Frequency and type of serious infections in fludarabine-refractory B-cell chronic lymphocytic leukemia and small lymphocytic lymphoma

Implications for clinical trials in this patient population

Abstract

BACKGROUND

Treatments for fludarabine-refractory chronic lymphocytic leukemia (CLL)/small lymphocytic lymphoma (SLL) are limited. Most new therapies being examined in fludarabine-refractory patients have shown a high frequency of serious infection. Little data exist regarding the frequency of infections in this population treated with noninvestigational best supportive care therapies.

METHODS

The infectious courses of 27 patients with fludarabine-refractory CLL/SLL were retrospectively reviewed. Fludarabine-refractoriness was defined as either relapse within six months of completion of or failure to respond to fludarabine treatment. Infections were documented after patients met National Cancer Institute criteria for further treatment. Serious infections were defined as infections mandating admission to the hospital for intravenous antibiotics.

RESULTS

Patient characteristics included: median age 67 years (range, 40–83), median 3 chemotherapy treatments (range, 1–8), and hypogammaglobulinemia in 73% of patients. Pneumocystis carinii prophylaxis was given to 89% of patients. Serious infections developed in 24 out of 27 patients (89%). Patients had a median of 2 admissions (range, 0–11) for serious infection occurring at a median of 4 months (range, 0–21) from onset of fludarabine-refractoriness. The median frequency of admission for infection was 0.17 per month. Most common sites for infection in decreasing frequency were: respiratory tract, urinary tract, blood/sepsis, and soft tissues. Bacteria caused 69 out of 88 infections (78.4%); viruses (varicella-zoster and herpes simplex) caused 11 out of 88 (12.5%); fungi caused 4 out of 88 (4.5%); and opportunistic infections caused 4 out of 88 (4.5%). Median survival was 13.0 months (range, 1–44+).

CONCLUSIONS

The frequency of serious infections in patients with fludarabine-refractory CLL/SLL is high. These findings are relevant to trials with new and highly effective agents for which the incidence of serious infections after treatment might otherwise appear to be prohibitively high. Cancer 2002;94:2033–9. © 2002 American Cancer Society.

DOI 10.1002/cncr.10437

Chronic lymphocytic leukemia (CLL) is one of the most common of the adult leukemias, affecting between 1.8-3 people per 100,000 population.1 The incidence increases with age and affects approximately twice as many males as females. The clinical course of CLL is unpredictable, with survival from initial diagnosis varying from 1 to greater than 20 years.2 While the majority of patients initially present with limited disease not necessitating immediate therapy, most patients progress to require treatment. Survival for advanced disease is short, with median survival less than three years in patients with poor prognostic factors.3 Infection remains a major cause of morbidity and mortality in patients with CLL/small lymphocytic lymphoma (SLL).

Alkylator-based therapy has recently been challenged, with consideration being given to fludarabine as a first line therapy. This is partly due to three randomized trials showing that fludarabine is superior to alkylator-based therapies.4–8 Fludarabine can provide a long time to disease progression even when it is given in advanced disease, with a median survival of 74 months in responders.9–14 Once patients become fludarabine-refractory, the prognosis is poor. To date, development of treatments for this subset of patients has been limited, though there are many ongoing trials examining the efficacy of new agents.15

It is concerning that many trials have shown a high frequency of serious infection in patients with fludarabine-refractory disease.16–19 This high frequency of infection has threatened the development of otherwise promising and effective new agents. To our knowledge, until now there has been little data on the frequency of infection in fludarabine-refractory CLL treated with ineffective investigational agents or commercially available therapies. It was our purpose to collect such data to gain a better understanding of the baseline frequency and types of infection in patients with CLL/SLL regardless of therapy administered. This data in turn can be used as a baseline for comparison for future trials with new investigational agents which otherwise might have been withdrawn because of an apparently high infection rate.

