INFECTIOUS TOXICITY OF DEXAMETHASONE DURING ALL REMISSION-INDUCTION CHEMOTHERAPY: Report of Two Cases and Literature Review

Abstract

Traditionally, children with acute lymphoblastic leukemia receive prednisone, as part of multiagent remission-induction chemotherapy. Recently, many cooperative groups use dexamethasone instead of prednisone during induction. We describe the infectious toxicities experienced by the first two patients in our institution treated with dexamethasone (10 mg/m²/day for 4 weeks with gradual tapering) during induction according to the dexamethasone arm of BFM 2000 and review the relevant literature that suggests an increased risk of infectious complications with dexamethasone. Only a prospective two-arm ALL dexamethasone study at two dose levels (6 and 10 mg/m²/day) will clarify if indeed the higher dose of dexamethasone during induction is more effective and without unacceptable toxicity.

Keywords:

• acute lymphoblastic leukemia
• corticosteroids
• dexamethasone
• induction chemotherapy
• infections
• prednisone

Glucocorticoids play an important role in the treatment of children with acute lymphoblastic leukemia (ALL) for several decades, since they cause cell-cycle arrest and apoptosis of lymphoblasts [1, 2]. Traditionally, children with ALL receive prednisone (PRD) for 4 consecutive weeks in addition to vincristine and asparaginase, with or without an anthracycline and intrathecal chemotherapy, during remission-induction. Many cooperative groups have shown that substitution of dexamethasone (DXM) for PRD resulted in better CNS protection against leukemia relapse [3–5].

Based upon that, we have recently started to use the DXM arm of the ALL-BFM 2000 study to all newly diagnosed children with ALL in our institution. This two-arm protocol employs 10 mg/m²/day in its DXM arm (the other arm continues to use PRD at 60 mg/m²/day) for 4 consecutive weeks followed by gradual tapering after a 7-day PRD prephase, in contrast to the previous ALL-BFM 95 study, where PRD at 60 mg/m²/day was exclusively used during induction. The first two patients with ALL treated according to BFM 2000 in our institution developed serious infections with Candida tropicalis and Staphylococcus aureus, in addition to developing profound myopathy with inability to walk without assistance. A description of their complications and a review of the literature regarding the glucocorticoids used for remission-induction is the subject of this report.

CASE 1

A 17-year-old girl was admitted to the Pediatric Hematology/Oncology Department of the University Hospital of Heraklion, in the island of Crete, Greece, with superior vena cava syndrome. For the last month, she complained of cough, worsening respiratory distress and 15% unintentional weight loss. A chest X-ray showed widening of the mediastinum and bilateral pleural effusions. A CBC revealed a WBC count of 7,100/mm³ with 11% blasts, hemoglobin 12.6 g/dL, and platelets 179,000/mm³. The serum chemistries were normal except for an elevated LDH to 3335 IU/L. A bone marrow aspiration showed 90% infiltration with lymphoblasts of T immunophenotype. CSF examination showed no blasts. After one intrathecal dose of methotrexate and a 7-day PRD prephase, she had a port-a-cath central venous catheter (CVC) placed. She received induction chemotherapy according to the DXM arm of BFM 2000. The bone marrow examination on the 33rd day of induction showed persistent infiltration with 40% lymphoblasts, while the CBC of that day revealed WBC 700/mm³ with an ANC 400/mm³, hemoglobin 8.2 g/dL and platelets 18,000/mm³. Because of the lack of a suitable donor for allogeneic transplantation, she received multiagent chemotherapy, which included DXM 20 mg/m²/day for 6 consecutive days, vincristine 1.5 mg/m² on days 1 and 6, high-dose methotrexate 5 g/m² on day 1, cyclophosphamide 200 mg/m² on days 2–4, cytarabine 2 g/m² bid on day 5, and high-dose E. coli asparaginase 25,000 IU/m² on days 6 and 11. The patient demonstrated delayed methotrexate clearance for which she received additional hydration and leucovorin rescue. On the 5th day of the multiagent chemotherapy block and while pancytopenic (WBC 300/mm³, ANC 100/mm³, hemoglobin 8.3 g/dL, platelets 8,000/mm³), she developed fever to 38.9°C. Conventional and quantitative blood cultures were obtained from the CVC and two peripheral veins and she was empirically started on intravenous antibiotics with ceftazidime (150 mg/kg/day) and amikacin (15 mg/kg/day). She did not defervesce and 5 days later liposomal amphotericin B was added at 1.5 mg/ kg/day, at a time when she had developed grade 4 mucositis. Eight days later, s.c. filgrastim at 5 μg/kg/day was added, in addition to teicoplanin. Despite discontinuation the next day of both ceftazidime and amikacin and addition of imipenem to the teicoplanin and liposomal amphotericin B combination, she continued with fever in the mid 38°C range. On day +15 after the initiation of the multiagent chemotherapy block, cultures obtained from the CVC grew Candida tropicalis. The same fungus was isolated two days later from cultures obtained from peripheral veins. The CVC was immediately removed and the dose of liposomal amphotericin B increased over 2 days to 5 mg/kg/day. Over the next 2 days, she developed mental status changes and a life-threatening episode of upper gastrointestinal bleeding for which she was intubated and admitted in the ICU. She continued with fever, not-responding to the use of antibiotics and antifungals for a total of 45 days, until she recovered from her bone marrow aplasia, when an ultrasound of the upper abdomen showed several lesions in the liver and spleen consistent with disseminated candidiasis. The patient recovered from the pancytopenia, she was discharged from the ICU and currently remains in complete remission from her leukemia 6 months after diagnosis, while she continues to receive oral fluconazole. Her main current clinical problem is muscle wasting noticed after her ICU admission, which is however slowly improving with physical therapy.

