British Journal of Haematology, 1992. 81, 288-295

The effect of macrophage colony-stimulating factor on haemopoietic recovery after autologous bone marrow transplantation

ASIM KHWAJA, KWEE YONG, H. MARK JONES, RAJESH CHOPRA, ANDREW K. MCMILLAN, ANTHONY H. GOLDSTONE, KEITH G. PATTERSON, CATHERINE MATHESON, KAREN RUTHVEN,* STEPHAN B. A B R A M S O N ~ A N D DAVID C. LINCH Department of Haematology, University College and Middlesex School of Medicine, London, and Alpha Therapeutic Corporation, *Theword, Norfolk, U. K., and tLos Angeles, California, U.S.A.

Received 27 November 1991; accepted for publication 24 January 1992

Summary. Macrophage colony-stimulating factor (M-CSF) is active in the late stages of monocyte maturation, activates mature monocyte-macrophages and enhances their produc- tion of various other cytokines. We have examined the effects of a 2 1 d course of escalating doses of M-CSF purified from human urine (hM-CSF) on recovery following autologous bone marrow transplantation (ABMT) in 20 patients with malignant lymphomas. Four patients were treated at each dose level of 4, 8, 16, 32 and 64 x 1 0 " U/m2/d and results compared to 46 concurrent controls. There was no signifi- cant difference in recovery to an absolute neutrophil count (ANC) of 0.5 x 10y/l (median 20 d in hM-CSF group versus 22 in controls) or in recovery of platelets to 50 x 1 Oy/l ( 3 2 d versus 39 d, 0.05 <P<0.1) ; hM-CSF patients received a median of 81 platelet units following ABMT (controls 11 2 units, P=NS). hM-CSF patients had a median of 5.5 d with fever > 37.5"C (control 8, P=NS), received parenteral

antibiotics for 14.5 d (control 17, P=NS) and had a 50% incidence of bacteraemia (control 48%). hM-CSF treated patients were discharged by a median of day 29 following transplantation (control 3 3 , P<0.05) . Platelet and neutro- phi1 recovery correlated significantly with the number of marrow mononuclear cells (MNC) reinfused in the hM-CSF group (P=0.05 and P = 0 * 0 1 4 respectively) but not in controls. Subgroup analysis showed that hM-CSF patients receiving > 2 x lo8 MNC/kg body weight reached an ANC of 0 . 5 ~ 109/1 by a median of day 16.5 (control 18.5, NS). became platelet transfusion independent by day 1 7 (control 29, P<O.05) and reached a platelet count of 50 x 109/1 by day 21 (control 40, P < O G ) . No significant toxicity attribu- table to hM-CSF treatment was seen. These results suggest that hM-CSF accelerates platelet recovery following ABMT and that relatively large marrow innocula are required to see this effect.

Several trials involving the use of haemopoietic growth factors following high-dose cytotoxic therapy and bone marrow transplantation (BMT) have been reported. Treat- ment with granulocyte and granulocyte-macrophage colony stimulating factors (G and GM-CSF) causes acceleration of myeloid recoverypost BMT (Sheridan et al. 1989: Brandt et al. 1988: Devereux r t a]. 1989: Nernunaitis et al, 1988) but no consistent effect on platelet recovery has been described.

Macrophage CSF (M-CSF) is a homodimeric molecule active in the terminal stages of monocyte/macrophage maturation which also has significant effects on mature monocyte function (Clark & Kamen, 1987) both in vitro and

Correspondence: Dr A. Khwaja. Department of Haematology. University College and Middlesex School of Medicine, 98 Chenies Mews, London WClE hHX.

in vivo (Khwaja et al, 1991) and is available in both purified (from human urine) and recombinant forms.

Monocytes/macrophages produce several haemopoietic growth factors and cytokines on stimulation, including interleukin-1 (IL-1) (Durum et al, 1985), IL-6 (Horii et al, 1988) and G-CSF (Motoyoshi et al, 1989) and may play a significant role in the local control of haemopoiesis in the bone marrow microenvironment. M-CSF has been shown to enhance monocyte/macrophage production of several mol- ecules including IL-1 (Moore et d , 198O), IL-6 (Navarro et al, 1989), G-CSF (Motoyoshiet al. 1989). tumour necrosis factor (TNF) and interferon (Warren & Kalph. 1986). In clinical trials, Komiyama et a1 showed that administration of purified human M-CSF (hM-CSF) led to increases in neutrophil counts in patients with childhood neutropenias (Komiyama e l al, 1988) and hM-CSF has been reported to ameliorate the

288

Effect of M-CSF after ABMT 289 Table 1. Comparison of patient characteristics of hM-CSF treated and control groups.

