RELATIONSHIP BETWEEN BACTERIAL BINDING TO LYMPHOCYTES AND CLINICAL FEATURES IN

CHRONIC LYMPHOCYTIC LEUKEMIA

RODNEY NELSON, MD,* AUREL BRATESCU, DVM,? AND MARIUS TEODORESCU, MI), PHD?

In previous studies we showed that spontaneous bacterial adherence can be used to identify human lymphocyte subpopulations and to demonstrate variable binding patterns in chronic lymphocytic leukemia (CLL). In this study, 10 strains of bacteria of different genera and species were used in blood smears from 24 CLL patients to determine the percentages of lymphocytes that bind bacteria. From these percentages, binding indices were calculated. The symptoms and other laboratory tests were independently recorded and the stages determined. When the two sets of data were compared, relatively low binding indices were found in symptomatic patients or in Stages I11 and IV; relatively high binding indices were found in asymptomatic patients or in Stages I and 11. We suggest that with progression of leukemia, lymphocytes with less “lectio” recognition potential are selected and escape any control mechanism of proliferation.

Cancer 44:1665-1670, 1979.

HRONIC LYMPHOCYTIC LEUKEMIA (CLL) is C a disease of variable and unpredictable course. This disease is usually a malignant proliferation of B lymphocytes, but a rare T lymphocyte variant has also been described.2 It appears that the identification of a CLL as being the result of a malignant proliferation of B- or T-cells can be correlated with the clinical features of an individual patient.2 In spite of numerous attempts, there has been no good correlation between lymphocyte mark- ers and clinical features for the majority of patients with typical B-cell CLL. Also, the generally accepted view is that in a particular patient the leukemic cell population is homogeneous, particularly when looking at the surface Ig by immunofluorescence.’ Davis, however, has recently presented evidence

From the Department of Medicine and the Department of Microbiology and Immunology, University of Illinois at the Medical Center, Chicago, Illinois.

Supported by grant lROl CA 21399 from the National Cancer Institute.

* Department of Medicine. t Department of Microbiology and Immunology. Address for reprints: Marius Teodorescu, MD, PhD,

Department of Microbiology and Immunology, Uni- versity of Illinois at the Medical Center, Chicago, IL 60012.

The authors thank Ana Teodorescu for her technical assistance.

Accepted for publication November 24, 1978.

suggesting heterogeneity of CLL lympho- c y t e ~ . ~

We have shown that bacteria can bind spontaneously to normal human lymphocyte subpopulationss and to leukemic B-cells.’O This spontaneous binding of bacteria revealed marked heterogeneity in the lymphocyte population of individual CLL patients and variability among different CLL patients. Here, we report possible correlations between bacterial binding and clinical features in a group of patients with CLL.

MATERIALS AND METHODS

Patient Population

We studied 24 patients with CLL. This diagnosis was based on peripheral lympho- cytosis of greater than 4000/mm3 persistent over a period of many weeks. (In fact the lowest lymphocyte count was 9,000/mm3.) No attempt at selection was made and patients were studied as they were seen in the clinics at the University of Illinois Hospital and the West Side Veterans Administration Hospital for routine visits. From the charts of these patients it was determined whether the patient’s presentation at the time of diagnosis of CLL was symptomatic or asymptomatic in relation to CLL. Patients with symptoms

0008-543X/79/1100/1665 $0.80 0 American Cancer Society

1665

1666 CANCER Noziernber 1979 V O l . 44

FIG. 1. A lymphocyte labeled by B . globigti in a blood smear.

related to anemia, thrombocytopenia, in- fection associated with hypogammaglobu- linemia, or cosmetically or mechanically troublesome adenopathy or splenomegaly were designated as symptomatic. The pres- ence or absence of splenomegaly was deter- mined on the day of study as was the complete blood count including platelets. The most recent determination of quantitative immuno- globulins was also recorded (IgG, IgA, IgM). This information was obtained without knowledge of bacterial binding results.

