- Email: [email protected]
Central nervous system involvement in acute nonlymphocytic leukemia
Central Nervaus System Involvement in Acute Nonlymphocytic Leukemia A Prospective Study of Adults in Remission
BRUCE A. PETERSON, M.D. RICHARD D. BRUNNING, M.D. CLARA D. BLOOMFIELD, M.D. DAVID D. HURD, M.D. JANE A. GAU, R.N., B.S.N, GRACE T. PENG, M.S. ANNE I. GOLDMAN, Ph.D. Minneapolis,
To identify adults with acute nonlymphocytic leukemia at risk for the development of central nervous system involvement, we performed periodic cerebrospinal fluid examinations on patients in remission. Among 58 consecutive patients monitored during first remission, central nervous system leukemia developed in nine (16 percent). Four patients, including one who was symptomatic, had central nervous system leukemia detected simultaneously with marrow relapse. Five additional patients were asymptomatic and continue to have bone marrow remission. Following central nervous system and systemic treatment, two of these flve patients have never had relapse, and three had relapse in the bone marrow five, 10, and 21 months later. Factors at diagnosls associated with the subsequent development of central nervous system leukemia were elevated leukocyte count, serum lysozyme and lactate dehydrogenase, extramedullary infiltration including splenomegaly, and monocytic (FAB M4 or M5a) morphology. In six of 17 patients (35 percent) with monocytlc morphology, central nervous system leukemia developed compared with only three of 41 patients (7 percent) with other subtypes (p = 0.02). Discriminant analysis identified leukocyte count, splenomegaly, and M4 or M5a morphology as the most important risk factors and led to a mathematical formula that correctly identified 90 percent of the patients. Although the risk of central nervous system leukemia in adults with acute nonlymphocytic leukemia is too low to justify routine prophylaxis, those patients recognized to be at greater risk should receive prophylaxis or be monitored closely with periodic lumbar punctures.
Minnesota
From the Division of Oncology, Department of Medicine, the Department of Laboratory Medicine and Pathology, and the School of Public Health, Division of Biometry, University of Minnesota Health Sciences Center, Minneapolis, Minnesota. This work was supported in part by the Coleman Leukemia Research Fund and the Masonic Memorial Hospital Fund, Inc. Requests for reprints should be addressed to Dr. Bruce A. Peterson, Box 348 UMHC, Division of Oncology, Department of Medicine, University of Minnesota Hospital and Clinic, Minneapolis, Minnesota 55455. Manuscript submitted December 4, 1986, and accepted March 11, 1987.
464
September
1967
The
American
Journal
The high risk of central nervous system relapse in acute lymphoblastic leukemia, particularly in children, became apparent as the results of systemic chemotherapy improved and bone marrow remissions increased in both frequency and duration [I]. Protection was afforded by the development of effective prophylactic central nervous system treatment, and now the majority of children with acute lymphoblastic leukemia who receive appropriate management appear to have been cured. Although it has also been suggested that the importance of the central nervous system as a sanctuary has grown in acute nonlymphocytic leukemia as complete remission rates have risen and durations of response have increased [l-3], studies of central nervous system prophylaxis in adults have not shown benefit with respect to improved remission duration or survival [4-71. This may reflect, in part, the relatively low incidence of involvement in adults with acute nonlymphocytic leukemia. Thus, it is unlikely that administering prophylaxis to every
of Medicine
Volume
63
CENTRAL
patient who enters remission would demonstrably improve therapeutic results for the group as a whole. However, for the individual in whom central nervous system leukemia develops, the consequences are usually grave. We sought to identify those persons who might benefit either from close central nervous system surveillance or prophylactic treatment after the completion of successful induction therapy. In view of the possibilities that: (1) admixing of involved blood or vertebral bone marrow with cerebrospinal fluid during lumbar puncture at diagnosis could result in spurious findings [8]; (2) malignant cells in the cerebrospinal fluid of asymptomatic patients at diagnosis are of uncertain clinical importance [9, IO]; and (3) cytarabine or other drugs that cross the blood brain barrier used systemically during induction chemotherapy may reduce or eradicate the leukemic cells present in the cerebrospinal fluid [ 1 I], we elected to study patients by repeated cerebrospinal fluid examinations during the interval from bone marrow remission to first relapse. PATIENTS AND METHODS Between January 1975 and June 1982, 111 adults with de novo acute nonlymphocytic leukemia received initial treatment at the Coleman Leukemia Research and Treatment Center of the University of Minnesota. Diagnostic subtypes were assigned using criteria of the French-American-British classification [ 121. Induction treatment consisted of cytarabine 100 mg/m* as a continuous infusion for seven days plus either daunorubicin [ 131 or doxorubicin [IO] for three days, and in some cases additional drugs [ 141. Daily daunorubicin plus prednisone for five days [ 151 was used for patients with hypergranular (M3) or microgranular variant (M3v) promyelocytic leukemia, All patients who entered complete remission received maintenance chemotherapy with weekly oral 8-thioguanine and cytarabine [ 181. The criteria described by the Cancer and Leukemia Group B were applied to define remission status [lo]. Seventy-one patients had complete remission. Of these, 13 patients were excluded from study for the following reasons: allogeneic bone marrow transplantation in first remission, two patients; no follow-up cerebrospinal fluid examinations either because of technical problems related to lumbar puncture or the patient was followed elsewhere, four patients; prophylactic central nervous system treatment was instituted at the discretion of the primary physician, seven patients. Fifty-eight patients were available for this study. Specimens of cerebrospinal fluid from lumbar punctures were collected in three tubes and measurements of protein, glucose, and cell count were obtained. Culture specimens were also generally obtained. The specimen in the third tube, the same portion used for cell count, was subjected to cytologic examinatlon. A few drops were cytocentrifuged, and the resultant preparation was stained with the WrightGiemsa stain. The unequivocal indentification of leukemic cells was considered a positive result. The first cerebrospinal fluid examination was planned for the time patients initially entered complete remission;
September
NERVOUS
SYSTEM
LEUKEMIA-PETERSON
ET AL
thereafter, repeat examinations were scheduled at three- to four-month intervals during remission, usually to coincide with routine follow-up bone marrow evaluations, and at the time of relapse in the bone marrow. Relapse in the bone marrow was diagnosed when the marrow myeloblast percentage rose above 5 percent or there were any clearly leukemic forms in the blood or bone marrow. Patients with central nervous system leukemia detected during remission received methotrexate, 12 mg intrathecally, every other day for a minimum of six doses or two doses beyond the first negative result on cytologic examination plus 2,400 rads of cranial irradiation. If patients did not clear with methotrexate or if chemical meningitis developed, cytarabine, 60 mg, was substituted. This therapy was to be followed by six additional monthly injections of either methotrexate or cytarabine and a course of systemic chemotherapy similar to the original induction therapy. Univariate analyses were done to identify clinical factors at diagnosis associated with the development of central nervous system leukemia during complete remission. Hypotheses of no difference in the distribution of discrete variables between two groups were tested using a chisquare test with Yates correction and in the distribution of continuous variables using the Mann-Whitney rank test. Fisher’s exact test was used when the numbers were less than 21. The significance level was a! <0.05. Stepwise discriminant analysis was done to select the risk factors that, in combination, were related to central nervous system leukemia [ 171. The criterion used to select the best set of discriminating variables was Rao’s V, a generalized. distance measure. RESULTS Fifty-eight adults with acute nonlymphocytic leukemia ranging in age from 18 to 79 years had complete remission and underwent cerebrospinal fluid examinations for central nervous system leukemia. Twenty-six were men and 32 were women. Distribution of cases by FrenchAmerican-British subtype was as follows: Ml, eight patients; M2, 27 patients; M3, three patients; M3 variant, one patient; M4, 15 patients; M5a, two patients; and M6, two patients. Four patients (French-American-BritishM3, M3v) had received daunorubicin plus prednisone, and 54 patients received cytarabine and an anthracycline. Overall, the duration of bone marrow complete remission ranged from one to 122 or more months (median, 11 months). Forty-six patients have had relapses in the bone marrow. The median number of diagnostic lumbar punctures was three (range, one to 14). Nine patients (16 percent) had central nervous system leukemia detected at or between the times of first bone marrow remission and relapse (Table I). The median time from complete remission to first detecting central nervous system leukemia for the nine patients was IO months, and the median duration of bone marrow remission in these patients was 16 months. Relapses in the bone marrow and central nervous
1987
The American
Journal
of Medicine
Volume
83
465
CENTRAL
NERVOUS
TABLE I
Patient
SYSTEM
LEUKEMIA-PETERSON
Characteristics
Age (years)
ET AL
at Diagnosis
of Patlents
with Subsequent
Gender
FAB Subtype
Leukocytes (per ~1)
LDH” w
Lysozymet (MN
M M F F M M F F M
M4 M4 M4 M5a M3v M8 M4Eo M2 M4
170,000 149,000 60,000 23,400 40,600 63,500 2,750 248,800 50,700
521 1,806 767 1,599 1,405 584 1,491 1,419
102.0 120.0 62.0 6.6 27.5 27.3 92.0 57.6
55 34 28 18 20 71 66 25 66
FAB = French-American-British classification; * Normal range, 205 to 475 IU. + Normal range, 2.0 to 11.9 fig/ml. t Died in remission.
