Clinical significance of central nervous system involvement at ... - Nature

Leukemia (2003) 17, 2090–2096 & 2003 Nature Publishing Group All rights reserved 0887-6924/03 $25.00 www.nature.com/leu

Clinical significance of central nervous system involvement at diagnosis of pediatric acute myeloid leukemia: a single institution’s experience BL Abbott1, JE Rubnitz1, X Tong1, DK Srivastava1, C-H Pui1, RC Ribeiro1 and BI Razzouk1 1

St Jude Children’s Research Hospital, University of Tennessee, Memphis, TN, USA

To determine the clinical significance of central nervous system (CNS) involvement at the time of diagnosis of pediatric acute myeloid leukemia (AML), we analyzed clinical features and outcomes of 290 patients treated consecutively on four institutional trials (AML80, AML83, AML87, and AML91). CNS status was classified as CNS1 (no blast cells in CSF; n ¼ 205), CNS2 (o5 WBC/ll CSF with blast cells; n ¼ 37), or CNS3 (X5 WBC/ll CSF with blast cells, or signs of CNS involvement; n ¼ 48). Patients with CNS3 status were significantly younger than others (P ¼ 0.016) and significantly more likely to have the favorable cytogenetic features t(9;11), t(8;21), or inv(16) (Po0.001). The CNS3 group had a significantly greater probability (7s.e.) of 5-year event-free survival (43.777.0%) than did the CNS1 (27.873.2%, P ¼ 0.015) and CNS2 (24.377.5%, P ¼ 0.032) groups. However, after adjustment for favorable genetic features, there was no significant difference in EFS between the CNS3 and the combined CNS1 þ CNS2 groups (P ¼ 0.075). In all, 10 of 151 patients treated on AML80 and AML83, but none of 139 treated on AML87 and AML91, had primary CNS relapse. CNS involvement had no adverse prognostic significance, and patients with CNS2 status had similar outcome to CNS1 patients in this large group of pediatric patients with AML, treated at a single institution. Leukemia (2003) 17, 2090–2096. doi:10.1038/sj.leu.2403131 Published online 18 September 2003 Keywords: AML; pediatric; CNS; radiotherapy

Introduction Central nervous system (CNS) involvement at the time of diagnosis has traditionally been considered as an adverse prognostic feature in acute myeloid leukemia (AML), and is usually treated with intensified CNS-directed therapy (radiation and/or frequent intrathecal chemotherapy).1–5 However, most of the available data on CNS involvement and its treatment in pediatric acute leukemia is derived from studies of acute lymphoblastic leukemia (ALL). Further, there is little information about the appropriate threshold (white blood cell (WBC) count in cytocentrifuged CSF) needed to diagnose CNS involvement in AML. The threshold used by an Italian research group,4 The Pediatric Oncology Group (POG),6 and St Jude7 is the presence of any blast cells (regardless of cell count). The Children’s Cancer Group (CCG)8 and the Berlin–Frankfurt–Mu¨nster (BFM) group3 use a threshold of 5 and 10 WBC/ml, respectively. There is growing evidence that children with ALL require more intensive CNS-directed therapy if any lymphoblasts at all are present in the Correspondence: BI Razzouk, Department of Hematology–Oncology, St Jude Children’s Research Hospital, 332 N Lauderdale Avenue, Memphis, TN 38105-2794, USA; Fax: þ 1 901 521 9005 This study was supported by Grants CA 21765 and CA 20180 from the National Cancer Institute, by a Center of Excellence Grant from the State of Tennessee, and by the American Lebanese Syrian Associated Charities (ALSAC). C-H Pui is the American Cancer Society-FM Kirby Clinical Research Professor. This work was Presented in part at the 43rd annual meeting of the American Society of Hematology, December 2001, Orlando, FL, USA. Received 16 June 2003; accepted 31 July 2003; Published online 18 September 2003

CSF at diagnosis9,10 (Nachman et al, Med Pediatr Oncol 2002; 39: 277; abstract). Two previous studies have addressed the prognostic significance of this finding in pediatric AML8 (Odom et al, ASCO Proc 1989; 8: 218; abstract). In one study, five of six children who had blast cells in the CSF at diagnosis but who had a low CSF leukocyte count experienced CNS relapse (Odom et al, ASCO Proc 1989; 8: 218; abstract). However, in a CCG study of 142 patients with newly diagnosed AML, 13 patients who had low CSF leukocyte count with blasts at diagnosis had outcomes similar to those of patients with no CSF blasts.8 The treatment of CNS disease in pediatric AML has been reviewed elsewhere.5 In recent years, most pediatric oncology groups have relied increasingly on intrathecal (IT) chemotherapy for the prevention and treatment of CNS leukemia and have reduced the use of CNS irradiation.6 In addition, new antileukemic therapies (eg, high-dose cytarabine11 and cladribine12) offer increased CNS penetration. We analyzed the clinical features and outcome of pediatric patients with and without CNS involvement in a cohort of 290 patients treated on four institutional AML protocols. This analysis expands on that of a 1985 report7 by including patients with CNS disease who have received contemporary chemotherapy and limited radiotherapy.

