Transient Thrombocytopenia Produced by Administration of Macrophage Colony-Stimulating Factor: Investigations of the Mechanism

MACROPHAGE colony-stimulating factor (M-CSF; CSF-1) is essential for the survival, growth, and development of the monocyte/macrophage cell lineage.1-5 Experiments in normal6 and M-CSF–deficient (op/op) mice7,8suggest that M-CSF has a major regulatory role in the development and maintenance of mononuclear phagocytes in liver, spleen, and kidney. Most circulating M-CSF is cleared by binding to its receptor on the macrophages of the liver and spleen, with subsequent endocytosis of the receptor-ligand complex and intracellular degradation.9,10M-CSF has been shown to acutely increase monocyte levels when administered to various animals and man. The magnitude of reported increases has been variable, but dose/response relationships have been demonstrated.5,6,11-14 Some investigations also have reported a return toward normal of elevated monocyte levels with continued M-CSF administration.12,13,15,16 Bone marrow monocyte production has been shown to be increased¹²; therefore, the decrease in circulating monocytes during prolonged M-CSF administration may be due to increased movement of monocytes into tissues.17 M-CSF administration has been shown to cause increases in the weights of livers and spleens, with both livers and spleens exhibiting increased infiltrates of macrophages.18,19 No consistent changes in the total white blood cell (WBC), lymphocyte, or neutrophil counts are produced by M-CSF, but occasionally increases in these cells are reported.12,15,16 These effects may be secondary to the induction of other cytokines produced by the responding macrophages.20 A puzzling and major adverse effect of M-CSF administration has been the production of thrombocytopenia in all species studied,12,15,16,21,22 which sometimes resolves despite continued M-CSF administration.16,22-24Thrombocytopenia was dose-related, but was not sufficiently severe to cause bleeding. Several reports document a temporal relationship between the nadir of the platelet count and the maximum increase in monocyte counts.12,13,15,20,24 25 In this investigation, we have characterized the thrombocytopenic effect of M-CSF and have sought to determine the mechanism of this unpredicted physiologic result and its resolution despite continued administration of M-CSF.

Female Swiss Webster (SW) mice, 27 to 30 g, and female C57BL/6N (C57BL) mice, 22 to 25 g, (Simonsen Laboratories, Gilroy, CA) were used for these studies. Uninjected mice from the same shipments were used as normal controls. Mice were housed in an American Association for Accreditation of Laboratory Animal Care approved facility in filter cages and fed standard rodent chow and tap water ad libitum. All experimental protocols were approved by the Committee for Animal Experimentation of the VAMC. In conducting research using animals, the investigators adhered to the “Guide for the Care and Use of Laboratory Animals” prepared by the Committee on Care and Use of Laboratory Animals of the Institute of Laboratory Animal Resources, National Research Council (National Academy Press, Washington, DC, 1996). Blood samples were obtained from the retroorbital venous plexus, with the use of 70 μL heparinized EDTA-coated glass capillary tubes (Drummond Scientific Co, Broomall, PA), on the days indicated, immediately before M-CSF injection. Splenectomy was performed under anesthesia with methoxyflurane vapor (Metofane; Pitman-Moore, Inc, Mundelein, IL). Mice were allowed to recover from surgery for at least 1 month before experimentation. Mice were killed by cervical dislocation.

Recombinant human macrophage colony-stimulating factor (M-CSF; CSF-1) was a generous gift from the Cetus and Chiron Corporations, Emeryville, CA. Dilutions were made in pyrogen-free 0.9% saline for injection (Abbott Laboratories, Inc, North Chicago, IL). M-CSF was administered by intraperitoneal (IP) injection twice daily, 8 hours apart, at the doses indicated, in volumes of approximately 0.3 mL, beginning on day 1.