Clinical Staging

Rai et aL7 as follows: The stage was assigned after the criteria of

Stage 0: Lymphocytosis in the blood and

Stage I : Enlarged lymph nodes Stage 11: Enlarged spleen or liver

Stage 111: Hemoglobin less than 11 g/100

Stage IV: Platelets less than 100,000/mm3

marrow only

C adenopathy

ml

Duration of Disease and Treatment

The time in months between the first known peripheral lymphocytosis and the time of

study was determined. It was also determined which patients had received either alkylating agent therapy or prednisone during the 3 months prior to study.

Preparation of Bacterial Cell suspension^^^'^ The following strains of bacteria were used

in this study (in parentheses we give the collection numbers where these strains have been deposited): Arizona hinshawii (ATCC- 3124), E. coli-2 (NRRL-B-1 1010), E . coli-3 (NRRL-B-1101 I) , E. coli-5 (LJI-14), Bacillus globigzi (NRRL-B- 1 1007), Brucella melitensis (ATCC-31242), Sarcina lutea (NRRL-B-11012), Staphylococcus aureus-1 (ATCC-3 1240), S . aureus-2 (UI-57), and S. epidermidis (NRRL-B- 11013). A strain ofE. coli which did not bind to lymphocytes was coated with purified rabbit antihuman light chain antibodies and used to identify Ig-bearing cells.5 The bacteria were grown as suspensions in Difco antibiotic medium I11 (Difco, Detroit, Michigan) under continuous aeration and were killed with merthiolate (final concentration 0.05%) after 24-48 hours. The suspensions were treated overnight with 10% formaldehyde in phos- phate-buffered saline, pH 7 , washed with 0.15 M NaCI, and adjusted to about 10'O bacteridml (calculated by spectrophotometry).

TA

BL

E

1.

Perc

ent

Bin

ding

of

Bac

teri

a by

Lym

phoc

ytes

fro

m 2

4 Pa

tien

ts w

ith C

hrot

:ic

Lvm

phoc

ytic

Leu

kem

ia

z ? 01

Pati

ent

num

ber

Bac

teri

a 1

2 3

4 5

6 7

8 9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

B.

mel

itens

is

A. h

insh

awii

E. c

oli 2

E. c

oli

3

E. c

oli

5

B. g

lobi

gii

S. lu

tea

S. a

ureu

s 1

S. a

ureu

s 2

S. e

pide

rmid

is

Bin

ding

inde

x

48

51

82

60

50

65

70

47

60

88

87

83

60

84

88

89

73

14

85

90

88

65

91

94

87

82

19

88

90

87

73

89

89

86

72

14

48

47

57

11

23

16

37

18

9

70

51

61

44

7 69

57

36

14

85

12

70

59

16

25

75

9 53

94

35

57

39

43

11

80

7 52

93

34

71

64

29

39

80

51

54

12

29

34

12

15

16

26

6 50

71

53

69

48

45

51

69

40

34

70

89

95

94

68

25

27

70

67

18

62

44

83

61

15

48

47

64

63

70

83

7 81

77

88

65

77

89

7 88

71

88

55

86

76

7 90

72

87

11

34

16

0 13

2

27

25

73

42

1 14

7

49

12

12

79

9 2

6 29

11

13

71

10

1 13

24

22

53

72

3 6

32

45

3 15

22

1

5 6

16

31

51

61

6 35

33

52

87

98

88

81

33

88 6 28

54

26

59

40

28

56

43

89

85

23

9

83

83

38

12

81

81

36

11

32

5 26

1

38

14

14

5

19

6 44

0

25

15

41

7

30

51

36

6

5 3

18

4

44

37

33

10

64

72

86

82

25

46

26

46

30

15

49

84 1 3 2 7 27 5 9 10

11

17

Ant

i-k-

E. c

oli*

4

34

17

25

31

4 10

38

34

14

3

5 1

6 35

26

22

43

29

11

39

43

37

57

Ant

i-A

-E. c

olit

10

0

2 6

52

04

2

6 0

34

23

83

9

3 0

27

0

3 7

22

6

7 2

1

Ant

i-k

+ A

-E. c

olit

8

36

21

33

35

21

43

42

39

18

41

41

39

9 33

30

24

47

36

26

39

44

40

59

*,T,$

E. c

oli 0

coa

ted

with

ant

i-k

chai

n, a

nd a

nti-

A c

hain

, or

anti

-k a

nd a

nti-

A c

hain

an

tibo

dies

; the

bac

kgro

und

valu

es g

iven

by

E. c

oli 0

coa

ted

with

nor

mal

IgG

wer

e su

btra

cted

.

