BEFORE THE INCORPORATION of prophylactic CNS

Transcription

1 Improved Survival of Children With Isolated CNS Relapse of Acute Lymphoblastic Leukemia: A Pediatric Oncology Group Study By A. Kim Ritchey, Brad H. Pollock, Stephen J. Lauer, Yvonne Andejeski, and George R. Buchanan Purpose: Isolated meningeal relapse in children with acute lymphoblastic leukemia (ALL) usually has been followed by bone marrow relapse and limited survival. The purpose of this study was to prevent marrow relapse by administering intensive therapy before delayed craniospinal radiation. Patients and Methods: Eighty-three patients with ALL in first bone marrow remission with an isolated CNS relapse were treated with systemic chemotherapy known to enter into the CSF and intrathecal chemotherapy for 6 months. Craniospinal irradiation (24 Gy cranial/15 Gy spinal) was then administered, followed by 1.5 years of maintenance chemotherapy. Results: All 83 patients achieved a second remission. The 4-year event-free survival (EFS) rate was 71.1% 5.3%. There was a fourfold increased risk of relapse for children whose initial remission was less than 18 months. The 4-year EFS rate for patients with a first complete remission H 18 months was 83.3% 5.3%, BEFORE THE INCORPORATION of prophylactic CNS therapy into the treatment of acute lymphoblastic leukemia (ALL) in children, meningeal relapse was the major impediment to cure. CNS relapse usually presaged the inevitable bone marrow relapse. However, with modern treatment, only 5% to 10% of children with ALL relapse in the CNS. Unfortunately, for children who sustain a meningeal relapse, the outlook, until recently, has been poor. In the past, treatment for a child with CNS relapse of ALL consisted of intrathecal chemotherapy followed by cranial or craniospinal irradiation (CSI); systemic therapy has varied. Throughout the 1970s and 1980s, the outcomes reported for patients with an isolated CNS relapse have generally been poor, with subsequent bone marrow relapse the major obstacle to cure. 1-3 In 1990, the Pediatric Oncology Group (POG) initiated a study to address this problem. Our hypothesis was that intensive systemic chemotherapy using agents known to cross into the CSF would improve survival by preventing marrow relapse. Because the ability to deliver intensive chemotherapy can be compromised by poor marrow reserve after CSI, we deferred radiotherapy for 6 months. During the initial 6 months of treatment, an aggressive regimen of intrathecal and systemic chemotherapy was used. Systemic drugs with a history of use in the treatment of CNS leukemia were preferentially used. This report describes the improved outcome of 83 children treated with this approach. and for those with a first complete remission less than 18 months, it was 46.2% 10.2% (P.0002.) There was a low incidence of neurologic toxicity and an unexpectedly high rate of allergic reactions to L-asparaginase. Five patients developed secondary malignancies: two with acute nonlymphoblastic leukemia during therapy, one with myelodysplasia after therapy, and two with brain tumors 1.5 to 2 years after cessation of therapy. Conclusion: For children with ALL and an isolated CNS relapse, treatment that delays definitive craniospinal irradiation by 6 months to allow for more intensive systemic and intrathecal chemotherapy results in better EFS than has been previously reported. Using this approach, the long-term prognosis for children with first complete remission H 18 months is comparable to that at the time of original diagnosis of ALL. J Clin Oncol 17: by American Society of Clinical Oncology. Patients PATIENTS AND METHODS Between April 1990 and June 1993, 84 patients were registered onto the study (POG 9061). One patient was excluded from analysis because of the detection of overt testicular disease after registration but before initiation of induction chemotherapy. The follow-up period was to November 1, All patients had ALL in first bone marrow remission (during or after cessation of primary therapy) with an isolated CNS relapse. Criteria for relapse were CSF WBC count more than 5/µL with blasts present on cytocentrifuge examination using Wright stain. Slides from all but six patients were reviewed by one