Controls hM-CSF patients

Age (median)

Disease HD NHL

Previous treatment for HI) Chemotherapy: 1 regimen

2 regimens > 2 regimens

Radiotherapy

NHL status Chemosensitive Chemoresistant

Splenectomy

3 2

30 (65%) 16 (35%)

10 (63%) 6 (37%)

6 (1 3%)

32

13 (65%) 7 (35%)

neutropenia following conventional dose chemotherapy (Motoyoshi et al. 1986). In addition, Masaoka et a1 (1988) have reported that treatment with hM-CSF is associated with accelerated neutrophil recovery following allogeneic BMT for a variety of indications.

In order to examine the potential effects of indirect stimulation of haemopoiesis by M-CSF following BMT, we have looked at the effects of escalating doses of hM-CSF on haematological recovery in 2 0 patients undergoing autolo- gous BMT for malignant lymphomas. In addition, we have carried out pharmacokinetic studies on these patients and assessed the production of secondary cytokines following a single test infusion of hM-CSF prior to chemotherapy.

MATEKIALS AND METHODS

Patients. During the period between January 1989 and January 1 9 9 1, 2 0 patients with malignant lymphomas treated with high-dose chemotherapy (BEAM) and ABMT were entered into a trial of hM-CSF. There were 1 3 patients with Hodgkin's disease: they had failed to respond to front- line alternating chemotherapy or a t least two modalities of treatment or had relapsed after a t least two chemotherapy regimens. There were seven patients with intermediate or high grade non-Hodgkin's lymphoma, as classified by the International Working Formulation (N-HPLC, 1982). who had either failed to achieve coniplete remission 011 conventio- nal first-line therapy or had relapsed from complete remis- sion. All patients had active lymphoma at the time of ABMT. No patient had bone marrow involvement as detected by trephine biopsy and the Karnofsky score was greater than 70% in all patients. Patient characteristics are shown in Table I.

After recruiting four consecutive patients a t a given dose level, patients were followed to haematological recovery with concurrent evaluation of potential toxicity. lf no toxicity attributable to hM-CSF treatment was observed up to 4 weeks after the recovery of the last patient at a given dose. then

treatment was started at the next dose level. During this period of toxicity evaluation, patients with the same disorders undergoing ABMT without CSF treatment were used as controls. These 46 patients were well matched with regard to previous therapy and disease status with those receiving hM- CSF (Table T).

Preparation of hM-CSF. Human M-CSF with a specific activity of 1 .5 x 1 Ox U/mg, as measured in a mouse colony- forming assay, was purified from human urine and provided by Alpha Therapeutic Ltd (Thetford, U.K.)/Green Cross Corporation (Osaka, Japan). The endotoxin content of the preparation was below the lower limit of detection of a limulus lysate assay (less than 0.6 pg endotoxin/pg hM-CSF).

Transplant protocol and hM-CSF administration. All patients received a full test dose of hM-CSF in the week prior to commencing chemotherapy. Four patients were treated at each dose level of 4, 8, 16 , 32 and 64 x lo6 IJ/mL. 2 1 test doses were given as one patient at dose level 1 received a test dose but did not proceed to ABMT because of rapidly progressive disease. All patients received identical chemo- therapy (BEAM) in the form of BCNU 300 mg/mL on day - 6, etoposide (VPI 6) 200 mg/m2 on days - 5 to - 2, cytosine arabinoside 200 mg/mL twice daily on days - 5 to - 2, and melphalan 140 mg/m2 on day - 1. Bone marrow was harvested and cryopreserved as previously described (Linch et al, 1982) with no patient receiving < 1.0 x 10' marrow mononuclear cells (MNC) per kg body weight. No i n vitro purging procedures were employed and marrow was rein- fused on day 0. Treatment with hM-CSF was started on day + 1 and was administered as a daily 2 h intravenous infusion via a Hickman line for 21 d.

Treatment of infection. Patients developing a temperature of 38OC or greater during the neutropenic phase were com- menced on intravenous broad spectrum antibiotics. No prophylactic systemic antibiotics were used. Antibiotic treat- ment was usually stopped after five consecutive days without fever, even during continuing neutropenia. or was stopped sooner if neutrophil recovery had taken place. Patients developing secondary fevers following antibiotic administra- tion, or whose temperatures failed to resolve after 48-72 h of appropriate antibiotic treatment, received empirical intra- venous amphotericin B. Parenteral antifungal treatment was also given to patients with microbiologically proven or clinically suspected fungal infections.