Platelet counts, total WBC counts, and hematocrit values were determined in whole blood diluted 1:2 (vol/vol) in isotonic saline solution (Hematall, Fisher Scientific Co, Pittsburgh, PA) and analyzed with an automated flow cytometric whole blood counter (Technicon H-1 System, Technicon Instruments, Tarrytown, NY), as previously described.26 

The small size of rodent white blood cells makes the differentiation between atypical lymphocytes and monocytes imprecise by Wright's stain, the most common method reported. To unambiguously quantify the monocytes, we established a new method based on the receptor for M-CSF, which among cells of the blood is present only on cells of the monocyte/macrophage lineage.27 Differential WBC counts in peripheral blood were determined after incubating buffy coat cells for 1 hour at 4°C with 0.1 mL of supernatant from an M-CSF–producing cell line (Rat2 pAPtag1 MCSF clone C5) (a generous gift of Drs Larry Rohrschneider and Gary Myles, Fred Hutchinson Cancer Research Center, Seattle, WA). The supernatant contained a secreted fusion protein that consisted of alkaline phosphatase fused in frame to amino acids 33-180 of rmM-CSF,28 prepared according to the method of Flanagan and Leder.29 Cells were then cytofuged and fixed for 30 seconds in a solution of 4% paraformaldehyde, 1.4 mmol/L Na₂HPO₄, 7.3 mmol/L KH₂PO₄, and 45% acetone, pH 6.6, at 4°C. An Alkaline Phosphatase Substrate Kit IV (BCIP/BNT) (Vector Laboratories, Burlingame, CA) was used to produce a blue precipitated reaction product, that indicated M-CSF binding to its receptor on monocytes. Cells were counterstained using 0.5% neutral red, dehydrated, and mounted. One thousand nucleated blood cells per animal were enumerated to obtain the differential cell count. Lymphocytes and neutrophils were identified by standard morphologic criteria, and monocytes were identified by the presence of a blue precipitate in or on cells.

Platelet morphology was quantified as previously described.30 Briefly, blood was obtained by cardiac puncture and anticoagulated with acid-citrate dextrose containing prostaglandin E₁, pH 6.7. Platelet-rich plasma (PRP) was prepared by centrifugation. Platelet morphology was observed by phase-contrast microscopy. Differential counts of platelet forms were performed with platelets drifting slowly between the coverslip and an ordinary glass slide using 400x magnification.

At the end of selected experiments, livers, spleens, and lungs were removed from normal or M-CSF–treated mice and tissue weights were recorded.

Soft agar cultures of spleen and bone marrow cells from normal and M-CSF–treated mice were prepared for quantification of granulocyte-macrophage colony-forming cells (GM-CFC) and megakaryocyte colony-forming cells (Meg-CFC), as previously described,31except for the following modifications: 20% horse serum was used instead of fetal calf serum and 0.1 mL (instead of 0.2 mL) of pokeweed mitogen spleen cell conditioned medium was used as the source of growth factors in each 1-mL culture. Control values were bone marrow: GM-CFC, 116 colonies/5 × 10⁴ cells and Meg-CFC, 16 colonies/5 × 10⁴ cells; spleen: GM-CFC, 62 colonies/10⁶ cells, and Meg-CFC, 52 colonies/10⁶ cells.

The ploidy distribution (DNA content) of megakaryocytes from the bone marrow of C57BL mice was measured using two-color flow cytometry, as previously described,32 with the following modifications: a FACScan with Lysis II software (Becton-Dickinson, Inc, San Jose, CA) was used for analyses. C57BL mice were used because normal SW mice demonstrate a ploidy distribution that is too variable for precise studies of changes in ploidy (J. Levin, unpublished observation, March 1992).