Patients and methods

Patients Between 1980 and 1997, 290 patients younger than 21 years with previously untreated, newly diagnosed AML were consecutively enrolled on four successive clinical trials (AML80,13 AML83,14 AML87,15 and AML9116) at St Jude Children’s Research Hospital. The diagnosis of AML was based on the finding that leukemic myeloid blast cells composed more than 30% of the mononuclear cells in the bone marrow. After cytochemical and flow cytometric analyses, blast cells were classified according to the French–American–British (FAB) criteria17 and were analyzed cytogenetically;18 additionally, 172 patient samples were analyzed for MLL gene rearrangement. Aspirated bone marrow samples were examined approximately every 3 months during the treatment. Diagnosis of CNS leukemia required the identification of leukemic blast cells in Wright-stained cytocentrifuged samples of CSF by at least two observers, signs of cranial nerve palsy or meningeal involvement, or the detection of a nonhemorrhagic CNS mass or chloroma by computed tomography. On the AML80, AML83, and AML87 treatment protocols, the detection of any number of leukemic cells in the CSF was considered diagnostic of CNS involvement. On AML91, the detection of five or more WBCs per microliter of CSF plus the presence of morphologically identifiable blast cells was considered diagnostic of CNS involvement. For the purpose of our study, all patients were reclassified according to the scheme used in pediatric ALL: CNS1 (no blast cells in cytocentrifuged CSF at the time of diagnosis), CNS2 (1–4 WBC/ml of CSF with

CNS involvement in pediatric AML BL Abbott et al

2091 morphologically identifiable blast cells), or CNS3 (5 or more WBC/ml with blast cells). Patients who met other criteria for CNS involvement (cranial nerve palsy, signs of meningeal disease, or a nonhemorrhagic CNS mass or chloroma) were classified as CNS3, regardless of CSF findings. If the CSF was contaminated by blood (more than 10 erythrocytes per microliter) containing leukemic cells at the time of diagnosis, the lumbar puncture (LP) was considered traumatic, and patients were classified on the basis of a repeat CSF sample obtained several days later. The Institutional Review Board of St Jude Children’s Research Hospital approved this study, and informed consent was obtained from patients, parents, or legal guardians, as appropriate.

Treatment The AML80,13 AML83,14 AML87,15 and AML9116 protocols have been described elsewhere. Systemic therapy on the four protocols is summarized as follows. The remission induction regimens are outlined in Table 1. On AML80, induction therapy consisted of daunorubicin/cytarabine (Ara-C), plus etoposide (VP-16)/azacytidine for refractory disease. Patients in remission were eligible for allogeneic bone marrow transplantation if they had an HLA-compatible donor. Patients who did not undergo transplantation received sequential intensive postremission therapy for approximately 12 months.13 On AML83, induction therapy consisted of etoposide/cytarabine followed by daunorubicin/Ara-C/thioguanine and finally VP-16/azacytidine. Postremission therapy on AML-83 comprised 16 6-week treatment cycles, each of which contained three paired drug combinations.14 On AML87, induction therapy consisted of targeted dosages of etoposide and cytarabine followed by Ara-C/ daunorubicin, VP-16/amsacrine, and VP-16/azacytidine. Consolidation therapy comprised three additional cycles of combination chemotherapy; no maintenance therapy was given.15 On AML91, induction therapy consisted of one or two courses of 2-chlorodeoxyadenosine (2-CDA, cladribine) followed by one to two courses of daunomycin/Ara-C/VP-16. Postremission therapy consisted of bone marrow transplantation or one course of consolidation with high-dose cytarabine.16 Table 2 lists the cumulative doses of Ara-C, VP-16, and daunomycin given during remission induction therapy on the four treatment protocols.

Figure 1 shows a schema of CNS-directed therapy on the four protocols. On the AML80, AML83, and AML87 protocols, cranial radiation (24 Gy) was administered to patients with CNS involvement (CNS2 and CNS3) on completion of maintenance therapy (at approximately 15, 27, and 8 months after diagnosis, respectively). On the AML91 protocol, radiation therapy (6–24 Gy of cranial radiation) was reserved for patients with symptomatic CNS involvement. The wide range of radiation doses reflects the fact that cranial radiation was administered only for control of CNS symptoms. Total body irradiation (TBI; 12–14 Gy) was given as part of the preparative regimen for hematopoietic stem cell transplantation. All patients with CNS disease received IT chemotherapy weekly (for a minimum of 4 weeks) until blast cells were cleared from the CSF. IT chemotherapy was then given monthly until the completion of systemic chemotherapy. On the AML80 and AML83 protocols, patients received five doses of IT chemotherapy during the 2.5 weeks of cranial radiation. IT chemotherapy for patients with CNS involvement comprised up to 23 doses of methotrexate (MTX) on AML80, up to 35 doses of MTX on AML83, up to 11 doses of MTX, hydrocortisone, and cytarabine (MHA) on AML 87, and up to nine doses of MHA on AML91.