Administration of blood products. Packed red blood cells were given as required to maintain a haemoglobin > 10 g/dl. Random donor platelets were given prophylactically to maintain a platelet count > 20 x I OY/L All blood products were irradiated prior to administration.

hM-CSF and secondary cytokine measurement. Samples were taken into glass containers without anticoagulant, allowed to clot a t room temperature, separated within 2 h and stored at -70°C. hM-CSF was measured by ELISA (Green Cross Corporation, Osaka, Japan). The ELISA protocol followed was as described by Hanamura et a1 (1 988). TNFa, IL-I and 11,-6 were measured by ELISA (Medgenix. Brussels, Belgium) and G-CSF by ELISA (British Biotechnology, Cambridge, 1J.K.). The lower limit of detection was 2 0 pg/ml for the TNFa, IL-I /{ and IL-6 ELISAs, and 1 5 0 pg/ml for the G-CSF assay.

290 A. Khwaja et a1

- E . 3

3000

2000

1000

4mU

OmU

16mU

0 32mU

64mU

0 1 I I I I 0 60 120 180 240 300

time (minutes)

Fig 1. Serum hM-CSF concentrations as measured by ELISA following the administration of a single dose of hM-CSF by 2 h i.v. infusion. The baseline values prior to hM-CSF infusion are shown at time 0. Results shown are means of two patients each at 4,8 and 16 mU/mz. three patients at 32 mU/m2, and four at 64 mlJ/m’.

Statistical analysis. This was carried out using the ’Stat- view’ program on an Apple Macintosh. Comparison of haematological recovery, of the use of blood products, and of infective parameters was made by using Wilcoxon’s test for unpaired data. Comparisons of the incidence of bacteraemia was made using Fisher’s exact test.

RESULTS

Effects of hM-CSF test dose The effects of a single infusion of hM-CSF have been previously reported in detail (Khwaja e t al, 1991). In brief, small but consistent reductions in haemoglobin concentra- tion and platelet numbers were observed in the 2 h following i.v. administration of hM-CSF. The fall in platelets was transient and was presumably secondary to monocyte activation. Enhanced monocyte migration into skin win- dows, an increase in the f-met-leu-phe stimulated monocyte respiratory burst and augmented candida phagocytosis and killing were observed following the infusion. These effects were observed at all doses tested.

hM-CSF pharmacokinetics hM-CSF was measured by ELISA following the test dose, administered i.v., over 2 h. Sera from 13 patients (two patients each at 4.8 and 16, three at 32 and four at 64 x lo6 U/m2) showed that all had detectable circulating M-CSF prior to the test dose with a mean of 71 7 f 101 (SE) U/ml. This is similar to the findings of Hanamura et a1 (1988) in healthy volunteers. Fig 1 shows that the maximal serum concentra- tion obtained was related to the dose given. Only patients

receiving 64 x 1 O h U/m2 had serum levels significantly above their baseline value at 4 h after the end of the infusion. Clearance was clearly biphasic at the 32 and 64 x lo6 U/m2 levels, with a distribution t l i r of 10 min and clearance tliL of 360 min at the 64 x loh U/m2 dose.

Secondary cytokine release Measurements of serum TNFa, IL-lB, IL-6 and G-CSF were made in the four patients receiving 64 x loo U/m2 hM-CSF before and at 2, 9 and 24 h following the test infusion, and of I L - l j alone in eight others. No significant change in the serum concentrations of any of these cytokines was observed at any of the time points tested following hM-CSF infusion.

Huematological recovery after ABMT The haematological recovery of individual patients receiving hM-CSF is shown in Table 11 and is compared with the median recovery of 46 concurrent control patients. Treatment with hM-CSF had no effect on the period of most severe neutrope- nia (absolute neutrophil count [ANC] <0.1 x 10y/l) which was a median of 10 d in both treatment and control groups. The median time taken to reach an ANC of 0.5 x l O Y / l was also similar in both groups, being 20 d with hM-CSF therapy and 22 d in controls. No fall in the ANC on discontinuation of hM-CSF infusions was detected in any patient. The median time to reach a monocyte count of 0.2 x lOy/1 was 26 d in both hM-CSF patients and controls. There was a trend towards a more rapid regeneration of platelets following ABMT in the patients treated with hM-CSF who became independent of platelet transfusion by day 26 (control 29) and took a median of 32 d to reach a platelet count of

Effect of M-CSF after ABMT 291

Table 11. Haematological recovery following AHMT in 20 patients receiving hM-CSF.

Trial no.

01 02 0 3 0 5 06 0 7 08 0 9 10 11 12 1 3 14 15 16 1 7 18 19 20 21

hM-CSF Median Mean f SE

Control Median Mean i SE

UPN no.

