Liver damage in patients with colony-stimulating factor-producing tumors

Liver Damage in Patients With Colony-Stimulating Factor-Producing Tumors AKIRASUZUKI,M.D., Ph.D., TAKAYUKITAKAHASHI,M.D., Ph.D., YOSHIAKIOKUNO,M.D., Ph.D., SHUJISEKO,M.D., Ph.D., YOSHIHIROFUKUDA,M.D., Ph.D., KISHIKONAKAMURA, MANABUFUKUMOTO,M.D., Ph.D., Kyot0,Japan, YOSHITERUKONAKA,M.D., Ph.D., Osaka, Japan, HIROOIMURA,M.D., Ph.D., Kyoto, Japan

cholestasis, were reproduced in the liver of mice transplanted with KHC287 or CHU-2. CONCLUSION:These results indicate that patients with CSF-producing tumors have characteristic liver damage, and suggest a new paraneoplastic syndrome of leuhocytosis and liver dnmnne.

PIJRPOSBWe have demonstrated that colonystimulating factor (CSF)-producing tumor cell lines produce not only CSF but also iuterleuhin1 (IL-l) and interleuhin-6 (IL-6). Clinically, we have observed that patients bearing such tumors present with liver dysfunction and fever in addition to marhed leuhocytosik The purpose of this study was to determiue whether or not the liver damage was specifically related to CSF-producing tumors. PATIENTS AND MEI’EOIB Clinicopathologic examinations were performed in six autopsied patients with CSF-producing tumors. We also transplanted two tumor cell lines (KHC287 and CHU-2), which produce granulocyte (G)-CSF, IL-l, and IL6, to nude mice. RESULT&of the six patients, five had G-CSFand one had granulocyte/macrophage (GM)CSF-producing tumors. IL-1 and IL-6 concentrations in plasma or culture supematant were elevated iu these patients. Biochemical examinations revealed high serum enzyme levels of the biliary system in contrast to normal or slight in. CrsaseSilltranearmaaee levelsinall patients studied. Serum direct bilirubin was elevated in five of the six patienta Three common pathologic changes of the liver were found: (1) focal necrosis associated with neutrophil infiltration in the centrilobular zones, (2) fibrous change and enlargement of the portal area associated with neutrophil infiltration, and (3) intrahepatic cholestasia The same pathologic chang* except for From the Second Department of Internal Medicine (AS, TT, YO, SS, YF, HI). and the First Department of Pathology (MF). Faculty of Medicine, Kyoto University, Kyoto, Japan, the College of Medical Technology (KN), Kyoto University, Kyoto, Japan, and the Department of Internal Medicine, Kitano Hospital (YK), Osaka, Japan. This work was supported in part by the Research Committees on Idiopathic Hematopoietic Disorders and B Cell Malignancies, Ministry of Health and Welfare, and a Grant-in-aid, Special Project, Researches on Cancer Bioscience, Japan. Requests for reprints should be addressed to Takayuki Takahashi, M.D., Second Department of Internal Medicine, Faculty of Medicine, Kyoto University, 64 Shogoin Kawaharacho. Sakyo-ku, Kyoto 606, Japan. Manuscript submitted January SD. 1992, and accepted in revised form August 19, 1992.

I

t is well known that some malignant tumors produce colony-stimulating factors (CSF) and cause marked leukocytosis. In recent years, several studies have shown that some CSF-producing tumors produce multiple cytokines, including interleukin-1 (IL-l) and interleukin-6 (IL-6), in addition to CSF [l-3]. More recently, we reported a newly established lung carcinoma cell line that produces granulocyte (G)-CSF, IL-l, and IL-6 [4]. Furthermore, we showed the co-production of IL-1 and IL-6 in six tumor cell lines that had been reported as CSF producers [5]. Although the major clinical manifestation in patients with CSF-producing tumors is excessive leukocytosis, we have also observed fever and a high level of serum C-reactive protein (CFW) in such patients. Moreover, these patients showed abnormal serum biochemistry in liver function tests and some of them had progressive jaundice. Characteristic pathologic changes were noted in the liver of autopsied patients. These findings prompted us to investigate whether or not the liver damage was specifically related to the CSF-producing tumors.