METHODS

We performed a retrospective review of records of patients with the diagnosis of CLL/SLL (low-grade lymphoproliferative disorders) treated with fludarabine at the Walter Reed Army Medical Center between July 1990 and December 1998 and who had been determined to be fludarabine-refractory. With few exceptions, all patients were treated for their serious infections at Walter Reed Army Medical Center. Patients were determined to be fludarabine-refractory utilizing Revised National Cancer Institute-sponsored Working Group Guidelines,20 i.e. failure to respond after two cycles of fludarabine or relapse/progression of disease within six months of completion of fludarabine. The infectious courses of patients were retrospectively gathered via a review of available records, including inpatient and outpatient charts, computerized lab records (blood counts, chemistries, bone marrow pathology, flow cytometry reports, culture results), computerized radiologic reports, and computerized medication records. Serious infections were identified as those requiring both admission to the hospital and treatment with intravenous antibiotics. Positive cultures were sought but not required as criteria for serious infection. Infections treated with oral antibiotics alone or treated through outpatient services were not included in the current analysis. Immunoglobulins were recorded if documented prior to or within three months of onset of fludarabine-refractoriness. If a patient was found to have several different sites of infection (e.g., urinary infection and pneumonia), or several distinctly different infections requiring intravenous antibiotics (e.g. bacterial peritonitis, candidemia, and disseminated varicella in one patient), then this was recorded as a single admission with multiple infections. Some patients were admitted relatively soon after discharge with a recurrence of their infection. These events were recorded as separate infections because, during clinical trials, such instances of infection in the same anatomic region were counted as separate serious adverse events.

Statistical Considerations

The association of the frequency of infection (infections per month of followup) and age, time from original diagnosis to treatment, initial response to fludarabine, number of cycles of fludarabine, immunoglobulin levels, and t-cell subset population counts was explored using the Spearman rank correlation coefficient. The relationship of infections with steroid usage and history of splenectomy was evaluated using the Wilcoxon rank sum test. Admission rates (admissions per month of followup) before and after fludarabine-refractoriness were compared using the Wilcoxon signed-ranks test. The frequency of viral infections and use of prophylactic acyclovir was explored using the Wilcoxon rank sum test. A two-sided Type I error rate (α) was used for all statistical tests, and no attempt was made to adjust the alpha level for the number of statistical tests within the current study. Thus, some tests noted as significant may owe their outcomes to chance alone, and this fact should be kept in mind when interpreting significant test statistics with P values only slightly smaller than 0.05.

RESULTS

Twenty seven patients were identified as being fludarabine-refractory. Their clinical features are detailed in Table 1. The median age was 67 years, with 21 males and 6 females. The median time from original diagnosis to fludarabine-refractoriness was 53 months. Twenty four of the 27 patients had CLL/SLL and 3 had B-cell prolymphocytic leukemia. These patients were heavily pretreated, receiving a median of two courses of chemotherapy (not including fludarabine). Twenty four of 27 patients (89%) had received chlorambucil or cyclophosphamide either as pulse or continuous cycle therapy, and most patients were alkylator-refractory. Many, but not all, cycles of alkylators were administered with steroids. Patients received a median of 4 cycles (range, 1–21) of fludarabine. One third of patients initially had partial responses to fludarabine, but the remainder had stable or progressive disease. Nineteen of 27 patients (70%) developed serious infections prior to fludarabine-refractoriness. The median frequency of admission for infection prior to fludarabine-refractoriness was 0.03 admissions/month. Hypogammaglobulinemia was noted in 19 of 26 evaluable patients (73%). At the time patients were deemed fludarabine-refractory, the median CD4 count was 478 × 10⁶/mL and the median CD8 level was 786 × 10⁶/mL. Five of 27 patients (18.5%) had undergone splenectomy. Only one patient received prolonged steroids (hydrocortisone) as part of a regimen with ketoconazole for hormone-refractory prostate carcinoma.