Normal SW mice, or SW mice treated with M-CSF, were used as platelet donors and recipients for these studies. Platelet survival studies were performed as previously described.33 Briefly, platelets pooled from donor mice were fluorescently labeled with 5-chloromethylfluorescein diacetate (CMFDA) (Molecular Probes, Eugene, OR) and injected into the tail veins of recipient mice. After infusion of labeled platelets, blood samples were obtained from the retroorbital venous plexus at 2, 4, and 6 hours and then approximately every 12 hours for the next 4 days. Blood samples were analyzed by flow cytometry, using a FACScan, to determine the proportion of labeled platelets present at each time point. Survival curves were constructed, and the circulating half-life (T_1/2) of the labeled platelets was determined graphically. In addition, platelet survival times were determined by using the multiple hit model (gamma function)34,35 and the best fit estimate, derived from the use of both linear and exponential sum of squares calculations.33 

The rate of disappearance of carbon particles from the circulating blood of normal or M-CSF–treated mice was measured as previously described.36 Briefly, a solution of India Ink (Difco Laboratories, Detroit, MI) was injected intravenously into either normal or M-CSF–treated mice. Blood samples were obtained at various times from 1 to 11 minutes after injection. Blood was lysed, and duplicate aliquots of each sample were read spectrophotometrically to determine the absorbance at 620 nm. The background values, obtained from lysed blood from mice not injected with carbon particles, were subtracted. Resultant values were used to determine the disappearance rate (T_1/2) of carbon particles from the blood.

Serum samples were obtained 4 and 8 hours after administration of 4 mg/kg/d M-CSF on days 1, 3, and 5 for determination of circulating M-CSF levels. Samples were stored at −70°C until assayed. Radioimmunoassays were performed in duplicate in a two-step procedure as previously described by Stanley37 with modifications.38 39 

For detection of murine anti-human M-CSF antibodies, 4 mg/kg/d M-CSF was administered for 5 days. Serum samples were obtained on days 5, 10, 12, and 15, and stored at −70°C. Samples were assayed using conditions described for the M-CSF radioimmunoassay37 with 10% normal rabbit serum replacing anti-CSF–antiserum.

Statistical analyses were performed with a two-tailed Student'st-test, using StatView (Abacus Concepts, Berkeley, CA).

The effect of different doses of M-CSF on the platelet counts of normal mice was examined. M-CSF was administered to SW mice for 5 days, in 2 daily IP injections, at doses of 2, 4, or 8 mg/kg/d. The platelet counts gradually decreased in a dose-dependent manner during the period of administration, reached a nadir on days 4 to 5, and immediately began to increase to above normal levels on discontinuation of treatment (Fig 1A). Blood sampling alone had no effect on the platelet count (data not shown).

To further investigate the relationship of dosage and duration of M-CSF administration to thrombocytopenia, 2, 4, or 8 mg/kg/d M-CSF was administered to normal SW mice for a period of 11 days. Platelet levels decreased in a dose-dependent manner up to day 5, after which levels began to increase despite continued administration (Fig 1B). The increase of platelet levels during the latter period of M-CSF administration also was dose-related; the largest dose allowed the least platelet recovery. At these doses, rebound thrombocytosis did not occur during the treatment period. To rule out the production of a neutralizing antibody against M-CSF as a potential cause of refractoriness to M-CSF, serum was obtained on days 5, 10, 12, and 15 from mice that had been treated with M-CSF for 5 days and analyzed for the presence of antibodies to M-CSF. No antibodies against M-CSF were detected after 5 days of M-CSF administration. Although anti–M-CSF antibodies subsequently became detectable 10 days after initiation of M-CSF administration in most animals, the titers were low.

To determine the effect of long-term administration and repeated exposure to M-CSF, mice received a dose of 8 mg/kg/d for 11 days (Fig2). Platelet levels increased after reaching their nadir on day 5, but rebound thrombocytosis did not occur during this first course of treatment until after M-CSF administration was discontinued. Following various lengths of time between repeat challenges of M-CSF, despite the detection of low levels of anti–M-CSF antibodies 10 to 15 days after the initial injection, thrombocytopenia still occurred within 5 days of M-CSF treatment, and rebound thrombocytosis now occurred during administration of M-CSF.

