British Jourrial of Haematology, 1989. 73, 475-479

Mechanism of action of a interferon in chronic granulocytic leukaemia: evidence for preferential inhibition of late progenitors

D. W. G A L V A N I A N D J . C . CAWLEY University Department of Haernatology. Royal Liverpool Hospital

Received 29 March 1989; accepted for publication 25 July 1989

Summary. The effect of a interferon ( a IFN) on colony forming unit, granulocyte-macrophage (CFU-GM) formation by nor- mal bone marrow (BM) as compared with chronic granulocy- tic leukaemia (CGL) BM and peripheral blood (PB) was tested in semi-solid assay systems employing either 5637CM or recombinant granulocyte-macrophage colony stimulating factor (GM-CSF) to support growth. aIFN (> 125 U/ml) caused consistent inhibition ( P = 0.02) ofday-7 (late progeni- tor) colonies, but had little or no effect on either day-7 clusters or day-I 4 colonies/clusters. This selective effect on day-7 colonies was quantitatively similar for both normal and CGI,

( P > O . 5 ) . Similar results were obtained whether or not the mononuclear preparations were depleted of potential acces- sory cells, suggesting that the aIFN-suppression is directly mediated. Morphological examination of colonies and clus- ters showed that IFN had no effect on cell maturation and that colony inhibition is not, therefore, a consequence of blocked maturation. Since the late-progenitor compartment is preferentially expanded in CGL. we suggest that our demonstration that aIFN seleclively inhibits this compart- ment is relevant to the clinical effects of the cytokine in the disease.

The clinical effects of a interferon (aIFN) in chronic granulo- cytic leukaemia (CGL) are now becoming clear. The cytokine is highly effective in controlling the peripheral white cell count once initial cytoreduction has been achieved with conventional chemotherapy (Talpaz et al. 1986). It is less successful in reducing very leukaemic disease and in the accelerated phase (Gastl et al. 1987). More normal meta- phases appear than with standard therapy. but even in well controlled disease, the Philadelphia (Ph) chromosome is only occasionally completely removed (Talpaz et al, 1987).

The mechanism of action of aIFN in CGL. however, remains unclear. Work in this laboratory suggests that an indirect effect via natural killer (NK) cells is unlikely (Galvani et nl, 1989). We have shown aIFN has no effect on lymphocyte activated killer (LAK) or T-cell cytotoxicity in hairy cell leukaemia, a disease which is highly responsive to rIFN (Griffiths & Cawley. 1988).

A direct effect of aIFN on CGL progenitors therefore seems more likely and it is now well established that rIFN directly

Correspondence: Dr D. W. Galvani, Department of Haematology. 3rd Floor Duncan Bldg, Royal Liverpool Hospital, Prescot Street, Liverpool L69 3BX.

inhibits normal progenitors (Broxmeyer c>t al. 1983). The demonstration of increased 2'5'-oligoadenylate synthetase in CGL patients responding to aIFN therapy (Rosenblum rt nl, 1986) is in accord with this hypothesis. The c-ablibcr rearrangement disappears in parallel with the (Ph) chromo- some in a proportion of responsive cases (Yoffe rt al, 1987). but how aIFN achieves this reduction in the leukaemic clone remains obscure.

Previous colony work has shown that a t high concentra- tions aIFN has a general suppressive effect on progenitors of normal marrow detected as day-1 4 granulocyte-macrophage colony forming units (CFU-GM) (Broxmeyer rt al, 1983; Rigby et al. 1985: Beran ef al. 1988). Previous work in CGL has concentrated on day-14 CFU-GM which at the high doses of IFN employed, were shown to be suppressed to the same extent as normal day-14 colonies (Dowding et al. 1987: Oladipupo Williams Pt al. 1981). The day-14 CFU-GM corresponds to a n early progenitor, the effects of aIFN on the more mature (late) progenitor that corresponds to day-7 CFU-GM is less well defined.