413 414 41 9 432 448 44 9 450 451 46 7 472 475 484 51 1 512 516 523 539 548 551 558

hM-CSF dose x 1 O6 lJ/mL _~

4 4 4 4 8 8 8 8

16 I 6 16 16 32 32 32 32 64 64 64 64

Disease

NHL NHL HD HD NHL NHL NHL NHL NHL HD HD HD HD HD HD HD HD HD HD HD

Days to recovery

Marrow WBC Neutrophils dose x > 1 .o >0 .5 10X/kg x 109/1 x IOY/ I

1.50 1.1 5 2.30 2.20 1.10 1.13 1 .00 2.70 2.00 2.73 2.28 2.30 1.67 1.50 1.20 3 . 5 0 1.70 1.80 2.80 1 .97

32 23 12 13 13 15 26 13 12 10 16 18 16 25 32 10 19 22 28 24

35 22 20 13 19 15 29 14 21 16 20 17 28 28 33 11 19 22 30 24

1.87 17 2 0 1 , 9 3 4 ~ 0 . 7 1 9 1 1 . 6 22+1.6

1.80 18 22 1 . 9 8 ~ t 0 . 7 2 1 1 1 . 3 24+1.6

Platelets > 50 x 1 0 9 / 1

61 NE 26 18 23 31 42 19 29 27 NE 21 38 42 42 I 6 48 33 69 42

32 3 5 1 3 . 5

39 4 2 1 3 . 4

Table 111. Comparison of haematological recovery (median days) and post- transplant hospitalization between hM-CSF treated patients and controls accord- ing to marrow mononuclear cell dose reinfused.

Median days to reach

Neutrophils Platelets Days of No. of 1 0 . 5 x loy/] >SO x 10'//1 hospitalization patients

Controls <2x108/kg 24 38 34 29 > 2 x 10'/kg 19 4 0 26 1 7

hM-CSF < 2 x 1OX/kg 24 42 31 12 > 2 x 1 08/kg 1 7* 211- 2 l t 8

*P<0.02 5 compared to hM-CSF patients receiving < 2 x 10' marrow cells/kg. tP<O.05 compared to control patients receiving > 2 x 10' marrow cells/kg.

50 x lOy/1 (control 39). This trend did not reach statistical significance (0.05 <P<O.l) . There was no association hM-CSF group and 2/46 in the controls. between dose of hM-CSF received and haematological recov- ery. There were no engraftment failures (defined as failure to

reach a n ANC of 0.1 x 10y/l by day 28 past ABMT) in the

The mean number of marrow MNC reinfused was similar in both groups ( I .93 f 0.7 x 10X/kg body weight in the

292 A. Khwaja et a1

60 -.

O.S. Yo 40 -.

20

Table IV. Comparison of transplant related morbidity between hM-CSF treated patients and controls with further breakdown according to marrow mononuclear cell dose reinfused.

I l l I I I I I I I I I I l l 1

CONTROLS IIIIII I 1 1 II I 1 I I

P.F.S. CONTROLS p = 0.380

% 40--

p = 0.092

-. 20 -.

9 -

hM-CSF patients Controls

Day of last platelet transfusion Platelet units transfused Days with fever> 37.5 "C Days of parenteral antibiotics

Parenteral antifungals % of patients no. of days

Days of hospitalization

26 81

5 .5 14.5

55 13

29*

29 112

8 1 7

58 18

33

hM-CSFz2 x 10' MNC per kg

17* 29 44 72 4 6

11.5 16

Control>2 x lo8 MNC per kg

38 65 13 18

21* 26

*P<O.05 compared with control patients.

hM-CSF group and 1.98f0.7 x 10X/kg in controls). There was no correlation between the number of MNC harvested and the number of previous treatment courses. Recovery of both neutrophils and platelets correlated significantly with the number of marrow cells reinfused in the hM-CSF patients (P=0.014 and P=O.O5 respectively), but not in the control group, the latter being in accord with our previous observa- tions (Anderson et al, 1986). Those patients treated with hM-CSF who received more than 2 x 10' marrow MNC/kg had improved haematological recovery when compared to patients receiving less than 2 x 10' MNC/kg. and had significantly accelerated platelet regeneration when com- pared to the 17 control patients who received > 2 x 1 Ox MNC (Table 111). hM-CSF treated patients receiving > 2 x 10' MNCjkg reached an ANC of 0 .5 x 10y/l by a median of day 16.5 (control 18.5, NS), became platelet transfusion inde- pendent by day 1 7 (control 29, P < 0 . 0 5 ) and reached a platelet count of 50 x 10y/l by day 2 1 (control 40, P<O.O5).