Statistical methods The duration of survival was defined as the interval between diagnosis and the time of the last follow-up or of death from any cause. An adverse event was defined as relapse, evidence of disease progression, or death from any cause. The duration of event-free survival (EFS) was defined as the interval between diagnosis and the first adverse event. Patients who did not enter complete remission (CR) were assigned an EFS value of zero. Data were censored at the time of the last follow-up if no adverse event was observed. The probability of survival was estimated by the Kaplan–Meier method and was compared among subsets of patients by the log-rank test. Since the EFS estimates were similar in the CNS1 and CNS2 groups, we combined these patients into one group (‘CNS1 þ 2’) and performed further analysis to identify factors that affect EFS. The factors affecting EFS within each CNS subtype were assessed by the log-rank test. Stratified analysis was performed

Table 1

Initial remission induction therapy in four AML study protocols

Study

Window therapy

Remission induction therapy 2

AML80

None

(1) Daunorubicin 45 mg/m i.v. 3 days, Ara-C 200 mg/m2 s.c. 7 days (2) Daunorubicin 45 mg/m2 i.v. 2 days, Ara-C 100 mg/m2 s.c. 5 days (3) VP-16 250 mg/m2 i.v. 6 days, 5-azacytidine 300 mg/m2 i.v. 4 days

AML83

None

(1) VP-16 200 mg/m2 i.v. 3 days, Ara-C 200 mg/m2 s.c. 4 days (2) Daunorubicin 50 mg/m2 i.v. 2 days, 6-TG 100 mg/m2 p.o. b.i.d. 5 days, Ara-C 200 mg/m2 s.c. 5 days (3) VP-16 250 mg/m2 i.v. 5 days, 5-azacytidine 300 mg/m2 i.v. 2 days

AML87

None

(1) (2) (3) (4) (5) (6)

AML91

2-CDA 8.9 mg/m2 i.v. 5 days (1–2 courses)

(1) Daunorubicin 30 mg/m2 i.v. 3 days, Ara-C 250 mg/m2 i.v. 5 days, VP-16 200 mg/m2 i.v. 2 days

Ara-C 500 mg/m2 s.c. 4 days, VP-16 500 mg/m2 i.v. 4 days Ara-C 500 mg/m2 s.c. 5 days,daunorubicin 50 mg/m2 i.v. 2 days VP-16 500 mg/m2 i.v. 3 days, m-AMSA 125 mg/m2 i.v. 3 days VP-16 500 mg/m2 i.v. 5 days, 5-azacytidine 300 mg/m2 i.v. 2 days Ara-C 500 mg/m2 s.c. 5 days, daunorubicin 50 mg/m2 i.v. 2 days VP-16 500 mg/m2 i.v. 4 days, Ara-C 500 mg/m2i.v. 4 days

(2) Daunorubicin 30 mg/m2 i.v. 3 days, Ara-C 250 mg/m2 i.v. 5 days, VP-16 200 mg/m2 i.v. 2 days Ara-C: cytarabine; 2-CDA: 2-chlorodoeoxyadenosine; VP-16: etoposide; m-AMSA: m-amsacarine; s.c.: subcutaneous; i.v.: intravenous; p.o.: oral. Leukemia


2092 Table 2 Cumulative doses of Ara-C and VP-16 given during remission induction therapy by treatment protocol Protocol

Ara-C (mg/m2)

VP-16 (mg/m2)

Daunomycin (mg/m2)

AML-80 AML-83 AML-87 AML-91a

2400 1800 9000 2500

1500 1850 8000 800

225 100 200 180

a

One to two courses of 2-chlorodeoxyadenosine (2-CDA) were given in the ‘up-front window’ before remission induction therapy.

who had a traumatic first and second LP and had blast cells in the CSF were classified as CNS3. Patients with CNS3 status were significantly younger at the time of diagnosis than patients with CNS1 or CNS2 status (P ¼ 0.016). The 85 patients with CNS involvement had CSF leukocyte counts ranging from 0 to 420/ml (median 13/ml). CNS involvement was diagnosed on the basis of demonstrable blast cells in the CSF in 82 patients, an intracerebral mass in one, spinal chloroma in one, and facial nerve palsy in one. A total of 20 patients had CNS signs and symptoms at presentation: chloromas were identified in 14 patients, cranial nerve palsy in six, and meningeal signs in one (four patients had more than one sign or symptom). Race and sex did not differ significantly among the three groups. As shown in Table 3, the incidence of CNS2 and CNS 3 status was relatively high in patients with AML FAB M4 and M5 subtypes, as reported previously.5,8 Furthermore, the incidence of CNS2 status was higher in patients treated on the AML91 protocol. This finding may reflect increased scrutiny of CSF samples in this patient group during a study of CNS2 status in ALL at our institution.9

Cytogenetics and MLL rearrangement As shown in Table 3, the favorable cytogenetic features inv(16), t(9;11), and t(8;21) were present in the leukemic cells of 26 of the 48 patients (54%) in the CNS3 group and in 55 of the 242 patients (23%) in the CNS1 þ 2 group (Po0.001). Among the 172 patients for whom data were available, MLL gene rearrangement was detected in eight of the 22 patients (36%) in the CNS3 group and in 34 of the 128 patients (27%) in the CNS1 þ 2 group (P ¼ 0.44). Figure 1 Schema of CNS-directed therapy for patients with CNS involvement at diagnosis of AML. IT chemotherapy consisted of methotrexate (MTX) alone or methotrexate, hydrocortisone, and cytarabine (MHA). Cranial: cranial irradiation.

to compare the survival of the CNS3 and CNS1 þ 2 groups to ensure that age, protocol, and cytogenetic features did not confound the comparison. However, because some subgroups would comprise very few patients if all factors were stratified simultaneously, we stratified each factor independently to compare survival in the CNS3 and CNS1 þ 2 groups. The cumulative incidence of CNS relapse was estimated by Prentice’s method.19 The association between CNS status and outcome was tested by Gray’s method.20 The duration of risk of CNS relapse was defined as the interval between diagnosis and the date of CNS relapse, death, or last contact, whichever came first. Death from any cause was considered a competing event. The differences between categorical variables were examined by using two-tailed Fisher’s or w2 exact test. The differences between continuous variables were examined by using the Wilcoxon–Mann–Whitney rank test or the Kruskal–Wallis test. SAS release 8.1 software (SAS Institute, Cary, NC, USA) was used to perform the statistical analyses.