W * n 4

P > E r

4

0

U

1668 CANCER November 1979 Vol. 44

TABLE 2. Clinical Features and Bacterial Binding: Comparison of Stages 0, I, I1 with Stages 111, IV

and 0.2% sodium azide. The cell suspension was adjusted with the same medium up to the initial volume of blood. T o each sample of 0.2 ml of blood cell suspension we added 0.05 ml from the suspension of one bacterial strain to reach a ratio of about 10 bacteriahed cell. The mixture was centrifuged at 900 x g at 4 C for 6 minutes to promote the binding of bacteria to lymphocytes. The pelleted cells were resuspended in 2.5 ml of MEM-BSA and recentrifuged three times at 150 x g at 4 C for 10 minutes. After each centrifugation, the supernatant layer which contained the unbound bacteria was dis- carded. After the final centrifugation, the pellet was resuspended in MEM-BSA to give a final volume of 0.2 ml. Smears were prepared and stained with Wright stain and lympho- cytes with bacteria attached were easily scored (Fig. 1).

The background of unbound bacteria in the blood smears was usually very low and the number of bacteria associated with each cell was relatively high (Fig. 1). T o provide a standardized methodology which corrects for the random association of bacteria with blood cells, a cell was considered labeled based on the following criterion. Two hundred erythro- cytes of each slide (1-2 microscopic fields) were screened for their association with bacteria. The largest number of bacteria (X) associated with a single erythrocyte was a measure of random association; thus, a lymphocyte associated with X + 1 or more bacteria was scored as labeled. Three separate counts of one hundred lymphocytes were performed and the results were expressed as means. A bacterial binding index was made for each patient by determining the mean binding for the 10 bacterial strains studied.

Stages Stages

n = 14 n = 10 values 0, I , I1 111, IV p

Age No. w/symptoms Mean hemoglobin

(g/100 ml) Mean IgC (mg/100 ml)

(n = 700-1800) Mean IgA (mg/100 ml)

(n = 60-330) Mean IgM (mgi100 ml)

(n = 50-150) No. wialkylating

therapy No. wiprednisone

therapy Median duration (mos) Mean total lymphocyte

Mean bacterial binding count/mm"

index (%)

67 6

63 8 <0.05

12.9

950

9.5

600

10.05

10.05

80 50 <0.05

35

6

35

8

>0.10

= 0.07

2 48

8 18

C0.05 <0.05

32700

49.3

28400

35.0

>0.10

= 0.07

n = normal range.

Binding Bacteria to Lymphocytes

This procedure was carried out as pre- viously de~cr ibed .~ Briefly, the blood was collected in heparinized tubes and the blood cells were washed three times at 4 C in Eagle's minimal essential medium (MEM) supple- mented with 6% bovine serum albumin (BSA)

TABLE 3. Clinical Features and Bacterial Binding: Comparison of Symptomatic with Asymptomatic

Groups of Patients

Asympto- Sympto- matic matic

n = l O n = 1 4

Age 67.4 54.8 Mean hemoglobin

(g/100 ml) 12.8 10.6 Mean platelets

( X 103/mP) 219 132 Mean IgG (mg/100 ml)

(n = 700-1800) 970 695 Mean IgA (mgi100 ml)

Mean IgM (mg/100 ml)

No. w/alkylating

No. w/prednisone

Median duration prior

Mean lymphocyte

Mean bacterial binding

(n = 60-330) 80 55

(n = 50-150) 46 27

treatment 3 11

treatment 1 9

to study (mos) 44 25

count at study/mm3 36700 25800

index (%) 56.6 35.1

P values

C0.05

Statistical Methods

Tests for significance were Student's t test for continuous variables and Fischer's exact test for proportions. Differences in duration of disease at the time of study between groups was tested by Spearman's Rank Test.

c0.05

<0.05

(0.05

c0.05

<0.05 RESULTS

10.05

The 24 patients studied represent a wide spectrum of duration and severity of chronic lymphocytic leukemia and showed a wide heterogeneity in the bacterial binding by their lymphocytes (Table 1). Fourteen patients were in Stages 0, I, and 11, and 10 patients were in Stages 111 and IV (Table 2). The

<0.05

<0.05

10.10

= 0.001

No. 5 BACTERIAL BINDING TO LEUKEMIC LYMPHOCYTES * Nelson et al.