CNS
= central
nervous
September
1967
The American
Journal
of Medicine
Extramedullary Spleen Nodes
+ + + + + -
system;
Nervous
System Involvement
Infiltration Llver Glngiva
+
+
+
+ + + -t -
+ -
+ + -
-t +
-
-
LDH = lactate
Time lrom Remission to CNS Flndlngs (months)
Over all Duration of Marrow Remission (months)
2 0 IO 3 16 18 11 2 15
3 5 IO 13 16 18 32 80+ 84(+)$
dehydrogenase.
common in the group with central nervous system leukemia when considered individually. Hepatomegaly and skin infiltration, categories with very few patients, were not. Various laboratory findings were also associated with the risk of developing central nervous system leukemia (Table II). The nine patients were characterized by significantly higher median values of initial leukocyte count (6O,OOO/kl versus 6,900/@) and serum levels of lactate dehydrogenase (1,212 IU versus 524 IU) and lysozyme (60 pg/ml versus 9.7 pg/ml). Only one of the nine patients with results that became positive had a leukocyte count under 2O,OOO/pl at diagnosis and only one additional patient had a count of under 4O,OOO/pl. The elevated levels of serum lysozyme are accounted for, in part, by the subtype of leukemia. Diagnoses by French-American-British classification in the nine patients were: M2, one patient; M3, one patient (microgranular variant): M4, five patients; M5a, one patient; and M6, one patient. Thus, the leukemia in six of the nine patients had monocytic morphologic and cytochemical features. The patient (Patient 8) classified as French-American-British M2 also had an elevated lysozyme level. The association of central nervous system leukemia and French-American-British subtype was significant when either all six French-American-British categories were considered separately or the monocytic leukemias, M4 and M5a, were compared with the other subtypes in aggregate. In six of 17 patients (35 percent) with M4 or M5a leukemia central nervous system leukemia developed, compared with only three of 41 patients (7 percent) who had other subtypes (p = 0.02). The only French-American-British subtype in which central nervous system leukemia was not detected was M 1. One patient (Patient 7) had bone marrow eosinophilia (14 percent). Since central nervous system leukemia was common
system were considered to occur simultaneously if the relapse occurring in the central nervous system preceded that in the marrow by less than one month. Four of the nine patients had simultaneous relapses two to 18 months (median, 13 months) after first entering remission. One (Patient 3) presented with a major symptomatic event, hemiparesis, but the others were neurologically asymptomatic . Five patients had malignant cells detected in the cerebrospinal fluid while they were both asymptomatic and fulfilled all criteria for complete remission. One of these patients (Patient 2) had cerebrospinal fluid blasts at the time of initially entering remission; the other four patients had isolated central nervous system recurrences from two to 15 months (median, seven months) into remission. Following retreatment, two of these patients did not have either subsequent bone marrow or central nervous system relapses after an additional 69 and 78 or more months in continuous remission, respectively. The other three patients had bone marrow relapses five, 10, and 2 1 months later and two had central nervous system relapses again prior to first marrow relapse. None of 22 patients followed in first remission for more than 16 months has had an isolated central nervous system relapse. Of the nine patients who had central nervous system leukemia, five were men and four were women. All but one of the patients had received cytarabine during the initial induction period. Although the median age of those with central nervous system leukemia, 34 years versus 54 years, was lower, this was not statistically significant (p = 0.19) (Table II). However, evidence for infiltrative disease at any extramedullary site was associated with the subsequent development of central nervous system leukemia. Specifically, gingival hypertrophy, lymphade nopathy, and splenomegaly were all significantly more
466
Central
Volume
63
CENTRAL
TABLE II
NERVOUS
SYSTEM
LEUKEMIA-PETERSON
ET AL
Comparison of Clinical and Laboratory Characteristics at Diagnosis of Adults with or without Subsequent Central Nervous System Leukemia CentralNervous System Positive (n = 9)
Median age (years) Gender Male Female Extramedullary infiltration Gingival Nodes Skin Liver Spleen Z 1 site FAB Ml 2 3 4 5a 8 Ml-3,6 M4-5a Median hemoglobin (g/dl) Median leukocytes/PI Median platelets/~1 Median lactate dehydrogenase (IU) Median lysozyme @g/ml) FAB
= French-American-British
All FAB Subtypes (n = 58) Negative (n = 49)
Positive (n = 6)
p Value
FAB M4 and M5 (II = 17) Negative (n = 11)
34
54
0.19
45
54
5
21 28
0.73
3 3
6 5
1 10 1 2 3 14
0.007 0.002 0.71 0.21 0.0006 0.002
8 26 3 10 1 1 38 11 9.6 6,900 56,000 524 (n = 40)
0.04
0.90 0.0005 0.20 0.009
9.7 (n = 45)
0.009
4
0 1 1 5 1 1 3 6 10.0 60,000 49,000 1,212 60
(n = 8)
(n = 8)
p Value 0.69
0.03 0.02 0.35 0.03 0.02
5 1
10 1
0.60
10.1 56,000 40,000 1,093
9.6 8,800 70,000 676 (n = IO)
0.19 0.03 0.07 0.16
62 (n = 5)
41 (n = 9)
0.32
0.02
classification.