Results

Patient characteristics Of the 290 patients, 85 (29%) had CNS involvement (CNS2, n ¼ 37; CNS3, n ¼ 48) at the time of diagnosis. Table 3 lists the presenting clinical features of this study group. Two patients Leukemia

Patient outcome The probability of 5-year EFS was significantly higher in the CNS3 (43.777.0% (s.e.)) than in either the CNS1 (27.873.2%, P ¼ 0.015) or CNS2 (24.377.5%, P ¼ 0.032) groups (overall P ¼ 0.04) (Figure 2). EFS did not differ significantly between the CNS1 and CNS2 groups (P ¼ 0.77), which were combined to form the CNS1 þ 2 group for further comparison. EFS differed significantly between the CNS3 and the CNS1 þ 2 groups after adjusting for treatment protocol (P ¼ 0.012) and for age at diagnosis (p2 vs 42 years, P ¼ 0.019). However, no significant difference in EFS was detected between the CNS3 and CNS1 þ 2 groups after adjusting for cytogenetic features (favorable vs others; P ¼ 0.075). This finding is explained by the greater frequency of favorable cytogenetic features21,22 in the CNS3 group (54%) than in the CNS1 þ 2 group (23%). It is noteworthy that t(9;11) is associated with favorable prognosis in the context of therapy used at our institution.22 However, this fact has been disputed by other investigators in the context of different chemotherapy regimens.23 Analysis according to treatment protocol showed that in the AML80, AML87, and AML91 protocols (but not in the AML83 protocol), the CNS3 group had a higher probability of EFS than did the CNS1 þ 2 group. No statistically significant difference in EFS was seen between the CNS 3 patients who had CNS signs and symptoms at diagnosis (5-year rate ¼ 60.0710.5%, n ¼ 20) and those who did not (32.178.4%, n ¼ 28) (P ¼ 0.09). In patients with CNS 3 status, the 5-year EFS estimate (%7s.e.) was higher in patients treated on AML87 and AML91 (57.1716.7 and 70.0713.6, respectively) than in patients treated on AML80 and AML83 (45.0710.6 and 9.176.1, respectively). However, no statistically significant difference could be shown because of the small number of patients.


2093 Table 3

Presenting clinical features

Feature

Category

Total (N ¼ 290)

CNS1 (N ¼ 205)

CNS2 (N ¼ 37)

CNS3 (N ¼ 48)

Age (years)

Median Range o2 years 2–9 years X9 years

8.3 0.003–20.7 57 (19.7%) 107 (36.9%) 126 (43.4%)

7.3 0.4–18.8 9 (24.3%) 14 (37.8%) 14 (37.8%)

4.6 0.2–19.5 16 (33.3%) 15 (31.3%) 17 (35.4%)

Sex

Male Female

157 (54.1%) 133 (45.9%)

114 (55.6%) 91 (44.4%)

17 (45.9%) 20 (54.1%)

26 (54.2%) 22 (45.8%)

Race

White Black Other

238 (82.1%) 49 (16.9%) 3 (1.0%)

167 (81.5%) 36 (17.6%) 2 (0.9%)

33 (89.2%) 4 (10.8%) 0 (0.0%)

38 (79.2%) 9 (18.7%) 1 (2.1%)

FAB type

M0 M1 M2 M3 M4 M5 M7 Other

4 52 76 17 48 61 20 9

(1.4%) (18.1%) (26.5%) (5.9%) (16.7%) (21.3%) (7.0%) (3.1%)

2 41 56 15 28 34 20 7

(1.0%) (20.2%) (27.6%) (7.4%) (13.8%) (16.8%) (9.9%) (3.4%)

2 (5.4%) 5 (13.5%) 10 (27.0%) 0 7 (18.9%) 11 (29.7%) 0 2 (5.4%)

Protocol

AML80 AML83 AML87 AML91

84 67 60 79

(29.0%) (23.1%) (20.7%) (27.2%)

60 46 50 49

(29.3%) (22.4%) (24.4%) (23.9%)

4 10 3 20

(10.8%) (27.0%) (8.11%) (54.0%)

20 11 7 10

(41.7%) (22.9%) (14.6%) (20.8%)

Cytogenetic features

t(8;21) inv(16) t(9;11) t(15;17)

39 19 23 13

(13.4%) (6.6%) (7.9%) (4.5%)

27 6 10 12

(13.2%) (2.9%) (4.9%) (5.9%)

4 5 3 0

(10.8%) (13.5%) (8.1%) (0.0%)