PATIENTS AND METHODS Patients Six patients with malignant tumors were examined in this study; these patients were referred to us from 1936 to 1999, because of marked and unexplained leukocytosis (over 15,69O/~L). I

Cell Culture Mononuclear cells were separated from biopsy specimens of a lymph node (Patient 1) and tumor mass (Patient 2). Separation was performed by cen-

February 1993

The Amerlcen Journal of Medlclne

Volume 94

125

LIVER DAMAGE IN CSF-PRODUCING

TUMORS

/ SUZUKI

ETAL

trifugation (4OOg, 30 minutes) on 60% Percoll (Pharmacia Fine Chemicals, Uppsala, Sweden); the cells were then suspended in RPMI-1640 medium (Nissui, Tokyo, Japan) with 10% fetal calf serum (FCS) (Hyclone, Logan, UT) and incubated in a 60 X 15-mm plastic dish (Falcon, Lincoln Park, NJ) for 2 hours in 5% carbon dioxide in air at 37OC to remove adherent cells. Nonadherent mononuclear cells (NAC) were cultured in RPMI-1640 medium with 10% FCS at an initial cell density of 5 X lO%nL in 5% carbon dioxide at 37OC. The culture supernatant was collected on Day 7 of culture. In the case of established tumor cell lines, cells were cultured for 4 days at an initial cell density of 2 X lOs/mL. As a control for the culture supernatant of lymphoma cells (Patient l), NAC from lymph nodes of patients diagnosed with chronic lymphadenitis were cultured for 7 days at an initial cell density of 5 X 105/mL, after informed consent was obtained to collect samples from biopsy specimens. As a control for the culture supernatants of tumor mass, fibroblasts separated from the same lymph node samples were used. After two subcultures, fibroblasts were cultured for 7 days at an initial cell density of 5 X l@/mL.

(BALB/c, nu/nu); tumor size and number of white blood cells were measured until a maximum of 49 days. The mice were then killed by cervical dislocation and were used for pathologic studies.

RESULTS

Transplantation of CSF-Fbducing Cell Lines Into Nude Mice Cells (5 X lo6 or 107) from KHC287 and CHU-2 were injected subcutaneously into nude mice

Clinical Findings Table I shows the clinical data of six patients. None of the patients had been treated with antitumor agents or by irradiation before admission and none of them had a history of liver disease or had received red blood cell or platelet transfusions. Some of them had received antibiotics because of leukocytosis and fever. However, since results of bacteriologic examinations were all negative, antibiotics were discontinued except during the terminal stage. All patients had spike fever of over 38OC every day or once every few days. Leukocytosis was neutrophil-dominant and persistent throughout the course of the disease. Serum CRP levels were markedly elevated in almost all patients. Patients 1, 3, and 4 received parenteral hyperalimentation during the terminal stage. An autopsy was performed in all patients and revealed no major bacterial or fungal infection that could explain the persistent leukocytosis. Table II shows the results of serum biochemical tests for liver function on admission. All patients showed high enzyme levels of the biliary system such as alkaline phosphatase (ALP), y-glutamyl transpeptidase (r-GTP), and leucine aminopeptidase (LAP), in contrast to normal or mild increases of glutamic oxaloacetic transaminase (GOT) and glutamic pyruvic transaminase (GPT) levels. Tumor metastases to the bone were demonstrated by the autopsy in Patients 2 and 5. This pattern persisted throughout the course of the disease, although GOT and GPT levels increased in some patients during the terminal stage. Mild increases of bilirubin levels were observed in four (Patients 1,2, 3, and 5) of the six patients at the time of admission. This elevation was direct bilirubin-dominant and was progressive in three patients (Patients 1,2, and 3). Patient 4 developed mild jaundice during the terminal stage. Although Patients 1 through 5 had blood transfusions in the terminal stage, the jaundice had already manifested prior to the transfusions in all five patients. Also, transfused red cells did not cause hemolysis in these patients. None of the patients, however, developed hepatic coma regardless of abnormal liver test results or jaundice. In Patient 1, serum creatinine levels were slightly high (1.3 to 2.3 mg/dL); however, in the remaining five patients, the serum creatinine levels remained within the normal limits (0.6 to 1.1 mg/dL) throughout the course of the disease.