In the present study we compare the effects of therapeutic concentrations of aIFN on day-7 versus day-14 colony/ cluster formation in normals and CGL and show that the cytokine is particularly inhibitory to day-7 colonies. Although comparable inhibition was demonstrated in nor-

4 7 5

476 D. W. Galvani and 1. C. Cawley mals and CGL, this observation bears relevance to the possible mechanism of action of aIFN in CGL.

PATIENTS AND METHODS

Patients. A total of 1 3 patients with typical chronic phase CGL were studied. Three patients were studied prior to commencing therapy; five patients during chronic phase controlled by hydroxyurea: and five patients whilst receiving aIFN (Wellferon) therapy. Although patients receiving aIFN clinically achieved haematological control, no patient had > 30% reduction in Ph metaphases in their marrow. Further, data relating to patients receiving the cytokine therapeuti- cally is kept entirely separate from the other CGL patients. Marked resistance to the effects ofuIFN were not seen in-vivo or in-vitro in this study.

Bone marrow aspiration and venepuncture were per- formed following informed consent. Normal bone marrow was obtained from patients undergoing thoracic or routine surgery following informed consent in accord with the Hospital Ethical Committee.

Intrvferon. Most experiments employed lymphoblastoid aIFN (Wellferon) as this was the agent used in the clinic. Wellferon therapy produces a serum level of uIFN of approxi- mately 100-200 U/ml (A. W. Nethersell. personal communi- cation). we therefore focused attention on these therapeutic concentrations in in-vitro assays. As Wellfcron contains human plasma protein (HPP) for stabilization purposes, experiments were performed using HPP concentrations equivalent to those found in the Wellferon preparation: no significant effect upon colony/cluster growth was observed. Furthermore, in three experiments recombinant human IFN- a?.,. (Roferon) was employed and this demonstrated identical inhibition to Wellferon (dose for dose).

Cell separation. Mononuclear cell preparation. Mononuclear cells from bone marrow (BM) and peripheral blood (PB) were separated by Ficoll-Hypaque centrifugation. The interface cells were then washed and prepared for plating in the colony assay.

Removal of T cells and nionocgtes. In those experiments employing purified myeloid progenitors, T-lymphocytes were removed by rosetting with aminoethyl isothiouronium bro- mide (AET)-treated sheep red blood cells and monocytes were removed by plastic adherence. Purity was checked by staining with monoclonal antibodies (Mabs) to CD3 and CD14 and < 1% of cells were positive for either Mab.

Further removal of potential accessory cells. In two experi- ments T and B lymphocytes, NK cells and monocytes were selectively removed by incubating mononuclear cells with Mabs to CD3 (Leu4), CD19 (LeulZ), Leu7 and CD14 (Leu- M 3 ) respectively, a t 4°C for 2 0 min. Cells were then incubated with sheep anti-mouse immunoglobulin coated magnetic beads (Dynal) a t 4 O C for 20 min. Bead-positive cells were removed by exposure to a magnet for 5 min: the bead- negative population was then washed before culture. Purity was checked following incubation overnight to allow re- expression of antigens on potentially contaminating cells. However, indirect immunofluorescence following a repeat standard two-layer antibody labelling technique revealed no

contaminating cells (< 1%) and < 1% lymphoid or monocy- tic cells were identifiable morphologically (the preparations consisted of immature erythroid and myeloid forms).

CFU-GM assay. BM cells were plated at 1 x lo5 cells/plate. PB cells were seeded at 2 x 1 Oi cells/plate. Cells were grown in 35 mm diameter standard tissue culture dishes (Flow) containing 1 ml of a mixture of RPMI medium, 0.3% agar (Difco), 20% fetal calf serum (Flow), L-glutamine and strepto- mycin/penicillin. Clonal growth was stimulated using 10% 563 7 conditioned medium ( 563 7CM). Alternatively, growth was stimulated with 100 U/ml of recombinant human granulocyte-macrophage colony stimulating factor (GM- CSF) (Glaxo). The choice of these concentrations of growth factors had been determined by previous experiments to optimize growth. All cultures were performed in triplicate. When cultures were continuously exposed to aIFN. the cytokine was added to the cell mixture at initial plating. Cultures were incubated at 3 7°C in a humidified atmosphere of 5% C 0 2 and plates scored at 7 and 14 d for colonies ( > 40 cells) and clusters (3-39 cells).