The effect of M-CSF, 4 mg/kg/d, on the blood cell counts of normal mice was documented. Blood samples were obtained on days 3 and 5 of M-CSF treatment, and days 8 and 10, after completion of the 5 days of M-CSF treatment. Total nucleated WBC counts, platelet counts, and hematocrit values were obtained. Differential cell counts were performed on cytocentrifuged buffy coat preparations in which monocytes had been specifically stained (see Materials and Methods) (Fig3), and neutrophils and lymphocytes were identified by standard morphologic characteristics. Total monocytes were significantly increased over control levels of 0.28 ± 0.4 × 10³/μL (mean ± 1 standard error [SE]) on days 3 (0.85 ± 0.17 × 10³/μL) (P < .05) and 5 (0.99 ± 0.25 × 10³/μL) (P < .05) of M-CSF treatment (Fig 4). The platelet count was significantly and maximally decreased from normal levels at the time of maximum absolute monocytosis. The total WBC count was significantly increased only on day 8, from the control level of 6.0 ± 0.3 × 10³/μL to 8.9 ± 0.8 × 10³/μL (P < .05). The hematocrit fell from an initial value of 52.0% ± 0.7% to a nadir of 44.6% ± 1.3% on day 5 of M-CSF treatment, and then gradually returned to normal.

The possible role of splenic platelet sequestration in the production of thrombocytopenia after M-CSF administration was examined. Splenectomized animals were given M-CSF, 2 mg/kg/d, and the platelet counts were compared with those of intact animals treated with the same dosage. Platelet counts fell to equivalent levels in intact and splenectomized SW mice during M-CSF treatment, and splenectomy did not affect either the degree of the thrombocytopenia or subsequent rebound thrombocytosis produced (Fig 5).

Soft agar cultures were performed to determine the total number of GM-CFC and Meg-CFC in spleen and bone marrow. M-CSF, 2 mg/kg/d, was administered to mice for up to 5 days, and total colony numbers were determined on days 2, 4, 5, 6, and 8. Both GM and Meg colony numbers remained normal on day 2 and then were increased twofold to sixfold over controls in the spleens of M-CSF–treated mice on days 4 to 8. The maximum increase in the spleen occurred on days 5 and 6. However, there were no differences between the total numbers of detectable GM-CFC or Meg-CFC present in the bone marrows of M-CSF–treated and normal mice (data not shown).

The effect of M-CSF on the ploidy distribution of megakaryocytes in C57BL mice was analyzed. M-CSF, 2 mg/kg/d, was administered on days 1 to 5. DNA content (ploidy) of bone marrow megakaryocytes was analyzed on days 3 to 7 (Fig 6). On days 4 to 7, the proportion of 32N, 64N, and 128N megakaryocytes was significantly increased over control levels (Fig 6C through F). However, although the proportion of 16N megakaryocytes decreased on days 6 and 7 as the higher ploidy megakaryocytes were increasing, 16N megakaryocytes remained the modal class. A higher dose of M-CSF, 8 mg/kg/d, which produced more severe thrombocytopenia, also significantly increased the proportion of 32N, 64N, and 128N classes on days 6 and 7 (Fig 6G through H), and in addition, 32N became the modal class on day 7.

To more thoroughly investigate the possible cause of thrombocytopenia and the role of the spleen, platelet survival in M-CSF–treated eusplenic and splenectomized mice was determined. M-CSF, 4 mg/kg/d, was administered to groups of eusplenic and asplenic SW mice starting on day 1. On day 4, pooled platelets harvested from normal mice and labeled with CMFDA were injected intravenously into the treated mice and a control group. M-CSF injections continued through day 5, and platelet survival was measured for 93 hours after injection of labeled platelets (Fig 7). The circulating half-life (T_1/2) of the labeled platelets in the control group was 32.9 ± 2.3 hours, and platelet survival, as measured by the multiple hit model (gamma function) was 2.58 ± 0.18 days. The T_1/2 of the labeled platelets was markedly reduced to 19.0 ± 0.7 hours in the eusplenic M-CSF–treated mice, and to 16.3 ± 1.7 hours in the asplenic mice (for both,P < .001 v control). Platelet survival was significantly shorter in M-CSF–treated recipient animals (1.19 ± 0.07 days in eusplenics and 1.06 ± 0.15 days in asplenics [for both, P < .001 v control]). There was no difference between the circulating half-life (T_1/2) or platelet survival in eusplenic and asplenic animals. Platelet survival also was examined in animals in which M-CSF, 4 mg/kg/d, was administered for 8 days before the injection of labeled platelets and was continued throughout the experiment. The T_1/2 of the labeled platelets in these animals was 26.2 ± 3.1 hours, and the platelet survival was 2.07 ± 0.43 days (P < .01 v control) (data not shown). Similar results were obtained for all these experimental groups when platelet survival estimates also were calculated using the best fit method (data not shown).