X

z (3

W 60- a

- 50-

z 40- a z m 33-

patients in Stages 0, I, and I1 had fewer symptoms at presentation and had received less intensive therapy than patients in Stages 111 and IV. Patients in Stages 0, I, and I1 also were under observation longer at the time of study than those in Stages 111 and IV. More patients in the early stages were observed without previous therapy. However, no difference in total lymphocyte count between the two groups was noted. Patients with advanced disease (Stages 111 and IV) had lower mean bacterial binding index to their lymphocytes than did patients in Stages 0, I , and I1 (49.3% vs. 35.0%). This difference very nearly achieves statistical significance in spite of the small number of patients studied (p = 0.07). Thus, mean bacterial binding appears to correlate with the clinical stage.

When the patients were divided according to their symptom status at presentation, several differences emerged (Table 3 and Fig. 2). The 10 patients who were asymptomatic at presentation had higher hemoglobin concentrations, platelet counts, and immuno- globulin levels than did the 14 patients who were symptomatic for CLL at the time of presentation. At the time of study, the asymptomatic group had received less therapy but had been under clinical surveillance longer than had the symptomatic group. The mean lymphocyte count at the time of study of the asymptomatic group was not signifi- cantly different than that of the symptomatic group. A striking difference in the mean bacterial binding index was noted between the two groups (56.6% for the asymptomatic group vs. 35.1% for the symptomatic group, p = 0.001). Thus, the mean bacterial binding index correlated significantly with the symp- tom status of CLL patients. In addition, the binding of three individual bacteria (S. epidermidis, S. aureus-1, and S . lutea) was particularly associated with symptom status at diagnosis. For S. epidermidk, 11 out of 14 patients had a binding index less than 15% and eight out of 10 asymptomatic patients had a binding index greater than 15%.

DISCUSSION

We showed here that CLL lymphocyte populations from individual patients are heterogeneous in their ability to bind various bacteria and that a correlation appears to exist between this binding property and the clinical stage or symptom status of CLL patients. The patients with Stage 0, I , and I1 were less symptomatic than those in Stages

I00

90

. 0. . . 0. . . 0

1669

. 0 . . 0 . ....

*O] 10 , i Asymptomotic Symptomatic

SYMPTOM STATUS

FIG. 2. The relationship between the symptom status and the binding index for bacteria.

111 and IV. However, it appeared that the difference in the mean bacterial binding index between symptomatic and asympto- matic patients was more striking than when differences between stages were considered. Three important issues can be raised by the observation of spontaneous bacterial binding to lymphocytes in CLL. First, does the hetero- geneity of bacterial binding reflect the under- lying heterogeneity of lymphocytes that have not been recognized by more traditional surface marker studies? Second, does dimin- ished spontaneous binding of bacteria in more advanced cases of CLL reflect a proliferation of more primitive and undifferentiated cells? Third, can spontaneous bacterial binding be used as a clinical tool for prognosis and to help decide on therapy at the time of diagnosis of CLL?

The heterogeneity of the lymphocytes in CLL has been recognized only r e ~ e n t l y . ~ The pattern and degree of fluorescence of lympho- cyte membranes treated with fluorescent anti- immunoglobulin suggested that the malignant lymphocyte clones are relatively homoge- n0us.I However, the pattern of binding bacteria shows that the population is hetero- geneous since some cells bind and others do not bind certain bacteria. A test carried out on CLL patients in which double labeling with bacteria was performed (data not shown)

1670 CANCER November 1979 VOl. 44

indicated that the cells binding one bacterial strain include those binding another, anti- genically and genetically, unrelated strain. For example, if E. coli-3 bound to 60% of the cells, the remaining 40% did not bind any bacteria. Also, no relationship appears to exist between the surface Ig and the bacterial binding, since treating the cells with anti-light chain antibodies under capping conditions does not affect the binding pattern (unpub- lished results). (The capping process was monitored on normal B-cells under the same conditions.) Even if capping of CLL lym- phocytes did not occur, some inhibition of binding by steric hindrance was expected if the binding of bacteria took place through Ig receptors. Also, the percentage of lympho- cytes binding bacteria sometimes far exceeds the percentage of Ig-bearing lymphocytes (Table 1) .

