Infections in Patients with Chronic Lymphocytic Leukemia Treated with Fludarabine

Methods

Patients

We identified 402 consecutive patients with chronic lymphocytic leukemia who were treated with fludarabine-containing regimens at the University of Texas M.D. Anderson Cancer Center from November 1984 through March 1992. We reviewed the patient records of the leukemia section to determine the clinicopathologic features of infections in this patient population.

Four patient groups were evaluated: previously treated patients and patients with newly diagnosed disease who received fludarabine alone or fludarabine plus prednisone. Patients were enrolled if they had either advanced disease (Rai stage III or IV) or earlier-stage disease (Rai stage 0-II) with evidence of active disease, defined by an increase in leukocyte count or tumor burden (including lymphadenopathy, hepatomegaly, splenomegaly, or bone marrow infiltration).

All patients received intravenous fludarabine, 30 mg/m² of body surface area per day for 5 days, with or without oral prednisone, 30 mg/m² per day for the same duration. Courses were repeated at 4-week intervals unless toxicity necessitated a delay. We used the response criteria outlined by the National Cancer Institute-Sponsored Working Group [6]. Complete responders were further stratified according to the presence or absence of lymphoid marrow nodules or infiltrates into a complete remission group and a complete remission with nodules group [7].

A workup consisting of history; physical examination; and complete hematologic and biochemical evaluation, including complete blood counts with differential and platelet counts and sequential multiple analysis, was conducted in all patients [8]. Bone marrow aspirates and biopsy samples were also obtained. We measured CD4 counts before the first course of chemotherapy and after three cycles. All febrile episodes were reviewed. Mild and moderate infections, such as those requiring no therapy or topical or oral therapy, were not included. Major infections were the focus of the analysis and included life-threatening episodes, such as disseminated infections (including bloodstream infections), pneumonias, meningitis, and fevers of undetermined origin that required parenteral antibiotic therapy in a hospital setting. More than one major infection could occur in a given patient. A fever of undetermined origin was defined as persistence of fever (body temperature > 101 °F) over 4 days, during which conventional diagnostic measures were unrevealing. Neutropenia was defined as a neutrophil count less than 1000 cells/mL.

Statistical Analysis

Associations between patient characteristics and major infection were evaluated by using the chi-square test. The level of significance between groups was set at a P value of 0.05 or less. Cut-points for quantitative variables were those that defined abnormal levels or others in common use. Clinically important variables were included in a multivariate logistic regression model from which adjusted odds ratios and 95% CIs were derived. Analyses were performed with the software programs Statistica (Statsoft, Tulsa, Oklahoma) and SAS/STAT (SAS, Inc., Cary, North Carolina).

Results

Table 1 shows the baseline characteristics of the 402 patients by treatment group. Patients previously treated for chronic lymphocytic leukemia were older and had more advanced Rai stage, worse performance status, lower levels of serum albumin and immunoglobulins, lower absolute neutrophil count, and lower platelet and circulating CD4 counts than those who not previously treated. Patients who received fludarabine alone differed substantially from those who received fludarabine plus prednisone. In the previously treated group, a higher proportion of patients who received fludarabine alone had lower serum IgM levels and higher CD4 counts than patients who received fludarabine plus prednisone. However, the latter group of patients was more likely to have refractory disease. Patients not previously treated for chronic lymphocytic leukemia who received fludarabine alone had a higher Rai stage and lower absolute neutrophil counts than those who received fludarabine plus prednisone.

The median number of previous regimens in the previously treated group was 2.5. Eighty-seven percent of these patients had received alkylating agents (chlorambucil or cyclophosphamide). Combination chemotherapy usually included an alkylating agent with an anthracycline and was administered to 22% of patients. Other treatments included cytosine arabinoside, cisplatin, and recombinant interferon-α.

Major infections occurred in all patient groups (Table 2). However, major infections developed in more previously treated patients (144 of 248 patients [58%]) than patients who were not treated previously (53 of 154 patients [34%]) (P < 0.001). Listeriosis or pneumocystosis occurred in 12 of 170 (7%) previously treated patients receiving fludarabine plus prednisone, 0 of 78 previously treated patients receiving fludarabine alone, and 2 of 154 (1%) previously untreated patients receiving fludarabine plus prednisone (P = 0.003). No infections caused by Streptococcus pneumoniae or Haemophilus influenzae were noted. In one pretreated patient, leprosy was reactivated during fludarabine therapy.