These results indicated faster clearance of platelets from the circulation in mice that had received M-CSF, that the spleen was not the primary mediator of this process, and that with prolonged administration of M-CSF, platelet survival remained significantly shortened, at a time when the platelet counts had returned to normal levels (Fig 1B). However, it was unclear whether the normal donor platelets were modified by exposure to M-CSF present in the circulation of the recipient mice, and thus had demonstrated shortened survival. Therefore, we performed additional platelet survival studies, using platelets from M-CSF–treated donors or normal donors. Donor mice received M-CSF, 4 mg/kg/d, for 2 days. On the morning of the third day, a dose of 2 mg/kg M-CSF was administered, and 2 hours later platelets were harvested from these and a group of untreated mice. Each pool of platelets was labeled with CMFDA and injected into normal mice, and survival of the untreated and M-CSF exposed platelets was compared (Fig8). The M-CSF–exposed donor platelets exhibited a normal or slightly prolonged T_1/2 in normal mice in comparison to untreated platelets, indicating that the M-CSF–treated platelets had not been altered by exposure to M-CSF in a manner that rendered them more susceptible to clearance.

To evaluate the effect of M-CSF on the monocyte/macrophage system, carbon clearance experiments were performed in mice that had been treated with M-CSF, 4 mg/kg/d, for various lengths of time. Absorbance (at 620 nm) due to the presence of carbon particles was monitored beginning at 1 minute after bolus injection to ensure homogeneous intravascular distribution of particles. Clearance of intravenously injected India Ink particles was significantly faster in mice that had been treated with M-CSF for 3, 5, or 9 days (Fig9). The T_1/2 of the carbon particles in mice treated for 3 days was 3.7 ± 0.4 minutes; for 5 days, 2.8 ± 0.2 minutes; and for 9 days, 3.9 ± 0.5 minutes, compared with the control value of 12.0 ± 0.7 minutes (for all,P < .0001). Furthermore, the measured 1-minute absorbance values (A620) were 1.091 for the control, 0.717 for the 3-day group, 0.566 for the 5-day group, and 0.637 for the day 9 group, indicating that there was greater clearance of carbon particles during the first minute in M-CSF–treated mice than in controls. Carbon clearance remained faster than normal in mice treated for 9 days with M-CSF, indicating that the monocyte/macrophage system remained hyperactive despite recovery of the platelet count.

In selected experiments, livers, spleens, and lungs were removed from M-CSF–treated animals after sacrifice. After administration of M–CSF, 4 mg/kg/d for 5 days, the livers and spleens of treated mice were significantly larger than those of the control animals. On day 5, liver weights were increased to 2.0 g (n = 14), from the control value of 1.3 g (n = 19) (P < .001), but by day 9, they were 1.4 g, almost normal. Spleen weights were increased to 250 mg on day 5 (n = 14), in contrast to the control value of 136 mg (n = 19) (P < .001) and remained increased (208 mg) on day 9 (n = 9). Total lung weights were not significantly increased on day 5 or day 9 from the control value of 157 mg (n = 8). After 9 days of administration of the same dose of M-CSF, liver weights were comparably increased to 2.0 g (n = 4), and spleen weights were increased further to 289 mg (n = 4).