Blood product use There was no statistically significant difference in platelet transfusion between the treatment groups with hM-CSF patients requiring a median of 81 platelet units in the post transplant period, compared with 112 units for controls. Patients treated with hM-CSF who received > 2 x 10' MNC/ kg required a median of 4 4 platelet units (control 72, NS). Red cell transfusions were similar with each group requiring a median of 8 units per patient. No patient in either group received granulocyte transfusions.

lnfections All patients in the hM-CSF group and 4 5 of 4 6 patients in the control group had suspected bacterial infection. The median number of days with pyrexia >37.5OC was 5.5 in the hM-CSF group and 8 in controls (NS). 10/20 (50%) of the hM-CSF patients and 22/46 (48%) controls had positive blood culture proven infections. Similar patterns of infecting

Effect of M-CSF after ABMT 29 3

proven aspergillus pneumonia, and one from adult respira- tory distress syndrome secondary to bacterial sepsis. Of the other treatment related deaths, one patient died on day + 11 of hepatorenal failure of uncertain origin and another on day + 4 6 of respiratory failure following intestinal pneumonitis, probably due to the conditioning chemotherapy and prior mantle radiotherapy.

We have previously documented gut toxicity with the BEAM regimen (Matthey et a!, 1989) and particular care was taken to record episodes of abdominal pain and diarrhoea in hM-CSF treated patients and controls. As stated above, one patient had severe colitis and died on day + 32 ( 1 1 d after discontinuing hM-CSF) of gut infarction. Three other patients receiving hM-CSF had abdominal symptoms to give a n overall incidence of 20%. 11 control patients had severe abdominal symptoms (24%). 70'34 of hM-CSF patients required total parenteral nutrition for a median of 10 d, as did 65% of controls for a median 11 .5 d. No difference was detected between the two groups with regard to serum bilirubin (median maximum level observed 3 0 pmol/l for hM-CSF and 26 pmol/l for controls) or plasma creatinine levels (103 mmol/l and 1 0 4 mmol/l respectively).

organisms were seen in both groups with the predominant isolate being the coagulase negative staphylococcus (4/10 and 1 1 / 2 2 respectively). There was no significant difference between the two groups with regard to the use of parenteral antibiotics (1 4.5 d in hM-CSF patients and 1 7 d in controls), or the use of intravenous antifungal therapy (5 5% of patients in the M-CSF group for a median of 1 3 d versus 58% of controls for median 18 d). The incidence of significant pneumonitis, as determined by examination of chest X-ray appearance, oxygen saturation and clinical signs, was also similar with 7/20 ( 3 5%) so affected in the hM-CSF group and 19/46 (41%) in the control arm.

Inpatient stay The median inpatient stay following ABMT for the hM-CSF group as a whole was 29 d and for controls 3 3 d (P<0.05) . hM-CSF treated patients receiving > 2 x lo8 MNCjkg had a median inpatient stay of 21 d (control 26, P=0.05) .

Toxicity No adverse events occurred during 2 1 test infusions of hM-CSF. One patient had episodes of flushing associated with hM-CSF infusion following ABMT-these were not asso- ciated with any changes in vital signs and resolved sponta- neously while the patient continued to receive hM-CSF.

There were two early deaths in the hM-CSF patients giving a treatment related mortality (before day +90) of 10%. Patient 0 2 developed interstitial pneumonitis attributed to fungal infection on day +8. This required treatment with intravenous amphotericin B and 5 flucytosine, and transfer to the intensive care unit for continuous positive airways pressure ventilation. He subsequently developed progressive bilirubinaemia in association with normal hepatic enzymes, liver enlargement and ascites from day +14, with a maximum bilirubin of 1041 pmolll and died with a presumed diagnosis of hepatic veno-occlusive disease (VOD) on day + 34. Permission for post mortem examination was refused.

Patient 12 developed pneumonitis in the first week after ABMT and received treatment with broad spectrum anti- biotics and intravenous amphotericin B. At the same time he developed a n unusually severe syndrome of myositis. pre- dominantly affecting the legs, and toxic colitis. A maximum creatine kinase of 2510 was recorded on day +9. No myoglobinuria was found. The myositis resolved sponta- neously by day f l l , whilst the patient continued on hM-CSF. Having reached a normal neutrophil level and while still platelet dependent, he developed sudden onset abdominal pain on day + 3 2 and was found a t laparotomy to have infarcted the majority of his large intestine. Despite resection of the affected bowel, further areas of small intestine were observed to appear ischaemic and he died a few hours after surgery. Examination of the affected colon showed ischaemic colitis with no evident cause. Permission for post mortem examination was refused.

There were six early deaths in the control group to give a treatment related mortality of 13%. Four of these patients died of infectious causes before day + 14, prior to neutrophil recovery: two of these died from overwhelming bacterial infection leading to multi-organ failure, one from autopsy

Overall survival and relapse rates Figs 2(a) and 2(b) show that there is no significant difference in either overall or disease progression free survival between the two treatment groups.

