than 0.1% B-ly 7 (CD 103) positive cells and no B-cell clonality in flow cytometry analyses of bone marrow aspirates were required |8|. Partial remission required normal blood counts as above, and reduction of organomegaly and bone marrow hairy cells by at least 50%, with less than 5% circulating hairy cells.

Statistical analysis Parametric and non-parametric description of data, student's T-test for independent samples, and correlation analyses were performed using the Statistica/Mac software (StatSoft, Tulsa, OK, U.S.A.).

Results Of the 102 patients, 52 had no post-treatment infection, 13 had minor infection, and 37 had febrile neutropenia qualifying for major infection. Pretreatment factors related to subsequent infectious complications are shown in Table 1. Parameters yielding a p-value of less

373 Table 1. Pretreatment parameters related to subsequent infection. Number of patients (n-102)

No or minor 65

Major 37

p-value

Hemoglobin, g/1 LDL cholesterol, mmol/1 Total cholesterol, mmol/1

115± 21 3.310.9 5.011.1

104120 2.610.9 4.41 1.1

0.014 0.015 0.033

Number of patients (n-102)

No 52

Any 50

p-value

Hemoglobin, g/1 LDL cholesterol, mmol/1 Total cholesterol, mmol/1 BM myelopoiesis, % Albumin, g/1 BM hairy cells, % C-reactive protein, mg/1

119121 3.31 0.8 5.110.9 32121 39.014.8 33124 6.3115.1

104119 2.711.0 4.511.2 21113 36.515.9 45123 15.0124

0.00028 0.014 0.015 0.017 0.028 0.037 0.04

Lymphocytes

0,0 4 84

Neutrophils

BM denotes bone marrow. P-values achieved from T-test for independent samples.

than 0.05 with the T-test for independent samples are shown. No correction of p-values for multiple analyses has been performed. Correlation coefficients (r-values) between indicated pretreatment parameters are shown in Table 2. Median and quartile blood counts from start of cladribine therapy for patients receiving GM-CSF and for matched controls are shown in Fig. 1. There was no difference with respect to hemoglobin, neutrophils, monocytes, and platelets. However, the lymphocyte counts in the GM-CSF-treated patients were significantly higher than in the matched control patients, and on the same level as the overall patient population. Similar data were achieved for patients receiving cladribine 2 mg/sqm/day [21]. Side effects from GM-CSF included frequent chills, rashes, and eosinophilia. Final response, incidence of post-treatment infection, and lymphocyte subsets in relation to treatment are shown in Tables 3, 4, and 5. Patients given cladribine with GM-CSF had significantly higher CD4, CD8,

0

28 56 Days From Start of Cladribine Therapy

Fig. 1. Blood counts following standard dose cladribine treatment in patients with pretreatment pancytopenia (hemoglobin <120 g/liter, neutrophils <1 x 1071, and platelets <100 x lOVliter). Thick lines indicate median values, thin lines show quartile values. Solid lines indicate patients treated with GM-CSF (n - 12). Dotted lines indicate patients treated without GM-CSF (n - 15). a) Lymphocytes (x 10~Vliter); p < 0.05 from day 10 through day 160; p < 0.01 from day 21 through day 35. b) Neutrophils (x 10~*/liter); no significant difference.

and CD20 cell counts at 6 months after treatment than did control patients (p - 0.018, p = 0.028, and p 0.032,respectively)and higher CD8 and NK-cell levels at 3 months (p - 0.05, and p - 0.036, respectively).

Table 2. Correlation coefficients between pretreatment parameters related to subsequent infection (r-values). Hemoglobin

Hemoglobin Neutrophils Platelets C-reactive protein Albumin LDL cholesterol BM cellularity BM hairy cells BM myelopoiesis

0.221 0.31 b -0.25* 0.48c 0.25 -0.32" -O.38c 0.42c

LDL cholesterol

BM myelopoiesis

BM hairy cells

0.25 0.02 0.54c -0.29' 0.17

0.42c 0.26' 0.29b -0.13 0.09 0.07 -0.3 l b -0.70 C

-0.38 c -0.11 -0.25' 0.05 -0.13 -0.05 0.35"

-0.18 -0.05 0.07

Discussion Cladribine therapy of HCL yields a high complete remission rate, with a prolonged remission duration. Table 3. Remission status following one course of cladribine therapy according to concomitant GM-CSF treatment. Number of patients (percentage).