CASE 2

A 3-year-old boy was referred to our department from his pediatrician for pallor and anorexia. On admission, he had organomegaly and scattered petechiae. His initial CBC showed a WBC count of 25,900/mm³ with 65% lymphoblasts, hemoglobin 7.9 g/dL and platelets 42,000/mm³. Morphologic and immunophenotypic examination of the blood and bone marrow were diagnostic of early-pre-B ALL. CSF examination was negative for lymphoblasts. A Hickman-type CVC was placed and after a 7-day PRD prephase, he was started on chemotherapy according to the DXM arm of BFM 2000. On the 12th day of chemotherapy, he was noticed to have a paronychia without pus of the right thumb due to nail-biting. Two days later and while pancytopenic (WBC 300/mm³, hemoglobin 7.1 g/dL, platelets 5,000/mm³), he developed jaundice and fever to 39.2°C for which he was empirically treated with ceftazidime (150 mg/kg/day) and amikacin (15 mg/kg/day). Serum chemistries showed total bilirubin 6.08 mg/dL, 80% conjugated, SGOT 87 IU/L, and SGPT 223 IU/L. Blood cultures from the CVC and a peripheral vein on that day grew comparable numbers (cfu/mL) of a β-lactamase producing strain of S. aureus. Intravenous teicoplanin was added the same day and he defervesced promptly. A cardiac ultrasound the next day was negative for heart valve lesions. A new set of blood cultures 2 days into the teicoplanin therapy were negative. He completed a 14-day course of teicoplanin, in addition to 7 days of ceftazidime and amikacin and recovered completely without complications. At the end of the induction chemotherapy, he had profound muscle weakness with inability to rise from the bed and walk well. Over the ensuing 4 weeks his weakness improved and he is currently in remission from his leukemia and receiving maintenance chemotherapy.

DISCUSSION

The first group to address the issue of the glucocorticoid of choice in the treatment of ALL was Cancer and Leukemia Group B (CALGB) in the 1970's, when it completed a study in 646 children and adolescents with ALL who were randomized to receive either PRD or DXM, as part of their remission-induction chemotherapy [3]. The 493 evaluable patients who achieved remission received the same corticosteroid as pulses throughout maintenance. Although both steroids appeared equally effective in protecting the bone marrow from leukemia relapse, there were 33 CNS relapses among 231 children treated with DXM (14.3%) compared to 67 CNS relapses among 262 patients treated with PRD (25.6%) (p =. 017) [3]. Moreover, the Dutch Childhood Leukemia Study Group-DCLSG in its moderately intensive ALL-6 study for non-high risk patients used DXM at 6 mg/m²/day during induction and maintenance [4]. At 10 years, the EFS for all patients was 81.5 ± 2.8%, the survival rate 84.8 ± 2.7%, and the cumulative incidence of isolated CNS relapse only 1.1 ± 0.8%, one of the lowest reported to date [4]. Finally, the Children's Cancer Group in its 1922 trial conducted between 1991 and 1995 compared DXM and PRD in induction and maintenance for standard-risk patients with ALL in supposedly equatoxic doses (DXM 6 mg/m²/day vs. PRD 40 mg/m²/day), while all patients received high-dose DXM (10 mg/m²/day for 21 days) in delayed intensification [5–7]. Although the day 7 marrow response and the day 28 remission induction rates were comparable, the EFS at 7 years was significantly higher on the DXM compared to the PRD arm (84.2% vs. 76.7%,p =. 002) [5–7].

