Medical and Pediatric Oncology 19:269-275 (1991)
Lower Incidence of Meningeal Leukemia When Prednisone Is Replaced by Dexamethasone in the Treatment of Acute Lymphocytic Leukemia Barbara Jones, MD*, Arnold 1. Freeman, MD, JonathanJ.Shuster, PhD, Claude Jacquillat, MD, Marise Weil, MD, Carl Pochedly, MD, Lucius Sinks, MD, Louise Chevalier, MD, Harold M. Maurer, MD, Kjell Koch, MD, Geoffrey Falkson, MD, Richard Patterson, MD, Barbara Seligman, MD, JurgSartorius, MD*, Faith Kung, MD, Farid Haurani, MD, Marie Stuart, MD, E. Omer Burgert, MD, Frederick Ruymann, MD, Arthur Sawitsky, MD, Edwin Forman, MD, Hansjuerg Pluess, MD, John Truman, MD, Nasrollah Hakami, MD, Oliver Glidewell, MA, Arvin S. Clicksman, MD, and James F. Holland, MD In 1971, Cancer and Leukemia Grou B (CALGB) mounted a study of acute IympEocytic leukemia (ALL) that compared the effects of the two steroid hormones dexamethasome and prednisone. Six-hundredforty-six children and adolescents with ALL were randomized to receive either prednisone or dexamethasone as art of their remission induction therapy. T e 493 evaluable patients who achieved complete remission received the same steroid as pulses throughout remission. Specific central nervous system (CNS) therapy was randomized to either six injections of intrathecal methotrexate (IT MTX) alone or to six injections of IT MTX with cranial radiation (2,400 cGy). Both cranial radiation and dexamethasone offered increased protection against CNS relapse as the first site of failure over IT MTX alone. There were 30 CNS relapses among
R
238 patients (12.6%) receiving cranial radiation plus IT MTX, whereas there were 70 CNS relapses among 225 ( P < 0.001) (22.5%) in those who received IT MTX alone. Similarly, there were 33 CNS relapses among 231 (14.3%) children treated with dexamethasone, whereas there were 67 CNS relapses among 262 (25.6%) treated with prednisone ( P = 0.017). Both steroids appeared equal in protecting the bone marrow. Recent national studies have shown significant improvements in preventing CNS relapse over the results in the present report. However, this finding warrants further investigation and, with further documentation, could lead to the substitution of prednisone by dexamethasone to aid further in preventing CNS relapse. This may be particularly important in patients at higher risk for CNS relapse.
Key words: Cancer and Leukemia Group B, CNS leukemia, steroid therapy
From Morgantown, West Virginia (B.J.); Roswell Park Memorial Institute, Buffalo, New York (L.S., A.I.F.); Department of Statistics, University of Florida, Gainesville, FL (J.J.S.); Institute De Recherches Sur Les Leucemies, Hospital Saint-Louis, Paris, France (C.J., M.W.); Mount Sinai Hospital, New York, New York (C.P.); Montreal Children’s Hospital, Montreal, P.Q., Canada (L.C.); Medical College of Virginia, Richmond, (H.M.M.); University of Miami Medical Center, Miami (K.K.); University of Pretoria, Pretoria, South Africa (G.F.); Bowman Gray School of Medicine, Winston-Salem, North Carolina (R.P.); State University of New York, Jewish Hospital of Brooklyn, Brooklyn (B .S.);Universitats-Kinderspital, Basel, Switzerland (J.S.); University of California, San Diego (F.K.); Thomas Jefferson Medical College, Philadelphia (F.H.); State University of New York Upstate Medical Center, Syracuse (M.S.); Mayo Clinic, Rochester, Minnesota (E.O.B.); Walter Reed Army Medical Center, Washington, DC (F.R.); Long Island-Jewish Hillside Medical Center, New Hyde Park, New York (A.S.); Brown University and Rhode
0 1991 Wiley-Liss, Inc.
Island Hospital, Providence (E.F., A.S.G.); Kinderspital, Zurich, Switzerland (H.P.); Massachusetts General Hospital, Boston (J.T.); University of Missouri, Columbia (N.H.); Mount Sinai Medical School, New York, NY (J.F.H.). *Deceased. The authors of this paper were members of Cancer and Leukemia Group B. Received April 16, 1990; accepted February 21, 1991. Address reprint requests to Dr. Arnold Freeman, Section of HematologylOncology, Children’s Mercy Hospital, 2401 Gillham Rd., Kansas City, MO 64108. Presented in part at the meeting of the American Association for Cancer Research, San Diego, California, May 1975, and at the meeting of the American Society of Clinical Oncology, Toronto, Canada, May 1984.
