Pulmonary Leukostasis As the Single Worst Prognostic Factor in

Patients with Acute Myelocytic Leukemia and Hyperleukocytosis

THOMAS J. LESTER, M.D. Patients with acute myelocytic leukemia and hyperleukocytosis
JAMES W. JOHNSON, M.S. have a poor prognosis due to vascular leukostasis and infiltration
JANET CUTTNER, M.D. in the central nervous system and lungs. The clinical records of all
New York, New York patients with a new diagnosis of acute myelocytic leukemia and
initial white blood cell count greater than 100,000 X log/liter ad-
mitted to the Mount Sinai Medical Center between the years 1974
and 1983 were examined. Forty-three patients were identified, 22
of whom had clinical and/or pathologic evidence of leukostasis of
the central nervous system and/or lung. All patients received in-
duction chemotherapy with daunorubicin and a continuous infusion
of cytosine arabinoside. Thirty-five patients also underwent thera-
peutic leukapheresis prior to induction chemotherapy. The overall
remission induction rate in these 43 patients was 51 percent. Fifteen
patients had lung leukostasis; the remission rate for these patients
was 27 percent (three complete remlssions, one partial remission),
as compared with a remission rate of 64 percent (18 of 28) for those
without pulmonary leukostasis (x: = 5.53; p = 0.02). Thirteen pa-
tients had central nervous system infiltration; the remission rate for
these patients was 46 percent (five complete remissions, one partial
remission), as compared with 53 percent (16 of 30) for patients
without central nervous system involvement (x: = 0.19; p = 0.67).
The median survival of 21 patients without leukostasls was 10.8
months, as compared with 15.4 months for seven patients with
central nervous system involvement and no lung leukostasis and 0.2
months for 15 patients with pulmonary leukostasis (xz = 19.9; p
<O.OOl). Thus, pulmonary leukostasis was found to be the single
worst prognostic factor in this group of patients.

Current management of acute myelocytic leukemia consisting of

chemotherapy with cytosine arabinoside and an anthracycline, empiric
antibiotics, and supportive platelet and red cell transfusions results
in a remission induction rate of 60 to 70 percent [i-3]. Stratification
of patients according to prognostic factors at presentation is essential
From the Polly Annenberg Levee Division of He-
to identify groups for more aggressive therapy.
matology, Department of Medicine, Mount Sinai Patients with acute myelocytic leukemia and hyperleukocytosis are
Medical Center, New York, New York. This work known to have a poor prognosis [4-61. Leukocyte aggregates in the
was supported in part by the Nancy Rosengarten lung and central nervous system with organ infiltration and vascular
Leukemia Fund. Requests for reprints should be rupture contribute in a significant manner to the high mortality rate.
addressed to Dr. Janet Cuttner, Mount Sinai
Medical Center, One Gustave L. Levy Place, New The relative contribution of leukemic involvement in these vital organs
York, New York 10029. Manuscript accepted to the overall response rate and survival has not previously been re-
November 19, 1984. ported.

July 1985 The American Journal of Medlclne Volume 79 43

PULMONARY LEUKOSTASIS IN ACUTE MYELOCYTIC LEUKEMIA-LESTER ET AL

PATIENTS AND METHODS treated with cytosine arabinoside and 6-thioguanine.