126

94

Enzymelinked lmmunosorbant Assay(E&A) Concentrations of G-CSF in the culture supernatant or plasma were measured by ELISA as previously described [6]. Concentrations of granulocyte/ macrophage (GM)-CSF, IL-la, IL-la, and IL-6 were measured by ELISA kits purchased from Genzyme Co. (Boston, MA) (for GM-CSF and IL-6) and Ohtsuka Pharm. Co. (Tokyo, Japan) (for IL-l), respectively. Concentrations of the tumor necrosis factor cy (TNFa) were measured by an ELISA kit purchased from Ohtsuka Pharm. Co. CalI Lines The lung carcinoma cell line, KHC287 [4], was established in our laboratory in 1987, from a patient included in this study (Patient 5). CHU-2 [7] was kindly provided by Professor S. Asano of the University of Tokyo. In our previous studies, KHC287 and CHU-2 were shown to produce G-CSF, IL-la, IL-l& and IL-6 both at the protein and mRNA levels [4,5]. A lung carcinoma cell line (LUY-2) and a uterine carcinoma cell line (UTM-2) were also established in our laboratory in 1990. LUY-2 and UTM-2 produced negligible amounts of G-CSF, IL1, and IL-6, and were used for control studies.

Februaty

1993

The American

Journal

of Medlclne

Volume

LIVER DAMAGE IN CSF-PRODUCING TUMORS / SUZUKI ET AL

TABLE I Clinical Data of Six Patients With CSF-ProducingTumors Peak WBC Count (/pL)

Tumcr

Treatment During tie 3 Weeks Befere Autopsy

Patient

Sex

Age

1

M

55

IBL-like T lymphoma

55,200

94.0

28.3

Fever, jaundice

Betamethasone, cyclophosphamide, vincristine, flomoxef sodium, RBC and platelet transfusion

2

M

75

Malignant fibrous histiocytoma

38,000

78.5

29.5

Fever, jaundice

Irradiation to neck tumor, prednisolone, ceftazidime, RBC and platelet transfusion

3

F

72

Pancreas carcinoma

115,000

92.5

22.4

Fever, jaundice

Plasma exchange to reduce jaundice, flomoxef sodium, RBC and platelet transfusion

4

M

71

Stomach carcinoma

63,000

89.0

18.9

Fever, mild jaundice

Ceftazidime, clindamycin phosphate, RBC transfusion

5

M

50

Lungcarcinoma

123,200

94.0

25.2

Fever, mild jaundice

Piperacillin sodium, RBC transfusion

6

M

59

Lung carcinoma

127,000

94.5

26.5

Fever

Diclofenac sodium,, betamethasone, cefbzoxime sodium

Symptoms

3C= red blc#_!cell; WBC = white blood cell; CSF = colony-stimulabngfactor; CRP = C-reactive protein; IBL = immunoblastic lymphadenopathy. ‘he normal CRP range is Mow 0.3 mg/dL.

Concentrations of Cytokines in Plasma and Culture Supernatant As shown in Table III, plasma concentrations of either G-CSF or GM-CSF were elevated in all six patients. Similar results were observed in all patients when plasma samples obtained at different time points were examined. Concentrations of either G-CSF or GM-CSF were also definitely elevated in the culture supernatants of tumor cells from all three patients examined. The plasma or culture supernatants generated a reasonable number of granulocyte or granulocyte/macrophage colonies

when added (10% to 30%)to the assay of GM-progenitors (CFU-GM) without exogeneous CSF (data not shown) [a]. From these results, Patient 1 and the remaining five patients were considered to have GM-CSFand G-CSF-producing tumors, respectively. As shown in Table IV, plasma IL-la levels were elevated in all five patients examined, while IL-l/3 was not detectable in any of these patients. High plasma IL-6 levels were observed in four of the five patients examined. These results were reproduced with plasma samples obtained at different time

TABLE II Tests for Liver Functionat the Time oi Admissionin Patients With CSF-ProducingTumors* D-Bil

T-Bil (mg/dL)

Patient

GOT (12-32)

: :

E ::

:

z:,

LDH (226475)

ALP M-70)

22 21 :;

331 302 242 610

460 157 966 393

104 28

610 281

318 35

GPT (5-26)