Table 1. Patient Characteristics

Characteristic	No. (range, unless otherwise indicated)
CLL: chronic lymphocytic leukemia; SLL: small lymphocytic lymphoma; PLL: prolymphocytic leukemia; AIHA: autoimmune hemolytic anemia; ITP: idiopathic thrombocytopenic purpura; Ig: immunoglobulin; IVIG: intravenous immunoglobulin; PCP: Pneumocystis carinii pneumonia.
Median age	67 years (40–83)
Male:female	21:6
Diagnosis
CLL/SLL	24
PLL	3
Median time diagnosis to fludarabine-refractoriness	53 months (3–260)
Median pretreatments (not including fludarabine)	2 (0–7)
0 regimens (%)	3/27 (11)
1 regimen (%)	6/27 (22)
2 regimens (%)	7/27 (26)
≥ 3 regimens (%)	11/27 (41)
Median fludarabine cycles	4 (1–21)
Original response to fludarabine
Complete response (%)	0/27 (0)
Partial response (%)	9/27 (33)
Stable disease (%)	6/27 (22)
Progressive disease (%)	12/27 (45)
Serious infections prior to fludarabine-refractoriness
Frequency	0.03 admissions/month followup
Bacterial infections (%)	49/53 (92)
Viral infections (%)	3/53 (6)
Fungal infection (%)	1/53 (2)
Opportunistic infections (%)	0/53 (0)
AIHA	8/27
ITP	1/27
Hypogammaglobulinemia at refractoriness (%)	19/26 (73)
IgG mg/dL median	613 (145–1980)
IgA mg/dL median	51 (< 6.7–219)
IgM mg/dL median	69 (< 5–3600)
CD4/CD8 counts (13 evaluated patients) at refractoriness
CD4/cu mm median	478 (90–2424)
CD8/cu mm median	786 (222–2929)
Splenectomy (%)	5/27 (18.5)
Chronic steroid usage (%)	1/27 (4)
Median chemotherapeutic regimens administered after fludarabine-refractoriness	1 (0–7)
IVIG	12/27 (44%)
Prophylaxis for hypogammaglobulinemia (%)	5/27 (18.5)
Treatment during serious infection (%)	3/27 (11)
Treatment autoimmune disease (3 AIHA, 1 ITP) (%)	4/27 (15)
PCP prophylaxis (%)	24/27 (89)

Twelve of 27 patients (44%) received intravenous immunoglobulin treatments, 5 as prophylaxis against infections, 3 as adjunctive therapy during serious infections, and 4 for the treatment of autoimmune cytopenias. Thirteen of 27 (48%) received colony-stimulating factors during followup either as prophylactic treatment of neutropenia or as treatment during serious infection. Twenty four of 27 (89%) received Pneumocystis carinii prophylaxis during all or part of the followup period. Ten patients received acyclovir prophylaxis against viral infections during the followup period, and four more participated in an ongoing protocol comparing acyclovir versus placebo in the prevention of viral infection. This trial is blinded, so as of publication, information regarding therapy was not available.

After onset of fludarabine-refractoriness, patients continued to receive treatments in the form of radiation therapy, single agent chemotherapy, and combination chemotherapy regimens. Survival is depicted in Figure 1. The median number of treatment regimens was two (range, 0–7). A variety of single and combination chemotherapeutic regimens were used as salvage in fludarabine-refractory patients. Sixty one salvage therapies were administered (39 single agent and 22 combination) with a combined response rate (complete response + partial response) of 7 out of 61 (11.4%). Immunotherapy with rituximab composed three of the seven total responses (eight cycles of rituximab administered), consistent with other recently published data in low-grade lymphoma patients.21

Figure 1. Survival in fludarabine-refractory patients.

Infections

Twenty four of 27 fludarabine-refractory CLL/SLL patients (89%) developed infections serious enough to warrant hospitalization for intravenous antibiotics. The demographics of these infections are shown in Table 2. Infections treated as an outpatient or with oral antibiotics alone were noted much more frequently than inpatient admissions, though these were not included in the current analysis. There was a median of 2 admissions per patient (range, 0–11), with the first serious infection occurring at a median time of 4 months (range, 0–21) from the onset of fludarabine-refractoriness, as shown in Figure 2. The median frequency of admission for infection after onset of fludarabine-refractoriness was 0.17 admissions/month. The median length of stay per admission was seven days (range, 1–71 days), and the median time between admissions was 42 days (range, 1–547 days). Eighty eight distinct infections developed in patients in the current series after the onset of fludarabine-refractoriness, summarized in Table 2. The majority of these infections (78.5%) were associated with bacterial organisms. Pneumonia, febrile neutropenia, and urinary infections represented the most frequent causes of bacterial infection, though other typical (sepsis, line infections, soft-tissue infection) and atypical (endophthalmitis, ascending cholangitis) bacterial infections developed as well. Viral infections (11 out of 88 infections; 12.5%) developed in 8 patients. Six admissions for intravenous acyclovir therapy were due to herpes simplex or varicella infections localized to a dermatome or oral-genital region. However, five admissions were for severe viral infections with dissemination or major organ involvement. Fungal (4 out of 88; 4.5%) and opportunistic (4 out of 88; 4.5%) infections only accounted for a small minority of infections. During the followup period, 23 of 27 patients died, 11 out of 23 (48%) clearly due to infectious causes and 12 out of 23 (52%) due to disease progression.