Possible risk factors for major infection were examined in univariate analysis by using Pearson chi-square tests. Rai stage; previous antineoplastic therapy; refractoriness to fludarabine-based therapy; levels of serum β₂-µglobulin, albumin, creatinine, and immunoglobulins; absolute granulocyte count; bone marrow lymphocyte count; platelet count; therapy with prednisone; and Zubrod performance status were evaluated (Table 3). Factors associated with risk for developing major infection were advanced Rai stage (P < 0.001), previous antineoplastic therapy (P < 0.001), refractoriness to fludarabine-based therapy (P < 0.001), high serum β₂-µglobulin level (P < 0.001), low scrum albumin level (P = 0.024), low granulocyte count (P = 0.003), and high serum creatinine concentration (P = 0.008).

Rai stage, previous treatment, serum albumin level, serum creatinine concentration, Zubrod performance status, platelet count, absolute granulocyte count, and concomitant steroid use were included in a multivariable logistic regression model for which the dependent variable was development of major infection. Serum β₂-µglobulin was not measured in 188 patients; therefore, we omitted this variable. We also omitted response to therapy because this variable could not be known at the start of the study. Results are summarized in Table 4.

In 158 of the pretreated patients, baseline CD4 counts and counts at the end of treatment cycle were available. All 158 patients had received at least three cycles of fludarabine therapy. One hundred thirty of these patients received fludarabine plus prednisone; the remainder received fludarabine alone. Five (26%) of the 19 patients with CD4 counts less than 50 cells/mL after the third treatment cycle developed zoster infections compared with 9 (6%) of 139 patients with a CD4 count greater than 50 cells/mL (P = 0.01). In addition, 3 of 19 patients (16%) with CD4 counts less than 50 cells/mL and 7 of 139 patients (5%) with CD4 counts greater than 50 cells/mL developed reactivation of mucocutaneous herpes simplex virus infection (P = 0.1). A similar association between low CD4 counts and late-onset listeriosis was previously reported in this patient population [9].

Discussion

This study represents the first attempt to systematically characterize fludarabine-associated infectious complications in persons with advanced chronic lymphocytic leukemia, a large and well-defined patient population. Encapsulated bacteria were previously thought to be the predominant pathogens in patients with this disease [10], but our study revealed the emergence of a new spectrum of pathogens (Listeria monocytogenes, Pneumocystis carinii, cytomegalovirus, herpes simplex virus, varicella zoster virus, and mycobacteria) typically associated with T-cell dysfunction. A similar pattern has been observed by others in case reports and smaller series. Sanders and colleagues [11] reported P. carinii pneumonia in two patients with advanced chronic lymphocytic leukemia receiving monthly fludarabine. One of the two patients also developed disseminated Mycobacterium bovis infection. Bastion and associates [12] reported a recurrent case of P. carinii pneumonia in a patient with refractory chronic lymphocytic leukemia treated with fludarabine. Using continuously infused fludarabine (at a different dosage than we used) in 51 patients with advanced chronic lymphocytic leukemia, Puccio and coworkers [8] reported 10 fatal opportunistic infections, including P. carinii pneumonia and mycobacterial infection, which occurred mostly during the first three cycles of therapy. Bergman and colleagues [13] reported infections by P. carinii and Aspergillus and eight infection-related deaths in 19 of their patients. Three fatal infections occurred in 15 patients in a report by Whelan and coworkers [14]. At the lower single dose of 20 mg/m², Grever and associates reported only one fatal infection in 32 patients, caused by pseudomembranous colitis and sepsis [15]. Twenty-one percent of fludarabine courses were complicated by secondary infections in the study by Hiddemann and colleagues [16].

Many other investigators [17-25] also reported a similar pattern of opportunistic infections. Redman, Hochster, and Dimopoulos and their colleagues described infections in patients receiving fludarabine therapy for low-grade lymphoma and Waldenstrom macroglobulinemia [26-28]. In the study by Redman and colleagues, several episodes of bloodstream infections by bacteria and fungi were reported in addition to P. carinii infection, pulmonary aspergillosis, cytomegalovirus infection, and eight dermatomal zoster infections [26]. A small number of our previously treated patients experienced overwhelming polymicrobial infections in a manner reminiscent of AIDS [5], severe combined immunodeficiency, or allogeneic bone marrow transplantation with severe graft-versus-host disease.