Oil Admission

Peak

(mg/dL) On Admission Peak

416 68

276 150

1.8 1.7

13.9 51.9

1.3 1.1

9.8 40.0

% 248 104

155 154 318 124

0.2 1.3 1.9 0.3

45.6 5.8 2.2 1.0

0.1 0.6 1.3 0.1

36.5 3.7 1.4 0.5

CSF = colony-stimulating factor; GOT = glutamic oxaloacetic transaminase; GPT = glutamic pyruvic transaminase; LDH = lactate dehydrogenase; ALP = alkaline phasphatase; 7.GTP = glutam transpeptidase; LAP = leucine aminopeptidas; T-Bil = total bilirubin; D-Bil = direct bilirubin. *Thevalues in parentheses are normal ranges (Ill/L). Normal ranges of T-Bil and D-Bil are 0.1-0.9 and 0.1-0.4 mg/dL, respectively.

February

1993

The Amsrlcen

Journel

ol Medlclne

Volume

94

127

LIVER DAMAGE IN CSF-PRODUCING TUMORS / SUZUKI ET AL

TABLE

points from all six patients. In the tumor cell culture supernatants, IL-la, IL-l/?, and IL-6 concentrations were elevated in all three patients examined. TNFa was not detectable (below 7 pg/mL) in the culture supernatant of the KHC287 cell line derived from Patient 5 and in all three plasma samples (from Patients 2, 3, and 6) examined.

III

CSF Concentrations

in Patients’

Plasma

Samples

or in Tumor

Culture

Supematants*

Plasma G-CSF GWSF

Patient

Culture Supematant G-CSF GMXSF

1

t60

2

205

cl0

2,592

cl0

3

86

t10

ND

ND

4

225


ND

ND

5

85

t10

6

237

t10

CHU-2

-

-

LUY-2

-

-

73

t10

UTM2

-

-

81


t60

t10

Lymph nodeNACtl-2

-

-

t60

t10

Fibroblasts5 l-2

-

-

t60

t10

Normal plasma

l-3

97

~60

1,800

173,000'


ND

ND

44,600

180

-

-

)= notdone. Other abbreviations asintext.

oncentrations of G-Wand GM-C% were measured by ELISA. All data are shown in pgiml. The ver limits of significant detection in ELISA kits for G-W and GM-W were 60 pg/mL and 10 pgimL, respectively. G CSFand GM-CSFconcentrations in the culture supematant of an established cell line (KHC287) from Patient 5. #Plastic nonadherent mononuclear cells fmm the lymph node of two different patients with chmnic I phadenitfs. r”Fibmblasts from the same lymph nodes as described above. The third generation of the fibroblasts was cultured

Pathologic Findings in the liver at Autopsy Tumor metastases to the liver were found on macroscopic examination in Patients 2, 4, and 6; however, metastatic tumors were mostly solitary and were not extensively disseminated in the liver. Extrahepatic occlusion of the bile duct was not observed in any of the patients; common bile duct obstruction was not found even in Patient 3 because of pancreatic tail involvement of the tumor. Microscopically, common pathologic changes of the liver were found in all six patients. First, focal necrosis in the centrilobular zones was observed, which was associated with the infiltration of mature neutrophils. This necrosis could be subdivided into three categories: (1) focal infiltration of neutrophils with degeneration of liver cells (Figure l), (2) neutrophi1 infiltration with partially necrotic liver cells (Figure 2), and (3) neutrophil infiltration with completely necrotic liver cells (Figure 3). The focal necrosis was clearly observed even in Patients 1, 2, and 4, who had lower peak white blood cell counts, although the necrotic area was less frequent and the

IV

TABLE

IL-1 and IL-6 Concentrations

in Patients’

Plasma

IL-la

Patient

Samples

or in Tumor

Plasma IL-1g

Culture

Supernatants*

IL-6

IL-la

Culture Supernatant IL-q3

IL.6

1

8.0

t15

1,320

63.5

17.0

5,400

2

15.0

t15

390

31.2

34.5

1,080

3

9.3

<15


ND

ND

ND

4

7.8

t15

167

ND

ND

ND

ND

5 6

6.8

ND

ND

t15

620

325'

53'

ND

ND

182,000' ND

CHU-2

-

-

-

260

62

LUY-2

-

-

-

c3

cl5

tlO0

UTM-2

-

-

-

t3

t15

cl00

<3

Normal plasma l-3

<15

LymphnodeNACl-2

-

-

Fibroblasts l-2

-

-

17,300

-

-

-

<3

<15


-

<3

cl5

tlO0


-

I= not done. Other abbreviations as m text. bncentrationsof IL-la, IL-16,and IL-6 weremeasured by ELISA. All data areshown in pglmt. Thelower limitsofsignificantdetection in ELISA kitsfor IL-la, IL-la, and IL-6were3, 15,and 100 pg/mL. .-la, IL-16, and IL-6 concentrations in the culture supematant ofanestablished cell line fKHC287) from Patient 5.