in patients with acute myelomonocytic or monocytic leukemia, this group of 17 patients was analyzed separately to see if any specific features distinguished those in whom central nervous system leukemia developed from those in whom it did not (Table II). Although serum levels of both lysozyme and lactate dehydrogenase were higher in the central nervous system-positive group, the differences observed were not significant. Higher leukocyte counts, however, were found in the subgroup in whom central nervous system leukemia developed (median, 56,0OO/~l versus 8,8OO/pl; p = 0.03), as was gingival hypertrophy, lymphadenopathy, and splenomegaly. Significant factors selected by stepwise discriminant analysis and listed in the order of their discriminant effect were leukocyte count (F = 12.51; p X0.005), splenomegaly (F = 9.85; p <0.005), and diagnosis M4 or M5a (F = 7.40; p
be at increased risk of the development of central nervous system leukemia. When applied to the study population, this equation identifies with 90 percent accuracy in those in whom central nervous system leukemia actually did develop. COMMENTS With the induction of more frequent and longer remissions in acute nonlymphocytic leukemia and a rising proportion of patients with cure, the risk of central nervous system leukemia and the attendant consequences of neurologic sequelae or possible systemic relapse assume added importance. However, data relevant to the development of a rational strategy of intervention are lacking. Most studies repot-t cases identified only through the recognition of neurologic symptoms; large numbers of asymptomatic adults have not been systematically evaluated. The incidence of overt symptomatic involvement during the entire clinical course in adults is low, generally ranging from 5 to 8 percent [3,8,21], and the termination of complete remission by an overt central nervous system relapse is a rare event [ 18,20-241. Thus, although routine prophylactic treatment of the central nervous system has
September 1987
The AmericanJournalof
Medicine
Volume
83
467
CENTRAL
NERVOUS
SYSTEM
LEUKEMIA-PETERSON
ET AL
variables are interrelated. In a stepwise multivariate analysis, the leukocyte count was the most discriminating variable, but both splenomegaly and morphology added further information. When applied to the study population, these three variables could accurately identify patients at highest risk. Confirmation requires the prospective application of these results to another group of patients. One of the most commonly reported associations of central nervous system leukemia in patients with acute nonlymphocytic leukemia is with monocytic or myelomonocytic leukemia [9,18,21,25,26], but not everyone has recognized such an association [3,27]. In preliminary reports, we pointed out the association of monocytic morphology and elevated serum lysozyme with central nervous system leukemia [I 1,281. All seven cases of central nervous system leukemia documented at the time of diagnosis by Meyer et al [9] had acute myelomonocytic leukemia (M4) by French-American-British criteria. Recently, in an analysis of surface markers in acute nonlymphocytic leukemia, the leukemic cells of eight of 11 cases with central nervous system involvement at diagnosis reacted with MY4, a monoclonal antibody to monocyte surface antigens [29]. In the current study, the association of morphology and central nervous system leukemia was significant. Six of the nine patients with central nervous system leukemia had monocytic subtypes of leukemia, and overall, in six of 17 (35 percent) cases with M4 and M5a leukemia, central nervous system involvement developed. The observed association of monocytic leukemia with the risk of developing central nervous system leukemia may be related to leukemic cell numbers or properties. Within the French-American-British subtypes at greatest risk, a high leukocyte count at diagnosis and evidence of tissue infiltration were related to subsequent central nervous system involvement. The relationship of extreme leukocytosis to monocytic leukemia has been previously reported [30], suggesting extramedullaty involvement could simply be on a random basis. The high leukocyte counts found in eight of our nine patients, including those without a monocytic component, would support this explanation. Similarly, our finding of elevated serum levels of lactate dehydrogenase in patients at risk may reflect a high tumor cell burden [27]. However, the statistical association of subsequent central nervous system leukemia with initial infiltration at other tissue sites, more commonly seen in monocytic leukemias, also suggests that normal functional properties retained by leukemic cells of myelomonocytic or monocytic origin may facilitate the adherence of such cells to vascular endothelium, passage through the vessel wall, and migration into tissue [31]. Holmes et al [32] have reported that a particular cytogenetic abnormality, inv( 16)(p 13q22), and bone marrow eosinophilia in patients with acute myelomonocytic leukemia were associated with relapse in the central nervous
been used in adults, the absence of a demonstrable benefit is not unexpected [4-71. A practical alternative to generally applied prophylaxis is clearly desirable. When the cerebrospinal fluid of newly diagnosed and untreated adults was routinely examined, Meyer et al [9] found central nervous system involvement in seven of 39 adults (18 percent). The actual incidence is most likely lower since in a much larger cooperative group trial, cytologic examinations with positive results were reported in only 8 percent of patients at diagnosis [IO]. Followup information on patients in both reports is limited, leaving the clinical relevance of the findings unknown. An evaluation restricted to the first month of remission in 44 adults found two (4 percent) with involved cerebrospinal fluid [5]. Within the interval from complete remission to first bone marrow relapse, nine of 58 adults (16 percent) in the present series had central nervous system involvement documented through periodic evaluation of cerebrospinal fluid. This rate may be lower than the actual incidence because seven patients were given central nervous system prophylaxis and are not included in the analysis. All cases were identified within 18 months of first entering remission. Five asymptomatic cases were detected during complete remission and four others, three without symptoms, occurred within the month preceding bone marrow relapse. The single symptomatic patient, alone, would represent an incidence similar to that found by others when clinical events are used to identify patients [ 18,20-241. Thus, since the cerebrospinal fluid was directly examined, involvement in this cohort of patients was detected more frequently. Although we detected a significant risk of central nervous system relapse through periodic surveillance, the great majority of adults with acute nonlymphocytic leukemia remain unaffected. Identification of those with early involvement could permit individualized intervention. However, multiple lumbar punctures as utilized in this study are burdensome, and single lumbar punctures, regardless of when performed, are not sufficient. In a preliminary study, we found that in three of four patients who had positive results during induction, the results became negative upon entering remission [ 1 I], an indication that an initial lumbar puncture at diagnosis may not be accurate. In the current study, only one of nine patients in whom central nervous system involvement developed had positive results upon first entering remission. Alternative strategies, such as the determination of risk factors, could define who is at greatest risk of the development of central nervous system leukemia. In an univariate analysis, we found an association of central nervous system involvement with an elevated initial leukocyte count, serum lactate dehydrogenase or lysozyme, M4 or M5a French-American-British subtype, and extramedullary tissue infiltration. Several of these
466
September 1987
The
American
Journal
of Medicine
Volume
83
CENTRAL
system [32]. Most of the patients in our study were diagnosed before results from adequate banded chromosome analyses were routinely obtained. However, only one of the nine patients with central nervous system leukemia had bone marrow eosinophilia (4 percent or more) at diagnosis, and neither of the two adults who were included in the present study and also in the original report of the association of chromosome 16 abnormalities and eosinophilia ever had central nervous system leukemia [33]. Thus, although rearrangement of 16q22 may help identify some patients at risk, other factors are also important. In this study, asymptomatic patients with positive cerebrospinal fluid dytologies were detected through routine surveillance and two of these patients following specific treatment have not had either a subsequent central nervous system or marrow relapse. Factors related to the
NERVOUS
SYSTEM
LEUKEMIA-PETERSON
ET AL
risk of the development of central nervous system leukemia were identified, and these factors have been developed into a simple, highly predictive formula. Thus, patients at greatest risk can be recognized, and surveillance or prophylaxis can be specifically planned. Since current practice in induction or intensification therapy often employs drugs, such as high-dose cytarabine, which are effective in crossing the blood brain barrier and which may provide collateral central nervous system protection [25,34], additional prophylaxis may not be necessary. However, if patients known to be at greatest risk do not receive central nervous system prophylaxis, they should be closely monitored. ACKNOWLEDGMENT We gratefully acknowledge the assistance of Ms. Carol Thomas in the preparation of the manuscript.