8 8 10 1

(16.7%) (16.7%) (20.8%) (2.1%)

MLL gene rearrangementa

Not peformed Germ line Rearranged

22 (12.8%) 108 (62.8%) 42 (24.4%)

12 (11.1%) 71 (65.7%) 25 (23.2%)

2 (5.9%) 23 (67.6%) 9 (26.5%)

8 (26.7%) 14 (46.6%) 8 (26.7%)

Outcome

Dead Alive

185 (63.8%) 105 (36.2%)

137 (66.8%) 68 (33.2%)

24 (64.9%) 13 (35.1%)

24 (50.0%) 24 (50.0%)

9.4 0.003–20.7 32 (15.6%) 78 (38.1%) 95 (46.3%)

6 10 2 13 16

0 (12.8%) (21.3%) (4.3%) (27.7%) (34.0%) 0 0

a

Information was available for 172 patients.

Prognostic factor analysis The CNS3 and CNS1 þ 2 groups were analyzed separately to identify prognostic factors. As CNS involvement was defined and treated differently in the AML91 protocol, we adjusted the analysis for treatment era (earlier era: AML80, AML83, and AML87; later era: AML91). Age (o2 vs 2–9 vs X9 years), race (white vs other), sex, and WBC count at diagnosis (p50 000/ml vs 450 000/ml) were not significantly prognostic of EFS within the CNS3 or CNS1 þ 2 groups after adjustment for treatment era (data not shown). Moreover, the presence of CNS signs and symptoms at diagnosis was not significantly prognostic of EFS in the CNS3 group after adjustment for treatment era (P ¼ 0.17).

CNS radiotherapy Figure 2 Probability of EFS of pediatric patients with AML according to CNS status at diagnosis. CNS3 vs CNS1, P ¼ 0.015; CNS3 vs CNS2, P ¼ 0.032. The mean 5-year EFS estimate (7s.e.) for each group was CNS1: 27.873.2%, CNS2: 24.377.5%, CNS3: 43.777.0%.

Of the 48 patients in the CNS3 group, 26 received cranial radiation (6–48 Gy). An additional three patients received TBI as part of the conditioning regimen for hematopoietic stem cell transplantation. Patients receiving TBI were considered to be in the radiation group. Leukemia


2094 Among the 55 patients with CNS2 or CNS3 status in the earlier era (AML80, AML83, and AML87), 26 (15 with CNS3 and 11 with CNS2 status) did not receive irradiation because they died or experienced relapse before radiation therapy was scheduled to occur (8–27 months after diagnosis). Notably, one patient with CNS2 status (treated on AML83), whose parents made a decision against radiation, remained alive and free of disease at the time of this report, 13 years after diagnosis. In total, 29 patients (six with CNS2 and 23 with CNS3 status) received radiation. Of these, four CNS2 patients (66.7%) died while 12 CNS3 patients (52.1%) remained in continuous complete remission. In the AML91 protocol, radiation was administered only to patients who had symptomatic CNS involvement. Table 2 shows the outcomes of the 10 CNS3 patients who were treated on the AML91 protocol. Importantly, four of the five patients who did not receive radiotherapy had remained in continuous complete remission for 7.6, 9.2, 9.3, and 9.6 years at the time of this report, and three of the five patients who received cranial irradiation for symptomatic disease were alive and free of disease, 8.3, 9.3, and 9.9 years after diagnosis.

CNS relapse Primary CNS relapse (isolated CNS recurrence of leukemia) occurred in 10 of the 290 patients, all of whom were treated on the AML80 (n ¼ 3) and AML83 (n ¼ 7) protocols. Only two of the 10, both of which had CNS3 status, received irradiation before they experienced a CNS relapse. No CNS relapse was observed in patients treated on the AML87 and AML91 protocols, all of whom were followed for at least 5 years. The overall 5-year cumulative incidence of CNS relapse was 3.471.1%. This figure was 6.672.0% on AML80 and AML83 vs 0 on AML87 and AML91 (P ¼ 0.002). The 5-year cumulative incidence of CNS relapse on the AML80 and AML83 combined was 4.772.1% for the 106 CNS1 patients, 7.177.2% for the 14 CNS2 patients, and 12.976.1% for the 31 CNS3 patients (P ¼ 0.26) (Figure 3). Although no statistically significant difference could be demonstrated due to small numbers, the cumulative incidence of CNS relapse was appreciably higher in the CNS3 group than in the others. None of the four CNS3 patients who had a CNS relapse had CNS signs and symptoms at diagnosis. One patient in the CNS3 group (3.2%) and three patients in the combined CNS1 þ 2 group (2.5%) had relapse in both the CNS and bone marrow.