128

February 1993

The American Journal of Medlclne

Volume 94

LIVER DAMAGE IN CSF-PRGDUCING TUMGR9 / SUZUKI ET AL

Flgure

1. Infiltration

lar zones panded,

of the and

liver

of mature

liver

in Patient

cells

are

eosin stain: original magnification

neutrophils

in the

5. Sinusoidal

degenerated. X200,

centrilobu-

spaces

(Hematoxylin

reduced

are exand

Figure

2. Neutrophil

infiltration

cells in Patient

5. (Hematoxylin

nification

reduced

X200,

and

partially

and eosin

necrotic

stain;

original

liver mag-

by 5%.)

by 5%.)

neutrophil infiltration was less prominent, when compared with that in patients who had peak white blood cell counts of over lOO,OOO/~L.Interestingly, in some areas, the infiltrated cells were mature monocytes. Generally, liver cells were slightly degenerated, even in the non-necrotic areas; however, no piecemeal liver cell necrosis was observed even in the peripoti area. Second, the portal area was fibrous, enlarged, and associated with the infiltration of mature neutrophils (Figure 4). Third, intrahepatocellular bile pigment or canalicular bile plugs were seen mainly in the centrilobular zones, indicating intrahepatic cholestasis, although these changes were minute in Patients 5 and 6. Liver involvement of tumors was not observed, even at the microscopic level, other than in Patients 24, and 6. Pathob@ Findings in the liver in TumorTransphtad Mice KHC287 from Patient 5 and CHU-2 were successfully transplanted to nude mice, and formed a subcutaneous tumor. The white blood cell count in the mice increased to 500 X 103/pL in proportion to the tumor size, and the leukocytosis was neutroPhil-dominant. LUY-2 and UTM-2 were also transplantable; however, the white blood cell count did not increase in these mice. Macroscopically, neither liver metastasis of the tumor nor hepatomegaly was observed in these mice. Marked splenomegaly was observed in the mice transplanted with KHC287 and CHU-2. Microscopically, focal necrosis was observed in the centrilobular zones in KHC287- and CHU-2transplanted mice, as it was in the livers of the six autopsied patients. Furthermore, as we had observed in the patients, neutrophil infiltration with the three possible categories of liver cell necrosis was also seen in these mice (Figures 5, 6, and 7).

Figure 3. Neutrophil

infiltration

cells in Patient

5. (Hematoxylin

nification

reduced

X200,

and focal

necrosis

and eosin

stain;

of the liver original

mag-

by 5%.)

The findings were similar to those in the patients, with infiltrated cells being predominantly monocytic in some areas. Fibrosis and enlargement of the portal area were also seen in the liver of the mice transplanted with KHC287 or CHU-2 (Figure 8). Cells infiltrated were neutrophils. In contrast to the liver of the patients, however, no bile retention was seen in any mice transplanted with KHC287 or CHU-2. Focal necrosis and fibrous portal areas were not observed in the mice transplanted with LUY-2 or UTM-2. In addition, no liver metastases of transplanted KHC287 or CHU-2 cells were observed, even microscopically. Concentrations of human G-CSF measured by ELISA were 850 and 2,680 pg/mL in the sera of KHC287- and CHU-2-transplanted mice, respectively. G-CSF levels in the sera of LUY-2- or UTM2-transplanted mice were all below the minimum detection limits of the ELISA kit.

Februrry 1993

The American Journal of Medlclne

Volume 94

129

LIVER DAMAGE IN CSF-PRODUCING TUMORS / SUZUKI ET AL

Figure

4. Portal

enlarged.

area

Mature

duct; arrow 2, portal inal magnification

in Patient

neutrophils vein.