of the investigators (S.L.). CNS leukemia was confirmed in all cases except two (insufficient cells present on the submitted From the Department of Pediatrics, West Virginia University Health Sciences Center, Morgantown, WV; Pediatric Oncology Group Statistical Offıce and Department of Statistics, University of Florida, Gainesville, FL; Department of Pediatrics, Emory University School of Medicine, Atlanta, GA; Department of Radiation Therapy, Walter Reed Medical Center, Washington, DC; and Department of Pediatrics, University of Texas Southwestern Medical Center at Dallas, Dallas, TX. Submitted March 15, 1999; accepted July 22, Supported in part by the following grants from the National Cancer Institute: CA-03161, CA-05587, CA-07431, CA-11233, CA-15525, CA-15989, CA-20549, CA-25408, CA-28383, CA-28476, CA-29139, CA-29293, CA-29691, CA-30969, CA-32053, CA-33587, CA-33603, CA-33625, CA-53128, CA-69177, and CA Address reprint requests to A. Kim Ritchey, MD (#9061), c/o Pediatric Oncology Group, 645 N Michigan Ave, Suite 910, Chicago, IL by American Society of Clinical Oncology X/99/ Journal of Clinical Oncology, Vol 17, No 12 (December), 1999: pp

2 3746 RITCHEY ET AL slide). All patients met the following additional eligibility criteria: age between 1 and 22 years at the time of relapse, no prior brain irradiation, total prior anthracycline dose less than 375 mg/m 2, and written informed consent according to the guidelines of the respective institutional review boards. Treatment for ALL before relapse varied, although most patients had been enrolled onto front-line POG protocols. One patient was enrolled onto POG 8035, 4 39 onto POG 8602 (AlinC 14), 5 26 onto POG 9005/9006 (AlinC 15), 6,7 or their pilot studies, and 17 were enrolled onto other protocols. The POG front-line ALL protocols consisted primarily of antimetabolite-based therapy with extended triple intrathecal chemotherapy for CNS prophylaxis. Cranial radiation has not been used as initial CNS prophylaxis in any of the aforementioned POG front-line trials. The general design of the recent POG trials for newly diagnosed ALL includes standard three- or four-drug induction, an intensification phase that lasts approximately 6 months during which a randomized question is asked, and maintenance with oral 6-mercaptopurine (MP) and intramuscular methotrexate (MTX) plus intermittent vincristine and prednisone pulses for a total of 2.5 to 3 years. In the POG 8062 intensification phase, the addition of intensive L-asparaginase or cytarabine was added to the standard regimen of intermediate-dose MTX (1 g/m 2 administered intravenously over 24 hours). In POG 9005 (for standardrisk patients), an intensive regimen of intermediate-dose MTX with or without intravenous MP was compared with an intensive oral MTX regimen with intravenous MP. In the protocol for high-risk patients, POG 9006, standard intermediate-dose MTX plus intravenous MP was compared with alternating chemotherapy (MTX and MP; teniposide and cytarabine; daunomycin, cytarabine, vincristine, prednisone, and asparaginase). Treatment The treatment schema for POG 9061 is summarized intable 1. Asymptomatic patients were enrolled onto an initial therapeutic window before the beginning of induction therapy. The objective of this window therapy was to determine the clinical impact and pharmacokinetics of intravenous MP in the CSF. After a bolus MP injection of 200 mg/m 2, the drug was infused on a continuous basis for 6, 8, 10, 12, 16, 21, 28, or 36 hours at a dose of 100 mg/m 2 /hr. Once the accrual objectives of the window had been met, all subsequent patients enrolled onto the study proceeded directly to induction therapy.atotal of 51 patients participated in the window therapy. The specific results of the window investigation have been presented previously and are not discussed further here. 8 Induction of CNS remission consisted of weekly triple intrathecal therapy with MTX, hydrocortisone, and cytarabine in age-adjusted doses, as follows: age 1 to 2 years: MTX 8 mg, hydrocortisone 8 mg, cytarabine 16 mg; age 2 to 3 years: MTX 10 mg, hydrocortisone 10 mg, cytarabine 20 mg; age 3 to 8 years: MTX 12 mg, hydrocortisone 12 mg, cytarabine 24 mg; age 9 years: MTX 15 mg, hydrocortisone 