Proplatelet formation was quantified as a potential indicator of stimulation of platelet production. M–CSF, 4 mg/kg/d, was administered for up to 5 days, and platelet differentials were performed on days 3, 5, and 9. At these times, when platelet levels were either falling (day 3), at their nadir (day 5), or increasing (day 9), the proportion of proplatelets was not increased (Table 1). However, the absolute numbers of proplatelets circulating on day 5 were significantly decreased from control levels, and conversely, were significantly increased on day 9 (Table 1).

Our studies have confirmed previous reports that M-CSF can produce thrombocytopenia.11,12,15,16,21,22 In addition, we have attempted to define the mechanisms of the production of thrombocytopenia by M-CSF and of the recovery of platelet levels during its continued administration. Treatment with M-CSF for 5 days produced dose-dependent thrombocytopenia, with rapid recovery of normal platelet levels and subsequent development of rebound thrombocytosis occurring after discontinuation of treatment. However, in experiments in which M-CSF was administered over a longer period, platelet levels surprisingly began to recover after 5 days, despite continued administration of M-CSF, strongly suggesting that thrombocytopenia was not the result of suppression of platelet production. Additional evidence for this hypothesis was that the ploidy distribution of megakaryocytes in the bone marrow remained essentially normal during the period of falling platelet levels, which reached their nadir on days 4 to 5. In contrast, in a model of bone marrow damage, produced by administration of 5-fluorouracil (5-FU) to mice, the proportion of higher ploidy megakaryocytes decreased initially, with 8N becoming the modal class on day 4.40 After M-CSF, we did not observe a decrease in the proportion of 16N and 32N megakaryocytes, but rather observed a slight right shift in ploidy. No changes were observed in the total GM-CFC and Meg-CFC in the bone marrow following M-CSF, in contrast to the marked increases observed during recovery from bone marrow damage produced by 5-FU.41 The rapidity of the recovery to normal or above normal platelet levels after termination of M-CSF treatment also argues against bone marrow suppression.

Additionally, we saw evidence of delayed stimulation of platelet production. The increased frequency of high ploidy megakaryocytes on days 4 to 7 is consistent with previous observations that ploidy levels shift upward in response to different degrees of thrombocytopenia produced by peripheral destruction of platelets, and that maximal changes in ploidy occur 48 to 72 hours after the stimulus of thrombocytopenia, preceding the recovery of normal platelet levels.42 The delayed increase in splenic Meg-CFC (and GM-CFC) is also consistent with the previously reported response of normal murine hematopoiesis to the stimulus of peripheral platelet destruction.31 43 Preliminary data have indicated that the mean platelet volume is increased after 3 or 5 days of M-CSF administration, consistent with stimulation of thrombopoiesis. The occurrence of rebound thrombocytosis during recovery of platelet levels, as well as the less severe nadir of thrombocytopenia observed after repeated exposure to M-CSF are consistent with the development of an expanded pool of megakaryocytes that resulted from this stimulation.

There was a gradual decrease in maximum observed M-CSF levels during the treatment period; however, M-CSF remained detectable after 5 days of administration, and no anti-M–CSF antibodies were present at this time. Clearance rates increased after repeated exposure to M-CSF, in agreement with previous reports.11,13,14,44 These observations are consistent with an increased monocyte/macrophage population and the proposed mechanism of monocyte/macrophage-mediated clearance of M-CSF.11 However, M-CSF remains efficacious after continued administration. Increased macrophage activity was still present after prolonged administration of M-CSF, as evidenced by our platelet survival and carbon clearance data. The ability to repeatedly produce thrombocytopenia during multiple courses of M-CSF administration further indicated that development of neutralizing antibody against M-CSF was not responsible for the phenomenon of recovery of platelet levels during short-term M-CSF administration.