The relatively low percentages of Ig- bearing lymphocytes in the CLL patients described here compared with those generally reported using fluorescent antibody are hard to explain without a parallel investigation. In normal individuals5 the antibody-coated bac- teria yield higher values (16% on the average) than those reported for fluorescent antibodies (6% on the a ~ e r a g e ) . ~ Therefore, our method appears to be more sensitive. A possible explanation may be that only recently F(ab)’, antibodies have been used to identify Ig- bearing cells;4 the binding of aggregates through Fc receptors may cause false positive results. The binding of antibody-coated

bacteria is not through Fc receptors as we have previously d ~ c u m e n t e d . ~ Moreover, the bind- ing of bacteria to lymphocytes is determined based on a built-in standard for background ‘(see Methods) thereby eliminating any sub- jective estimate of what is a positive cell.

At this point, it is very difficult to explain why lymphocytes from symptomatic CLL patients bind bacteria less than those that were asymptomatic. The effect of therapy on the surface markers of CLL lymphocytes was not assessed due to the small and un- selected nature of the patient population studied here. However, the percentage of cells bearing surface Ig and the percentage of cells that bound bacteria varied inde- pendently, suggesting that if therapy had any bearing on the results, the effect was selective (e.g. patient # 1 compared with patient #24). One patient with particularly low bacterial binding index did not receive any recent therapy (not presented here).

It appears from our recent studies that the bacteria bind due to an interaction between a carbohydrate on their cell wall and a protein with lectin-like properties on the lymphocyte,6 probably as part of the process of intercellular recognition. When the normal cell lines are transformed into malignant lines their surface lectins are reduced.s Therefore, w e may speculate that in more acute forms of CLL, or in advanced stages, the malignant clone may fail to differentiate, does not develop any more lectins, and is no longer controlled by cell contact mechanisms.

REFERENCES

1 . Aisenberg, A. C., and Bloch, K. J.: Immuno- globulins on the surface of neoplastic lymphocytes. N. Engl. J . Med. 287:272, 1972.

2. Brouet, J., Sasportes. M., Flandrin, G., Preud’- homme, J., and Seligmann, M.: CLL of T cell origin: immunological and clinical evaluation in eleven patients. Lancet ii:890, 1975.

3. Davis, S. A.: The variable pattern of circulating lymphocyte subpopulations in CLL. N . Engl. J . M e d . 294:1150, 1976.

4. Hoffman, T. , Wang, C. Y., Winchester, R. Y., Ferrarini, M., and Kunkel, H. G.: Human lymphocytes bearing “Ia-like” antigens; absence in patients with infantile agammaglobulinemia. J . Immunol. 1 19: 1520, 1977.

5. Mayer, E. P. , Bratescu, A., Dray, S., and Teodorescu, M.: Enumeration of human B lymphocytes in stained blood smears by their binding of bacteria. Clin. Immunol. Immunopathol. 9137, 1978.

6. Mayer, E. P., and Teodorescu, M.: Selection of lymphocyte-binding mutants of Escherichia coli for the identification of lymphocyte subpopulations. Fed. Proc. 57:1654, 1978.

7. Rai, K. R., Savitsky, A , , Cronkite, E. P., Chanana, A,, Levy, R. N., and Pasternak, B. S.: Clinical staging of chronic lymphocytic leukemia. Blood 46:219, 1975.

8. Simpson, D. L., Tharne, D. R., and Lah, M. M.: Lectins: endogenous carbohydrate binding proteins from vertebrate tissues: functional role in recognition process. Life Sciences 22:727, 1978.

9. Teodorescu, M., Mayer, E. P., and Dray, S.: Identification of five human lymphocyte subpopulations by their differential binding of various strains of bacteria. Cell. Immunol. 29:353, 1977.

10. Teodorescu, M., Mayer, E. P., and Dray, S.: Enumeration and identification of human leukemic lymphocytes by their natural binding of bacteria. Cancer Res. 37:1715, 1977.