15 mg, cytarabine 30 mg. The final concentration of MTX and hydrocortisone was 1.5 mg/ml, and the final concentration of cytarabine was 3 mg/ml. CNS remission was defined as two subsequent lumbar punctures with no blasts evident on cytocentrifuge analysis of CSF. If remission was not achieved after six intrathecal treatments, the patient s treatment was considered a failure. The systemic (nonintrathecal) induction chemotherapy consisted of dexamethasone 10 mg/m 2 /d administered orally or intravenously in three divided doses for 28 days, vincristine 1.5 mg/m 2 (maximum dose, 2 mg) administered intravenously weekly for 4 weeks, and daunorubicin 25 mg/m 2 administered intravenously weekly for 3 weeks. During the following 6-week consolidation phase, cytarabine 3 g/m 2 as a 3-hour intravenous infusion every 12 hours for 4 doses followed 3 hours later by L-asparaginase 10,000 IU/m 2 was administered intramuscularly on the first and third week. L-Asparaginase 10,000 IU/m 2 was also given intramuscularly on the first and fourth days of weeks 2 and 5 of consolidation. Table 1. Protocol Design: POG 9061 Induction (weeks 1-4) DEX: 10 mg/m 2 orally daily 28 VCR: 1.5 mg/m 2 IV weekly 4 DNR: 25 mg/m 2 weekly 3 TIT: MTX/HC/Ara-C (age-adjusted dose) IT weekly 4 Consolidation (weeks 5-10) Ara-C: 3 g/m 2 IV every 12 hours 4, followed by L-asp* 10,000 IU/m 2 IM weeks 5 and 8 L-asp*: 10,000 IU/m 2 IM on days 1 and 4 of weeks 6 and 9 TIT: MTX/HC/Ara-C (age-adjusted dose) IT week 10 Intensification (weeks 11-22) MTX: 1,000 mg/m 2 IV over 24 hours, followed by MP: 1,000 mg/m 2 IV over 8 hours weeks 11, 14, 17, and 20 VP-16: 300 mg/m 2 IV, followed by CYC: 500 mg/m 2 IV weeks 12, 15, 18, and 21 TIT: MTX/HC/Ara-C (age-adjusted dose) IT weeks 13, 16, 19, and 22 Irradiation (weeks 23-26) Cranial radiation: 2.4 Gy in 1.5-Gy daily fractions Spinal radiation: 1.5 Gy in 1.5-Gy daily fractions DEX: 10 mg/m 2 orally daily 21 VCR: 1.5 mg/m 2 IV weekly 3 L-asp*: 10,000 IU/m 2 IM thrice weekly 9 Maintenance (weeks ) MP: 75 mg/m 2 orally daily 42 MTX: 20 mg/m 2 IM weekly 6 Alternate with VCR: 1.5 mg/m 2 IV weekly 4 CYC: 300 mg/m 2 IV weekly 4 Abbreviations: DEX, dexamethasone; VCR, vincristine; DNR, daunorubicin; TIT, triple intrathecal therapy; HC, hydrocortisone; Ara-C, cytarabine; L-asp, L-asparaginase; VP-16, etoposide; CYC, cyclophosphamide; IV, intravenously; IT, intrathecally; IM, intramuscularly. *L-Asparaginase modifications noted in Methods. Throughout the 12-week intensification phase, chemotherapy was administered weekly. During the first week, MTX 1 g/m 2 was given as an intravenous infusion over 24 hours (with leucovorin rescue) followed immediately by MP 1 g/m 2 by intravenous infusion over 8 hours. The second-week drugs were etoposide 300 mg/m 2 administered intravenously over 1 hour followed by cyclophosphamide 500 mg/m 2 administered as an intravenous push (with mesna). During the third week, three intrathecal drugs in age-adjusted doses were given. This three-week cycle was repeated four times during intensification. The 4-week irradiation phase consisted of simultaneous radiation and chemotherapy. Radiation was delivered to the entire craniospinal axis using megavoltage equipment. The cranium was treated by two equally weighted opposed lateral beams matched to a spinal field treated by a single posterior beam. All fields were treated at 1.5 Gy per daily fraction with the cranium receiving 2.4 Gy and the spine 1.5 Gy total dose. Review of radiation therapy was performed by one of the investigators (Y.A.) and the Quality Assurance Review Center (Providence, RI). Chemotherapy consisted of dexamethasone 10 mg/m 2 /d administered orally or intravenously divided in three doses for 21 days, vincristine 1.5 mg/m 2 (maximum dose, 2 mg) administered intravenously weekly for 3 weeks, and L-asparaginase 10,000 IU/m 2 administered intramuscularly three times per week for 3 weeks. The 76-week maintenance treatment completed the planned 102 weeks of total therapy. MP at 75 mg/m 2 /d orally and MTX 20 mg/m 2 /wk intramuscularly were given for 6 weeks, alternating with 4 weekly doses of intravenous vincristine 1.5 mg/m 2 /wk (maximum dose, 1.5 mg) plus intravenous cyclophosphamide 300 mg/m 2 /wk.