REFERENCES 1.
2.
3.
4.
5.
6.
8.
9.
10.
11.
12.
Evans AE, Gilbert ES, Zandstra R: The increasing incidence of central nervous system leukemia in children (children’s cancer study group A). Cancer 1970; 26: 404-409. Dawson DM, Rosenthal DS, Moloney WC: Neurological complications of acute leukemia in adults: changing rate. Ann Intern Med 1973; 79: 541-544. Wolk RW, Masse SR, Conklin R, Freireich EJ: The incidence of central nervous system leukemia in adults with acute leukemia. Cancer 1974; 33: 863-869. Cassileth PA, Katz ME: Chemotherapy for adult acute nonlymphocytic leukemia with daunorubicin and cytosine arabinoside. Cancer Treat Rep 1977; 61: 1441-1445. Eppinger-Helft M, Pavlovsky S, Hidalgo G, et al: Chemoimmunotherapy with corynebacterium parvum in acute myelocytic leukemia. Cancer 1980; 45: 280-284. Gale RP, Foon KA, Cline MJ, Zighelboim J, and the UCLA Acute Leukemia Study Group: Intensive chemotherapy for acute myelogenous leukemia. Ann Intern Med 1981; 94: 753-757. Wiernik PH, Schimpff SC, Schiffer CA, et al: Randomized clinical comparison of daunorubicin (NSC-82151) alone with a combination of daunorubicin, cytosine arabinoside (NSC-63878), 6-thioguanine (NSC-752), and pyrimethamine (NSC-3061) for the treatment of acute nonlymphocytic leukemia. Cancer Treat Rep 1976; 60: 41-53. Kruskall MS, Carter SR, Ritz LP: Contamination of cerebrospinal fluid by vertebral bone-marrow cells during lumbar puncture. N Engl J Med 1983; 308: 697-700. Meyer RJ, Ferreira PPC, Cuttner J, Greenberg ML, Goldberg J, Holland JF: Central nervous system involvement at presentation in acute granulocytic leukemia. A prospective cytocentrifuge study. Am J Med 1980; 68: 691-694. Yates J, Glidewell 0, Wiernik P, et al: Cytosine arabinoside with daunorubicin or Adriamycin for therapy of acute myelocytic leukemia. A CALGB study. Blood 1982; 60: 454-462. Peterson BA, Brunning RD, Bloomfield CD, et al: Central nervous system (CNS) surveillance during complete remission (CR) in acute nonlymphocytic leukemia (ANLL) (abstr). Blood 1983; Gl(suppl 1): 206. Bennett JM, Catovsky D, Daniel M-T, et al: Proposals for the
September
13.
14.
15.
16.
17. 18.
19.
20.
21.
22.
23.
1987
classification of the acute leukaemias. French-AmericanBritish (FAB) co-operative group. Br J Haematol 1976; 33: 451-458. Yates JW, Wallace HJ Jr, Ellison RR, Holland FJ: Cytosine arabinoside (NSC-63878) and daunorubicin (NSC-83142) therapy in acute nonlymphocytic leukemia. Cancer Chemother Rep 1973; 57: 485-488. Peterson BA, Bloomfield CD, Bosl GJ, Gibbs G, Malloy M: Intensive five-drug combination chemotherapy for adult acute non-lymphocytic leukemia. Cancer 1980; 46: 663668. Peterson BA, Bloomfield CD, Theologides A, Kennedy BJ: Daunorubicin-prednisone in the treatment of acute nonlymphocytic leukemia. Cancer Treat Rep 1981; 65(suppl 4): 29-33. Peterson BA, Bloomfield CD: Long-term disease-free survival in acute nonlymphocytic leukemia. Blood 1981; 57: 1144-1147. Gooley WW, Lohnes PR: Multivariate data analysis. New York: Wiley, 1971. Bennett JM, Catovsky D, Daniel M-T, et al: Central nervous system (CNS) involvement in the acute myeloid leukemias (AML): the eastern Cooperative oncology group (ECOG) experience (abstr). Blood 1984; 64: 144. Pavlovsky S, Eppinger-Helft M, Muriel FS: Factors that influence the appearance of central nervous system leukemia. Blood 1983; 42: 935-938. Pippard MJ, Callender ST, Sheldon PWE: Infiltration of central nervous system in adult acute myeloid leukaemia. Br Med J [Clin Res] 1979; 1: 227-229. Weil M, Jacquillat Cl, Gemon-Auclerc MF, et al: Acute granulocytic leukemia. Treatment of the disease. Arch Intern Med 1976; 136: 1389-1395. Armitage JO, Burns CP: Maintenance therapy of adult acute nonlymphocytic leukemia. An argument against the need for central nervous system prophylaxis. Cancer 1978: 41: 697-700. Armitage JO, Burns CP, Dick FR, Aunan SB, Slymen DJ: Results of aggressive chemotherapy for acute nonlymphoblastic leukemia in adults. Cancer Treat Rep 1982; 66: 1917-1924.
The American
Journal
of Medicine
Volume
83
469
CENTRAL
24.
25.
26.
27.
28.
29.
470
NERVOUS
SYSTEM
LEUKEMIA-PETERSON
ET AL
Clarkson BD, Dowling MD, Gee TS, Cunningham IB, Burchenal JH: Treatment of acute leukemia in adults. Cancer 1975; 36: 775-795. Weinstein HJ, Mayer RJ, Rosenthal DS, Coral FS, Camitta BM, Gelber RD: Chemotherapy for acute myelogenous leukemia in children and adults: VAPA update. Blood 1983; 62: 315-319. Tobelem G, Jacquillat C, Chastang C, et al: Acute monoblastic leukemia: a clinical and biologic study of 74 cases. Blood 1980; 55: 7 i-76. Stewart DJ, Keating MJ, McCredie KB, et al: Natural history of central nervous system acute leukemia in adults. Cancer 1981; 47: 184-196. Peterson BA, Bloomfield CD: Asymptomatic central nervous system (CNS) leukemia in adults with acute non-lymphocytic leukemia (ANLL) in extended remission (abstr). Proc Am Assoc Cancer Res Am Sot Clin Oncol 1977; 18: 341. Griffin JD, Davis R, Nelson D, et al: Use of surface marker analysis to predict outcome of adult acute myeloblastic
September
1987
The
American
Journal
of Medicine
30.
31.
32.
33.
34.
Volume
83
leukemia. Blood 1986; 68: 1232-1241. Cuttner J, Conjalka MS, Reilly M, et al: Association of monocytic leukemia in patients with extreme leukocytosis. Am J Med 1980; 69: 555-558. Straus DJ, Mertelsmann R, Koziner B, et al: The acute monocytic leukemias: multidisciplinary studies in 45 patients. Medicine (Baltimore) 1980; 59: 409-425. Holmes R, Keating MJ, Cork A, et al: A unique pattern of central nervous system leukemia in acute myelomonocytic leukemia associated with inv(16)(p13q22). Blood 1985; 65: 1071-1078. Arthur DC, Bloomfield CD: Partial deletion of the long arm of chromosome 16 and bone marrow eosinophilia in acute nonlymphocytic leukemia: a new association. Blood 1983; 61: 994-998. Mayer RJ, Schiffer CA, Peterson BA, et al: Intensive postremission therapy in adults with acute nonlymphocytic leukemia with Ara-C by continuous infusion or bolus administration: preliminary results of a CALGB phase I study. Semin Oncol 1985; 12(suppl3): 84-90.