Figure 3 Cumulative incidence of CNS relapse for patients with CNS1, CNS2, or CNS3 status who were treated on the AML80 and AML83 protocols (overall P ¼ 0.26). Leukemia

Discussion In this single-institution study, we have shown that CNS disease at diagnosis is not an adverse prognostic factor in pediatric AML. This finding confirms that of an earlier study at our institution,7 which also showed that the presence of CNS leukemia at diagnosis did not adversely affect the rate of remission induction or the duration of complete remission.7 A CCG study conducted between April 1988 and October 1989 suggested that CNS leukemia is associated with an adverse prognosis in pediatric AML,8 but this association was not confirmed in a subsequent CCG study conducted between 1990 and 1995.24 This study could not specifically address the role of cranial irradiation in the treatment of CNS involvement or in systemic control of leukemia in pediatric AML, because 26 of the 55 patients treated on protocols AML80, AML83, and AML87 (47%) died or had relapses before radiotherapy was administered (8–27 months after diagnosis). Moreover, cranial irradiation was given at different time periods and to different subgroups of patients with CNS involvement (symptomatic vs nonsymptomatic) on different treatment protocols. One potential risk of reducing the use of CNS radiation in patients with CNS disease is that it may increase the probability of primary CNS or hematologic relapse. It was suggested that the CNS is a sanctuary site in which leukemic cells are relatively protected from antineoplastic drugs that do not penetrate the blood–brain barrier well.5 In our study, primary CNS relapse was relatively infrequent, occurring only among patients treated on the AML80 and AML83 protocols, in which cranial irradiation was administered at 15 and 23 months post diagnosis, respectively. No primary CNS relapse was observed on the two more recent treatment protocols, when radiation therapy was administered 8 months after diagnosis (AML87) or was reserved for patients with symptomatic CNS disease (AML 91). As shown in Table 2, the treatment regimen on AML87 included higher doses of cytarabine and etoposide, indicative of a more intensive systemic chemotherapy effect that could have contributed to better overall outcome. Moreover, the two later regimens included chemotherapeutic agents that are well known to have greater CNS penetration (high-dose cytarabine11 on AML 87 and cladribine12 on AML91). This decrease in primary CNS relapse created a trend toward higher survival estimates in the CNS3 group in the two latter treatment protocols, although the small number of patients in each subgroup diminished our power to detect a statistically significant difference. In AML91, four of five CNS3 patients who received no cranial irradiation because of the absence of symptoms remained in durable complete remission (Table 4). Despite the absence of statistical power to compare these small subgroups, our results suggest that long-term remission of CNS leukemia can be achieved without the use of CNS radiotherapy. This finding further suggests that cranial irradiation may not be necessary in the context of contemporary therapy for AML patients with asymptomatic CNS involvement at diagnosis. Future studies should investigate the outcome of patients with AML who have CNS involvement at diagnosis and are treated without radiotherapy. These studies should explore the potential role of systemic therapy with agents that offer increased CNS penetration and further evaluate the proper timing of cranial irradiation (if given) in patients who do and do not have symptoms. The importance of this information increases with the accumulation of evidence that cranial radiation is a source of long-term mortality and morbidity, including the risk of secondary neoplasms25,26 and long-term endocrine and neurological sequelae.27–29 Ideally, this issue should be addressed by


2095 Table 4

Clinical and biological characteristics and outcome of CNS3 patients treated on the AML91 protocol

Patient

Age at Dx

1 2 3 4 5 6 7 8 9 10

4.4 2.8 1.0 1.3 10.7 9.3 4.3 1.0 0.6 14.0

Sex

FAB subtype

WBC at Dx (x109/l)

CSF WBC

CSF blasts (%)

F M F F F F M M F M

M1 M4 M5 M4 M3 M5 M2 M5 M5 M2

192.4 13.9 5.4 50.8 36.8 4.1 31.3 55.0 9.4 109.2

39 13 3 7 13 5 0 1 5 0

60 19 49 16 19 1 14 13 8 1

Cytogenetic abnormality

CNS symptoms

Cranial radiation (dose, Gy)

Relapse

t(3;5) inv(16) t(9;11) inv(16) t(1;16) t(9;11) t(8;21) t(10;11) t(9;11) t(8;21)

None None Thoracic cord chloroma None Intracranial chloroma Intracranial chloroma Intracranial chloroma Facial nerve palsy None Facial nerve palsy

No No No No No Yes Yes Yes Yes Yes

Yesa No No No No Yesa Yesa No No No

(9) (8) (6) (18.6) (12)

Dx: diagnosis; M ¼ male; F ¼ female; CR: complete remission; CSF ¼ cerebrospinal fluid; CNS ¼ central nervous system. a All relapses were hematological.

a clinical trial in which the use of CNS radiotherapy is randomly assigned. Such a trial seems unlikely, given the known risks of radiation therapy and the growing evidence that modern systemic antileukemic therapy is capable of successfully eradicating CNS leukemia. Moreover, the outcomes (including CNS relapse) of patients treated in the recent Children’s Cancer Group,30 Pediatric Oncology Group (Becton et al, Blood 2001; 98: 461 a; abstract), and Medical Research Council31 trials, none of which include cranial radiation, are similar to those of patients treated on the Berlin–Frankfurt–Mu¨nster32 trial that used cranial radiation for prevention and treatment of CNS involvement. Another finding of our study is that the clinical behavior of pediatric AML with CNS2 status (low number of WBCs in cytocentrifuged CSF, with leukemic blast cells) does not differ from that of CNS-negative AML. This confirms the findings previously shown in a CCG pilot study.8 These results differ from those of an earlier study that showed CNS relapse in five of six children who had CNS2 status at diagnosis of AML (Odom et al, ASCO Proc 1989; 8: 218; abstract). This disparity may have been caused by different treatment. Additional studies are needed to determine the prognostic and therapeutic relevance of CNS2 and CNS3 status in the context of other treatment regimens for pediatric AML. Finally, the use of modern chemotherapeutic agents that offer greater penetration of the CNS, such as high-dose cytarabine and cladribine, may have improved the eradication of low-level CNS leukemia without requiring additional CNS-directed therapy in our more recent trials. Another interesting finding of our study was the significant association between CNS involvement and the t(9;11) and inv(16) cytogenetic abnormalities. The cytogenetic abnormality t(9;11) is associated with a favorable prognosis in the context of therapy used at our institution,22 although other investigators have reported different findings in the context of other chemotherapy regimens.23 Since these features are favorable prognostic factors in our experience, they may explain the higher survival rates of patients with AML and CNS involvement. CNS involvement is also reportedly more frequent in infants and in patients with AML M4 and M5 subtypes,1,7 a finding confirmed in this study, and most cases of AML with t(9;11)22 or inv(16)33 are of the M5 or M4 AML subtype, respectively.