3 ISOLATED CNS RELAPSE OF ALL IN CHILDREN 3747 During the study, there were two changes in protocol therapy. Because of a high incidence of severe local reactions to intramuscular Escherichia coli L-asparaginase, polyethylene glycol (PEG) L-asparaginase was substituted after approximately 20 patients had been enrolled onto the study. During consolidation, intramuscular PEG L-asparaginase 25,000 IU/m 2 replaced the dose of L-asparaginase given after the cytarabine. The two doses of L-asparaginase during the second and fifth weeks were omitted. During the irradiation phase, PEG L-asparaginase 25,000 IU/m 2 was administered intramuscularly in the first and third week instead of the nine doses of thrice-weekly L-asparaginase. The second change to the protocol was the addition of filgrastim after cytarabine during the consolidation phase because of a high frequency of hospital admissions for fever and neutropenia and one death from sepsis. This protocol change was initiated after approximately 40 patients had been enrolled onto the study. This study was performed after approval by local human investigations committees, and informed consent was obtained from each subject or subject s guardian before initiation of treatment. Statistical Methods All statistical analyses were performed using the Statistical Analysis System, version Event-free survival (EFS) curves were generated using the method of Kaplan and Meier. 10 Cox proportional hazards regression was used to calculate the proportional hazards ratio (PHR; an estimate of the relative risk of failure) for each of the putative prognostic factors. 11 We calculated 95% confidence intervals (CIs) for the PHR estimates. RESULTS Patient Characteristics Table 2 lists the characteristics of the 83 patients at both original diagnosis of ALL and relapse. No patients had received prior radiation. Risk classification for patients with B-precursor disease, based on WBC count greater than 50,000/µL and/or age 10 years at diagnosis, was retrospectively assigned using the uniform approach developed by the National Cancer Institute and the major pediatric oncology clinical trials groups. 12 There were 48 patients classified as standard risk and 35 as high risk. Patients with T-lineage ALL (n 3) were classified as high risk. CNS status at diagnosis of ALL was also determined. CNS 1 was defined as WBC count less than 5 cells/µl of CSF and no identifiable blasts on cytocentrifuge examination; CNS 2 was defined as WBC count less than 5 cells/µl of CSF and blasts present on cytocentrifuge examination; and CNS 3 was defined as WBC count 5 cells/µl with blasts present on cytocentrifuge examination. Most patients (75%) were asymptomatic at the time of discovery of their CNS relapse. Of the 21 symptomatic patients, nine had only headache and/or vomiting, whereas the remainder had additional neurologic findings (papilledema, transient sixth nerve palsy, hyperphagia). Outcome All 83 patients achieved a CNS remission after induction therapy. Fifty-five patients continue in their second remission and are more than 3 years off therapy. The Table 2. Patient Characteristics: POG 9061 No. % At original diagnosis Age, years Median 4.1 Range WBC, 10 3 / L Median 15 Range NCI risk classification Standard High Sex Male Female Race White Nonwhite CNS status Primary therapy POG POG 9005/6 26 Other 18 At relapse Age, years Median 6.5 Range CR1 duration 12 months months months months CR1 duration 18 months months CSF WBC, / L Median 61 Mean 444 Range 5-5, / L / L ,000/ L 7 8 1,000/ L 9 11 Symptoms/signs present No 62 Yes 21 Abbreviations: NCI, National Cancer Institute; CR1, first complete remission. 4-year EFS rate ( SE) for the 83 study patients was 71.1% 5.3%, as shown in Fig 1. Figure 2 shows that the EFS was lower for patients with first remission duration less than 18 months (4-year EFS rate, 46.2% 10.2%) than for patients with first remissions 18 months (4-year EFS rate, 83.3% 5.3%; P.0005, log-rank test).

4 3748 RITCHEY ET AL Fig 1. EFS of 83 patients enrolled onto POG The 4-year cumulative EFS rate was 71.1% 5.3%. Twenty-eight patients were removed from the study because of relapse (n 14), toxicity (n 5), secondary malignancy (n 5), bone marrow transplantation in remission (n 3), or because they were lost to follow-up (n 1). All of the 14 patients who experienced relapse were within 29 months of enrollment onto the study (Table 3.) Only two of the 14 relapses occurred before radiation at 6 months, one in a patient with T-lineage ALL who had a brief first remission duration (7 months) and a marrow relapse just before irradiation at 6 months. The other patient was originally diagnosed with B-lineage ALL at age 17 months, had been off therapy for 2.75 years, and had symptoms of Fig 2. EFS for patients by duration of first remission. The 4-year cumulative EFS rate for patients with first remission H 18 months was 83.3% 5.3%, and for patients with a first remission less than 18 months it was 46.2% 10.2%.