BRUCE A. PETERSON, M.D. RICHARD D. BRUNNING, M.D. CLARA D. BLOOMFIELD, M.D. DAVID D. HURD, M.D. JANE A. GAU, R.N., B.S.N, GRACE T. PENG, M.S. ANNE I. GOLDMAN, Ph.D. Minneapolis,
To identify adults with acute nonlymphocytic leukemia at risk for the development of central nervous system involvement, we performed periodic cerebrospinal fluid examinations on patients in remission. Among 58 consecutive patients monitored during first remission, central nervous system leukemia developed in nine (16 percent). Four patients, including one who was symptomatic, had central nervous system leukemia detected simultaneously with marrow relapse. Five additional patients were asymptomatic and continue to have bone marrow remission. Following central nervous system and systemic treatment, two of these flve patients have never had relapse, and three had relapse in the bone marrow five, 10, and 21 months later. Factors at diagnosls associated with the subsequent development of central nervous system leukemia were elevated leukocyte count, serum lysozyme and lactate dehydrogenase, extramedullary infiltration including splenomegaly, and monocytic (FAB M4 or M5a) morphology. In six of 17 patients (35 percent) with monocytlc morphology, central nervous system leukemia developed compared with only three of 41 patients (7 percent) with other subtypes (p = 0.02). Discriminant analysis identified leukocyte count, splenomegaly, and M4 or M5a morphology as the most important risk factors and led to a mathematical formula that correctly identified 90 percent of the patients. Although the risk of central nervous system leukemia in adults with acute nonlymphocytic leukemia is too low to justify routine prophylaxis, those patients recognized to be at greater risk should receive prophylaxis or be monitored closely with periodic lumbar punctures.
Minnesota
From the Division of Oncology, Department of Medicine, the Department of Laboratory Medicine and Pathology, and the School of Public Health, Division of Biometry, University of Minnesota Health Sciences Center, Minneapolis, Minnesota. This work was supported in part by the Coleman Leukemia Research Fund and the Masonic Memorial Hospital Fund, Inc. Requests for reprints should be addressed to Dr. Bruce A. Peterson, Box 348 UMHC, Division of Oncology, Department of Medicine, University of Minnesota Hospital and Clinic, Minneapolis, Minnesota 55455. Manuscript submitted December 4, 1986, and accepted March 11, 1987.
464
September
1967
The
American
Journal
The high risk of central nervous system relapse in acute lymphoblastic leukemia, particularly in children, became apparent as the results of systemic chemotherapy improved and bone marrow remissions increased in both frequency and duration [I]. Protection was afforded by the development of effective prophylactic central nervous system treatment, and now the majority of children with acute lymphoblastic leukemia who receive appropriate management appear to have been cured. Although it has also been suggested that the importance of the central nervous system as a sanctuary has grown in acute nonlymphocytic leukemia as complete remission rates have risen and durations of response have increased [l-3], studies of central nervous system prophylaxis in adults have not shown benefit with respect to improved remission duration or survival [4-71. This may reflect, in part, the relatively low incidence of involvement in adults with acute nonlymphocytic leukemia. Thus, it is unlikely that administering prophylaxis to every
of Medicine
Volume
63
CENTRAL
patient who enters remission would demonstrably improve therapeutic results for the group as a whole. However, for the individual in whom central nervous system leukemia develops, the consequences are usually grave. We sought to identify those persons who might benefit either from close central nervous system surveillance or prophylactic treatment after the completion of successful induction therapy. In view of the possibilities that: (1) admixing of involved blood or vertebral bone marrow with cerebrospinal fluid during lumbar puncture at diagnosis could result in spurious findings [8]; (2) malignant cells in the cerebrospinal fluid of asymptomatic patients at diagnosis are of uncertain clinical importance [9, IO]; and (3) cytarabine or other drugs that cross the blood brain barrier used systemically during induction chemotherapy may reduce or eradicate the leukemic cells present in the cerebrospinal fluid [ 1 I], we elected to study patients by repeated cerebrospinal fluid examinations during the interval from bone marrow remission to first relapse. PATIENTS AND METHODS Between January 1975 and June 1982, 111 adults with de novo acute nonlymphocytic leukemia received initial treatment at the Coleman Leukemia Research and Treatment Center of the University of Minnesota. Diagnostic subtypes were assigned using criteria of the French-American-British classification [ 121. Induction treatment consisted of cytarabine 100 mg/m* as a continuous infusion for seven days plus either daunorubicin [ 131 or doxorubicin [IO] for three days, and in some cases additional drugs [ 141. Daily daunorubicin plus prednisone for five days [ 151 was used for patients with hypergranular (M3) or microgranular variant (M3v) promyelocytic leukemia, All patients who entered complete remission received maintenance chemotherapy with weekly oral 8-thioguanine and cytarabine [ 181. The criteria described by the Cancer and Leukemia Group B were applied to define remission status [lo]. Seventy-one patients had complete remission. Of these, 13 patients were excluded from study for the following reasons: allogeneic bone marrow transplantation in first remission, two patients; no follow-up cerebrospinal fluid examinations either because of technical problems related to lumbar puncture or the patient was followed elsewhere, four patients; prophylactic central nervous system treatment was instituted at the discretion of the primary physician, seven patients. Fifty-eight patients were available for this study. Specimens of cerebrospinal fluid from lumbar punctures were collected in three tubes and measurements of protein, glucose, and cell count were obtained. Culture specimens were also generally obtained. The specimen in the third tube, the same portion used for cell count, was subjected to cytologic examinatlon. A few drops were cytocentrifuged, and the resultant preparation was stained with the WrightGiemsa stain. The unequivocal indentification of leukemic cells was considered a positive result. The first cerebrospinal fluid examination was planned for the time patients initially entered complete remission;
September
NERVOUS
SYSTEM
LEUKEMIA-PETERSON
ET AL
thereafter, repeat examinations were scheduled at three- to four-month intervals during remission, usually to coincide with routine follow-up bone marrow evaluations, and at the time of relapse in the bone marrow. Relapse in the bone marrow was diagnosed when the marrow myeloblast percentage rose above 5 percent or there were any clearly leukemic forms in the blood or bone marrow. Patients with central nervous system leukemia detected during remission received methotrexate, 12 mg intrathecally, every other day for a minimum of six doses or two doses beyond the first negative result on cytologic examination plus 2,400 rads of cranial irradiation. If patients did not clear with methotrexate or if chemical meningitis developed, cytarabine, 60 mg, was substituted. This therapy was to be followed by six additional monthly injections of either methotrexate or cytarabine and a course of systemic chemotherapy similar to the original induction therapy. Univariate analyses were done to identify clinical factors at diagnosis associated with the development of central nervous system leukemia during complete remission. Hypotheses of no difference in the distribution of discrete variables between two groups were tested using a chisquare test with Yates correction and in the distribution of continuous variables using the Mann-Whitney rank test. Fisher’s exact test was used when the numbers were less than 21. The significance level was a! <0.05. Stepwise discriminant analysis was done to select the risk factors that, in combination, were related to central nervous system leukemia [ 171. The criterion used to select the best set of discriminating variables was Rao’s V, a generalized. distance measure. RESULTS Fifty-eight adults with acute nonlymphocytic leukemia ranging in age from 18 to 79 years had complete remission and underwent cerebrospinal fluid examinations for central nervous system leukemia. Twenty-six were men and 32 were women. Distribution of cases by FrenchAmerican-British subtype was as follows: Ml, eight patients; M2, 27 patients; M3, three patients; M3 variant, one patient; M4, 15 patients; M5a, two patients; and M6, two patients. Four patients (French-American-BritishM3, M3v) had received daunorubicin plus prednisone, and 54 patients received cytarabine and an anthracycline. Overall, the duration of bone marrow complete remission ranged from one to 122 or more months (median, 11 months). Forty-six patients have had relapses in the bone marrow. The median number of diagnostic lumbar punctures was three (range, one to 14). Nine patients (16 percent) had central nervous system leukemia detected at or between the times of first bone marrow remission and relapse (Table I). The median time from complete remission to first detecting central nervous system leukemia for the nine patients was IO months, and the median duration of bone marrow remission in these patients was 16 months. Relapses in the bone marrow and central nervous
1987
The American
Journal
of Medicine
Volume
83
465
CENTRAL
NERVOUS
TABLE I
Patient
SYSTEM
LEUKEMIA-PETERSON
Characteristics
Age (years)
ET AL
at Diagnosis
of Patlents
with Subsequent
Gender
FAB Subtype
Leukocytes (per ~1)
LDH” w
Lysozymet (MN
M M F F M M F F M
M4 M4 M4 M5a M3v M8 M4Eo M2 M4
170,000 149,000 60,000 23,400 40,600 63,500 2,750 248,800 50,700
521 1,806 767 1,599 1,405 584 1,491 1,419
102.0 120.0 62.0 6.6 27.5 27.3 92.0 57.6
55 34 28 18 20 71 66 25 66
FAB = French-American-British classification; * Normal range, 205 to 475 IU. + Normal range, 2.0 to 11.9 fig/ml. t Died in remission.