Although these studies offer adequate justification for wider use of DXM, it remains unclear whether the substitution of DXM for PRD is necessary during remission-induction chemotherapy for all patients with ALL. First, 98–99% of the children with ALL will achieve remission by the end of a 4-week induction multiagent chemotherapy, irrespective of the steroid used [8]. Second, early and extensive use of intrathecal chmotherapy offers adequate prophylaxis against CNS leukemia, such that isolated CNS relapses now occur in < 4% of the patients [9, 10]. Finally, most modern protocols employ DXM during intensification and in conjunction with vincristine during maintenance, which may make the universal use of DXM during remission-induction chemotherapy unnecessary, especially in standard-risk patients with ALL.

DXM has increased half-life compared to PRD, which results in higher steady-state concentrations [1, 2]. DXM binding to the cortisol-binding globulin in plasma is the lowest of the commonly-used glucocorticoids, which allows a higher proportion of the drug to be free and biologically active. Moreover, DXM crosses the blood-brain barrier better than PRD [11, 12].

Until recently, BFM and other cooperative groups, like the Italian AIEOP, used only PRD in induction [13–18]. Lately, based on the results of the studies summarized above, BFM has introduced DXM at 10 mg/m²/day in a randomized two-arm study which compares it with PRD. Nevertheless, it continues to administer a 7-day PRD prephase, since the peripheral blood blast response to the latter has repeatedly been shown to have great prognostic significance [19–22].

The DXM dose chosen for induction in BFM 2000 is 40% higher than the commonly used dose of the drug (6 mg/m²/day) in several other studies, both in Europe (e.g., ALL-6) [23] and the US (e.g., CCG-1922 and DFW-I [24] protocol of the University of Texas Southwestern Medical Center at Dallas). Although 4-week courses of DXM at 10 mg/m²/day have been used during reintensification in ALL-BFM 81 and 83 [25, 26], it is likely that such a regimen has a different toxicity profile when used prior to achievement of hematologic remission. Moreover, the use of DXM during reintensification was limited to 3 weeks during the subsequent ALL-BFM 86, 90, and 95 studies and currently most cooperative groups, e.g., FRALLE [27], CCG [5], EORTC [28], use 3 instead of 4 weeks of DXM during reintensification.

Higher doses of glucocorticoids have increased antileukemic efficacy, as shown by Dana Farber Cancer Institute-DFCI in its 91-01Pilot study, where increasing DXM doses to 18 or 150 mg/m²/day elicited better clearance of the bone marrow and peripheral blood from leukemic lymphoblasts compared to standard doses of DXM (6 mg/m²/day) or PRD (40 mg/m²/day) [29]. However, this very high-dose DXM regimen was administered as an upfront window for 3 days only. Moreover, Hurwitz et al. compared the frequency of septic complications in 91-01Pilot, which included in addition to the upfront window DXM at 6 mg/m²/day for 28 days, with that of leukemic children treated in previous or subsequent DFCI protocols that used PRD [30]. Sixteen of the 38 children (42%) treated on the DFCI 91-01Pilot study had documented gram positive or negative sepsis during remission induction, including 4 toxic deaths (11%). In contrast, there were only 4 induction deaths among 369 children (1%) treated on the previous 87-01 protocol (p =. 0035) and only 1 induction death among 377 children (< 1%) treated on the subsequent 91-01 study (p =. 0003), both of which used PRD at 40 mg/m²/day during induction [30]. The 3-day glucocorticoid window that preceded induction in 91-01Pilot exposed 8 of the 38 randomised patients to the highest dose of DXM (150 mg/m²/day), and 5 of them (63%) developed serious infectious complications. Despite that, there was no statistically significant difference in the incidence of infectious complications among the 4 corticosteroid regimens used in the upfront window, i.e., PRD at 40 mg/m²/day vs. DXM at 6, 18 or 150 mg/m²/day, likely due to the small number of patients in each group [30]. The 4 septic deaths in 91-01Pilot were due to S. aureus, E.coli, Klebsiella (one each), while the responsible pathogen was not identified in the fourth case. The remaining 12 septic non-lethal episodes were due to S. aureus in 3 cases, S. epidermidis in 2, S. pneumoniae in 2, Enterococcus in 1, mucor in 1 and were polymicrobial in the remaining 3 cases [30]. Since other institutions, like UTSW Medical Center at Dallas, use DXM-containing induction similar to that of 91-01Pilot with 6 mg/m²/day without witnessing such toxicity (N. Winick, MD, personal communication), it is likely that the 3-day corticosteroid window in 91-01Pilot contributed to the very high incidence of infectious complications. In addition, in June 2000 the Pediatric Oncology Group-POG amended the induction chemotherapy in its 9905 protocol by reusing PRD at 40 mg/m²/day due to the unacceptable infectious toxicity associated with DXM in the 4-drug induction used in this and other protocols of the group (N. Winick, MD, personal communication).