X200,

4. The

are

(Hematoxylin

reduced

area

infiltrated.

is fibrous Arrow

and

1, bile

and eosin stain; orig-

Figure 5. Neutrophil mouse

(Hematoxylin reduced

by 5%.)

transplanted and

infiltration

in the centrilobular

with

CHU2.

Liver

eosin

stain;

original

cells are

zone

in a

degenerated.

magnification

X200,

by 5%.)

COMMENTS

Figure

6. Neutrophil

infiltration

and partially

necrotic

liver

cells in a CHU-2-transplanted mouse. The arrow indicates central vein. (Hematoxylin and eosin stain; original magnification X200.

Figure

reduced

by 5%.)

7. Neutrophil

infiltration

in a CHU-2-transplanted vein. (Hematoxylin and X200, reduced by 5%.) 130

February 1993

and focally

necrotic

liver cells

mouse. The arrow indicates central eosin stain; original magnification

The American Journal of Medicine

We have shown here that patients with CSF-producing tumors have common abnormalities in liver function and have characteristic histopathologic changes of the liver. These pathologic changes, except for intrahepatic cholestasis, were reproduced in mice transplanted with KHC287 or CHU-2 cell lines, which had been derived from patients with CSF-producing tumors. The latter observations suggest that the abnormal liver function and pathologic changes could be attributed to CSF-producing tumors. Metastatic tumors to the liver should be taken into consideration as capable of causing such pathologic changes of the liver. However, these pathologic changes were commonly observed both in patients with metastatic disease and in those without it. Furthermore, tumor-transplanted mice reproduced these pathologic changes without metastatic disease to the liver, From these facts, it is unlikely that metastatic tumors to the liver contributed to these histologic characteristics of the liver. The cause of the focal necrosis of the liver is unclear. However, neutrophil infiltration at the site of necrosis, which was seen both in livers of the six patients and in those of tumor-transplanted mice, gives us a clue. That is, massive neutrophil infiltration in the parenchyma was associated with degenerated, partially necrotic, or totally necrotic liver cells. Such different steps of liver cell injury may reflect the process that finally leads to complete focal necrosis. From this point of view, the focal necrosis might have been caused by massive neutrophi1 infiltration. As a possible explanation for liver cell damage, superoxide produced by infiltrated neutrophils might have played an important role,

Volume 94

LIVER DAMAGE IN CSF-PRODUCING TUMOR9 / SUZUKI ET AL

since IL-l stimulates IL-8 production by fibroblasta or macrophages [9], and IL-8 stimulates chemotaxis and superoxide production by neutrophils [lo]. In addition, IL-6 activates neutrophils and enhances their cytotoxicity Ill]. In this study, high plasma IL-la and IL-6 levels were observed in all five and in four of five patients examined, respectively. Furthermore, we observed continuous IL-la, IL-l& and IL-6 production by KHC287, which had been derived from Patient 5. This mechanism for the liver cell injury appears to be similar to that for the adult respiratory distress syndrome, in which neutrophils play a pathogenic role through superoxide production [12,13]. The reason why monocytes infiltrated into the parenchyma is unknown, since none of the six patients had monocytosis in the peripheral blood. IL1, which was persistently released from tumors, might have activated the monocyte/macrophage (MOD%) function in the tissues as well as in the liver. It is likely, therefore, that activated Ma/Me also caused liver cell injury through the production of superoxide [14] or tumor necrosis factor [15]. Regarding the fibrosis of the portal area, IL-l, which was persistently produced by the tumors, might have played an important role here, since IL1 promotes the proliferation of fibroblasts [16]. GCSF has also been reported to enhance fibroblast proliferation [17]. Therefore, fibrosis of the portal area could have been induced by IL-l and/or GCSF. It is unclear whether or not intrahepatic cholestasis was specifically related to CSF-producing tumors, because it was not reproduced in the KHC287- or CHU-2-transplanted nude mice. The tumors in patients who developed progressive jaundice might have produced some additional factors, other than IL-l, IL-6, and CSF, that caused cholestasis in uiuo [18]. Alternatively, both focal necrosis and fibrosis of the portal area might have induced a precholestatic state that easily developed into actual cholestasis, under the influence of factors such as drugs. High serum levels of ALP, y-GTP, and LAP, which were noted in all patients throughout the course of the disease, suggest latent bile duct damage. We studied the relationship between jaundice and drugs or treatment in each patient; however, no precise relationship could be determined. None of the six patients had any drug allergies, such as skin eruptions. Further studies in a larger number of patients are required to elucidate whether or not jaundice is caused by CSF-producing tumors per se and whether it is cytokine-mediated. A cell line derived from a patient with severe jaundice would also be a good tool for elucidating this problem. The liver damage in the present study appeared to be related to multi-cytokine production (CSF,