270
Jones et al.
INTRODUCTION
In the last 20 years there has been a dramatic improvement in the cure rate of children with acute lymphocytic leukemia. In part, this is the result of early specific therapy to the central nervous system (CNS) to irradicate nondetectable leukemic cells. This early CNS-directed treatment not only significantly decreased the incidence of meningeal relapse but also increased the overall remission duration and survival [ 11. A study initiated in 1968 by Cancer and Leukemia Group B [l] demonstrated a decrease in meningeal relapse from 50% to 23% by the use of intrathecal methotrexate alone. In the now classic work, Aur and his colleagues at St. Jude’s [2] significantly decreased the occurrence of meningeal relapse and improved the overall cure rate by treatment with 2,400 cGy cranial radiation plus intrathecal methotrexate [3]. In 1971, Cancer and Leukemia Group B (CALGB) embarked on a study of childhood acute lymphocytic leukemia (protocol 71 11) where the steroid used for induction therapy, and as “pulses” with vincristine throughout maintenance therapy, was randomized between prednisone and dexamethasone. This was based on data showing less impediment of transcapillary granulocyte migration by the latter drug [4]. The specific CNS therapy in protocol 71 11 was randomized to weekly intrathecal methotrexate compared with weekly intrathecal methotrexate plus cranial radiation. The effects of these treatments on the frequency of meningeal relapse are the subject of this report. MATERIALS AND METHODS
In 1971, CALGB Centers instituted a treatment protocol for patients to age 20 years at diagnosis with previously untreated acute lymphocytic leukemia. Initially, all patients were randomized to one of eight treatment plans (Fig. 1). At diagnosis, all atients were randomized to receive vincristine 2 mglm /week intravenously and either prednisone 40 mg/m2/day or dexamethasone 6 mg/m2/day orally in three divided doses. Those patients who achieved complete remission (CR) received pulses of the same steroid during maintenance. A major question was the effect of L-asparaginase during induction. Thus, these patients were also randomized to receive asparaginase 1,000 units/kg/day intravenously for 10 days prior to, simultaneously with, or subsequent to their 3 week course of vincristine. The fourth group served as a control receiving 4 weeks of vincristine and steroid but no asparaginase. The beneficial effects of subsequent L-asparaginase therapy have been reported [5] and will not be discussed further here. Prior to July 12, 1971, those patients achieving
rl
complete remission received as maintenance therapy methotrexate 15 mg/m2/day for 5 days intramuscularly followed by 9 days with no therapy and a second 5 day course of methotrexate (Fig. 1, arm A). Following the second 9 day rest period, they received mercaptopurine 600 mg/m2/dayintravenously for 5 days, a 9 day rest, and a second 5 day course of mercaptopurine. In July 1971, the results of the prior CALGB study (protocol 6801) demonstrated the superiority of maintenance with daily oral mercaptopurine and weekly oral methotrexate. Consequently, a second maintenance arm was added to protocol 7 111 such that all patients who achieved CR after July 1971 were randomized either to the parenteral regimen described above or to daily oral mercapto urine 90 mg/m2, weekly oral methotrexate 15 mgim , and monthly reinforcements consisting of a single dose of vincristine and a 7 day course of corticosteroid (Fig. 1, arm B). Drug doses in both regimens were decreased for significant toxicity, At the completion of 2 months of therapy, patients who had not presented with CNS leukemia or who had not developed CNS leukemia during these 2 months were randomized for specific treatment of the CNS. Six doses of intrathecal methotrexate, 12 mg/m2, were given at weekly intervals for the first 3 weeks with or without 2,400 cGy of cranial radiation and then given every other week during the beginning of continuation therapy. External beam radiation therapy was given to the whole brain, including the upper spinal cord down to the second cervical vertebrae. The radiation field included the entire brain, extending anteriorly to include the frontal lobes, posterior half of the eyeball, optic disc, and nerves superiorly to the vertex and posteriorly to the occiput. The inferior line of the field was extended to 0.5 cm below the skull base and posteriorly to the second cervical vertebrae. The anterior halves of both eyes were shielded. Two hundred cGy were delivered per day 5 days per week for a total dosage of 2,400 cGy. After completing 1 year of therapy, all patients still in complete remission on arm A (pulsed parenteral courses; see Fig. 1) were switched to daily oral mercaptopurine and weekly oral methotrexate (arm B) in the same doses utilized by arm B during the first year of therapy. “Pulses” were also administered consisting of two doses of vincristine and 2 weeks of the same induction steroid at each 3 months of maintenance. All patients remaining in CR continued antileukemic therapy for 5 years, at which time they randomly continued for 2 additional years or discontinued therapy. CR status was terminated by bone marrow relapse (>25% blast cells), the occurrence of meningeal leukemia (two or more definitely identified blast cells on cytologic preparations of cerebrospinal fluid or 10 or more celldper microliter not attributable to chemical
f
Dexamethasone Reduces CNS Leukemia FIRST QUARTER
I
I I
I
I
= 0 0 0 0 0
0
0
0
-/’. .