Two sets of experiments were performed in which non- purified BM cells were pre-incubated with concentrations of Wellferon for 2 , 6. 12 and 24 h before being washed three times with RPMI and then plated in the CFU-GM assay.

In certain experiments using methylcellulose (0.8'j/,) instead of agar, aIFN was added at the beginning of culture and the same dose was again added to the culture system at day 7. This additional manipulation was performed in case the activity of the aIFN had been lost during the first 7 d of culture.

Morphology. Colonies and clusters were removed from culture plates with a Gilson 100 111 pipette using an Olympus inverted microscope and transferred to microtitre plates for washing. Individual colonies were washed and cytocentri- fuged on to a slide. Three to four clusters were pooled and washed together before cytocentrifugation. The best morpho- logical quality was obtained with 0.Srj: methylcellulose rather than 0.3% agar. Slides were stained with a standard May Grunwald Giemsa (MGG) stain and 100 cell differential counts performed. In selected instances alkaline phosphatase anti-alkaline phosphatase (APAAP) technique was per- formed to confirm macrophage numbers (Dako-Mac Mab).

Thymidine incorporation. Purified or non-purified cells were placed in flat-bottomed microtitre plates at 1 x lo5 cells in 200 p1 ofmedia (RPMI, 10% fetal calf serum. 10% 5637CM, bglutamine. streptomycin/penicillin). aIFN was added to cultures to give final concentrations of 1-3000 U/ml. Each dose was analysed in triplicate. Following 72 h of culture, 20 111 of %H labelled thymidine (Amersham) was added to each well (18.5 kBq). 12 h later, cells were harvested and filters dried and immersed in scintillation fluid prior to counting on a 11 counter. Thymidine uptake was expressed in counts per minute (cpm).

Statistics. The effect of uIFN upon colony and cluster formation was plotted and regression analysis was performed (after the method of Armitage) and the correlation coeffi- cient(s) used to calculate a t-value. Standard probability tables were then used to calculate P.

When normal results were compared to CGL results (either

Action of CI lnterftron in CGL 4 7 7 Table 11. Effect of aIFN on purified progenitors from normals and CGL.* Results expressed as percentage ( f SEM) of control plates without n1FN.t

colony inhibition or thymidine uptake). a standard Student’s t-test was performed.

RESULTS

Effct of alFN on coloriy fornintion by iirlfractionated Bhl mononucleur cells aIFN caused a consistent dose-dependent inhibition of day-i colony growth from normal BM and CGL BM and PB; this was true whether 5637CM or human rGM-CSF was used to support growth (Table I). A less marked, non-significant reduction of day-14 colonies was also observed. Addition of aIFN at day-7 to colonies grown in methyl cellulose had no effect on day 14 colony or (cluster) growth, either in normals or CGL (data not shown).

Normal and CGL day-7 colonies were similarly inhibited ( P > O . 5 ) over a range of concentrations of aIFN (Table I) . Similar inhibition was observed when cells were exposed to aIFN for 2 . 6 , 12 and 24 h and then washed before plating in agar (data not shown). The inhibitory effect of aIFN was, however. most marked in those samples exposed to aIFN for the longest period of time.

When similar experiments were performed in methyl

Table I. Effect of rIFN on normal and CGL progenitors.* Results expressed as percentage (1SEM) of control plates without a1FN.t

~

Day 7 Day 14 GtlFN Source of (U/ml) material Colonies Clusters Colonies Clusters

0 NormalBM 100 100 100 100 -

CGL BM + PB 1 2 5 NormalBM 6 0 1 8 1 1 6 f 1 6 87112 77+12

CGL BM 571.12 9 9 1 1 3 7 6 f l 1 1 4 1 f 4 5 CGL PB 71f 7 76f17 87f 5 137f15

250 NormalBM 54f 4 1 1 9 f 2 5 8 6 f 3 140f35 CGL BM 37&10 6 8 1 h 92f 7 152143 CGL PB 5 2 1 8 9 4 f 2 6 7 8 f 1 2 l 0 l f 7