Figure 8. Fibrous and enlarged filtration

in a KHC287-transplanted

portal

area

with neutrophil

mouse.

duct; arrow 2, portal vein. (Hematoxylin inal magnification X200, reduced by 5%.)

Arrow

in-

1, bile

and eosin stain; orig-

IL-l, and IL-61 by tumors in the six patients. Recently, IL-6-transgenic mice [19] and G-CSF [20] and GM-CSF [21] gene transfection into murine hemopoietic stem cells have been reported. However, focal necrosis of the liver parenchyma, fibrosis of the portal area, and bile retention were not described in these mice, although neutrophil accumulation in the portal area was described in mice with G-CSF or GM-CSF gene transfection [20,21]. Also, in the clinical trials with G-CSF or GM-CSF for granulopenic patients, any CSF-induced liver damage has not been reported [22-241. These transgene experiments and clinical trials with CSF, therefore, suggest that the combined effects of CSF, IL-l, and IL-6 may be responsible for inducing liver damage in CSF-producing tumors, although IL-l seems most important. As a result of the present study, we urge clinicians to be aware of a new paraneoplastic syndrome of liver damage and leukocytosis, which is presumably caused by tumor production of multiple cytokines. This also suggests a new disease entity of cytokinerelated organ damage. The liver damage may predispose to the development of jaundice. Attention must be given to using drugs in order to obviate further liver damage in those patients.

ACKNOWLEDGMENT We are grateful to Professor S. Asano (University of Tokyo) for providing CHU-2 to us. We are also grateful to Dr. Y. Ihara. Dr. T. Okada, and Dr. T. Nishimura for clinical cooperation. We thank Ms. M. Aogaki, Mr. I. Watanabe. and Mr. M. Fukushima. in the Assay Technology Research Center, Chugai Pharm. Co., for measuring G-CSF concentrations.

REFERENCES 1. Demetri GD, Zenzie BW, Rheinwald JG, Griiin JD. Expression of colony-stimulating factor genes by normal human mesothelial cells and human malignant mesothelial cell lines in vitro. Blood 1989; 74: 9406.

February 1993

The Americen Journal of Medlclne

Volume 94

131

LIVER DAMAGE IN CSF-PRODUCING

TUMORS

2 Sato K. Fujii Y. Kakiuchi T. et a/. Paraneoplastic and Ieukocytosis caused by squamous carcinoma thyroid hormonerelated

protein, inter&kin

/ SUZUKI ET AL

syndrome of hypercalcemia celfs (T3M1)

producing para-

la. and granufocyte colony-stimu-

ases, and cationic proteins. J Clin Invest 1987; 79: 669-74.

lating factor. Cancer Res 1989; 49: 47406. 3. Mochiiki

DY, Eisenman JR, Conlon PJ. Larsen AD,Tushinski RI. lnterleukin 1

regulates hamatopoietic

13. Holman RG. Maier RV. Superoxide production by neutrophils in a model of distress syndrome. Arch Surg 1988: 123: 1491-5. 14. Henson PM. Johnston JR6 Tissue injury in inflammation; oxidants, protein-

adult respiratory

activity, a role previously ascribed to hemopoietin

1.

15. Philip R. Epstein LB. Tumor necrosis factor as immunomodulator tor of monocyte

cytotoxicity

induced by itself, y-interferon

and media-

and interleukin

1.