-
SECOND OUARTER
‘00 .
.
271
I1
m
THIRD 8 FOURTH QUARTERS REPEAT SECOND OUARTER
B
I
I
0
. 0 . 0 . 0 0 0 0 0 0 0 0 0
I
I
m
BI m
I
Fig. 1. Treatment schema. The top arm is arm A (parenteral courses) and the lower arm is arm B (daily oral therapy).
arachnoiditis), or other definitely identified extramedullary disease such as testicular relapse, or death while in CR. Immunophenotype (cell surface markers) was not performed on these patients. Spinal taps were not routinely performed at the time of diagnosis, nor were spinal taps pei-jormed at the time of hematologic relapse. Surveillance spinal taps were also not performed on these patients after they completed their CNS phase of therapy. Informed consent was obtained from patients, their parents or legal guardians, depending on legal circumstances prior to beginning therapy. These protocols were evaluated and approved by the institutionalreview boards at all participating centers. CALGB protocol 71 11 closed to accrual in March 1974, and follow-up was completed in September 1981 (latest update) under a contract with the National Institutes of Health. In 1979, the pediatric division of CALGB was disbanded. Data have been filed on tapes that are not compatible with current computers utilized by the Pediatric Oncology Group where these data are now stored. Thus, without independent funding, a current analysis of data is not feasible. STATISTICAL METHODS
Complete response rates were compared via the Shuster-Downing chi-squared method [6]. Steroid comparisons were adjusted for imbalances in steroid assignments. Prognostic factors were developed using a forward
stepwise Cox regression [7] with time to first treatment failure as the dependent variable. Independent variables studied were age (cut at 1 , 4 , 7 , and 10 years completed), white count (cut at 10,000,20,000,50,000and 100,000 per kl), sex, and race. Comparison of treatments was made by the log rank test [8], with adjustment as indicated in the comparisons quoted. Figures are Kaplan-Meier curves. Note that a lack of statistically significant difference should not be equated to no difference. The terms significant and nonsignificant are arbitrarily distinguished by P < 0.05 and P > 0.05, respectively.
RESULTS
Table I describes which patients are included in various analyses. Of the 673 patients who entered the study between February 5 , 1971, and March 18, 1974, 646 were evaluable for the induction phase. The overall remission rate was 85.7%. Of the 646 patients whose data were analyzed for determining risk groups, 554 patients achieved complete remission and 425 patients were evaluable for analysis of maintenance therapy (Table I). Excluded from this analysis were 49 patients who relapsed prior to the CNS randomization, 50 patients who received parenteral maintenance therapy prior to the amendment randomizing the maintenance to two regimens, and 30 patients with inadequate data or noncompliance to some aspects of the study.
272
Jones et al.
TABLE 1. Patient Accounting Parameter Total accrual Induction Ineligible Protocol violation Early loss Inadequate records Nonrandom entry Evaluable Maintenance CR (out of 646) Failed prior to CNS phase Inadequate data for risk grouping Regimen A, prior to amendment Disqualified in maintenance (with data) Disqualified in maintenance (no data) Fully evaluable Patients used in Induction analysis CNS (steroid/XRT) (554-49)a Maintenance analysis Steroid-curve (Figure 2)
No.
673 7
8 2 7 3 646 554 (85.7%) 49
3 50 15 12 425 646 493 425 55 1
a(554-49), 554 (CR)-49 patients failed prior to CNS phase.