The clinical records of all patients with newly diagnosed acute Thirty-five patients (81 percent) underwent therapeutic
myelocytic leukemia and initial white blood cell counts leukapheresis. Twenty-one patients (49 percent) had
greater than 100,000 X log/liter admitted to the Mount Sinai no response to treatment, 19 (44 percent) had complete
Medical Center between 1974 and 1983 were examined for remission, and three (seven percent) had a partial re-
prognostic variables. The degree of leukemic infiltration of sponse and were discharged from the hospital with
the liver, spleen, gums, lung, and central nervous system was evidence of persistent disease.
assessed. Leukemic blasts in the cerebrospinal fluid by cy- There was no significant difference in response rate
tospin analysis were used as evidence of infiltration. Intra- or survival related to the presence of organomegaly.
cranial hemorrhage in the presence of thrombocytopenia
Ten of 12 patients (83 percent) with gingival hypertrophy
without blasts in cerebrospinal fluid was not taken as evi-
attained remission status, as compared with 12 of 31
dence of leukostasis. The diagnosis of pulmonary leukostasis
was made by a constellation of clinical criteria as follows: (31 percent) without hypertrophy (XT = 6.89; p = 0.01).
dyspnea with tachypnea, diffuse pulmonary rales, nonhyp- The distributions of the degree of hyperleukocytosis in
ercapneic respiratory failure, and diffuse interstitial infiltrates patients with response versus that for patients without
on chest radiography. As in the central nervous system, response were compared using the chi-square ap-
pulmonary hemorrhage in and of itself, in the presence of proximation to the Kruskal-Wallis tests; there was no
thrombocytopenia, was not used as evidence of leukostasis. significant difference.
All patients were categorized according to the French- Twenty-two patients (51 percent) had leukemic in-
American-British (FAB) classification [7]. induction che- volvement of the lung and/or central nervous system;
motherapy consisted of seven days of continuous-infusion sixteen (73 percent) of these were classified as having
cytosine arabinoside and three days of daunorubicin. Patients
monocytic leukemia (M4 or M5). Twenty-one patients
who did not have a remission received a second course of
the same chemotherapy for five and two days, respectively. (49 percent) had no evidence of disease in the lungs or
One patient was treated with cytosine arabinoside and 6- central nervous system; eleven (52 percent) of these
thioguanine. Many of the patients underwent therapeutic had monocytic leukemia.
leukapheresis as part of an independent study [8]. Patients Fifteen patients (35 percent) had pulmonary leuko-
received antibiotics, red blood cells, and platelet transfusions stasis; four (27 percent) of these patients had remission
as indicated clinically. The results of treatment were recorded (three complete, one partial) as compared with 18 of
as complete, partial, or no remission as defined by the Cancer 28 (64 percent) without pulmonary leukostasis (x: =
and Leukemia Group B criteria. A few patients died before 5.53; p = 0.02). Six patients (14 percent) had leukemic
any treatment could be administered. The time to response involvement of the lung and central nervous system; one
and the number of courses of chemotherapy required were
(17 percent) had partial remission and five (83 percent)
recorded. No patient with a presenting white blood cell count
greater than 100,000 X 10g/liter was excluded from analysis. had no response. This group did not differ significantly
Follow-up data on all patients are available-several remain in response rate from those patients with pulmonary
alive and are examined periodically at Mount Sinai Hos- leukostasis only. Seven patients (16 percent) had in-
pital. filtration of the central nervous system without pulmo-
nary leukostasis; five (71 percent) had complete re-
RESULTS mission and two (29 percent) had no response. The
We identified 43 patients with a white blood cell count remission induction rate in those patients with central
greater than 100,000 X log/liter on presentation (see nervous system involvement was not significantly dif-
Table I). Twenty-five patients (58 percent) were male. ferent from those without central nervous system in-
The mean age was 48.7 years (SD = 18.3 years). The volvement (x: = 0.19; p = 0.67).
youngest patient was 12 years old, the oldest 76. The Forty patients have died; the median survival for the
mean white blood cell count was 195,000 X log/liter entire group was 4.4 months (see Figure 1). Three
(SD = 92,000 X log/liter). The highest recorded white deaths occurred within 24 hours of diagnosis; all of
blood cell count was 490,000 X log/liter. these patients had pulmonary leukostasis. Eight patients
Twelve patients (28 percent) had gingival hypertro- died within five days of diagnosis. Seven of these had
phy, 19 (44 percent) had splenomegaly, 21 (49 percent) pulmonary leukostasis, which was confirmed on post-
had hepatomegaly, 13 (30 percent) had central nervous mortem examination in two. Three patients are alive and
system infiltration, and 15 (35 percent) had clinical have complete remission eight, 20, and 68 months after
evidence of pulmonary leukostasis. Seven patients (16 diagnosis. Median survival in the 35 patients who un-
percent) received no chemotherapy, 25 (58 percent) derwent therapeutic leukapheresis was 6.4 months, and
received the seven-day cytosine arabinosideithreeday median survival was 0.6 months for those who did not
daunorubicin course, 10 (23 percent) required a second undergo leukapheresis. Comparison of the survival
course of the five-day cytosine arabinoside/two-day curves shows that survival was improved for those
daunorubicin regimen, and one patient (2 percent) was undergoing leukapheresis (Wilcoxon x: = 2.67, p =

44 July 1985 The American Journal of Medicine Volume 79

 PULMONARY LEUKOSTASIS IN ACUTE MYELOCYTIC LEUKEMIA-LESTER ET AL

TABLE I Data on Patients with Acute Myelocytic Leukemia and Hyperleukocytosis

Presence* (t) or Absence (0)

Initial Central Response
White Blood Nervous Lysoryme to
Patient Cell Count Lung System FAB Level Initial
Number Age/Sex (X 10g/titer) Leukostasis Infiltration Classification Wml) Treatment’