Proc Nab Acad Sci U S A 1987; 84: 5267-71. 4. Suzuki A, Takahashi T. Okuno Y, et a/. Analysis of abnormal expression of G-

Nature

CSF gene in a novel tumor cell line (KHC287)

of fibroblast proliferation. J lmmunol 1982; 128: 2177-82. 17. Mora LM, Caceres JR, Santiago E, et a/. Evidence that G-CSF induce fibro-

elaborating

GCSF,

IL-1 and IL-6

with co-ampliicatfon of c-myc and c-ki-ras, Int J Cancer 1991; 48: 428-33. 5. Okuno Y. Takahashi T, Suzuki A. eta/. Co-production of interleukin 1 and 6 in tumor cefl lines elaborating

colony stimulating factors. Jpn J Cancer Res 1991;

82: 890-2. 6. Watari K. Asano S, Shirafuji N, et a/. Serum

1986: 323: 86-9.

16. Schmidt JA, Mizel S8, Cohen D, Green I. lnterleukin

blasts proliferation

as well as differentiation

of myeloid cells [abstract].

colony-stimulating

factor levels in healthy volunteers

and patients with various disorders as esti-

mated by enzyme immunoassays.

Blood 1989; 73: 117-22.

Blood

1990; 76 Suppl 1: 157A.

18. Miioguchi Y, Miyajima K, Sakagami Y, Yamamoto granulocyte

1. a potential regulator

of the cholestatic

S, Morisawa S. Detection

factor in the liver tissue of patients with acute intrahepatic

cholestasis. Ann Allergy 1986: 56: 304-7. 19. Suematsu S, Matsuda T, Aozasa K. eta/. IgGl plasmacytosis

in interleukin 6

7. Nagata S, Tsuchiya M. Asano S. et a/. Molecular doning and expression of

transgenic

cDNA for human

20. Chang JM. Metcalf D, Gonda TJ, Johnson GR. Long-term exposure to retro-

granulocyte

colony-stimulating

factor.

Nature

1986; 319:

mice. Proc Natl Acad Sci U S A 1989; 86: 7547-51.

415-8.

virally expressed granulocyte-colony-stimulating

8. Nakamura K, Takahashi T, Tsuyuoka R, et al. Identification of CSF activity in patients with malignant tumors associated with excessive leukocytosis. Jpn J

granulocytic

Clin Oncol 1991; 21: 395-9.

21. Johnson GR, Gonda TJ, Metcalf D. Hariharan IK, Cory S. A lethal myeloproliferative syndrome in mice transplanted with bone marrow cells infected with a

9. Matsushima

K. Morishita K. Yoshimura T. eta/. Molecular cloning of a human

monocytederived neutrophil chemotactic factor (MDNCF) and the induction of MDNCF mRNA by interleukin 1 and tumor necrosis factor. J Exp Med 1988; 167: 1883-93.

10. Schroder JM, Mrowietz U, Eishin M. Morita E, Christophers and partial biochemical phil-activating

characterization

E. Purification

of a human monocytederived,

peptide that lacks interleukin

1 activity. J lmmunol

neutro1987; 139:

and progenitor

cell hyperplasia

Clin Invest 1989; 84: 1488-96.

retrovirus expressinggranulocyte-macrophage

ll.Borish L. Rosenbaum recombinant

interleukin

R. Albury L. Clark S. Activation 6. Cell lmmunol

of neutrophils

by

1989; 121: 28D-9.

colony stimulating factor. EM80

J 1989; 8: 441-8. 22. Gabrilove JL, Jakubowski A, Scher H, et al. Effect of granulocyte colonystimulating factor on neutropenia and associated morbidity due to chemotherapy for transitional-cell

carcinoma

of the urothelium.

N Engl J Med 1988; 318:

1414-22. 23. Eguchi K, Sasaki S. Tamura T, eta/.

3474-83.

factor induces a nonneoplastic without tissue damage in mice. J

human granulocyte-colony-stimulating

Dose escalation study of recombinant factor (KRN86Dl)

in patients with ad-

12. Weiland JE. Davis W8. Holter JF, eta/. Lung neutrophils in the adult respira-

vanced malignancy. Cancer Res 1989; 49: 5221-4. 24. Antin JH, Smith BR, Holmes W. Rosenthal DS. Phase l/II study of recombi-

tory distress syndrome-cfinical

nant human granulocyte-macrophage

and pathophysiologic

signkicance.

Am Rev

Respir Dis 1986; 133: 218-25.

132

February

1993

The Amerlcen

mia and myelodysplastic

Journal

of Medicine

Volume

94

syndrome.

colony-stimulating

factor in aplastic ane-

Blood 1988; 72: 705-13.