Conclusions Our single-institution study has shown that CNS involvement is more common in patients whose AML has favorable cytogenetic features. In the context of our treatment, the CNS2 category (low

CSF leukocyte count with blast cells) has no prognostic significance. Although only a relatively small number of patients with CNS3 status were studied, our findings suggest that cranial irradiation may not be necessary even for these patients, in the context of contemporary chemotherapy regimens. Our recently completed frontline AML study (AML97),34 whose results are being analyzed, prospectively addresses this issue.

Acknowledgements We thank Sharon Naron for expert editorial assistance.

References 1 Dahl GV, Simone JV, Hustu HO, Mason C. Preventive central nervous system irradiation in children with acute nonlymphocytic leukemia. Cancer 1978; 42: 2187–2192. 2 Stavem P, Evansen SA. CNS involvement in acute myelomonoblastic leukemia. Br J Haematol 1984; 58: 383–384. 3 Creutzig U, Ritter J, Zimmermann M, Schellong G. Does cranial irradiation reduce the risk for bone marrow relapse in acute myelogenous leukemia? Unexpected results of the Childhood Acute Myelogenous Leukemia Study BFM-87. J Clin Oncol 1993; 11: 279–286. 4 Amadori S, Ceci A, Comelli A, Madan E, Masera G, Nespali L et al. Treatment of acute myelogenous leukemia in children: results of the Italian Cooperative Study AIEOP/LAM 8204. J Clin Oncol 1987; 9: 1356–1363. 5 Pinkel D, Woo S. Prevention and treatment of meningeal leukemia in children. Blood 1994; 84: 355–366. 6 Ravindranath Y, Steuber CP, Krischer J, Civin CI, Ducore J, Vega R et al. High-dose cytarabine for intensification of early therapy of childhood acute myeloid leukemia: a Pediatric Oncology Group study. J Clin Oncol 1991; 9: 572–580. 7 Pui CH, Dahl GV, Kalwinsky DK, Look AT, Mirro J, Dodge RK et al. Central nervous system leukemia in children with acute nonlymphoblastic leukemia. Blood 1985; 66: 1062–1067. 8 Woods WG, Kobrinsky N, Buckley J, Neudorf S, Sanders J, Miller L et al. Intensively timed induction therapy followed by autologous or allogeneic bone marrow transplantation for children with acute myeloid leukemia or myelodysplastic syndrome: a Children’s Cancer Group pilot study. J Clin Oncol 1993; 11: 1448–1457. 9 Mahmoud HH, Rivera GK, Hancock ML, Krance RA, Kun LE, Behm FG et al. Low leukocyte counts with blast cells in cerebrospinal fluid of children with newly diagnosed acute lymphoblastic leukemia. N Engl J Med 1993; 329: 314–319. 10 Pui C-H. Toward optimal central nervous system-directed treatment in childhood acute lymphoblastic leukemia. J Clin Oncol 2003; 21: 179–181. Leukemia