5 ISOLATED CNS RELAPSE OF ALL IN CHILDREN 3749 Relapse Site No. Table 3. Relapses on POG 9061 CR1 18 Months (n 26) Months on Study CR1 18 Months (n 57) hypothalamic syndrome for 3 months before documentation of CNS relapse. Her second CNS and initial marrow relapse occurred 1 month before scheduled irradiation. Eleven patients were removed from the study before completion of planned therapy for reasons other than relapse. Three patients received a bone marrow transplant while in remission, one withdrew because of fungal sepsis precluding protocol therapy, another developed ascending paralysis consistent with Guillain-Barré syndrome during induction, and two were lost to follow-up (one refused further therapy and one moved.) Two patients died during remission, one with septic shock during consolidation and one with congenital heart disease (Ebstein s anomaly) and endomyocarditis during maintenance therapy. Two patients developed secondary acute nonlymphocytic leukemia that involved translocation of chromosome 11q23 during maintenance therapy after 9 and 10 months on study. One of these two patients had T-lineage ALL and had previously received considerable exposure to epipodophyllotoxin agents during initial therapy. Three patients had off-study events that occurred after completion of treatment. One developed a secondary myelodysplasia and died during bone marrow transplantation. Two patients died from brain tumors, a grade 3 anaplastic astrocytoma and a glioblastoma, that developed after they were 1.5 and 2 years off therapy, respectively. Participation in the MP window study immediately before induction therapy had no significant impact on EFS (PHR 1.40; 95% CI, 0.61 to 3.20; P.43). The addition of filgrastim to the protocol had no significant impact on EFS (PHR 1.33; 95% CI, 0.58 to 3.04; P.50). Prior frontline treatment (POG v non-pog) also had no significant impact on EFS (PHR 1.25; 95% CI, 0.43 to 3.67; P.68). Predictors of EFS Cox regression analysis was used to estimate the association between EFS and several clinical variables (Table 4). The only significant prognostic factor was the length of the initial remission. Children whose initial remission duration No. Months on Study Bone Marrow 5 6, 11, 12, 24, , 29 7 Testes CNS Combined 2* 7, All sites Total *Both patients with overt testicular relapse with 13% and 10% marrow blasts. Overt CNS relapse with 19% marrow blasts. was less than 18 months had a more than fourfold increased risk of failure than those with a first remission 18 months (PHR 4.32; 95% CI, 1.87 to 10.02). The prognostic significance of different thresholds for first remission duration was invariant over a 12- to 36-month range. No other factors that we examined were significantly associated with EFS. Toxicity Overall, the therapy was well tolerated, except for the toxicity that was anticipated from this intensive CNSdirected treatment program. Of primary concern was the possible development of neurologic toxicity. However, only six patients experienced significant neurologic symptoms during treatment. Three patients had symptoms likely to be therapy-related. One patient developed transient, excessive somnolence during induction and again during irradiation (most likely secondary to dexamethasone), one child had transient hemiparesis and aphasia after high-dose cytarabine during consolidation, and one child developed multifocal partial complex seizures 7 months after completing therapy. Two patients had symptoms thought possibly to be therapyrelated: one child had ascending paralysis soon after beginning induction that was consistent with Guillain-Barré syndrome, and one patient developed a facial nerve palsy after radiation during an episode of herpetic mucositis. Finally, one patient developed CNS depression and seizures during induction, 4 days after the first intrathecal therapy. The latter patient had a history of hyperphagia consistent with hypothalamic syndrome for 2 months before diagnosis and had increased intracranial pressure at diagnosis; her symptoms were believed to be caused by underlying CNS leukemia. There was an unexpectedly high rate of reactions to L-asparaginase. Of the 21 patients who received E coli L-asparaginase, 15 experienced severe local toxicity, including pain, marked swelling, and erythema, whereas one patient had a systemic reaction. On the basis of this toxicity, the protocol was amended to substitute PEG for E coli L-asparaginase, as previously noted. Unfortunately, not only did the reaction rate remain high (63%), but more than half Prognostic Factor Table 4. Prognostic Risk Factors: POG 9061 Risk Category Proportional Hazards Ratio 95% Confidence Interval CR1 duration 18 months WBC at diagnosis 50,000/mm Age at diagnosis 4 years Sex Male Race Nonwhite NCI Risk at diagnosis High risk CSF blasts at diagnosis Present CSF WBC at relapse 100/ L