For the first year of remission, the maintenance chemotherapy was randomized between oral and parenteral mercaptopurine and methotrexate. There were no detectable differences in remission duration, marrow remission, CNS remission, or survival comparing these two maintenance regimens. These arms have therefore been collapsed for the present evaluation of CNS relapse rate. Of particular interest is the incidence of meningeal leukemia as the first site of relapse (Tables I1 to IV). At the completion of 2 months of maintenance therapy, 493 patients still in CR were randomized for CNS therapy, with 255 receiving intrathecal methotrexate alone and 238 receiving 2,400 cGy cranial radiation in addition to intrathecal methotrexate. Meningeal leukemia occurred as the first site of failure in 30 of the 238 (12.6%) who were radiated and 70 of the 255 (27.5%) who were not radiated (P < 0.001). The patients who received cranial radiation had fewer meningeal relapses, and this was reflected in a longer complete remission duration ( P = 0.037). Unexpectedly, the use of dexamethasone also decreased the incidence of meningeal relapse. Only 33 of 23 1 (14.3%) who received dexamethasone experienced CNS relapse as compared with 67 of the 262 (27.5%) who received prednisone ( P = 0.017). Table 111 compares the relapse rates of those receiving dexamethasone as opposed to those receiving prednisone. The only statistical difference was the decrease of CNS relapse in those receiving dexamethesone. There
was no statistical difference in bone marrow relapse. Although the selection of steroid was randomized at the onset of the induction therapy, the study was not designed to test the effects of dexamethasone versus prednisone on the CNS relapse rate. It thus seemed prudent to look at the interaction of other variables with these steroids. These variables include the induction and maintenance therapies. Comparison of the induction remission results based on the steroid given is shown in Table 11. The remission rate was not significantly affected by the steroid used. Using a two-sided comparison of steroids adjusted for the induction therapy, P is 0.90 (Mantel-Haenszel test). Apart from the choice of steroid, the other induction drugs, including asparaginase, had no significant effect on the number of CNS relapses (P = 0.63 by the log rank test). Similarly, there were no significant differences in the rates of CNS relapse between the two maintenance regimens (parenteral vs. oral 6MP and methotrexate) during the first year of maintenance. After this study was instituted, it became apparent that presenting characteristics such as white count, blast count, and age affected the outcome. High initial white counts and an onset during infancy both predispose to CNS relapses. Because this study was not prospectively stratified by risk categories, a retrospective analysis was performed to determine whether there was pooling of good- or poor-risk patients into a particular treatment arm.All evaluable patients were analyzed as to prognostic features of initial white count, age, sex, and race. In this study, race and sex were not significant prognostically, but white count and age were important determinants of outcome. Based on relapse-free survival, the most effective prognostic subdivisions of the initial white cell count were counts < lO,OOO/pl, 10,000 to 1OO,OOO/pl, and > 100,0OO/p1. Using a similar analysis with age, only two groups emerge, namely, those under 11 years of age at diagnosis and those over 11 years. Combining these two characteristics, the good-prognosis group were all patients with a white cell count < lO,OOO/pJ and those children under 11 years of age with a white cell count between 10,000 and I00 ,OOO/pl. The poor-prognosis patients were all those with a white cell count > 100,000/ ~1 and those 11 years of age and older with a white cell count > 10,000. There was no observable pooling of high-risk patients in either steroid arm.Using the above criteria, there were 366 good-risk and 59 patients poor-risk patients in 425 evaluable patients. Figure 2 depicts the effects of the steroid used on the CNS remission rate for both good- and poor-risk patients (see also Table IV). Unlike Table IV, in which the analysis is restricted to patients post-CNS prophylaxis, Figure 2 includes all evaluable patients in marrow
Dexamethasone Reduces CNS Leukemia
273
TABLE 11. Effect of Induction Theraw on CNS Remission*
Chemo
No.
vs
38 52 54 195
AVS VAS VSA
Prednisone Percent M-CR M-CR at CNS 87 83 80 88
Dexamethasone Percent M-CR M-CR at CNS
No.
12 15 74 82
46 49 62 150
85 82 82 89
72 71 13 83
*VS, Vincristine, steroid; AVS, prior asparaginase; VAS, simultaneous asparaginase; VSA, subsequent asparaginase; M-CR, marrow remission; M-CR at CNS, percent still in M-CR at time of CNS treatment.
TABLE 111. Relapse Rate-Dexamethasone Vs. Predisone* Prednisone Observed Expecteda No. T relapses BM relapses CNS relapses Deaths
262 142 106 67 132
Dexamethasone Observed Expected 23 1 122 I05 33 118
145 117 55 136
Two-sided P value
119 93 45 114
0.70 0.12 0.017 0.61
*No., number; T, total; BM, bone marrow; CNS, central nervous system. aExpected, the number expected via the log rank test for equal target population survival curves.