Distinct types of infections occur at different times during the course of chronic lymphocytic leukemia. Infections are not particularly common in patients who were not previously treated or after therapy with alkylating agents. At this stage, infections are likely to be secondary to hypogammaglobulinemia and consist of sinusitis, pneumonia, and occasional bloodstream infections by encapsulated bacteria. Opsonization is required to kill these organisms (Streptococcus pneumoniae, Staphylococcus species, and Haemophilus influenzae). With the introduction of combination cytotoxic therapy (such as cyclophosphamide-containing regimens), neutropenia-related bacterial infections (gram-positive and gram-negative) and infections caused by herpes simplex viruses tend to occur more frequently. A similar pattern is also observed when patients with newly diagnosed chronic lymphocytic leukemia are treated with fludarabine alone. Because of the severe and cumulative immunosuppression resulting from previous cytotoxic therapy and the severe lympholytic activity of both fludarabine and corticosteroids, a new spectrum of infections is encountered in pretreated patients receiving fludarabine-containing regimens, including infections by fungi, viruses, mycobacteria, L. monocytogenes, and P. carinii. In these previously treated patients, immunosuppression (mainly lymphopenia and lymphocyte dysfunction) may persist for more than 1 year after the end of antineoplastic therapy [29]. Occasional infections, including those associated with defects in cell-mediated immunity, may occur, usually among previously treated patients with advanced disease and a low CD4 count.

Although this characterization of infection incidence, type, and timing is useful for clinical purposes, it is important to keep in mind that the pathogenesis of major opportunistic infections in patients with chronic lymphocytic leukemia is complex, multifactorial, and incompletely understood. There is concern that the T-cell-mediated dysregulation of advanced-stage chronic lymphocytic leukemia with abnormal CD4/CD8 ratios may be further aggravated by fludarabine therapy [2, 29-32]. O'Brien and colleagues [29] measured CD4 counts in 217 patients with chronic lymphocytic leukemia who received fludarabine plus prednisone and observed a rapid and serious decrease from a median of 1015 cells/mL at baseline to a low of 169 cells/mL after 3 months of therapy. Of interest, most of these infections occur during this early period. Wijermans and coworkers [30] investigated the influence of fludarabine on several aspects of immunocompetence in 17 patients with advanced chronic lymphocytic leukemia. A profound and rapid decrease in all T-cell subsets occurred in all cases. In most patients, CD4 counts remained depressed 12 months after the end of therapy. Opportunistic infections developed in 7 patients (and were fatal in 2 cases). The spectrum of infections included P. carinii pneumonia and cytomegalovirus, varicella-zoster virus, hepatitis A virus, and invasive fungal infections. Immunoglobulin levels tended to increase in almost all patients during therapy, and no severe cumulative myelotoxicity was observed. Hence, the investigators attributed the observed increase in the opportunistic infections to the adverse effect of fludarabine on the various T-cell subpopulations. A similar rapid reduction of CD4 counts after fludarabine therapy has been associated with opportunistic infections in patients with advanced chronic lymphocytic leukemia [13, 17]. The CD4 counts usually increase markedly after discontinuation of therapy. However, normal values may not return for up to 2 years [29].

Other immunologic consequences of fludarabine therapy also play a role in predisposing patients to opportunistic infections and were the subject of two comprehensive reviews [33, 34]. Of interest is the concomitant decrease in B-cell and monocyte counts after fludarabine therapy. A recent report by Girmenia and associates [35] described a patient who developed listeriosis 2 years after discontinuation of fludarabine therapy despite having a normal CD4 count. Corticosteroids have been clearly associated with such infections through multiple mechanisms, including lympholytic action and suppression of the macrophage-monocyte axis [36]. The use of higher doses of prednisone without fludarabine in comparable patients at our institution did not increase the risk for these unusual infections, especially listeriosis [37]. This finding implies that the potential synergistic effect of prednisone plus fludarabine lowers the threshold for such infections. Hence, glucocorticoids should be avoided in combination with fludarabine unless otherwise indicated (for example, in cases of autoimmune cytopenia).

Multivariable logistic regression analysis (Table 4) identified four baseline variables associated with an increased risk for a major infection: advanced Rai stage (III or IV), previous cytotoxic chemotherapy, serum creatinine concentrations of 1.4 mg/dL or more (≥ 123.76 µmol/L), and absolute granulocyte counts less than 1000 cells/µL. Response to fludarabine is also related to risk for infection; the lowest risk is seen in complete responders [29]. However, because response to therapy would not be known at the start, we omitted this variable from our statistical analysis.