6 3750 RITCHEY ET AL of the reactions were systemic, including anaphylaxis. However, there were no deaths or apparent long-term toxicity from either form of L-asparaginase. The most common and recurring toxicity was myelosuppression, especially during the consolidation phase. Sixtyone percent of patients were hospitalized at least once during this phase of treatment for fever and neutropenia, and one death occurred from Clostridium sepsis. After the addition of filgrastim to the protocol, there was a decrease in both the frequency of hospitalizations for fever and neutropenia as well as delays in therapy because of myelosuppression. Myelotoxicity was also common during intensification, irradiation, and especially early in maintenance; however, admissions to the hospital for fever and neutropenia were uncommon. Overall, treatment was delivered according to schedule. The average delay before the initiation of irradiation, scheduled at study week 23, was 4 weeks (range, 0 to 16 weeks). Three episodes of Candida sepsis were documented, one each during induction, consolidation, and intensification. One patient was removed from study, another relapsed shortly after Candida sepsis, and the third continued on study and remains in remission. Approximately half of the patients developed transient asymptomatic liver enzyme elevations during the last 2 weeks of induction therapy that was not associated with an increase in serum bilirubin concentration. Treatment was rarely delayed because of this toxicity. DISCUSSION This study of a large group of uniformly diagnosed and treated children with ALL and isolated CNS relapse achieved an excellent 4-year EFS rate of 71.1% 5.3%. The outcome for the children who had an initial remission that lasted more than 18 months was particularly striking, with a 4-year EFS rate of 83.3% 5.3%. These results are superior to those of other comparably sized series of children with isolated CNS leukemia and are even similar to the results of treatment of children with newly diagnosed ALL. Recent reports of initial treatment of children with ALL from the major pediatric oncology clinical trials groups indicate EFS rates of 75% to 85% for standard-risk patients and 65% to 75% for high-risk patients Studies of children with CNS relapse of ALL treated in the 1970s and early 1980s described continuous complete remission rates of 10% to 60%, with most reporting 25% to 50%. 1-3,17-20 CNS disease was reasonably well controlled by neuraxis radiation, with most failures occurring in the bone marrow. In the past 10 years, these children seem to have a better outcome, primarily as a result of more intensive chemotherapy. In two prior POG studies, Land et al 21 reported a 70% 4-year disease-free survival rate for 20 children treated with intensified systemic therapy plus craniospinal radiation, and Winick et al 22 reported a 4-year EFS rate of 46% for 120 children treated with intensified systemic therapy plus cranial radiation and extended intrathecal chemotherapy. From St. Jude Children s Research Hospital, Riberio et al 23 recently reported a 70% 5-year disease-free survival rate for 20 children who received intensive chemotherapy with CSI delayed in most patients for 1 to 4 months, and Kumar et al 24 reported a 56% 3-year EFS rate in 18 patients with CNS relapse treated in a similar fashion. We believe that the excellent results of our study can be attributed to the coordination of intensive intrathecal and systemic chemotherapy having a known CNS effect while delaying CSI to allow for the safe and timely administration of chemotherapy. CSI has been associated with subsequent marrow compromise and difficulty in delivering doseintensive chemotherapy. Prior studies suggested that delaying radiation would not result in an excessive number of early relapses, as long as intensive chemotherapy was administered during the preradiation phase Although there were two relapses before radiation in our study, the 6-month delay in administering radiation seemed effective in allowing for the administration of intensive chemotherapy. The choice of chemotherapeutic drugs given before CSI in this protocol was based on their ability to enter the CSF, prior use in the treatment of CNS leukemia, and their proven efficacy when used systemically. Use of drugs not previously received by most patients was also preferred. Dexamethasone was chosen because of better penetration into the CSF than prednisone. 28 High-dose cytarabine has been shown to have efficacy in the treatment of both CNS and systemic ALL Both intravenous MTX and MP cross into the CSF and have been associated with a low CNS relapse rate when used in during initial therapy of ALL. 32 The combination of etoposide and cyclophosphamide has been used effectively in the treatment of ALL in relapse with acceptable toxicity and is not used in the in treatment of ALL in POG front-line studies. Moreover, etoposide as administered during intensification has recently been shown to achieve cytotoxic levels in the CSF. 33 Because of the anticipated myelosuppression after CSI, the combinations of oral MP and MTX and intravenous vincristine and cyclophosphamide were chosen for maintenance therapy. Both combinations have been used in a previous POG relapse protocol with minimal toxicity. 22 There are other possible explanations for these good results. First, inclusion of patients with reactive pleocytosis in the CSF could favorably bias outcomes, but the diagnostic criteria for CNS leukemia used in the study are considered conservative, and all but eight patients had central review and confirmation of diagnosis. The correct