1 50/F 254,000 0 0 M2 Complete

2 54/M 118,000 + + M4 Complete
3 50/F 224,000 + + M2 28 None
4 51/F 253,000 + + None
5 48/M 211,000 0 0 M4 68 None
6 26/M 117,000 0 0 M2 24 Complete
7 75/F 115,000 0 0 M4 56 None
8 17/M 278,000 0 0 M2 Complete
9 25/M 256,000 0 0 M4 40 Complete
10 45/M 300,000 + 0 M4 58 Complete
11 51/M 215,000 + 0 M4 72 None
12 75/F 145,000 + 0 None
13 12/M 306,000 0 0 M2 Complete
14 53/F 102,000 0 + M5 None
15 74/M 129,000 0 0 M5 None
16 65/F 117,000 0 0 - None
17 66/M 201,000 + 0 M2 16 None
18 16/F 118,000 0 0 M4 115 Complete
19 22/M 346,000 + + M4 159 None
20 38/M 275,000 + 0 M4 193 None
21 45/M 157,000 0 0 M4 30 Complete
22 55/F 225,000 + 0 M4 56 Complete
23 60/M 263,000 + 0 M5 None
24 71/F 129,000 + 0 M4 64 Complete
25 28/M 172,000 0 + M4 100 Complete
26 58/M 126,000 0 0 M5 84 Complete
27 51/F 277,000 0 0 - None
28 62/M 143,000 0 0 M5 58 Partial
29 76/M 102,000 0 0 - 77 None
30 17/M 438,000 0 + M4 120 Complete
31 15/F 152,000 0 0 M2 24 Complete
32 61/M 160,000 0 0 M5 50 Complete
33 41/F 135,000 0 0 M2 33 None
34 76/F 215,000 + + M2 53 None
35 29/F 137,000 0 + M4 94 Complete
36 65/M 105,000 0 0 M4 58 Partial
37 47/M 128,000 0 0 M4 94 None
38 57/M 112,000 0 + M4 166 Complete
39 60/F 152,000 + + M2 12 Partial
40 50/M 100,000 0 0 M2 63 Complete
41 56/F 204,000 0 + M4 110 Complete
42 53/F 207,000 0 + M5 60 None
43 49/M 120,000 + 0 M5 123 None

l As defined in Patients and Methods.

FAB = French-American-British.

0.10; log rank XT = 3.58, p = 0.06). The effect of the
height of the initial white blood cell count above
lOO,OOO/~l on survival was examined using the Cox
regression model. The result showed no significant in-
fluence of white blood cell count on survival.
The survival curves for patients with lung leukostasis
and for those with central nervous system infiltration
are shown in Figures 2 and 3, respectively. Figure 4
shows the survival curves of three groups: those having
lung leukostasis with or without central nervous system
infiltration (15 patients, median survival 0.2 months);
those with central nervous system infiltration alone
(seven patients, median survival 15.4 months); and
those having neither lung nor central nervous system
involvement (2 1 patients, median survival 10.8 months).
Comparison of the curves in Figure 4 shows a highly
significant difference (Wilcoxon xg = 20.7, p <O.OOl;
log rank xz = 20.5, p <O.OOl).

July 1985 The American Journal of Medicine Volume 79 45

 PULMONARY LEUKOSTASIS IN ACUTE MYELOCYTIC LEUKEMIA-LESTER ET AL

0 Figure 1. Cumulative survival in all pa-

4 8 12 16 20 24 28 32 36 40 tients with acute myelocytic leukemia and
white blood cell count greater than
TIME IN MONTHS 100,000 X 10g/liter.

I I I I I I I I I
2 1.0 b
L

2 0.8
2

w 0.6 /Without pulmonary leukostasis

i &Bq
f 0.4
.?

r’ 0.2

= 0

Figure 2. Comparative survival in pa-

tients with and without pulmonary leuko-
TIME IN MONTHS 1 stasis.

Without CNS infiltration

_I!-I
With CNS

= 0.2 op-TIJ....
-Tljnftltrat~on -..on
3 t I I I I I
‘b-2
0-w-m
-I l
e-e e __-- _-_--___
== 0 I I I I I I _-_ ___-- -----_-_ I ,-I
4 8 12 I6
I2 16 20 2: 28 32 36 40 Figure 3. Comparative survival in pa-
tients with and without central nervous
TIME IN MONTHS system (CPJS) infiltration.