CNS
= central
nervous
September
1967
The American
Journal
of Medicine
Extramedullary Spleen Nodes
+ + + + + -
system;
Nervous
System Involvement
Infiltration Llver Glngiva
+
+
+
+ + + -t -
+ -
+ + -
-t +
-
-
LDH = lactate
Time lrom Remission to CNS Flndlngs (months)
Over all Duration of Marrow Remission (months)
2 0 IO 3 16 18 11 2 15
3 5 IO 13 16 18 32 80+ 84(+)$
dehydrogenase.
common in the group with central nervous system leukemia when considered individually. Hepatomegaly and skin infiltration, categories with very few patients, were not. Various laboratory findings were also associated with the risk of developing central nervous system leukemia (Table II). The nine patients were characterized by significantly higher median values of initial leukocyte count (6O,OOO/kl versus 6,900/@) and serum levels of lactate dehydrogenase (1,212 IU versus 524 IU) and lysozyme (60 pg/ml versus 9.7 pg/ml). Only one of the nine patients with results that became positive had a leukocyte count under 2O,OOO/pl at diagnosis and only one additional patient had a count of under 4O,OOO/pl. The elevated levels of serum lysozyme are accounted for, in part, by the subtype of leukemia. Diagnoses by French-American-British classification in the nine patients were: M2, one patient; M3, one patient (microgranular variant): M4, five patients; M5a, one patient; and M6, one patient. Thus, the leukemia in six of the nine patients had monocytic morphologic and cytochemical features. The patient (Patient 8) classified as French-American-British M2 also had an elevated lysozyme level. The association of central nervous system leukemia and French-American-British subtype was significant when either all six French-American-British categories were considered separately or the monocytic leukemias, M4 and M5a, were compared with the other subtypes in aggregate. In six of 17 patients (35 percent) with M4 or M5a leukemia central nervous system leukemia developed, compared with only three of 41 patients (7 percent) who had other subtypes (p = 0.02). The only French-American-British subtype in which central nervous system leukemia was not detected was M 1. One patient (Patient 7) had bone marrow eosinophilia (14 percent). Since central nervous system leukemia was common
system were considered to occur simultaneously if the relapse occurring in the central nervous system preceded that in the marrow by less than one month. Four of the nine patients had simultaneous relapses two to 18 months (median, 13 months) after first entering remission. One (Patient 3) presented with a major symptomatic event, hemiparesis, but the others were neurologically asymptomatic . Five patients had malignant cells detected in the cerebrospinal fluid while they were both asymptomatic and fulfilled all criteria for complete remission. One of these patients (Patient 2) had cerebrospinal fluid blasts at the time of initially entering remission; the other four patients had isolated central nervous system recurrences from two to 15 months (median, seven months) into remission. Following retreatment, two of these patients did not have either subsequent bone marrow or central nervous system relapses after an additional 69 and 78 or more months in continuous remission, respectively. The other three patients had bone marrow relapses five, 10, and 2 1 months later and two had central nervous system relapses again prior to first marrow relapse. None of 22 patients followed in first remission for more than 16 months has had an isolated central nervous system relapse. Of the nine patients who had central nervous system leukemia, five were men and four were women. All but one of the patients had received cytarabine during the initial induction period. Although the median age of those with central nervous system leukemia, 34 years versus 54 years, was lower, this was not statistically significant (p = 0.19) (Table II). However, evidence for infiltrative disease at any extramedullary site was associated with the subsequent development of central nervous system leukemia. Specifically, gingival hypertrophy, lymphade nopathy, and splenomegaly were all significantly more
466
Central
Volume
63
CENTRAL
TABLE II
NERVOUS
SYSTEM
LEUKEMIA-PETERSON
ET AL
Comparison of Clinical and Laboratory Characteristics at Diagnosis of Adults with or without Subsequent Central Nervous System Leukemia CentralNervous System Positive (n = 9)
Median age (years) Gender Male Female Extramedullary infiltration Gingival Nodes Skin Liver Spleen Z 1 site FAB Ml 2 3 4 5a 8 Ml-3,6 M4-5a Median hemoglobin (g/dl) Median leukocytes/PI Median platelets/~1 Median lactate dehydrogenase (IU) Median lysozyme @g/ml) FAB
= French-American-British
All FAB Subtypes (n = 58) Negative (n = 49)
Positive (n = 6)
p Value
FAB M4 and M5 (II = 17) Negative (n = 11)
34
54
0.19
45
54
5
21 28
0.73
3 3
6 5
1 10 1 2 3 14
0.007 0.002 0.71 0.21 0.0006 0.002
8 26 3 10 1 1 38 11 9.6 6,900 56,000 524 (n = 40)
0.04
0.90 0.0005 0.20 0.009
9.7 (n = 45)
0.009
4
0 1 1 5 1 1 3 6 10.0 60,000 49,000 1,212 60
(n = 8)
(n = 8)
p Value 0.69
0.03 0.02 0.35 0.03 0.02
5 1
10 1
0.60
10.1 56,000 40,000 1,093
9.6 8,800 70,000 676 (n = IO)
0.19 0.03 0.07 0.16
62 (n = 5)
41 (n = 9)
0.32
0.02
classification.