TABLE IV. CNS Relapse Rate* No. CRT iIT MTX 301238 IT MTX 701255 Dexamethasone 331231 Prednisone 671262 CRT effect Within group receiving pred Within group receiving dexa Steroid effect Within group without CRT Within group with CRT
Percent
P-Values (Two-sided)
12.6 27.5 14.3 25.5
Lower Incidence of Meningeal Leukemia When Prednisone Is Replaced by Dexamethasone in the Treatment of Acute Lymphocytic Leukemia Barbara Jones, MD*, Arnold 1. Freeman, MD, JonathanJ.Shuster, PhD, Claude Jacquillat, MD, Marise Weil, MD, Carl Pochedly, MD, Lucius Sinks, MD, Louise Chevalier, MD, Harold M. Maurer, MD, Kjell Koch, MD, Geoffrey Falkson, MD, Richard Patterson, MD, Barbara Seligman, MD, JurgSartorius, MD*, Faith Kung, MD, Farid Haurani, MD, Marie Stuart, MD, E. Omer Burgert, MD, Frederick Ruymann, MD, Arthur Sawitsky, MD, Edwin Forman, MD, Hansjuerg Pluess, MD, John Truman, MD, Nasrollah Hakami, MD, Oliver Glidewell, MA, Arvin S. Clicksman, MD, and James F. Holland, MD In 1971, Cancer and Leukemia Grou B (CALGB) mounted a study of acute IympEocytic leukemia (ALL) that compared the effects of the two steroid hormones dexamethasome and prednisone. Six-hundredforty-six children and adolescents with ALL were randomized to receive either prednisone or dexamethasone as art of their remission induction therapy. T e 493 evaluable patients who achieved complete remission received the same steroid as pulses throughout remission. Specific central nervous system (CNS) therapy was randomized to either six injections of intrathecal methotrexate (IT MTX) alone or to six injections of IT MTX with cranial radiation (2,400 cGy). Both cranial radiation and dexamethasone offered increased protection against CNS relapse as the first site of failure over IT MTX alone. There were 30 CNS relapses among
R
238 patients (12.6%) receiving cranial radiation plus IT MTX, whereas there were 70 CNS relapses among 225 ( P < 0.001) (22.5%) in those who received IT MTX alone. Similarly, there were 33 CNS relapses among 231 (14.3%) children treated with dexamethasone, whereas there were 67 CNS relapses among 262 (25.6%) treated with prednisone ( P = 0.017). Both steroids appeared equal in protecting the bone marrow. Recent national studies have shown significant improvements in preventing CNS relapse over the results in the present report. However, this finding warrants further investigation and, with further documentation, could lead to the substitution of prednisone by dexamethasone to aid further in preventing CNS relapse. This may be particularly important in patients at higher risk for CNS relapse.
Key words: Cancer and Leukemia Group B, CNS leukemia, steroid therapy
From Morgantown, West Virginia (B.J.); Roswell Park Memorial Institute, Buffalo, New York (L.S., A.I.F.); Department of Statistics, University of Florida, Gainesville, FL (J.J.S.); Institute De Recherches Sur Les Leucemies, Hospital Saint-Louis, Paris, France (C.J., M.W.); Mount Sinai Hospital, New York, New York (C.P.); Montreal Children’s Hospital, Montreal, P.Q., Canada (L.C.); Medical College of Virginia, Richmond, (H.M.M.); University of Miami Medical Center, Miami (K.K.); University of Pretoria, Pretoria, South Africa (G.F.); Bowman Gray School of Medicine, Winston-Salem, North Carolina (R.P.); State University of New York, Jewish Hospital of Brooklyn, Brooklyn (B .S.);Universitats-Kinderspital, Basel, Switzerland (J.S.); University of California, San Diego (F.K.); Thomas Jefferson Medical College, Philadelphia (F.H.); State University of New York Upstate Medical Center, Syracuse (M.S.); Mayo Clinic, Rochester, Minnesota (E.O.B.); Walter Reed Army Medical Center, Washington, DC (F.R.); Long Island-Jewish Hillside Medical Center, New Hyde Park, New York (A.S.); Brown University and Rhode
0 1991 Wiley-Liss, Inc.
Island Hospital, Providence (E.F., A.S.G.); Kinderspital, Zurich, Switzerland (H.P.); Massachusetts General Hospital, Boston (J.T.); University of Missouri, Columbia (N.H.); Mount Sinai Medical School, New York, NY (J.F.H.). *Deceased. The authors of this paper were members of Cancer and Leukemia Group B. Received April 16, 1990; accepted February 21, 1991. Address reprint requests to Dr. Arnold Freeman, Section of HematologylOncology, Children’s Mercy Hospital, 2401 Gillham Rd., Kansas City, MO 64108. Presented in part at the meeting of the American Association for Cancer Research, San Diego, California, May 1975, and at the meeting of the American Society of Clinical Oncology, Toronto, Canada, May 1984.