Although the BFM group has not yet published toxicity and efficacy results for its 2000 study, it is likely that all glucocorticoid-related side effects of induction chemotherapy, such as hypertension, hyperglycemia, myopathy, behavioral disturbances [31], osteoporosis, avascular necrosis of the bones [32, 33], and maybe infections will be increased by substitution of PRD at 60 mg/m²/day with DXM at 10 mg/m²/day. Although it is generally believed that DXM and PRD have equal glucocorticoid activity, when the latter is given at a sevenfold higher dose, this assumption may be incorrect. Kaspers et al. using bone marrow samples from patients with ALL showed marked variation in antileukemic activity of PRD and DXM in vitro [34]. In their study, although leukemic lymphoblasts shared cross-resistance to both agents (correlation coefficient = 0.85, p <. 000001), the individually calculated ratios of LC₅₀ for PRD and DXM showed a wide range from 0.7 to > 500 with a median of 16.2. In other words, the in vitro antileukemic activity of DXM was a median of 16.2-fold higher than that of PRD, which contrasts to the generally assumed dose equivalency of 7 [34]. Moreover, in terms of suppression to the endogenous production of cortisol, DXM is 17-fold more potent than PRD [35]. These in vitro and in vivo data are suggestive of much greater glucocorticoid exposure to even only 6 mg/m²/day of DXM compared to the highest dose of PRD (60 mg/m²/day) used during remission-induction chemotherapy.

We describe the serious infections of the first two patients with ALL who received the DXM arm of BFM 2000 in our institution. Although case 1 did not develop candidemia due to C. tropicalis during the first 4 weeks of induction chemotherapy, but rather 2 weeks later, we believe that exposure to high-doses of DXM for 31 consecutive days (26 days at 10 mg/m²/day and 6 days at 20 mg/m²/day) played a critical role in the development of the bloodstream infection. Other contributing factors were the mucositis that resulted from delayed methotrexate clearance, the extensive use of broad-spectrum antibiotics, the prolonged neutropenia and the protracted ICU hospitalization. Leukemia, prolonged neutropenia [36], broad-spectrum antibiotic usage [37], especially with imipenem, corticosteroid therapy, and total parenteral nutrition [37, 38] have all been implicated in the pathogenesis of Candida bloodstream infections. Regarding case 2, who developed bacteremia due to S. aureus midway into his induction, he recovered without complications. It is interesting that S. aureus bacteremia in patients with leukemia has not been associated with frequent deep-seated complications, such as endocarditis or osteomyelitis [39–41], like in otherwise healthy children, probably due to the prompt initiation of intravenous antibiotics in febrile neutropenic children with ALL. Moreover, platelet aggregation, cytokine release, and phagocyte activation play a central role in the pathogenesis of endocarditis, and corticosteroids downregulate these responses [42]. Despite that, S. aureus bacteremia remains a significant cause of morbidity and mortality in neutropenic patients with cancer. Over a 10-year study period that was recently reported, S. aureus caused 30 of 438 (7%) cases of bacteremia in neutropenic patients with cancer with an attributable mortality of 10% [43]. Moreover, acute leukemia as an underlying malignancy and severe oral mucositis were more frequent among patients with S. aureus bacteremia compared to those with gram-negative bacteremia [43].

After seeing the infectious and musculoskeletal complications described above, we have modified the DXM arm of the BFM 2000 protocol in our institution and we currently use only 6 mg/m²/day of DXM during remission-induction chemotherapy. Five patients have entered this modified protocol without undue toxicity, although in our anecdotal experience DXM at 6 mg/ m²/day may cause more pronounced steroid-induced myopathy and behavioral changes than PRD at 60 mg/m²/day.

In conclusion, our two cases in addition to the available literature suggest an increased risk of infectious complications, when DXM is used in induction. Although, we do not have the power to prove that DXM at 10 mg/m²/day for 4 consecutive weeks is too toxic during induction, from an infectious standpoint there is evidence to suggest that the dose of DXM cannot safely exceed 6 mg/m²/day, when used daily for 4 consecutive weeks as part of multiagent remission-induction chemotherapy. Although the BFM 2000 study is expected to clarify the issue of the most effective corticosteroid of induction at the doses chosen (PRD 60 mg/m²/day vs. DXM 10 mg/m²/day), only a prospective two-arm all DXM study at two dose levels (6 and 10 mg/m²/day) during induction will clarify if indeed the higher dose is more effective and without unacceptable toxicity.