All patients (n - 102)

Complete remission 74 (73%)

Partial remission 20 (20%)

Minor or no response 8 (8%)

-0.70 c Patients with pretreatment pancytopenia

LDL denotes low density lipoprotein, and BM denotes bone marrow. •p<0.05; b p<0.01; c p< 0.001.

GM-CSF ( n - 12) No GM-CSF ( n - 2 1 )

9(75%) 18(86%)

3(25%) 3(14%)

0(0%) 0(0%)

374

The main predictive factors for neutropenic fever were not previous therapy, history of infections, duration of the disease, or blood neutrophil or hairy cell Minor Major No Factor counts, but rather the hemoglobin value and the serum infection infection infection LDL-cholesterol (or total cholesterol). Furthermore, All patients (n - 102) 13 (13%) 37 (36%) 52 (51%) patients with subsequent neutropenic fevers also had Patients with pretreatment pancytopenia significantly higher percentages of hairy cells and lower percentages of myelopoietic cells in the bone marrow, GM-CSF ( n - 1 2 ) 2(17%) 3(25%) 7(58%) No GM-CSF ( n - 2 1 ) 7(33%) 4(19%) 10(48%) higher C-reactive protein and lower serum albumin than patients without posttreatment complications. Thus, an unfavourable ratio between hairy cells and Table 5. Lymphocyte subsets in relation to cladribine dose and myelopoietic cells in the bone marrow, and signs of treatment with GM-CSF (mean ± S.D.). tumor-induced inflammation predicted infection. There were significant correlations between the preMark- Time GM-CSP Control' Low-dose Standarder from n - 12 n-21 n-8 dose treatment hemoglobin value and all other predictive start of n-94 factors except for cholesterol, indicating that these two treatment parameters reflect different aspects relevant for neutroCD20 Baseline 1116± 1387 245815635 196713626 273716831 penic fever. However, the way in which hypocholester1 week 18± 13 818 80194* 22134* olemia is related to posttreatment fever remains unb 1 month n.d. 35161 48193 8±l clear. 3 months 631140 691111 61198 47±41 6 months 218± 192* 95191* 83156 1251155 There is a rapid and dramatic decrease of all blood 1 year 241± 197 1821144 n.d. 1821149 lymphoid subsets following cladribine treatment of CD4 Baseline 874±619 8311747 145011792 9531954 HCL [2], whereas neutrophil counts are less affected 1 week 124±93 105148 3961332* 1561135* b [8]. Since almost all febrile episodes present within the 1 month 345±285 n.d. 4381282 3781363 3 months 300±165 2071137 3341101 2831167 first weeks after start of treatment, the cladribine-in6 months 4071230* 2301137* 4711196 3231198 duced immune suppression during this period seems to 1 year 3491154 n.d. 4601218 4061168 be the most significant side effect of therapy. 6461521 CD8 Baseline 5731416 10611 1820 93811725 Neutropenia, monocytopenia and infections are 1 week 87135 53122 2571300 1061125 characteristic features of advanced HCL. Granulocyte 1 month n.d. 95115" 2551233 126170 3 months 4431287* 2391206* 209138 3271250 colony-stimulating factor (G-CSF) was previously 6 months 4991197* 3071202* 4571157 3631220 shown to increase neutrophils in three of four studied 1 year 5211222 4181168 n.d. 4691225 HCL patients [23]. Our decision to evaluate GM-CSF NK Baseline 1381119 1361135 1401103 2181421 for possible prevention of neutropenic fever was due to 1 week 31113 1113 21117 991169 its documented ability to increase both neutrophil and 1 month n.d. n.d. 64185 10418 3 months 2591266* 97180* 1101126 1531161 monocyte counts, and the fact that monocytes influ6 months 1911127 136191 120145 1591109 enced by GM-CSF are shown to have an augumented 1 year 2271190 2221117 n.d. 2251155 activity against Candida albicans [24], a significant NK indicates natural killer cells (CD3-negative, CD16 or CD56 positive), n.d. cause of infection in HCL. Since post-therapeutic neu- no data. 1 Patients with pretreatment pancytopenia receiving standard dose cladri- tropenic fever mostly presents early, GM-CSF was given in conjunction with cladribine. However, GMbine. b Data from 2 paUents only. CSF with cladribine did not improve the neutrophil * p < 0.05 when comparing GM-CSF vs. control, or low-dose vs. standard counts more rapidly than cladribine alone, nor did it dose. improve the monocyte counts or reduce the incidence of febrile neutropenia. Prophylactic use of GM-CSF We recently documented that continuous intravenous with cladribine in this setting thus does not seem indiinfusion is an unnecessary mode of administration for cated. G-CSF might be more effective than GM-CSF cladribine; subcutaneous injections give the same re- for the stimulation of granulocytopoiesis in HCL; howsponse rate with less toxicity [8]. In order to further im- ever, it is more likely that the lack of early improvement prove HCL therapy we found it most important to in our study was due to the simultaneous cladribine address the posttreatment infections. Neutropenic therapy. In addition, we selected patients with adfever frequently develops within a few weeks after clad- vanced disease and pancytopenia for GM-CSF, whereribine therapy, and causes significant morbidity and as no such criteria were applied to the G-CSF study additional treatment costs due to the unavoidable need [23]. Patients with advanced disease may have a refor hospitalization and intravenous antibiotics. Fatal duced neutrophil response also to G-CSF. opportunistic infections and bacterial septicemias seem To our surprise, GM-CSF significantly reduced more common following cladribine than during inter- cladribine toxicity to the normal lymphoid cell subsets. feron treatment. We therefore aimed to define pretreat- The protective effect of GM-CSF in normal T-cells ment factors associated with neutropenic fever follow- against cladribine-induced lymphotoxicity is a new obing cladribine treatment of HCL, and to evaluate the servation that, if confirmed, might prove useful. The effects of concomitant GM-CSF. Table 4. Post-cladribine neutropenic fever according to concomitant GM-CSF treatment. Number of patients (percentage).