2096 11 Blaney SM, Balis FM, Poplack DG. Current pharmacological treatment approaches to central nervous system leukaemia. Drugs 1991; 41: 702–716. 12 Liliemark J. The clinical pharmacokinetics of cladribine. Clin Pharmacokinet 1997; 32: 120–131. 13 Dahl GV, Kalwinsky DK, Mirro J, Look AT, Pui C-H, Murphy SB et al. Allogeneic bone marrow transplantation in a program of intensive sequential chemotherapy for children and young adults with acute nonlymphocytic leukemia in first remission. J Clin Oncol 1990; 8: 295–303. 14 Kalwinsky D, Mirro Jr J, Schell M, Behm F, Mason C, Dhal GV. Early intensification of chemotherapy for childhood acute nonlymphoblastic leukemia: Improved remission induction with a five-drug regimen including etoposide. J Clin Oncol 1988; 6: 1134–1143. 15 Arnaout MK, Radomski KM, Srivastava DK, Tong X, Belt JA, Raimondi SC et al. Treatment of childhood acute myelogenous leukemia with an intensive regimen (AML-87) that individualizes etoposide and cytarabine dosages: short- and long-term effects. Leukemia 2000; 14: 1736–1742. 16 Krance RA, Hurwitz CA, Head DR, Raimondi SC, Behm FG, Crews KR et al. Experience with 2-chlorodeoxyadenosine in previously untreated children with newly diagnosed acute myeloid leukemia and myelodysplastic diseases. J Clin Oncol 2001; 19: 2804–2811. 17 Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DA, Gralnick HR et al. Proposed revised criteria for the classification of acute myeloid leukemia. A report of the French–American–British Cooperative Group. Ann Intern Med 1985; 103: 620–625. 18 Williams DL, Harris S, Williams KJ, Brosius MJ, Lemonds W. A direct bone marrow chromosome technique for acute lymphoblastic leukemia. Cancer Genet Cytogenet 1984; 13: 239–257. 19 Prentice RL, Kalbfleisch JD, Peterson Jr AV, Flournoy N, Farewell VT, Breslow NE. The analysis of failure times in the presence of competing risks. Biometrics 1978; 34: 541–554. 20 Gray RJ. A class of k-sample tests for comparing the cumulative incidence of a risk. Ann Stat 1988; 16: 1141–1154. 21 Grimwade D, Walker H, Oliver F, Wheatley K, Harrison C, Harrison G, et al, on behalf of the Medical Research Council Adult Children’s Leukaemia Working Parties. The importance of diagnostic cytogenetics on outcome in AML: analysis of 1,612 patients entered into the MRC AML 10 trial. Blood 1998; 92: 2322–2333. 22 Rubnitz JE, Raimondi SC, Tong X, Srivastava DK, Razzouk BI, Shurtleff SA et al. Favorable impact of the t(9;11) in childhood acute myeloid leukemia. J Clin Oncol 2002; 20: 2302–2309. 23 Raimondi SC, Chang MN, Ravindranath Y, Behm FG, Gresik MV, Steuber CP et al. Chromosomal abnormalities in 478 children with acute myeloid leukemia: clinical characteristics and treatment outcome in a cooperative pediatric oncology group study – POG 8821. Blood 1999; 94: 3707–3716.

Leukemia

24 Woods WG, Kobrinsky N, Buckley JD, Lee JW, Sanders J, Neudrof S et al. Timed-sequential induction therapy improves postremission outcome in acute myeloid leukemia: a report from the Children’s Cancer Group. Blood 1996; 97: 4979–4989. 25 Walter AW, Hancock ML, Pui CH, Hudson MM, Ochs JS, Rivera GK et al. Secondary brain tumors in children treated for acute lymphoblastic leukemia at St. Jude Children’s Research Hospital. J Clin Oncol 1998; 16: 3761–3767. 26 Loning L, Zimmermann M, Reiter A, Kaatsch P, Henze G, Riehm H et al. Secondary neoplasms subsequent to Berlin–Frankfurt–Mu¨nster therapy of acute lymphoblastic leukemia in childhood: significantly lower risk without cranial radiotherapy. Blood 2000; 95: 2770–2775. 27 Hasle H, Helgestad J, Christensen JK, Jacobsen BB, Kamper J. Prolonged intrathecal chemotherapy replacing cranial irradiation in high-risk acute lymphatic leukemia: long-term follow-up with cerebral computed tomography scans and endocrinological studies. Eur J Pediatr 1995; 154: 24–29. 28 Leung W, Hudson MM, Strickland DK, Phipps S, Srivastava DK, Ribeiro RC et al. Late effects of treatment in survivors of childhood acute myeloid leukemia. J Clin Oncol 2002; 18: 3273–3279. 29 Reinhardt D, Thiele C, Creutzig U. Neuropsychological sequelae in children with AML treated with or without prophylactic CNSirradiation. Klin Padiatr 2002; 214: 22–29, (article in German). 30 Woods WG, Neudorf S, Gold S, Sanders J, Buckley JD, Barnard DR et al. A comparison of allogeneic bone marrow transplantation, autologous bone marrow transplantation, and aggressive chemotherapy in children with acute myeloid leukemia in remission: a report from the Children’s Cancer Group. Blood 2001; 97: 56–62. 31 Stevens RF, Hann IM, Wheatley K, Gray RG. Marked improvements in outcome with chemotherapy alone in pediatric acute myeloid leukemia: results of the United Kingdom Medical Research Council’s 10th AML trial. Br J Heamatol 1998; 101: 130–140. 32 Creutzig U, Ritter J, Zimmermann M, Reinhardt D, Hermann J, Berthhold F et al, for the Acute Myeloid Leukemia – Berlin– Frankfurt–Mu¨nster Study Group. Improved treatment results in high-risk pediatric acute myeloid leukemia patients after intensification with high-dose cytarabine and mitoxantrone: results of study Acute Myeloid Leukemia Berlin–Frankfurt–Mu¨nster 93. J Clin Oncol 2001; 19: 2705–2713. 33 Razzouk BI, Raimondi SC, Srivastava DK, Pritchard M, Behm FG, Tong X et al. Impact of treatment on the outcome of acute myeloid leukemia with inversion 16: a single institution’s experience. Leukemia 2001; 15: 1326–1330. 34 Crews KR, Ghandi V, Srivastava DK, Razzouk BI, Tong X, Behm FG et al. An interim comparison of a continuous infusion vs a short daily infusion of cytarabine given with cladribine for pediatric acute myeloid leukemia. J Clin Oncol 2002; 20: 4217–4224.