On the basis of our results and data from the literature, we devised a prophylactic strategy for patients with chronic lymphocytic leukemia treated with fludarabine (Figure 1). Simple measures, such as informing patients to avoid food items known to contain L. monocytogenes (unpasteurized milk, some cheeses, raw vegetables, and undercooked poultry or meat) are effective in preventing life-threatening listeriosis in patients at risk. A supply of amoxicillin and clavulanate to be self-administered by patients at the onset of signs and symptoms of infection (such as fever and chills.) may help avoid the excessive morbidity associated with bacterial infections, particularly those caused by encapsulated organisms. A supply of acyclovir to be commenced if mucocutaneous lesions appear may abort or shorten the course of herpetic infections and decrease the severity of postherpetic neuralgia. True antimicrobial prophylaxis should, however, be reserved for patients at highest risk for infection, particularly those with advanced Rai stage, renal dysfunction, an absolute granulocyte count less than 1000 cells/µL, and previous cytotoxic therapy. Persistent (>10 to 14 days) and profound (<100 neutrophils/mL) neutropenia is a known independent risk factor for bacterial and fungal infections. Hence, antifungal prophylaxis should be considered in this setting, particularly if mucositis and fungal colonization are present [38]. Because of the risk for aspergillosis, itraconazole would be the antifungal agent of choice.

Antibacterial prophylaxis is somewhat more controversial. Trimethoprim-sulfamethoxazole is effective against P. carinii pneumonia, listeriosis, and other bacterial infections known to occur in patients with chronic lymphocytic leukemia. Because the drug is also inexpensive and relatively well tolerated, it is suitable for long-term use. Acyclovir is active against herpes viruses and should be considered in patients with positive findings on herpes serology or previous herpetic infections, particularly when CD4 counts decrease to 50 cells/mL or less. The duration of antimicrobial prophylaxis is unclear. However, prophylaxis should not be considered after completion of fludarabine therapy in patients who have achieved a major response to fludarabine, because these patients have a very low risk for major infection [39]. On the other hand, patients with known risk factors for major infection (including those with low CD4 counts or concomitant steroid therapy) should continue to receive prophylaxis as long as these risk factors persist.

Given the high cost associated with the use of intravenous immunoglobulin and colony-stimulating factors, these agents should be used sparsely. Immunoglobulins should be reserved for patients with recurrent infections by encapsulated organisms, and colony-stimulating factors should be used according to the American Society of Clinical Oncology practice guidelines [40]. Thus, colony-stimulating factors should be reserved for patients with known risk factors for major infection [41]. Additional means of preventing opportunistic infections in such patients include restricting the use of corticosteroids (because their addition does not improve response rate) and limiting the length of therapy. Continuous therapy beyond 6 to 9 months, maintenance therapy, and a combination or sequence of purine analogues should be avoided because of cumulative immunosuppression [33].

The workup and management of infection are shown in Figure 2 and are based on a risk-adjusted approach, type of immunosuppression, and site and severity of the presumed infection. Whenever possible, trimethoprim-sulfamethoxazole should be part of any therapeutic regimen because of its activity against various pathogens observed in patients with chronic lymphocytic leukemia who receive fludarabine and because of its favorable pharmacokinetics. This is particularly important in patients with signs and symptoms that suggest pneumonia or meningitis (the latter may be caused by L. monocytogenes) [9].

Despite our positive findings, the study had several limitations. Because of the retrospective nature of the study, a standardized approach to the workup of febrile patients, including systematic and sequential CD4 counts and an aggressive approach to the diagnosis of pulmonary infiltrates (such as bronchoalveolar lavage or biopsy), was not followed. Several patients developed their infectious episode at institutions other than ours. Hence, accurate and complete analysis of their febrile episodes was not possible. Furthermore, we did not take a uniform approach to infection prophylaxis and therapy. This was complicated by the long duration of the study period, which allowed the potential introduction of unknown variables. Differences in key baseline characteristics among the patient groups (Table 1) may have accounted for some of the observed effects of therapy. Furthermore, additional differences in other aspects of treatment among the treatment groups may have been present but went undetected because of the retrospective nature of our study.

In conclusion, our findings should increase the index of suspicion for major infections (including atypical ones) in previously treated patients with chronic lymphocytic leukemia who are receiving fludarabine therapy. Atypical infections are more likely to occur among patients receiving concomitant corticosteroid therapy. Major infections are more common in previously treated patients and patients with advanced Rai stage, renal dysfunction, or an absolute granulocyte count less than 1000 cells/µL at baseline. Every effort should be made to prevent these infections. Whether CD4 counts could serve as a surrogate marker of an increasing risk for infections in this patient population is unclear. Continued follow-up of patients and further study are needed to determine the kinetics and frequency of immune reconstitution in these patients.

From the University of Texas M.D. Anderson Cancer Center, Houston, Texas.

Dr. Kontoyiannis: Infectious Disease Unit, Massachusetts General Hospital and Harvard Medical School, Fruit Street, Boston, MA 02114.

Drs. O'Brien, Kantarjian, and Keating and Ms. Lerner: Leukemia Section, Department of Hematology, University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030.

Dr. Robertson: Savannah Hematology/Oncology Associates PC, 4849 Paulsen Street, Suite 201, Savannah, GA 31405.