in patients with acute myelomonocytic or monocytic leukemia, this group of 17 patients was analyzed separately to see if any specific features distinguished those in whom central nervous system leukemia developed from those in whom it did not (Table II). Although serum levels of both lysozyme and lactate dehydrogenase were higher in the central nervous system-positive group, the differences observed were not significant. Higher leukocyte counts, however, were found in the subgroup in whom central nervous system leukemia developed (median, 56,0OO/~l versus 8,8OO/pl; p = 0.03), as was gingival hypertrophy, lymphadenopathy, and splenomegaly. Significant factors selected by stepwise discriminant analysis and listed in the order of their discriminant effect were leukocyte count (F = 12.51; p X0.005), splenomegaly (F = 9.85; p <0.005), and diagnosis M4 or M5a (F = 7.40; p
be at increased risk of the development of central nervous system leukemia. When applied to the study population, this equation identifies with 90 percent accuracy in those in whom central nervous system leukemia actually did develop. COMMENTS With the induction of more frequent and longer remissions in acute nonlymphocytic leukemia and a rising proportion of patients with cure, the risk of central nervous system leukemia and the attendant consequences of neurologic sequelae or possible systemic relapse assume added importance. However, data relevant to the development of a rational strategy of intervention are lacking. Most studies repot-t cases identified only through the recognition of neurologic symptoms; large numbers of asymptomatic adults have not been systematically evaluated. The incidence of overt symptomatic involvement during the entire clinical course in adults is low, generally ranging from 5 to 8 percent [3,8,21], and the termination of complete remission by an overt central nervous system relapse is a rare event [ 18,20-241. Thus, although routine prophylactic treatment of the central nervous system has
September 1987
The AmericanJournalof
Medicine
Volume
83
467
CENTRAL
NERVOUS
SYSTEM
LEUKEMIA-PETERSON
ET AL
variables are interrelated. In a stepwise multivariate analysis, the leukocyte count was the most discriminating variable, but both splenomegaly and morphology added further information. When applied to the study population, these three variables could accurately identify patients at highest risk. Confirmation requires the prospective application of these results to another group of patients. One of the most commonly reported associations of central nervous system leukemia in patients with acute nonlymphocytic leukemia is with monocytic or myelomonocytic leukemia [9,18,21,25,26], but not everyone has recognized such an association [3,27]. In preliminary reports, we pointed out the association of monocytic morphology and elevated serum lysozyme with central nervous system leukemia [I 1,281. All seven cases of central nervous system leukemia documented at the time of diagnosis by Meyer et al [9] had acute myelomonocytic leukemia (M4) by French-American-British criteria. Recently, in an analysis of surface markers in acute nonlymphocytic leukemia, the leukemic cells of eight of 11 cases with central nervous system involvement at diagnosis reacted with MY4, a monoclonal antibody to monocyte surface antigens [29]. In the current study, the association of morphology and central nervous system leukemia was significant. Six of the nine patients with central nervous system leukemia had monocytic subtypes of leukemia, and overall, in six of 17 (35 percent) cases with M4 and M5a leukemia, central nervous system involvement developed. The observed association of monocytic leukemia with the risk of developing central nervous system leukemia may be related to leukemic cell numbers or properties. Within the French-American-British subtypes at greatest risk, a high leukocyte count at diagnosis and evidence of tissue infiltration were related to subsequent central nervous system involvement. The relationship of extreme leukocytosis to monocytic leukemia has been previously reported [30], suggesting extramedullaty involvement could simply be on a random basis. The high leukocyte counts found in eight of our nine patients, including those without a monocytic component, would support this explanation. Similarly, our finding of elevated serum levels of lactate dehydrogenase in patients at risk may reflect a high tumor cell burden [27]. However, the statistical association of subsequent central nervous system leukemia with initial infiltration at other tissue sites, more commonly seen in monocytic leukemias, also suggests that normal functional properties retained by leukemic cells of myelomonocytic or monocytic origin may facilitate the adherence of such cells to vascular endothelium, passage through the vessel wall, and migration into tissue [31]. Holmes et al [32] have reported that a particular cytogenetic abnormality, inv( 16)(p 13q22), and bone marrow eosinophilia in patients with acute myelomonocytic leukemia were associated with relapse in the central nervous
been used in adults, the absence of a demonstrable benefit is not unexpected [4-71. A practical alternative to generally applied prophylaxis is clearly desirable. When the cerebrospinal fluid of newly diagnosed and untreated adults was routinely examined, Meyer et al [9] found central nervous system involvement in seven of 39 adults (18 percent). The actual incidence is most likely lower since in a much larger cooperative group trial, cytologic examinations with positive results were reported in only 8 percent of patients at diagnosis [IO]. Followup information on patients in both reports is limited, leaving the clinical relevance of the findings unknown. An evaluation restricted to the first month of remission in 44 adults found two (4 percent) with involved cerebrospinal fluid [5]. Within the interval from complete remission to first bone marrow relapse, nine of 58 adults (16 percent) in the present series had central nervous system involvement documented through periodic evaluation of cerebrospinal fluid. This rate may be lower than the actual incidence because seven patients were given central nervous system prophylaxis and are not included in the analysis. All cases were identified within 18 months of first entering remission. Five asymptomatic cases were detected during complete remission and four others, three without symptoms, occurred within the month preceding bone marrow relapse. The single symptomatic patient, alone, would represent an incidence similar to that found by others when clinical events are used to identify patients [ 18,20-241. Thus, since the cerebrospinal fluid was directly examined, involvement in this cohort of patients was detected more frequently. Although we detected a significant risk of central nervous system relapse through periodic surveillance, the great majority of adults with acute nonlymphocytic leukemia remain unaffected. Identification of those with early involvement could permit individualized intervention. However, multiple lumbar punctures as utilized in this study are burdensome, and single lumbar punctures, regardless of when performed, are not sufficient. In a preliminary study, we found that in three of four patients who had positive results during induction, the results became negative upon entering remission [ 1 I], an indication that an initial lumbar puncture at diagnosis may not be accurate. In the current study, only one of nine patients in whom central nervous system involvement developed had positive results upon first entering remission. Alternative strategies, such as the determination of risk factors, could define who is at greatest risk of the development of central nervous system leukemia. In an univariate analysis, we found an association of central nervous system involvement with an elevated initial leukocyte count, serum lactate dehydrogenase or lysozyme, M4 or M5a French-American-British subtype, and extramedullary tissue infiltration. Several of these
466
September 1987
The
American
Journal
of Medicine
Volume
83
CENTRAL
system [32]. Most of the patients in our study were diagnosed before results from adequate banded chromosome analyses were routinely obtained. However, only one of the nine patients with central nervous system leukemia had bone marrow eosinophilia (4 percent or more) at diagnosis, and neither of the two adults who were included in the present study and also in the original report of the association of chromosome 16 abnormalities and eosinophilia ever had central nervous system leukemia [33]. Thus, although rearrangement of 16q22 may help identify some patients at risk, other factors are also important. In this study, asymptomatic patients with positive cerebrospinal fluid dytologies were detected through routine surveillance and two of these patients following specific treatment have not had either a subsequent central nervous system or marrow relapse. Factors related to the
NERVOUS
SYSTEM
LEUKEMIA-PETERSON
ET AL
risk of the development of central nervous system leukemia were identified, and these factors have been developed into a simple, highly predictive formula. Thus, patients at greatest risk can be recognized, and surveillance or prophylaxis can be specifically planned. Since current practice in induction or intensification therapy often employs drugs, such as high-dose cytarabine, which are effective in crossing the blood brain barrier and which may provide collateral central nervous system protection [25,34], additional prophylaxis may not be necessary. However, if patients known to be at greatest risk do not receive central nervous system prophylaxis, they should be closely monitored. ACKNOWLEDGMENT We gratefully acknowledge the assistance of Ms. Carol Thomas in the preparation of the manuscript.
REFERENCES 1.