500 NormalBM 3331 8 107f17 73117 81f 8 CGL BM 32f14 85f 7 5 3 f 1 6 129f29 CGL PB 47f12 80f28 8 1 h l l 131114

1000 NormalBM 27f 7 119f32 621.21 86+15

CGL PB 23f 4 91f22 73f10 136f13 3000 NormalBM 18f 5 6 3 f 1 5 673~15 1183x36

CGL BM 2 6 f l l 83f 7 64h 9 1 3 8 3 2 7

CGL BM 23511 86325 7 0 f 2 1 182159 CGL PB 2 3 f 5 100424 78f15 131f10

Significance of inhibition ( P ) Normal BM 0.02 >0 .5 >0 .5 >0 ,5 CGL BM 0.02 >0.5 >0.5 >0.5 CGL PB 0.01 >0.5 >0.5 >0 .5

*Normal BM. n=4: CGL BM. n= 3: CGL PB. n=4. tColony numbers in control plates. Normal BM: day-7. 16-142

(mean=55), day-14 6-29 (mean=23). CGL BM: day-7 39-119 (mean=87). day-14 4-43 (mean=33). CGL PB: day-7 15-97 (mean=54), day-14 17-30 (mean=26).

Day 7 Day 14

aIFN Source of (Ujml) material Colonies Clusters Colonies Clusters

0 NormalBM 100 100 100 CGL PB

125 NormalBM 64h12 l l O f 4 9 6 f 8 CGLPB 53f14 95f 6 1 0 6 f l l

250 NormalBM 42f20 1 0 0 f 7 80&14 CGLPB 44f15 109f10 116113

500 NormalBM 36f16 9 4 f 6 9 3 1 1 3 CGLPB 26f14 99f 4 93f-11

1000 NormalBM 3 6 f 1 0 9 3 f 7 75114 CGLPB 22f 7 103f 6 98f 9

3000 NormalBM 1 9 1 9 9 1 f 1 6 5 6 f 1 0 CGL PB 9 % 4 9 1 f 1 0 108514

100

102* 9 l l 0 f l l

1 0 3 f 9 120f18 101f15 1133~12

93* 7 129f23 97f26

134 f 20

Significance of inhibition ( P ) NormalBM 0.01 r 0 . 5 0.5>P>0.1 1 0 . 5 CGL PB 0.01 >0.5 >0.5 >0 .5

*Normal BM, n = 4; CGL PB, n = 5. tColony numbers in control plates. Normal BM: day-7 30-95

(mean=47), day-I4 15-52 (mean=30). CGL PB: day-7 18-141) (mean=97); day-14 15-185 (niean=52).

cellulose and colonies and clusters removed for morphologi- cal examination, the composition of both clusters and colonies was similar in the presence or absence of rIFN (data not shown).

Efject q f a l F N o n colot~yforr~iation by pur$rd RMprogetiitor cells With 563 7CM as stimulus. nIFN again produced a consistent inhibition of day-7 colonies, from normal BM and CGL PB (Table 11); day- 14 colonies were unaffected by concentrations of nIFN up to 1000 U/ml (Table 11). In the two experiments employing magnetic beads to deplete the samples the effect of aIFN on colony formation had exactly similar characteristics ( P > 0 . 5 when results were compared to Table 11; e.g. 52% inhibition at 125 U/ml).

Identical inhibition of colonies was again observed when rGM-CSF was used instead of 563iCM to support growth.

In methyl cellulose culture the morphology of both colonies and clusters again remained unchanged whether or not aIFN was added. Thus, in normal colonies without rIFN ( n = 2 ) the differential was blasts 3%, promyelocytes and myelocytes 2%, rnetamyelocytes and neutrophils 18% and macrophages 77%. Whereas in normal colonies grown in 500 U/ml EIFN (r7=2) the differential was blasts 2%. promyelocytes and myelocytes 2%. metamyelocytes and neutrophils 20% and macrophages 76%. A similar cell differential was observed in experiments with CGL BM ( n = 2) and aIFN had no effect on the cell distribution observed.