375 cause of this protection is unknown, but might be found within the GM-CSF-induced cytokine cascade. However, episodes with neutropenic fever were not ameliorated. Side effects of treatment, such as chills and rashes, seemed more common than with cladribine alone, and some patients required premature withdrawal of GM-CSF. However, patients with active HCL are prone to cutaneous reactions from several drugs, especially antibiotics and allopurinol. Predictive factors may enable individualized cladribine therapy for HCL. Patients at high risk for neutropenic fever might benefit from treatment within clinical studies evaluating initial low-dose cladribine therapy [21], pretreatment with interferon, and the role of different colony-stimulating factors. Acknowledgments

The study was supported by The Nordic Cancer Union, the Swedish Cancer Society, The Swedish Society for Medical Science, the Karolinska Institute's Research Funds, The Medical Research Council, and Bergliot and Sigurd Skaugens Fond til Bekjempelse av Kreft. The granulocyte-macrophage colony-stimulating factor was provided by Schering-Plough, Stockholm, Sweden.

References 1. Bouroncle BA, Wiseman BK, Doan CA. Leukemic reticuloendotheliosis. Blood 1958; 13:609-30. 2. Juliusson G, Lenkei R, Liliemark J. Flow cytometry of blood and bone marrow cells in hairy-cell leukemia. Phenotype of hairy cells and lymphocyte subsets following treatment with 2-chlorodeoxyadenosine. Blood 1994; 83: 3672-81. 3. Haglund U, Juliusson G, Stellan B, Gahrton G. Hairy-cell leukemia is characterized by clonal chromosome abnormalities clustered in specific regions. Blood 1994; 83: 2637-45. 4. Golomb HM, Hadad LJ. Infectious complications in 127 patients with hairy-cell leukemia. Am J Hematol 1984; 16: 393401. 5. Paoletti M, Bitter MA, Vardiman JW. Hairy-cell leukemia: Morphologic, cytochemical and immunologic features. Clin LabMed 1988; 8: 179-95. 6. Beutler E. Cladribine (2-chlorodeoxyadenosine). Lancet 1992; 340:952-6. 7. Bryson HM, Sorkin EM. Cladribine. A review of its pharmacodynamic and pharmacokinetic properties and therapeutic potential in lymphoproliferative disorders. Drugs 1993; 46: 872-94. 8. Juliusson G, Heldal D, Hippe E et al. Subcutaneous injection of 2-chlorodeoxyadenosine for symptomatic hairy-cell leukemia. J Clin Oncol 1995; 13 (in press). 9. Juliusson G, Liliemark J. Rapid recovery from cytopenia in hairy-cell leukemia following treatment with 2-chloro-2'deoxyadenosine (CdA): Relation to opportunistic infections. Blood 1992; 79: 888-94.