270
Jones et al.
INTRODUCTION
In the last 20 years there has been a dramatic improvement in the cure rate of children with acute lymphocytic leukemia. In part, this is the result of early specific therapy to the central nervous system (CNS) to irradicate nondetectable leukemic cells. This early CNS-directed treatment not only significantly decreased the incidence of meningeal relapse but also increased the overall remission duration and survival [ 11. A study initiated in 1968 by Cancer and Leukemia Group B [l] demonstrated a decrease in meningeal relapse from 50% to 23% by the use of intrathecal methotrexate alone. In the now classic work, Aur and his colleagues at St. Jude’s [2] significantly decreased the occurrence of meningeal relapse and improved the overall cure rate by treatment with 2,400 cGy cranial radiation plus intrathecal methotrexate [3]. In 1971, Cancer and Leukemia Group B (CALGB) embarked on a study of childhood acute lymphocytic leukemia (protocol 71 11) where the steroid used for induction therapy, and as “pulses” with vincristine throughout maintenance therapy, was randomized between prednisone and dexamethasone. This was based on data showing less impediment of transcapillary granulocyte migration by the latter drug [4]. The specific CNS therapy in protocol 71 11 was randomized to weekly intrathecal methotrexate compared with weekly intrathecal methotrexate plus cranial radiation. The effects of these treatments on the frequency of meningeal relapse are the subject of this report. MATERIALS AND METHODS
In 1971, CALGB Centers instituted a treatment protocol for patients to age 20 years at diagnosis with previously untreated acute lymphocytic leukemia. Initially, all patients were randomized to one of eight treatment plans (Fig. 1). At diagnosis, all atients were randomized to receive vincristine 2 mglm /week intravenously and either prednisone 40 mg/m2/day or dexamethasone 6 mg/m2/day orally in three divided doses. Those patients who achieved complete remission (CR) received pulses of the same steroid during maintenance. A major question was the effect of L-asparaginase during induction. Thus, these patients were also randomized to receive asparaginase 1,000 units/kg/day intravenously for 10 days prior to, simultaneously with, or subsequent to their 3 week course of vincristine. The fourth group served as a control receiving 4 weeks of vincristine and steroid but no asparaginase. The beneficial effects of subsequent L-asparaginase therapy have been reported [5] and will not be discussed further here. Prior to July 12, 1971, those patients achieving
rl
complete remission received as maintenance therapy methotrexate 15 mg/m2/day for 5 days intramuscularly followed by 9 days with no therapy and a second 5 day course of methotrexate (Fig. 1, arm A). Following the second 9 day rest period, they received mercaptopurine 600 mg/m2/dayintravenously for 5 days, a 9 day rest, and a second 5 day course of mercaptopurine. In July 1971, the results of the prior CALGB study (protocol 6801) demonstrated the superiority of maintenance with daily oral mercaptopurine and weekly oral methotrexate. Consequently, a second maintenance arm was added to protocol 7 111 such that all patients who achieved CR after July 1971 were randomized either to the parenteral regimen described above or to daily oral mercapto urine 90 mg/m2, weekly oral methotrexate 15 mgim , and monthly reinforcements consisting of a single dose of vincristine and a 7 day course of corticosteroid (Fig. 1, arm B). Drug doses in both regimens were decreased for significant toxicity, At the completion of 2 months of therapy, patients who had not presented with CNS leukemia or who had not developed CNS leukemia during these 2 months were randomized for specific treatment of the CNS. Six doses of intrathecal methotrexate, 12 mg/m2, were given at weekly intervals for the first 3 weeks with or without 2,400 cGy of cranial radiation and then given every other week during the beginning of continuation therapy. External beam radiation therapy was given to the whole brain, including the upper spinal cord down to the second cervical vertebrae. The radiation field included the entire brain, extending anteriorly to include the frontal lobes, posterior half of the eyeball, optic disc, and nerves superiorly to the vertex and posteriorly to the occiput. The inferior line of the field was extended to 0.5 cm below the skull base and posteriorly to the second cervical vertebrae. The anterior halves of both eyes were shielded. Two hundred cGy were delivered per day 5 days per week for a total dosage of 2,400 cGy. After completing 1 year of therapy, all patients still in complete remission on arm A (pulsed parenteral courses; see Fig. 1) were switched to daily oral mercaptopurine and weekly oral methotrexate (arm B) in the same doses utilized by arm B during the first year of therapy. “Pulses” were also administered consisting of two doses of vincristine and 2 weeks of the same induction steroid at each 3 months of maintenance. All patients remaining in CR continued antileukemic therapy for 5 years, at which time they randomly continued for 2 additional years or discontinued therapy. CR status was terminated by bone marrow relapse (>25% blast cells), the occurrence of meningeal leukemia (two or more definitely identified blast cells on cytologic preparations of cerebrospinal fluid or 10 or more celldper microliter not attributable to chemical
f
Dexamethasone Reduces CNS Leukemia FIRST QUARTER
I
I I
I
I
= 0 0 0 0 0
0
0
0
-/’. .