478 D. W. Galvani and 1. C. Cawley

I 0 untreated

treated

Colonies clusters colonies clusters 0 AY -7 0 AY- I4

Fig 1. Effect of in-viw administration of aIFN on in-vitro purified progenitor growth (meanfSEM) from CGL BM. 5 6 3 i C M used as growth stimulus throughout. (Untreated n = 4, treated 17 = 5.) Mean ( f S E M ) day-7 colony formation on no d F N therapy= 3 0 f 7 ; during aIFN therapy= 3 f 1 (P<O.Ol). Mean (kSEM) day-14 colony formation on no aIFN therapy = 14 f 7: during aIFN ther- apy = 6 f 2 (P> 0.1).

Eflects of in-vivo administration of uIFN oil in-vitro colony formation As with in-vitro exposure to aIFN. day-7 colony formation (from purified BM) was reduced in CGL patients receiving aIFN in-vivo (Fig 1). Day-7 cluster and day-1 4 cluster/colony formation was similar in CGL patients on and off therapy (Fig 1).

Effects of alFN on 3H-tliyrnidine incorporation by purijied norrnal B M and CGL B M rIFN caused identical percentage inhibition of day-3 thymi- dine incorporation in normals and CGL. Thus, normal BM proliferation was reduced from 39 199* 5547 to 2 2 0 7 4 i 3 2 8 8 cpm at 1 2 5 U/ml crIFN, a percentage inhibi- tion of 44% (P=O.O5) . CGL BM proliferation was reduced f r o m 4 7 3 8 9 f 5 1 7 3 t o 2 5 9 9 3 f 4 0 9 3 a t 125U/mlaIFN,a percentage inhibition of 45% ( P = O . 0 5 ) . When normal and CGL values were compared there was no statistically signifi- cant difference between these groups ( P > 0 . 5 ) . A similar effect was seen at day 1. Following longer incubation in liquid culture (cultures fed and recharged with aIFN every 3 d), this inhibitiory effect of aIFN on jH-thymidine uptake became less marked.

DISCUSSION In the present study we have shown that aIFN at therapeutic concentrations has a selective suppressive effect on late

progenitors as manifested by reduction of day-7 colony formation. Day-7 cluster and day-14 colony and cluster formation were unaffected or minimally inhibited. Compar- able selective inhibition of day-7 colonies was observed in normal BM and CGL BM and PB. This same effect was demonstrated in BM cultured from CGL patients receiving aIFN therapy.

Identical results were obtained when material was de- pleted of potential accessory cells including T and B lympho- cytes, NK cells and nionocytes. This suggests that the inhibitory effect of aIFN on late progenitors is directly mediated.

Identical inhibition of day- 7 colonies occurred whether the cells were cultured continuously with ctIFN or briefly exposed to the cytokine before plating. The lack of inhibition late in culture is unlikely to be attributable to inactivation of aIFN during incubation since the material is stable and since addition of cytokine to methyl cellulose culture at day 7 had no further effect on day- 14 colony/cluster formation.

Our demonstration. that primitive progenitors are less inhibited by aIFN than are late precursors. is in keeping with the demonstration that the early progenitor is less responsive to regulation by colony stimulating factors (Johnson et nl. 1977; Eaves & Eaves, 1987). All these observations are difficult to reconcile with the concept that day-7 CFU are derived from day-14 CFU and that day-14 colonies will therefore have passed through a developmental stage equiva- lent to the late progenitor. The actual developmental sequence may therefore not be as simple as this and perhaps day-14 progenitors do not necessarily develop through a day- 7 phase.

Since aIFN can cause differentiation ofcell lines and since it has been reported to influence the maturation of bone marrow colonies (Verma et al, 1979). careful examination of possible morphological effects in our culture system seemed important. Like Broxmeyer et a1 (1983). we showed that aIFN had no effect on the cellular composition of colonies or clusters either in normals or CGL. Furthermore, reduction of day-7 colonies was not accompanied by a n increase in day-7 cluster formation. It therefore seems likely that the aIFN effect is on proliferation rather than differentiation.