Table 2. Infectious Demographics

Characteristic	No. of patients
Patients with serious infection (%)	24/27 (89%)
Median time to first infection	4 months (range, 0–21)
Median no. of admissions	2 (range, 0–11)
Median no. of infections	2 (range, 0–12)
Frequency of admissions/patient	0.17 admissions/month
Median length of admission	7 days (range, 1–71)
Median sum of hospital days	26 days (range, 0–80)
Median time between admissions	42 days (range, 1–547)
Median survival	13 months (range, 1–44+)

Type of infection	Source	No. (percentage)
Bacterial	69 (78.4%)
Pneumonia	25 (28.4%)
Febrile neutropenia	17 (19.3%)
Urosepsis/urinary infection	9 (10.2%)
Sepsis	7 (8.0%)
Line infection	3 (3.4%)
Cellulitis	2 (2.3%)
Abscess
Submandibular	1 (3.4%)
Peritonsilar	1 (1.1%)
Ascending cholangitis	1 (1.1%)
Lymphadenitis	1 (1.1%)
Endophthalmitis	1 (1.1%)
Fever of unknown origin	1 (1.1%)
Viral	11 (12.5%)
Varicella-zoster virus
Localized	4 (4.5%)
Disseminated	3 (3.4%)
Encephalitis	1 (1.1%)
Herpes simplex virus
Localized	2 (2.3%)
Pneumonitis	1 (1.1%)
Fungal	4 (4.5%)
Hematogenous candidiasis	2 (2.3%)
Pulmonary cryptococcosis	1 (1.1%)
Torulopsis glabrata pneumonia	1 (1.1%)
Opportunistic	Pneumocystis carinii	4 (4.5%)

Figure 2. Percentage of patients admitted for serious infection over time.

As noted previously, a number of patients received acyclovir as prophylaxis. Six of those 27 patients were given prophylaxis prior to any serious viral infection. None of these six patients developed serious viral infections. In contrast, 17 of 27 patients were not given prophylaxis and 8 of these 17 developed severe viral infections. There was a significant association between the viral infection frequency (viral infections per month of followup) and treatment with acyclovir (P = 0.048), suggesting a protective benefit of this therapy. Of the eight patients who developed serious viral infections, four were subsequently placed on acyclovir prophylaxis, and only one of these patients went on to develop another serious viral infection. Two of the four patients without acyclovir prophylaxis subsequently died of recurrent severe viral infection.

DISCUSSION

Infections in patients with fludarabine-refractory CLL have the potential to threaten the development of promising new therapies due to misattribution of infectious toxicity. An example of this can be seen with Campath-1H.22 Campath-1H is a highly effective therapy, though serious concerns arose during clinical trials because of the high prevalence of serious infections in fludarabine-refractory patients. The current data show that even in the absence of investigational therapy, the frequency of infections serious enough to require hospitalization for intravenous antibiotics in fludarabine-refractory CLL/SLL patients is extremely high (median 0.17 admissions/month). These data hopefully will improve our understanding of infection in these patients and therefore prevent early abandonment of highly effective agents due to a perceived high frequency of infection. It may be more appropriate to reserve critical evaluation of infectious toxicity until the novel agent can be compared to standard-of-care approaches in Phase III study design.

A number of factors help to explain the high risk of infections noted in patients in the current series. Fludarabine itself is well known to have its own infectious complications. In patients receiving fludarabine, advanced Rai stage and previous chemotherapy are known major independent risk factors for infection.23 Hypogammaglobulinemia is recognized as a risk factor for infection.24 Our patients shared many of these risk factors: advanced Rai stage in 100%, chemotherapy prior to fludarabine in 89%, and hypogammaglobulinemia in 73%. The high frequency of infection may be partially due to the subsequent immunosuppressive chemotherapies. Unfortunately, the majority of subsequent chemotherapeutic regimens resulted in no response from the malignancy.