DISCUSSION

In this phase 1/11 study we have examined the effects of escalating doses of hM-CSF on various parameters of recovery following ABMT. The overall group was composed of 6 6 patients with malignant lymphoma all receiving identical chemotherapy and unmanipulated bone marrow. hM-CSF was given to 2 0 patients by a daily 2 h intravenous infusion for 2 1 d, starting the day after marrow reinfusion. Overall recovery to WBC > 1 .0 x 1 OY/1 and to an ANC > 0.5 x 1 Oy/l was similar in both groups. Motoyoshi and colleagues reported that treatment with hM-CSF ameliorated the neu- tropenia seen following conventional dose chemotherapy for solid tumours (Motoyoshi eta!. 1986). Masaoka r t a1 (1 988) reported a more rapid neutrophil recovery in 3 3 patients undergoing allogeneic BMT for a variety of indications who received hM-CSF but not in 1 2 patients undergoing ABMT. N o details ofpreparative regimens were given and the median age of the group as a whole was only 1 5 years. These factors may account for the different responses seen. In our study, there was no significant difference in platelet recovery (32 d in hM-CSF group versus 39 d in controls to reach platelets > 50 x 10y/l) or in platelet transfusion requirement (81 versus 11 2 units respectively). Masaoka et a1 do not report the effect of hM-CSF treatment on platelet recovery in their trial. We were unable to detect any effect of increasing dose on haematological recovery but it should be noted that our lowest dose level corresponds to the maximum dose used by Motoyoshi et al and Masaoka et al.

We were able to show a marked and statistically significant correlation between the number of nucleated marrow cells

294 A. Khwaja et a1 infused and both platelet and neutrophil recovery in the hM-CSF treated patients (P=O.O5 and P=0.014 respect- ively) but not in the controls. Although retrospective sub- group analysis should be treated with a degree of caution, we found that hM-CSF treated patients receiving > 2 x 1 Ox marrow MNC/kg body weight reached an ANC of 0.5 x 109/1 by a median of day 16.5 and platelets > 50 x 1 09/1 by day 2 1. This patient group had a median inpatient stay after ABMT of only 21 d. M-CSF has been shown to enhance monocyte/ macrophage production of IL-1 (Moore et ul, 1980), IL-6 (Navarro et al, 1989). G-CSF and GM-CSF (Motoyoshi et ul. 1989) in vitro, and when given to cyclophosphamide treated mice led to an increased production of marrow CFU-M and CFU-GM and spleen CFU-GM, CFU-GEMM and BFU-E (Brox- meyer et al, 1987). As M-CSF did not affect CFU-GEMM or BFU-E production in vitro, the latter effects were presumed to be indirectly mediated by accessory cells. Our data suggest that hM-CSF may cause indirect enhancement of platelet recovery following ABMT and that large marrow cell inno- cula, containing both progenitor and accessory cells, are required to see a clear effect. We were unable to show any effect of a single hM-CSF test infusion on the production of secondary cytokines in sufficient quantities to be detectable in the peripheral circulation. It is possible that more prolonged exposure would be required to show such an effect, or that M-CSF exerts its effects locally in the bone marrow micro- environment.

We have also looked at the effect of hM-CSF treatment on infection following ABMT. Infection in this setting is often difficult to confirm microbiologically and therefore we have looked at other indicators in addition to culture results. No statistically significant reduction in the number of days with fever (5.5 in hM-CSF versus 8 in controls) or days of parenteral antibiotics (14.5 d versus 18 d) was seen. Positive blood culture rates were equivalent in the two groups with a similar spectrum of infecting organisms. The incidence of pneumonitis and antifungal use was also similar in both hM-CSF treated patient and controls. We have previously shown that hM-CSF will enhance monocyte migration and candidacidal activity in vivo (Khwaja et ul. 1991); however, it is clear from our data and those of others that the great majority of severe infections, in particular those causing patient death, arise in the period ofcomplete aplasia following ABMT. In common with other CSFs administered in this fashion, hM-CSF treatment did not affect the length of this period and it may be difficult to demonstrate a clear reduction in infectious complications in this setting without the benefit of large patient numbers, probably several hundred in each arm of a randomized trial. Recently, Nemunaitis et al (1 991) have shown that recombinant M-CSF may be useful in treating invasive fungal infection in bone marrow transplant recipients. However, the majority of their patients had either normal or near normal white blood cell counts at the time of receiving M-CSF and are therefore not directly comparable to our patient group. A small but significant reduction in the period of hospitalization was seen in the hM-CSF treated group in our trial (29 d versus 3 3 d) and this was most marked in those patients receiving > 2 x lo8 MNC/kg body weight (21 d).