We have no direct proof that colonies/clusters were all Ph positive and therefore part of the leukaemic clone. However. it is generally accepted that the vast majority of CFU-GM grown from CGL patients are indeed Ph positive (Moore & Metcalf, 1 9 73: Dube et al. 1984). Furthermore, Dowding et a1 (1987) recently showed that CGL colonies grown in the presence of high concentrations of aIFN remained predomin- antly Ph positive.

It is not clear from our data why aIFN has a selective direct effect on late (day-7) progenitors, whether from normals or CGL. However. there is evidence both in normals (Cashman at al, 1985; Jacobsen et al, 1978) and CGL (Eaves et al, 1986) , that more late progenitors are in cell cycle than are early precursors and they may therefore be more responsive to the well-known anti-proliferative effects of dFN.

It is now widely accepted that there are no fundamental differences in the kinetics of normal progenitors and their leukaemic counterparts (Andreeff. 1986); our limited 'H-

Action of s( lnterferon in CGL 479 Eaves. A.C.. Cashman. J.D., Gabourg. L.A.. Kalousek. D.G. & Eaves,

C.J. ( 1986) Unregulated proliferation of primitive CML progenitors in the presence of normal adherent cells. Proceedings qf the National Academy ojSciencesoJtlie Unitedstates ofAmerica. 83. 5 306-5310,

Eaves, C.J. &Eaves, A.C. (1987) Cell culture studies in CML. Chronic Myloid Leukaernia (ed. by J . M. Goldman). pp. 93 1-962. Bailliere Tindall. London.

Galvani. D.W., Owens. W.. Nethersell, A.B.W. & Cawley. J.C. (1989) The beneficial effects of nIFN in CGL are probably not due to NK cells. British lournal of Haernatology. 71, 2 3 3-2 3 7.

Gastl. G., Aulitzky. W.. Tilg, H.. Luttichau. I.. Konwalinka. G.. Huber. C. & Geissler. D. (1987) Effect of recombinant IFNn on the growth of haemopoietic progenitor cells in CML and its relationship to the clinical response. Irnrnunobiology, 175, 288-292.

Griffiths. S.D. & Cawley. J.C. (1988) n interferon andLAK cell activity in hairy-cell leukemia. Leukemia. 2, 377-381.

lacobsen. N., Broxmeyer. H.E.. Grossband. E. 8: Moore, M.A.S. (1 978) Diversity of human granulopoietic precursor cells: separation of cells that form colonies in diffusion chambers (CFU-d) from populations of colony-forming cells in vitro (CFU-c) by velocity sedimentation. Blood. 52, 221-232.

Johnson. G.R.. Dresch. C. & Metcalf. D. (1977) Heterogeneity in human neutrophil. macrophage and eosinophil progenitor cells demonstrated by velocity sedimentation separation. Blood. 50,

Moore, M.A.S. & Metcalf. M. (1 973) Cytogenetic analysis of human acute and chronic myeloid leukaemia cells cloned in agar culture. lnternntional lournal of Cancer. 11, 143-1 52.

Oladipupo Williams, C.K.. Svet-Moldavskaya, I.. Vilcek. J., Ohnunia. T. &Holland, J.F. (1981) Inhibitory effectsofhuman leukocyte and fibroblast interferons on normal and CML granulocyte progenitor cells. Oncology, 38, 3 56-360.

Rigby. W.R.C.. Ball, E.D.. Guyse. P.M. & Fauger. M.W. (1985) The effects of recombinant DNA-derived interferons on the growth of myeloid progenitor cells. Blood. 65, 858-862.

Rosenblum. M.G.. Maxwell, B.L.. Talpaz. M., Kelleher. P.]., McCredie. K.B. & Glutterman. J.U. (1986) In vivo sensitivity and resistance of CML cells to aIFN: correlation with receptor binding and induction of 2'. 5'-oligoadenylate synthetase. Cancer Research. 46, 4848- 4852.

Strife. A. & Clarkson. B. (1988) Biology of chronic myelogenous leukemia: Is discordant maturation the primary defect; Seminnrs in Hernutologg. 25* 1-19.