7 ISOLATED CNS RELAPSE OF ALL IN CHILDREN 3751 diagnosis of CNS relapse in these unconfirmed patients was likely inasmuch as their diagnostic CSF WBC count ranged from 11 to 5,160 cells/µl with blasts identified by local pathologists in each instance. Second, there has been a suggestion in the literature that patients who receive cranial radiation as part of their initial ALL therapy have a worse outcome after CNS relapse. 34 None of the patients on this study received cranial radiation as part of their primary treatment. Interpretation of the results in the literature is complicated by the small numbers of patients treated with a variety of regimens both before and after their CNS relapse. However, in the prospective randomized POG study reported by Land et al, 21 there was no significant difference in outcome between patients who had previously received cranial radiation versus those who had not. Finally, most of the patients in this study were initially treated according to antimetabolite-based POG protocols that include minimal to no exposure to anthracyclines, alkylating agents, and epipodophyllotoxins, all of which were used in our study. Whether children treated with other regimens that include more anthracyclines and alkylating agents during initial therapy would have an equally high rate of EFS on this study protocol is unknown. Previous studies of isolated CNS relapse have identified a number of prognostic factors, including initial remission duration, WBC count at diagnosis of ALL, age at diagnosis of ALL or at relapse, and race. 2,3,20,22 In this study population, however, the only clinical prognostic factor associated with EFS was length of first remission. It is possible that the failure of previously identified prognostic factors to be statistically significant may be a result of small numbers and not a lack of biologic or clinical significance. Alternatively, the effectiveness of intensified systemic therapy may have diminished the prognostic value of these clinical variables. Although patients with meningeal relapse of ALL are at high risk of CNS injury, there was a very low incidence of symptomatic leukoencephalopathy in our study. The degree of CNS injury associated with the treatment of ALL is related to three factors: the neurotoxicity of intrathecal and systemic chemotherapy, the deleterious effect of ionizing radiation, and damage from CNS leukemia itself. A number of studies have shown that CNS-active drugs (eg, MTX and cytarabine) when given in high doses intravenously are more damaging to the CNS when given after CNS irradiation. The low incidence of CNS toxicity noted in this study probably relates to the delivery of intensive therapy before as opposed to after irradiation. Long-term neurotoxicity of treatment is also of concern. Although there has not been an excess of reported late neurotoxicity, two study patients developed brain tumors after completion of therapy. Neurocognitive assessments were not incorporated into the study. In conclusion, isolated CNS relapse of childhood ALL no longer need carry the poor prognosis that was attributed to it in the past. This study provides evidence in a large group of children with isolated CNS disease that CNS-directed irradiation can safely be delayed for at least 6 months while intensive systemic and intrathecal therapy are administered. A high EFS rate was achieved with acceptable toxicity. This treatment approach addresses both the risk of marrow relapse and CNS relapse in a coordinated fashion and offers an excellent chance for long-term survival that parallels that at initial diagnosis of leukemia. The POG plans to build on the results of this study by using the initial remission duration to separate patients into high- and low-risk groups (first remission duration or 18 months.) To address the inferior EFS rate of the high-risk group, patients will receive intensified systemic therapy for 12 months before craniospinal radiation. To attempt to reduce neurotoxicity in the low-risk group, CNS radiation will be reduced to 1.8 Gy delivered only to the whole brain after a year of intensified systemic therapy. ACKNOWLEDGMENT We thank Dr Donald Pinkel, who has supported this study from its inception and provided critical analysis, the POG clinical research associates, and the many patients and their families who joined the study. 1. Willoughby MCN: Treatment of overt meningeal leukemia in children: Results of second MRC meningeal leukemia trial. Br J Med 1: , George SL, Ochs JJ, Mauer AM, Simone JV: The importance of an isolated central nervous system relapse in children with acute lymphoblastic leukemia. J Clin Oncol 3: , Ortega JA, Nesbit ME, Sather HN, et al: Long-term evaluation of a CNS prophylaxis trial: Treatment comparisons and outcome after CNS relapse in childhood ALL: A report from the Children s Cancer Study Group. J Clin Oncol 5: , Pullen J, Boyett J, Shuster JJ, et al: Extended triple intrathecal chemotherapy provides effective central nervous system preventive therapy for patients with B-progenitor acute lymphoblastic leukemia: A Pediatric Oncology Group study. J Clin Oncol 11: , 1993 REFERENCES 5. Harris MB, Shuster JJ, Pullen J, et al: Consolidation therapy with antimetabolite-based therapy in standard-risk acute lymphocytic leukemia of childhood: A Pediatric Oncology Group study. J Clin Oncol 16: , Mahoney DH, Nitschke R, Shuster J, et al: Comparison of intensive methotrexate/mercaptopurine (MTX/MP) vs. low-dose repetitive (LD) MTX/MP for lower risk ALL (LR-ALL): A Pediatric Oncology Group randomized phase III study. Proc Am Soc Clin Oncol 15: 366, 1996 (abstr) 7. Lauer SJ, Tolenado S, Winick N, et al: A comparison of early intensive methotrexate/mercaptopurine (MTX/MP) vs early intensive alternating chemotherapy for high-risk acute lymphoblastic leukemia (HR-ALL): A Pediatric Oncology Group randomized phase III study. Proc Am Soc Clin Oncol 14: 342, 1995 (abstr)