-
SECOND OUARTER
‘00 .
.
271
I1
m
THIRD 8 FOURTH QUARTERS REPEAT SECOND OUARTER
B
I
I
0
. 0 . 0 . 0 0 0 0 0 0 0 0 0
I
I
m
BI m
I
Fig. 1. Treatment schema. The top arm is arm A (parenteral courses) and the lower arm is arm B (daily oral therapy).
arachnoiditis), or other definitely identified extramedullary disease such as testicular relapse, or death while in CR. Immunophenotype (cell surface markers) was not performed on these patients. Spinal taps were not routinely performed at the time of diagnosis, nor were spinal taps pei-jormed at the time of hematologic relapse. Surveillance spinal taps were also not performed on these patients after they completed their CNS phase of therapy. Informed consent was obtained from patients, their parents or legal guardians, depending on legal circumstances prior to beginning therapy. These protocols were evaluated and approved by the institutionalreview boards at all participating centers. CALGB protocol 71 11 closed to accrual in March 1974, and follow-up was completed in September 1981 (latest update) under a contract with the National Institutes of Health. In 1979, the pediatric division of CALGB was disbanded. Data have been filed on tapes that are not compatible with current computers utilized by the Pediatric Oncology Group where these data are now stored. Thus, without independent funding, a current analysis of data is not feasible. STATISTICAL METHODS
Complete response rates were compared via the Shuster-Downing chi-squared method [6]. Steroid comparisons were adjusted for imbalances in steroid assignments. Prognostic factors were developed using a forward
stepwise Cox regression [7] with time to first treatment failure as the dependent variable. Independent variables studied were age (cut at 1 , 4 , 7 , and 10 years completed), white count (cut at 10,000,20,000,50,000and 100,000 per kl), sex, and race. Comparison of treatments was made by the log rank test [8], with adjustment as indicated in the comparisons quoted. Figures are Kaplan-Meier curves. Note that a lack of statistically significant difference should not be equated to no difference. The terms significant and nonsignificant are arbitrarily distinguished by P < 0.05 and P > 0.05, respectively.
RESULTS
Table I describes which patients are included in various analyses. Of the 673 patients who entered the study between February 5 , 1971, and March 18, 1974, 646 were evaluable for the induction phase. The overall remission rate was 85.7%. Of the 646 patients whose data were analyzed for determining risk groups, 554 patients achieved complete remission and 425 patients were evaluable for analysis of maintenance therapy (Table I). Excluded from this analysis were 49 patients who relapsed prior to the CNS randomization, 50 patients who received parenteral maintenance therapy prior to the amendment randomizing the maintenance to two regimens, and 30 patients with inadequate data or noncompliance to some aspects of the study.
272
Jones et al.
TABLE 1. Patient Accounting Parameter Total accrual Induction Ineligible Protocol violation Early loss Inadequate records Nonrandom entry Evaluable Maintenance CR (out of 646) Failed prior to CNS phase Inadequate data for risk grouping Regimen A, prior to amendment Disqualified in maintenance (with data) Disqualified in maintenance (no data) Fully evaluable Patients used in Induction analysis CNS (steroid/XRT) (554-49)a Maintenance analysis Steroid-curve (Figure 2)
No.
673 7
8 2 7 3 646 554 (85.7%) 49
3 50 15 12 425 646 493 425 55 1
a(554-49), 554 (CR)-49 patients failed prior to CNS phase.