Treatment with hM-CSF was apparently free of significant toxicity. No adverse events were seen during the administra- tion of a full test dose prior to conditioning chemotherapy. Two deaths occurred in patients receiving hM-CSF. One patient died of progressive hepatic VOD, a recognized compli- cation of BMT (McDonald et al, 1984). VOD with the BEAM protocol is a rare event, and this is the first fatal case in our experience of over 300 procedures. Although the possibility remains that hM-CSF may have been implicated in this death, we have been unable to document any adverse hepatotoxic or procoagulant effects of hM-CSF (K.Y., unpublished obser- vations) during test doses. The second death was in a patient who developed colitis during the aplastic phase which resolved spontaneously while still on hM-CSF and who developed sudden onset bowel infarction on day + 32, 11 d after his last dose of hM-CSF. BEAM is a regimen with significant bowel toxicity (Matthey et al, 1989) and we were unable to show any difference in the incidence of gastrointes- tinal symptoms and TPN requirements between hM-CSF treated patients and controls. The significance of the mode of this patient’s death and its relationship to hM-CSF treatment is unclear. There was no significant difference between the groups with regard to overall survival or relapse rates.

These results show that, in the overall treatment group, the administration of hM-CSF did not result in significant improvement in haematological recovery following BEAM and ABMT when compared with concurrent control patients. However, those patients receiving relatively large numbers of nucleated marrow cells had accelerated recovery, especially of platelets, and were able to be discharged from hospital significantly earlier than controls. These data suggest that the administration of a lineage-specific haernopoietic growth factor, as defined by in vitro assays, may lead to complex effects in vivo via its effects on accessory cells. Further prospective randomized studies of M-CSF in patients receiving large marrow innocula are justified.

REFERENCES

Anderson, C.C.. Coldstone. A.H., Souhami, R.L.. Linch, D.C.. Harper, P.G., McLennan. K.A.. Jones, M.. Machin. S.J., Jelliffe, A.M., Cawley, J.C. & Richards, J.D.M. (1984) Very high dose chemo- therapy with autologous bone marrow rescue in adult patients with resistant relapsed lymphoma. Cancer Chemotherapy Pharma-

Brandt. S.J.. Peters, W.P., Atwater, S.K.. Kurtzberg, J., Borowitz, M.J., Jones,R.B., Shpa1l.E.J.. Bast,R.C..Gilbert, C.J.&Oette.D.H. (1988) Effects of recombinant human granulocyte-macrophage colony- stimulating factor on hernatopoietic reconstitution after high-dose chemotherapy and autologous bone marrow transplantation. New England Journal of Medicine. 318, 869-876.

Broxmeyer, H.E., Williams, D.E.. Cooper, S . . Waheed, A. & Shadduck, R.K. (1987) The influence in vivo of murine colony-stimulating factor-1 in myeloid progenitor cells of mice recovering from sublethal dosages of cyclophosphamide. Blood, 69, 91 3-918.

Clark, S.C. & Kamen. R. (1987) The human hematopoietic colony- stimulating factors. Science, 236, 1229-1237.

Devereux, S., Lincb, D.C.. Gribben, J.G., McMillan, A,. Patterson, K. & Goldstone, A.H. (I 989) GM-CSF accelerates neutrophil recovery after autologous bone marrow transplantation for Hodgkin’s disease. Bone Marrow Transplantation, 4, 49-54.

cology, 16, 170-175.

Efleect of M-CSF after ABMT 295

Durum, S.K.. Schmidt, J.A. 6; Oppenheim. J.J. (1985) Interleukin I: an immunological perspective. Annual Review of Immunology. 3 , 263-285.

Hanamura. T.. Motoyoshi. K.. Yoshida, K., Saito, M., Miura, Y., Kawashima, T., Nishida. M. &Takaku, F. (1988) Quantitation and identification of human monocytic colony-stimulating factor in human serum by enzyme-linked immunosorbent assay. Blood, 72,

Horii. Y.. Muraguchi. A.. Suematsu. S., Matsuda, T., Yoshisaki, K., Hiranon, T. & Kishimoto. T. (1988) Regulation of BSF-2/IL-6 production by human mononuclear cells. Macrophage dependent synthesis of BSF-Z/IL-h by T cells. journal of’ Immunology 141, 1529-3 535.

Khwaja. A,. Johnson, B.. Addison. I.E.. Yong. K., Ruthven, K.. Abramson. S. & Linch. D.C. (1991) I n vivo effects of macrophage colony-stimulating factor on human monocyte function. British Journal of Haemntology, 77, 2 5 - 3 I .

Komiyama. A,. Ishiguro. A,, Kubo. T.. Yasukohchi, S., Yasui. K., Yanagisawa. M.. Yamada, S., Yamazaki. M. & Akabane, T. ( 1 988) Increases in neutrophil counts by purified human urinary colony- stimulating factor in chronic neutropenia of childhood. Blood, 71, 41-45.