The most likely cause for the thrombocytopenia appeared to be either peripheral destruction or platelet sequestration. We have documented a maximum increase in the level of circulating monocytes after M-CSF administration at the time of the platelet nadir, an inverse relationship that has been previously reported.12,13,25Evidence for increased removal of platelets from the circulation is the observation that platelet survival was decreased in M-CSF–treated mice. The role of the spleen in the sequestration or destruction of platelets in this model, though, is minimal or absent, because asplenic animals exhibited the same degree of thrombocytopenia as eusplenic mice and equivalently shortened platelet survival after M-CSF administration. However, in humans, M-CSF administration resulted in shortened platelet survival with marked platelet uptake occurring in the spleen.25 Because platelet survival remained shortened during prolonged M-CSF administration, at a time when platelet levels were increasing, it appeared that increased platelet production, stimulated by thrombocytopenia, had compensated for the increased platelet clearance. Exposure of platelets to M-CSF did not appear to produce a platelet defect, because platelets obtained from M-CSF–treated donors did not exhibit shorter survival than nonexposed platelets prepared in the same manner, after transfusion into normal mice.

M-CSF treatment increased the size of the liver and spleen in treated animals, a common result of stimulation of the monocyte/macrophage system, which may have contributed to the observed thrombocytopenia. Our detection of increased rates of carbon particle clearance after M-CSF administration is consistent with a mechanism for the development of thrombocytopenia that includes the production of activated macrophages. There is precedence for a nonimmune process of platelet removal by macrophages. In the reactive hemophagocytic syndrome, activated macrophages (designated histiocytes) have been shown to phagocytose blood cells, including platelets.45,46Interestingly, M-CSF levels are markedly increased in at least some patients with this syndrome.47 Following administration of M-CSF, macrophages from bone marrow aspirates in patients25and circulating macrophage-like cells in rabbits15 have been reported to demonstrate phagocytosis of platelets. We attempted to determine if Kupffer cells were responsible for the clearance of platelets, by injecting mice with carrageenan, because these cells can be selectively blocked with carrageenan.10 However, this reagent cannot be used to evaluate the role of Kupffer cell activity in production of thrombocytopenia, because we confirmed that carrageenan itself causes acute, severe thrombocytopenia.48 

Administration of other cytokines, such as granulocyte-macrophage colony-stimulating factor (GM-CSF),49,50G-CSF,51,52 interleukin (IL)-1,53,54IL-2,55,56 and stem cell factor (SCF)57,58 also has been shown to produce dose-dependent thrombocytopenia, which has been reported in some studies to resolve during continued administration,50,51,54,57 indicating that this is a phenomenon not unique to M-CSF. Although the mechanism(s) of thrombocytopenia associated with these other cytokines has not been established, descriptions of normal bone marrow function after cytokine treatment,49,51,55,59 data suggesting peripheral destruction of platelets,49-51,55 and reports of large platelets or increased mean platelet volume (MPV) at the platelet count nadir51,59 suggest that suppression of megakaryocytopoiesis is probably not the mechanism of the thrombocytopenia. In a single dog, GM-CSF–induced thrombocytopenia was not prevented by splenectomy, similar to our finding in mice, but thrombocytopenia persisted for 4 to 5 weeks after discontinuation of treatment,50 in contrast to our results in mice. However, the thrombocytopenia produced by IL-1 administration to mice was prevented by splenectomy,53 suggesting a different role for the spleen in that model.

We have concluded that thrombocytopenia produced by M-CSF was due to increased activity of the monocyte/macrophage system and that increased removal of platelets from the circulation, primarily due to phagocytosis by the liver (and perhaps bone marrow) macrophages, occurred as a result of some as yet unidentified nonimmune process. The return to normal platelet levels despite continued administration of M-CSF suggested that increased platelet production, stimulated by thrombocytopenia, was able to compensate for the increased rate of removal of platelets from the circulation. Further experiments are needed to determine whether M-CSF–induced thrombocytopenia can be prevented in normal or bone marrow damaged individuals by prior or concurrent administration of a stimulator of thrombopoiesis, such as thrombopoietin.60 61 This might eliminate the dose-limiting toxicity of M-CSF and permit further investigation of its clinical potential.

The authors thank Dr Jolanda Schreuers (formerly of Chiron Corp) for helpful discussions and assistance in making these studies possible, and Dr E. Richard Stanley, Albert Einstein College of Medicine, whose laboratory determined the M-CSF levels and anti-M–CSF antibody titers.