46 July 1965 The American Journal of Medicine Volume 79

 PULMONARY
1.o
0.8
0.6
0.4
0.2
Figure 4. Comparative survival in pa- 0
tients with and without leukemic involve-
ment of the central nervous system (Cl%)
and lung.
COMMENTS
Lichtman [9] has examined the effects of leukemic
leukocytosis on whole blood viscosity and the facility
with which leukemic blasts traverse channels of cap-
illary diameter. Patients with acute leukemia and hyp-
erleukocytosis are protected from marked alteration in
whole blood viscosity by the presence of anemia,
keeping the total cytocrit less than 45 percent, and the
rarity of leukocrits greater than 20 percent. The leu-
kemic myeloblast has a deformability index of 1.22,
approaching that of a rigid body (di = 1.0). We have
reported a similar protective effect of anemia on the
whole blood viscosity of patients with leukemia and
marked hyperleukocytosis [ IO]. Transfusion of red
blood cells to correct the anemia prior to cytoreductive
therapy may dramatically raise the whole blood vis-
cosity and initiate leukostasis.
In patients with acute myelocytic leukemia and hy-
perleukocytosis. the inelasticity of the myeloblast leads
to plugging of vascular channels of capillary dimensions
with leukocyte aggregates and thrombi. The vessels
become distended, anoxic, and leaky, leading to infil-
tration of myeloblasts into parenchyma and, in some
instances, to vascular rupture. Cellular enzymes and
procoagulant material may contribute to parenchymal
damage as blasts degenerate. Organs most frequently
cited as targets for leukostasis and leukemic infiltration
are the central nervous system and lung. It is not clear
whether the vascular beds in these areas are in some
way unique and susceptible to leukostasis, or merely
that they represent vital organs whose function is crucial
to survival and in which derangement is readily ap-
parent.
LEUKOSTASIS IN ACUTE MYELOCYTIC LEUKEMIA-LESTER ET AL
Without leukemic
involvement of
4 8 12 .I6 20 24 28 32 36 -40
TIME IN MONTHS
Vascular leukostasis and parenchymal infiltration are
frequently suspected clinically, but rarely documented
histologically antemortem. Myeloblasts in the cere-
brospinal fluid clearly identify patients with leukemic
infiltration of the meninges, and lumbar puncture, per-
formed by an experienced physician with appropriate
platelet coverage, is associated with low morbidity. The
documentation of leukostasis and leukemic infiltration
of the lung is more difficult. Obtaining pulmonary tissue
in a patient who is thrombocytopenic and often critically
ill demands careful consideration of risks and potential
benefit. Our approach has been to limit such invasive
procedures to patients at risk for opportunistic infection
in whom direct identification of the organism leads to
therapeutic intervention. Tenholder and Hooper [ 1 I]
investigated 98 pulmonary infiltrates in 139 patients with
acute leukemia. In 17 episodes of parenchymal infil-
trations before or within 72 hours of the institution of
chemotherapy, no opportunistic infections were found.
They concluded that the initial febrile illness in a patient
with acute leukemia and an abnormality on chest radi-
ography is not caused by opportunistic infection. Our
experience at Mount Sinai Hospital has been similar and
we did not obtain pulmonary tissue routinely in the pa-
tients under discussion.
The interpretation of hypoxemia as measured by
arterial blood gas analyses in patients with hyperleu-
kocytosis has been the subject of much debate. Chillar
et al [ 121 studied in vitro oxygen consumption at 37’C
in patients with acute leukemia and in granulocyte-rich
solutions from normal donors. They found rapid con-
sumption of oxygen proportional to the degree of leu-
kocytosis-the presence of immature granulocytes
seemed to accelerate this phenomenon. We recom-
July 1995 The American Journal of Medicine Volume 79 47
PULMONARY LEUKOSTASIS IN ACUTE MYELOCYTIC LEUKEMIA-LESTER ET At