Linch. D.C., Knott, L.J.. Patterson. K.G.. Cowan. D.A. & Harper, P.G. (1982) A comparison of three methods of bone marrow progenitor cell concentration prior to cryopreservation. Journal of Clinical Pathology, 35, 186-190.

Masaoka, T., Motoyoshi, K.. Takaku, F.. Kato. S.. Harada. M.. Kodera. Y.. Kanamaru, A.. Moriyama. Y.. Ohno. R.. Ohira, M., Shibata. H. & Tnoue, T. (1988) Administration of human urinary colony- stimulating factor after bone marrow transplantation. Bone Marrow Transplantation, 3, 121-127.

Matthey. F.M.. Khwaja. A,, Behrens. R. & Linch. D.C. (1989) Effects of high dose conditioning chemotherapy prior to autologous bone marrow transplantation on intestinal permeability. Bone Marrow Transplantation. 4, (Suppl. 2). 29.

McDonald, C.B.. Sharma. P.. Matthews, D.E.. Shulman, H.M. & Thomas. E.D. (1984) Venocclusive disease of the liver after bone marrow transplantation: diagnosis. incidence and predisposing factors. Hrpatologg. 4, 1 16-1 22.

Moore. R.N., Oppenheim, J.J., Farrar, 1.1.. Carter, C.S., Waheed, A. 8. Shadduck. R.K. (1 980) Production oflymphocyte activating factor (interlcukin-1 )by macrophages activated with colony-stimulating factors. Joirrrial of Immunology, 125, 1302-1 305.

886-892.

Motoyoshi. K., Takaku, F.. Maekawa, T., Miura, Y.. Furusawa. S., Hattori. M.. Nomura, T.. Mizoguchi. H., Ogawa. M.. Kinugasa, K., Tominaga. T.. Shimoyama. M.K.D., Ohta, K.. Taguchi. T.. Masaoka, ‘r. & Kimura. I. (1 986) Protective effect of partially purified human urinary colony-stimulating factor on granulocyto- penia after antitumor chemotherapy. Experimr~ntul Hematology. 14, 1069-1075.

Motoyoshi, K., Yoshida. K.. Hatake. K . , Saito. M.. Miura, Y.. Yanai, N.. Yamada, M., Kawashima. T., Wong. G.G.. Temple, P.A., Leary, A.C.. Witek-Giannotti, J.S., Fujisawa, M., Yuo. A,. Okabe, T. & Takaku, F. (1989) Recombinant and native urinary colony- stimulating factor directly augments granulocytic and granulo- cyte-macrophage colony-stimulating factor production of human peripheral blood monocytes. Experirnentul Hematology, 17, 68-71,

N-HLPC (Non-Hodgkins Lymphoma pathologic classification pro- ject) (1 982) National Cancer Institute sponsored study of classifi- cations of non-Hodgkins lymphomas: Summary and description of a working formulation for clinical usage. Cancer, 4 9 , 2 1 12-2 13 5.

Navarro. S., Debili, N.. Bernaudin. J.-F., Vainchenker. W. & Doly, 1. (1989) Regulation of the expression of II,A in human monocytes. Journal of Immunology, 142, 4339-4345.

Nemunaitis, J.. Singer, J.W., Buckner, C.D.. Hill, R. , Storb. R., Thomas, E.D. & Appelbaum. F.R. (1988) Use of recombinant human granulocyte-macrophage colony-stimulating factor in autologous bone marrow transplantation for lymphoid malignan- cies. Blood. 72, 834-836.

Nemunaitis. J., Meyers. J.D., Buckner, C.D., Shannon-Dorcy, K.. Mori. M.. Shulman. H.. Bianco. J.A.. Higano. C.S., Groves. E.. Storb, R., Hansen, 1.. Appelbaum. F.R. & Singer, J.W. ( 199 1 ) Phase I trial of recombinant human macrophage colony-stimulating factor in patients with invasive fungal infections. Blood, 78, 907-9 1 3.

Sheridan. W.P.. Morstyn, G.. Wolf, M.. Dodds. A.. Lusk. J., Maher, D., Layton, J.E., Green, M.D.. Souza. L.M. & Fox, R.M. (1989) Granulocyte colony-stimulating factor and neutrophil recovery after high-dose chemotherapy and autologous bone marrow transplantation. Lancet, ii. 891 -895.

Warren, M.K. & Ralph. P. (1986) Macrophage growth factor CSI7-1 stimulates human monocyte production of interferon. tumor necrosis factor. and colony stimulating activity. Journal of’frnmuno- logy. 137, 2281-2285.