Talpaz, M., Kantarjian. H.M.. McCredie, K.B.. Keating, M.J. & Gutterman. J.U. ( 1986) CML: hematologic remissions and cytoge- netic improvement induced by recombinant nIFN. New England Journal o/ Medicine, 314, 1065-1069,

Talpaz. M.. Kantarjian. H.M., McCredie. K.B.. Keating. M.].. Trujillo. 1. & Gutterman, 1.U. (1987) Clinical investigation of human EIFN in CML. Blood, 69, 1280-1288.

Verma. D.S.. Spiher. G., Gutterman. J.U.. Zander. A.R.. McCredie. K. & Dicke. J.A. (1 979) Human leukocyte interferon preparation blocks granulopoietic differentiation. Blood. 54, 142 3-42 7.

Yoffe. G.. Blick, M.. Kautarjian. H.. Spitzer. G.. Gutterman. J . & Talpaz. M. (1 987) Molecular analysis of interferon-induced sup- pression of Ph chromosome in patients with CML. Blood. 69.961- 963.

823-831.

thymidine incorporation data is in accord with this conclu- sion. Why, then, does the clinical administration of aIFN produce selective inhibiiton of the Ph clone in some patients? There is now considerable evidence that the late progenitor compartment in CGL is preferentially expanded as compared with earlier progenitors (Eaves et R Z , 1986; Strife & Clarkson. 1988) and our data also shows a modest increase in the late progenitor compartment. We therefore suggest that the present demonstration that ctIFN selectively inhibits late progenitors accounts for the clinical efficacy of the cytokine in CGL. Thus comparable percentage inhibition of normal and CGL late progenitors will produce a greater absolute reduction of CGL cells. During prolonged clinical administra- tion this effect will encourage the preferential restoration of normal haematopoiesis.

ACKNOWLEDGMENTS We are grateful to Wellcome for their support and supply of Wellferon. We thank Dr J. Carron and Dr E. G. H. Rhodes for guidance and advice regarding the culture assays and Dr J. M. Davies and Dr C. R. M. Hay for allowing us to study their patients. Members of the Liverpool University Department of Surgery have been helpful at all times in the procurement of normal bone marrow, in particular Professor R. Shields. Professor T. Cooke, Mr A. Kingsnorth and Mr S. Leinster. Finally, the expert secretarial help of Mrs Pamela Knowles is greatly appreciated.

REFERENCES Andreeff. M. (1 986) Cell kinetics in leukaemia. Seminars in Haeninto-

Beran. M.. McCredie, K.B.. Keating, M.J. & Gutterman. J.U. (1988) Anti-leukemic effect of recombinant TNF in vitro and its modula- tion by n and alFNs. Blood. 72, 728-738.

Broxmeyer. H.E.. Lu, L.. Platmer. E., Feit. C.. Juliano, L. & Rubin. B.Y. (1 98 3) Comparative analysis of the influences of human n, 8. and nIFNs on human multipotential (CFU-GEMM). erythroid (BFU-E) and granulocyte-macrophage (CFU-GM) progenitor cells. lournal of Inrmunulogy, 131, 1300-1305.

Cashman. J.. Eaves. A.C. &Eaves. C.J. (1985) Regulated proliferation of primitive hemopoietic progenitor cells in long-term human marrow cultures. Blood. 66, 1002-1005.

Degliantoni. G.. Perussia. B., Mangoni, L. & Trinchieri. G . (1985) Inhibition of BM colony formation by human NK cells and by NK cell-derived colony-inhibitory activity. Journal of Experimental Medicine. 161, 1 1 52-1168.

Dowding. C., Gordon, M.Y. & Coldman. J.M. (1987) Effect of recombinant nIFN on normal and CGL progenitors. Experimental Haemntology. 15, 441.

Dube, I.D.. Gupta. C.M.. Kalousek. D.K.. Eaves, C.J. & Eaves, A.C. (1984) Cytogenetic status of early myeloid progenitor compart- ments in Ph positive CML: persistence of Ph negative committed progenitors that are suppressed from differentiating in v i m British lourrial of Haemntology. 56. 63 3-644.

log$. 23, 300-314.