2.
3.
4.
5.
6.
8.
9.
10.
11.
12.
Evans AE, Gilbert ES, Zandstra R: The increasing incidence of central nervous system leukemia in children (children’s cancer study group A). Cancer 1970; 26: 404-409. Dawson DM, Rosenthal DS, Moloney WC: Neurological complications of acute leukemia in adults: changing rate. Ann Intern Med 1973; 79: 541-544. Wolk RW, Masse SR, Conklin R, Freireich EJ: The incidence of central nervous system leukemia in adults with acute leukemia. Cancer 1974; 33: 863-869. Cassileth PA, Katz ME: Chemotherapy for adult acute nonlymphocytic leukemia with daunorubicin and cytosine arabinoside. Cancer Treat Rep 1977; 61: 1441-1445. Eppinger-Helft M, Pavlovsky S, Hidalgo G, et al: Chemoimmunotherapy with corynebacterium parvum in acute myelocytic leukemia. Cancer 1980; 45: 280-284. Gale RP, Foon KA, Cline MJ, Zighelboim J, and the UCLA Acute Leukemia Study Group: Intensive chemotherapy for acute myelogenous leukemia. Ann Intern Med 1981; 94: 753-757. Wiernik PH, Schimpff SC, Schiffer CA, et al: Randomized clinical comparison of daunorubicin (NSC-82151) alone with a combination of daunorubicin, cytosine arabinoside (NSC-63878), 6-thioguanine (NSC-752), and pyrimethamine (NSC-3061) for the treatment of acute nonlymphocytic leukemia. Cancer Treat Rep 1976; 60: 41-53. Kruskall MS, Carter SR, Ritz LP: Contamination of cerebrospinal fluid by vertebral bone-marrow cells during lumbar puncture. N Engl J Med 1983; 308: 697-700. Meyer RJ, Ferreira PPC, Cuttner J, Greenberg ML, Goldberg J, Holland JF: Central nervous system involvement at presentation in acute granulocytic leukemia. A prospective cytocentrifuge study. Am J Med 1980; 68: 691-694. Yates J, Glidewell 0, Wiernik P, et al: Cytosine arabinoside with daunorubicin or Adriamycin for therapy of acute myelocytic leukemia. A CALGB study. Blood 1982; 60: 454-462. Peterson BA, Brunning RD, Bloomfield CD, et al: Central nervous system (CNS) surveillance during complete remission (CR) in acute nonlymphocytic leukemia (ANLL) (abstr). Blood 1983; Gl(suppl 1): 206. Bennett JM, Catovsky D, Daniel M-T, et al: Proposals for the
September
13.
14.
15.
16.
17. 18.
19.
20.
21.
22.
23.
1987
classification of the acute leukaemias. French-AmericanBritish (FAB) co-operative group. Br J Haematol 1976; 33: 451-458. Yates JW, Wallace HJ Jr, Ellison RR, Holland FJ: Cytosine arabinoside (NSC-63878) and daunorubicin (NSC-83142) therapy in acute nonlymphocytic leukemia. Cancer Chemother Rep 1973; 57: 485-488. Peterson BA, Bloomfield CD, Bosl GJ, Gibbs G, Malloy M: Intensive five-drug combination chemotherapy for adult acute non-lymphocytic leukemia. Cancer 1980; 46: 663668. Peterson BA, Bloomfield CD, Theologides A, Kennedy BJ: Daunorubicin-prednisone in the treatment of acute nonlymphocytic leukemia. Cancer Treat Rep 1981; 65(suppl 4): 29-33. Peterson BA, Bloomfield CD: Long-term disease-free survival in acute nonlymphocytic leukemia. Blood 1981; 57: 1144-1147. Gooley WW, Lohnes PR: Multivariate data analysis. New York: Wiley, 1971. Bennett JM, Catovsky D, Daniel M-T, et al: Central nervous system (CNS) involvement in the acute myeloid leukemias (AML): the eastern Cooperative oncology group (ECOG) experience (abstr). Blood 1984; 64: 144. Pavlovsky S, Eppinger-Helft M, Muriel FS: Factors that influence the appearance of central nervous system leukemia. Blood 1983; 42: 935-938. Pippard MJ, Callender ST, Sheldon PWE: Infiltration of central nervous system in adult acute myeloid leukaemia. Br Med J [Clin Res] 1979; 1: 227-229. Weil M, Jacquillat Cl, Gemon-Auclerc MF, et al: Acute granulocytic leukemia. Treatment of the disease. Arch Intern Med 1976; 136: 1389-1395. Armitage JO, Burns CP: Maintenance therapy of adult acute nonlymphocytic leukemia. An argument against the need for central nervous system prophylaxis. Cancer 1978: 41: 697-700. Armitage JO, Burns CP, Dick FR, Aunan SB, Slymen DJ: Results of aggressive chemotherapy for acute nonlymphoblastic leukemia in adults. Cancer Treat Rep 1982; 66: 1917-1924.
The American
Journal
of Medicine
Volume
83
469
CENTRAL
24.
25.
26.
27.
28.
29.
470
NERVOUS
SYSTEM
LEUKEMIA-PETERSON
ET AL
Clarkson BD, Dowling MD, Gee TS, Cunningham IB, Burchenal JH: Treatment of acute leukemia in adults. Cancer 1975; 36: 775-795. Weinstein HJ, Mayer RJ, Rosenthal DS, Coral FS, Camitta BM, Gelber RD: Chemotherapy for acute myelogenous leukemia in children and adults: VAPA update. Blood 1983; 62: 315-319. Tobelem G, Jacquillat C, Chastang C, et al: Acute monoblastic leukemia: a clinical and biologic study of 74 cases. Blood 1980; 55: 7 i-76. Stewart DJ, Keating MJ, McCredie KB, et al: Natural history of central nervous system acute leukemia in adults. Cancer 1981; 47: 184-196. Peterson BA, Bloomfield CD: Asymptomatic central nervous system (CNS) leukemia in adults with acute non-lymphocytic leukemia (ANLL) in extended remission (abstr). Proc Am Assoc Cancer Res Am Sot Clin Oncol 1977; 18: 341. Griffin JD, Davis R, Nelson D, et al: Use of surface marker analysis to predict outcome of adult acute myeloblastic
September
1987
The
American
Journal
of Medicine
30.
31.
32.
33.
34.
Volume
83
leukemia. Blood 1986; 68: 1232-1241. Cuttner J, Conjalka MS, Reilly M, et al: Association of monocytic leukemia in patients with extreme leukocytosis. Am J Med 1980; 69: 555-558. Straus DJ, Mertelsmann R, Koziner B, et al: The acute monocytic leukemias: multidisciplinary studies in 45 patients. Medicine (Baltimore) 1980; 59: 409-425. Holmes R, Keating MJ, Cork A, et al: A unique pattern of central nervous system leukemia in acute myelomonocytic leukemia associated with inv(16)(p13q22). Blood 1985; 65: 1071-1078. Arthur DC, Bloomfield CD: Partial deletion of the long arm of chromosome 16 and bone marrow eosinophilia in acute nonlymphocytic leukemia: a new association. Blood 1983; 61: 994-998. Mayer RJ, Schiffer CA, Peterson BA, et al: Intensive postremission therapy in adults with acute nonlymphocytic leukemia with Ara-C by continuous infusion or bolus administration: preliminary results of a CALGB phase I study. Semin Oncol 1985; 12(suppl3): 84-90.