For the first year of remission, the maintenance chemotherapy was randomized between oral and parenteral mercaptopurine and methotrexate. There were no detectable differences in remission duration, marrow remission, CNS remission, or survival comparing these two maintenance regimens. These arms have therefore been collapsed for the present evaluation of CNS relapse rate. Of particular interest is the incidence of meningeal leukemia as the first site of relapse (Tables I1 to IV). At the completion of 2 months of maintenance therapy, 493 patients still in CR were randomized for CNS therapy, with 255 receiving intrathecal methotrexate alone and 238 receiving 2,400 cGy cranial radiation in addition to intrathecal methotrexate. Meningeal leukemia occurred as the first site of failure in 30 of the 238 (12.6%) who were radiated and 70 of the 255 (27.5%) who were not radiated (P < 0.001). The patients who received cranial radiation had fewer meningeal relapses, and this was reflected in a longer complete remission duration ( P = 0.037). Unexpectedly, the use of dexamethasone also decreased the incidence of meningeal relapse. Only 33 of 23 1 (14.3%) who received dexamethasone experienced CNS relapse as compared with 67 of the 262 (27.5%) who received prednisone ( P = 0.017). Table 111 compares the relapse rates of those receiving dexamethasone as opposed to those receiving prednisone. The only statistical difference was the decrease of CNS relapse in those receiving dexamethesone. There
was no statistical difference in bone marrow relapse. Although the selection of steroid was randomized at the onset of the induction therapy, the study was not designed to test the effects of dexamethasone versus prednisone on the CNS relapse rate. It thus seemed prudent to look at the interaction of other variables with these steroids. These variables include the induction and maintenance therapies. Comparison of the induction remission results based on the steroid given is shown in Table 11. The remission rate was not significantly affected by the steroid used. Using a two-sided comparison of steroids adjusted for the induction therapy, P is 0.90 (Mantel-Haenszel test). Apart from the choice of steroid, the other induction drugs, including asparaginase, had no significant effect on the number of CNS relapses (P = 0.63 by the log rank test). Similarly, there were no significant differences in the rates of CNS relapse between the two maintenance regimens (parenteral vs. oral 6MP and methotrexate) during the first year of maintenance. After this study was instituted, it became apparent that presenting characteristics such as white count, blast count, and age affected the outcome. High initial white counts and an onset during infancy both predispose to CNS relapses. Because this study was not prospectively stratified by risk categories, a retrospective analysis was performed to determine whether there was pooling of good- or poor-risk patients into a particular treatment arm.All evaluable patients were analyzed as to prognostic features of initial white count, age, sex, and race. In this study, race and sex were not significant prognostically, but white count and age were important determinants of outcome. Based on relapse-free survival, the most effective prognostic subdivisions of the initial white cell count were counts < lO,OOO/pl, 10,000 to 1OO,OOO/pl, and > 100,0OO/p1. Using a similar analysis with age, only two groups emerge, namely, those under 11 years of age at diagnosis and those over 11 years. Combining these two characteristics, the good-prognosis group were all patients with a white cell count < lO,OOO/pJ and those children under 11 years of age with a white cell count between 10,000 and I00 ,OOO/pl. The poor-prognosis patients were all those with a white cell count > 100,000/ ~1 and those 11 years of age and older with a white cell count > 10,000. There was no observable pooling of high-risk patients in either steroid arm.Using the above criteria, there were 366 good-risk and 59 patients poor-risk patients in 425 evaluable patients. Figure 2 depicts the effects of the steroid used on the CNS remission rate for both good- and poor-risk patients (see also Table IV). Unlike Table IV, in which the analysis is restricted to patients post-CNS prophylaxis, Figure 2 includes all evaluable patients in marrow
Dexamethasone Reduces CNS Leukemia
273
TABLE 11. Effect of Induction Theraw on CNS Remission*
Chemo
No.
vs
38 52 54 195
AVS VAS VSA
Prednisone Percent M-CR M-CR at CNS 87 83 80 88
Dexamethasone Percent M-CR M-CR at CNS
No.
12 15 74 82
46 49 62 150
85 82 82 89
72 71 13 83
*VS, Vincristine, steroid; AVS, prior asparaginase; VAS, simultaneous asparaginase; VSA, subsequent asparaginase; M-CR, marrow remission; M-CR at CNS, percent still in M-CR at time of CNS treatment.
TABLE 111. Relapse Rate-Dexamethasone Vs. Predisone* Prednisone Observed Expecteda No. T relapses BM relapses CNS relapses Deaths
262 142 106 67 132
Dexamethasone Observed Expected 23 1 122 I05 33 118
145 117 55 136
Two-sided P value
119 93 45 114
0.70 0.12 0.017 0.61
*No., number; T, total; BM, bone marrow; CNS, central nervous system. aExpected, the number expected via the log rank test for equal target population survival curves.
TABLE IV. CNS Relapse Rate* No. CRT iIT MTX 301238 IT MTX 701255 Dexamethasone 331231 Prednisone 671262 CRT effect Within group receiving pred Within group receiving dexa Steroid effect Within group without CRT Within group with CRT
Percent
P-Values (Two-sided)
12.6 27.5 14.3 25.5