The frequency of 0.17 admissions/month for infection in our fludarabine-refractory patients is impressive because it was so much higher than the frequency (0.03 admissions/month for infection) seen in the same group of patients prior to their fludarabine-refractoriness (P < 0.0005). Infections occurred in almost all patients (89%) over the course of the illness, highlighting the inevitability of this complication. The morbidity was significant, with a median of 26 days in the hospital due to infection per patient. Even more telling was that the median survival was only 13 months, and 48% of this mortality was a direct consequence of infection.

As there were only a small number of patients, the current study had limited power to detect statistically significant associations between patient characteristics and the frequency of infections. There was a significant relationship between the frequency of serious infection and the immunoglobulin (Ig) M level (r = 0.48, P = 0.017). Neither age, time from original diagnosis, initial response to fludarabine, number of cycles of fludarabine, IgG level, IgA level, CD4 counts, CD8 counts, steroid usage, or splenectomy were significantly associated with an increased frequency of infection.

The majority (78.5%) of infections in our patient population were associated with bacteria. As with previous studies examining infection in CLL/SLL, the respiratory tract25 and the urinary tract26, 27 were the most susceptible sites. Neutropenic fevers were also frequent in our patients, though this is a reflection of the severity of the underlying CLL as well as the intensity of subsequent conventional chemotherapies. We acknowledge that the majority of infections were treated empirically and cultures were not “proven.” The assumption that infections such as pneumonia and febrile neutropenia were bacterial is based on previously established epidemiology for such infections and may therefore have underestimated the incidence of fungal, viral, or opportunistic infections in this group of patients.

Many of our patients with pneumonia received antibiotics that were active against both typical and atypical organisms because treatment was empiric without knowledge of specific organism. Therapy in 8 of 25 admissions for pneumonia included trimethoprim-sulfamethoxazole (oral or intravenous), though Pneumocystis carinii was not clearly identified as an etiologic agent. The difference between pulmonary infection, lymphocytic progression, or pulmonary chemotherapeutic toxicity even in retrospect was at times difficult to differentiate. For instance, one patient had pulmonary infiltrates on his last two admissions in conjunction with fevers, elevated neutrophil count, and a productive cough. This patient ultimately died of pulmonary lymphocytic progression, though it is difficult to discern whether this patient had an infection contributing to the severity of his illness.

Viral infections were the next most common cause of serious infection in our patients, which is similar to previously published literature.28 Only admissions for intravenous acyclovir were recorded, accounting for the high percentage of disseminated viral infections. It is notable that these severe infections were outnumbered by less severe/localized viral infections that were treated by outpatient therapy with oral acyclovir alone. A prospective analysis by our group has previously noted this.29 Serious viral infections may have represented an even higher proportion of admissions had not a number of our patients been receiving acyclovir prophylaxis. As noted above, none of the 6 patients taking acyclovir prophylaxis developed serious viral infections, whereas 8 of 17 patients (47%) not receiving anti-viral prophylaxis subsequently developed serious viral infections. Because this was not a prospective randomized trial, one must be cautious in interpreting these findings. However, there does seem to be a marked difference between the group who received acyclovir and the group that did not receive prophylaxis. Opportunistic and fungal infections represented only a small fraction of the number of admissions for serious infections in our patients. While the literature cites a high risk of opportunistic infections after fludarabine,30–31 only 2 of our 27 patients developed a Pneumocystis carinii pneumonia infection. It is likely that we did not see more opportunistic infections because the great majority (89%) of our patients received prophylaxis against Pneumocystis carinii. The typical regimen of prophylaxis included double strength trimethoprim-sulfamethoxazole, 1 tablet twice daily three times weekly or monthly inhaled pentamidine. Based on our prospective experience and this data, we have started giving viral prophylaxis using acyclovir 400–800 mg orally three times a day. It is also possible that some of these opportunistic infections may have been successfully treated with oral antibiotics, though these were not included in the current analysis.

CONCLUSIONS

The current data show that the frequency of infections serious enough to require hospitalization for intravenous antibiotics in patients with fludarabine-refractory CLL/SLL is extremely high. These infections occurred frequently even in the setting of conventional chemotherapy. As a result of this propensity for infection, caution should be exercised in interpreting the frequency of infections seen during development of promising new treatment modalities. Evaluation of infectious toxicity should be reserved for Phase III trials which allow physicians to use a treatment of choice if patients are not randomized to the experimental agent. This could be especially important for trials of otherwise highly effective agents in which the incidence of serious infections after treatment might otherwise appear to be prohibitively high.