10. Betticher DC, Fey MF, von Rohr A et al. High incidence of infections after 2-chlorodeoxyadenosine (2-CDA) therapy in patients with malignant lymphomas and chronic and acute leukaemias. Ann Oncol 1994; 5: 57-64. 11. Saven A, Piro L. Drug therapy: Newer purine analogues for the treatment of hairy-cell leukemia. N Engl J Med 1994; 330: 691-7. 12. Keating MJ. Immunosuppression with purine analogues - the flip side of the gold coin. Ann Oncol 1993; 4: 347-8 (Editorial). 13. Seymour JF, Kurzrock R, Freireich EJ, Estey EH. 2-Chlorodeoxyadenosine induced durable remissions and prolonged suppression of CD4+ lymphocyte counts in patients with hairy-cell leukemia. Blood 1994; 83: 2906-11. 14. Estey EH, Kurzrock R, Kantarjian HM et al. Treatment of hairy-cell leukemia with 2-chlorodeoxyadenosine. Blood 1992; 79: 882-7. 15. Tallman MS, Hakimian D, Variakojis D et al. A single cycle of 2-chlorodeoxyadenosine results in complete remission in the majority of patients with hairy-cell leukemia. Blood 1992; 80: 2203-9. 16. Lauria F, Benfenati D, Raspadori D et al. High complete remission rate in hairy-cell leukemia treated with 2-chlorodeoxyadenosine. Leukemia Lymphoma 1993; 11: 399-404. 17. Sorensen JM, Vena DA, Montello MJ et al. Treatment of hairy cell leukemia (HCL) with 2-chlorodeoxyadenosine (CdA) under the group C protocol mechanism. Proc Am Soc Clin Oncol 1993; 12: 302 (Abstr 988). 18. Ortho Biotech Inc, Raritan NJ. FDA drug registration file for cladribine (2-chlorodeoxyadenosine). Data presented in: Leustatin (Cladribine) Injection. Product Monograph, 1993. 19. Liliemark J, Albertioni F, Hassan M, Juliusson G. On the bioavailability of oral and subcutaneous 2-chloro-2'-deoxyadenosine in humans: Alternative routes of administration. J Clin Oncol 1992; 10:1514-8. 20. Liliemark J, Juliusson G. On the pharmacokinetics of 2-chloro2'-deoxyadenosine nucleotides in leukemic cells. Comparison of intermittent and continuous intravenous infusion and after oral administration in patients with leukemia. Clin Cancer Res 1995; 1 (in press). 21. Juliusson G, Lenkei R, Tjennfjord G et al. Low-dose cladribine for symptomatic hairy-cell leukaemia. Br J Haematol 1995; 89: 637-9. 22. Anonymous. Consensus resolution: Proposed criteria for evaluation of response to treatment in hairy-cell leukemia. Leukemia 1987; 1:405-8. 23. Glaspy JA, Baldwin GC, Robertson PA et al. Therapy for neutropenia in hairy-cell leukemia with recombinant human granulocyte colony-stimulating factor. Ann Intern Med 1988; 109: 789-95. 24. Smith PD, Lamerson CL, Banks SM et al. Granulocyte-macrophage colony-stimulating factor augments human monocyte fungicidal activity for Candida albicans. J Infect Dis 1990; 161: 999-1005. Received 20 December 1994; accepted 8 February 1995. Correspondence to: Dr. Gunnar Juliusson Department of Hematology University Hospital S-581 85 Linkoping Sweden