mend immediate arterial blood analysis performed in kapheresis before cytotoxic agents are administered.
the respiratory intensive care unit, in order to avoid We [8] have evaluated leukapheresis in 22 of the pa-
prolonged transport, mishandling by technicians, and tients described herein, and found that success in
the requisite oxygen consumption inherent in the lowering the white blood cell count was an important
cooling and rewarming of specimens. predictor of response. Fifteen of 17 patients with greater
The frequency of cases of monocytic leukemia in than 30 percent reduction in the initial white blood cell
those patients with infiltrations of parenchyma was not count had remission, whereas none of five with stable
unexpected [ 131. We [ 141 found seven of 39 consec- or increasing hyperleukocytosis had responses (p =
utive patients with newly diagnosed acute myelocytic 0.001).
leukemia to have central nervous system involvement Utilizing readily obtainable clinical data, i.e., symp-
as evidenced by cytocentrifuge examination of cere- toms, findings on physical examination, arterial blood
brospinal fluid. All seven patients had myelomonocytic gas values, and chest radiographic results, we have
leukemia (M4). Monocytic blasts seem particularly identified a subgroup of patients with acute myelocytic
prone to aggregate and invade parenchyma. leukemia and hyperleukocytosis-those with pulmonary
Four patients of Myers et al [ 151 and five patients of leukostasis-who are largely responsible for the high
Tryka et al [ 161 had acute myelocytic leukemia and mortality rate. This subgroup is in sharp contrast to the
acute respiratory failure following cytotoxic chemo- unaltered initial response rate of patients with infiltration
therapy. Pulmonary tissue demonstrated vascular en- of the central nervous system. Although therapeutic
gorgement/leukemic infiltration in all cases. They leukapheresis may improve survival in patients whose
proposed leukemic cell lysis with release of endoge- white blood cell count can be controlled, those with
nous pyrogen, proteolytic enzymes, and procoagulant rapid return of the leukocytosis to pretreatment levels
material as responsible for this clinical picture. Tryka seem not to be helped. Therapy in these patients should
et al demonstrated degenerating blasts in the pulmonary be directed at aggressive pulmonary support and re-
parenchyma by immunoperoxidase staining of biopsy duction of the hyperleukocytosis as rapidly as possible.
specimens. In contrast, the clinical picture of pulmonary Innovative approaches in removal of leukemic blasts
toxicity developed in only one of our patients following without exposing vital organs to the products of cellular
chemotherapeutic lysis. This may be related to our lysis may reduce the lethal consequences of current
treatment approach with aggressive therapeutic leu- cytotoxic therapy.

REFERENCES
1. Glucksberg H, Buckner CD, Fefer A, et al: Combination che- and blood in leukemia. J Clin Invest 1973; 52: 350-358.
motherapy for acute nonlymphoblastic leukemia in adults. 10. Worsley A, Cuttner J, Kochwa S: Whole blood viscosity in
Cancer Chemother Rep 1975; 59: 1131-1137. patients with leukemias and myeloproliferative disorders
2. Holland JF: Acute myelocytic leukemia. Arch Intern Med 1976; (abstr). Proc Am Assoc Cancer Res 1981; 22: 170.
136: 1377-1381. 11. Tenholder MF, Hooper RG: Pulmonary infiltrates in leukemia.
3. Gale RP, Cline MJ: High remission induction rate in acute Chest 1980; 78: 468-473.
myeloid leukemia. Lancet 1977; I: 497-499. 12. Chillar RK, Belman MJ, Farbstein M: Explanation for apparent
4. Wiernik PH, Serpick AA: Factors affecting remission and hypoxemia associated with extreme leukocytosis: leuko-
survival in adult acute nonlymphocytic leukemia. Medicine cyte oxygen consumption. Blood 1980; 55: 922-924.
(Baltimore) 1970; 49: 505-513. 13. Cuttner J, Conjalka MS, Reilly M, Meyer RJ, Holland JF: As-
5. Gunz FW, Burnes E: Prognosis of acute leukemia of adults. sociation of monocytic leukemia in patients who present
NZ Med J 1965; 64: 555-561. with extreme leukocytosis. Am J Med 1980; 69: 555-
6. Boggs DR, Wintrobe MM, Cartwright GE: To treat or not to treat 558.
acute granulocytic leukemia. Arch Intern Med 1969; 123: 14. Meyer RJ, Ferreira PPC, Cuttner J, Greenberg ML, Goldberg
568-570. J, Holland JF: Central nervous system involvement at pre-
7. Bennett JM, Catovsky D, Daniel MT, et al: Proposals for the sentation in acute granulocytic leukemia. Am J Med 1980;
classification of the acute leukemias. Br J Haematol 1976; 68: 691-694.
33: 451-458. 15. Myers TJ, Solon RC, Klatsky AU, Hild DH: Respiratory failure
8. Cuttner J, Holland JF, Norton L, Ambinder E, Button G, Meyer due to pulmonary leukostasis following chemotherapy of
RJ: Therapeutic leukapheresis for hyperleukocytosis in acute nonlymphocytic leukemia. Cancer 1983; 51:
acute myelocytic leukemia. Med Pediatr Oncol 1983; 11: 1808-1813.
76-78. 16. Tryka AF, Godleski JJ, Fanta CH: Leukemic cell lysis pneu-
9. Lichtman MA: Rheology of leukocytes, leukocyte suspensions mopathy. Cancer 1982; 50: 2763-2770.

49 July 1985 The American Journal of Medlclne Volume 79