8 3752 RITCHEY ET AL 8. Ritchey AK, Newman EM: Clinical impact and pharmacokinetics of intravenous (IV) 6 -mercaptopurine (MP) in children with isolated central nervous system (CNS) leukemia: A Pediatric Oncology Group study. Proc Am Soc Clin Oncol 13:318, 1994 (abstr) 9. SAS Institute Inc.: SAS/STAT User s Guide, Version 6, Fourth Edition, Volume 2. Cary, NC, SAS Institute Inc., 1989, p Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53: , Cox DR: Regression models and life tables. J Royal Stat Soc 34: , Smith M, Arthur D, Camitta B, et al: Uniform approach to risk classification and treatment assignment for children with acute lymphoblastic leukemia. J Clin Oncol 14:18-24, Reiter A, Schrappe M, Ludwig WD, et al: Chemotherapy in 998 unselected childhood acute lymphoblastic leukemia patients: Results and conclusions of the multicenter trial ALL-BFM 86. Blood 84: , Veerman AJ, Hahlen K, Kamps WA, et al: High cure rate with a moderately intensive treatment regimen in non-high-risk childhood acute lymphoblastic leukemia: Results of protocol ALL VI from the Dutch Childhood Leukemia Study Group. J Clin Oncol 14: , Harris MB, Shuster JJ, Pullen J, et al: Consolidation therapy with antimetabolite-based therapy in standard-risk acute lymphocytic leukemia of childhood: A Pediatric Oncology Group study. J Clin Oncol 16: , Nachman JB, Sather HN, Sensel MG, et al: Augmented postinduction therapy for children with high-risk acute lymphoblastic leukemia and a slow response to initial therapy. N Engl J Med 338: , Wells RJ, Weetman RM, Baehner RL: The impact of isolated central nervous system relapse following initial complete remission in childhood acute lymphocytic leukemia. J Pediatr 97: , Kun LE, Camitta B, Mulhern RK, et al: Treatment of meningeal relapse in childhood acute lymphoblastic leukemia: I. Results of craniospinal irradiation. J Clin Oncol 2: , Pinkerton CR, Chessels JM: Failed central nervous system prophylaxis in children with acute lymphoblastic leukemia: Treatment and outcome. Br J Haematol 57: , Behrendt H, VanLeeuwen EF, Schuwirth C, et al: The significance of an isolated central nervous system relapse, occurring as first relapse in children with acute lymphoblastic leukemia. Cancer 63: , Land VJ, Thomas PRM, Boyett JM, et al: Comparison of maintenance treatment regimens for first central nervous system relapse in children with acute lymphoblastic leukemia. Cancer 56:81-87, Winick NJ, Smith SD, Shuster J, et al: Treatment of CNS relapse in children with acute lymphoblastic leukemia: A Pediatric Oncology Group study. J Clin Oncol 11: , Riberio RC, Rivera GK, Hudson M, et al: An intensive retreatment protocol for children with an isolated CNS relapse of acute lymphoblastic leukemia. J Clin Oncol 13: , Kumar P, Kun LE, Hustu HO, et al: Survival outcome following isolated central nervous system relapse treated with additional chemotherapy and craniospinal irradiation in childhood acute lymphoblastic leukemia. Int J Radiat Oncol Biol Phys 31: , Hustu HO, Aur RJA: Extramedullary leukemia. Clin Haematol 7: , Green DM, Baski S, West CR, et al: The use of subcutaneous cerebrospinal fluid reservoirs for the prevention and treatment of meningeal relapse of acute lymphoblastic leukemia. Am J Pediatr Hematol/Oncol 4: , Mandell LR, Steinherz P, Fuks Z: Delayed central nervous system (CNS) radiation in childhood CNS acute lymphoblastic leukemia. Cancer 66: , Balis FM, Lester CM, Chrousos GP, et al: Differences in cerebrospinal fluid penetration of corticosteroids: Possible relationship to prevention of meningeal leukemia. J Clin Oncol 5: , Wells RJ, Fuesner J, Duney R, et al: Sequential high-dose cytosine arabinoside-asparaginase treatment in advanced childhood leukemia. J Clin Oncol 3: , Frick J, Ritch PS, Hansen RM, Anderson T: Successful treatment of meningeal leukemia using systemic high-dose cytosine arabinoside. J Clin Oncol 2: , Amadori S, Papa G, Avvisati G, et al: Sequential combination of systemic high-dose ara-c and asparaginase for the treatment of central nervous system leukemia and lymphoma. J Clin Oncol 2:98-101, Camitta B, Leventhal B, Lauer S, et al: Intermediate-dose intravenous methotrexate and mercaptopurine therapy for non-t, non-b acute lymphocytic leukemia of childhood: A Pediatric Oncology Group study. J Clin Oncol 7: , Relling MV, Mahmoud HH, Pui C-H, et al: Etoposide achieves potentially cytotoxic concentrations in CSF of children with acute lymphoblastic leukemia. J Clin Oncol 14: , Gelber RD, Sallan SE, Cohen HJ, et al: Central nervous system treatment in childhood acute lymphoblastic leukemia. Cancer 72: , 1993