Macrophage Colonystimulating Factor

M-CSF is approved for clinical use in some countries under the name Leuko-prol® (mirimostim). It was originally cloned in 1985 but was one of the first cytokines studied in the 1960s and had been purified from urine by 1975(46). As the name would suggest, M-CSF was first shown to stimulate the growth of bone marrow-derived monocyte/macrophage cells in vitro (7) but was subsequently found to play a role in inflammation (47), bone remodeling (48), reproduction (49), the central nervous system (50-52), and cancer (53-56).

PK studies using M-CSF created a new paradigm for understanding the relationship between the PK and PD of hematopoietic cytokines. This new understanding centers on the ability of these cytokines to stimulate the production of their appropriate target cells, in this case monocytes/macrophages, only then to have those very cells consume and ultimately clear the stimulator from the serum as their numbers increase. This model has been extended to TPO (57,58), EPO (59,60), G-CSF (61), and perhaps even GM-CSF (41,43,44), but rests on insight gained from the study of M-CSF (62).

Mice normally have detectable levels of M-CSF in their serum, and studies performed using radiolabeled M-CSF demonstrated the serum half-life of this cytokine to be about 10 minutes. Approximately 96% of the cleared M-CSF could be accounted for by splenic or hepatic macrophages, the remainder was eliminated in the urine. Upon analysis of a number of parameters, including the effect of lysosomal protease inhibitors, it was apparent that internalization and degradation in macrophages via the cell surface M-CSF receptor, c-fms, was the predominant mechanism of M-CSF clearance.

The implications of this mechanism are clear. First, the clearance of physiological amounts of cytokines can be quite rapid, being mediated by the normal population of receptor positive cells. Second, pharmacological levels of exogenous cytokine can quickly saturate this clearance mechanism, leading to prolonged exposure and increasing the relative contribution of nonspecific clearance mechanisms, for example, renal filtration. Third, as the PD response to the cytokine accumulates over time, the capacity of the selective clearance mechanism will increase, reducing the relative role of nonspecific pathways. Fourth, in the absence of a target cell response, the clearance of a cytokine might be rather slow, increasing as the response mounts. This model is very attractive to explain homeostatic regulation of cytokine levels and target cell populations, and has ramifications for therapeutic administration of recombinant cytokines that share much of their biology with their endogenous prototypes. Indeed, this exact mechanism was used to develop therapeutically enhanced versions of G-CSF, as is outlined later.


G-CSF was one of the earliest cytokines to be biologically and biochemically characterized by the Australian CSF pioneers at the Walter and Eliza Hall Institute of Medical Research in Melbourne under the guidance of such giants in the field as Don Metcalf and Richard Stanley. It is due only to the insight of these pioneers that human G-CSF could be purified (22) and cloned (15,16) elsewhere and subsequently developed into a major therapeutic drug that has been administered to several million cancer patients since its launch in 1991.

Some of the early studies were confounded by incomplete separation of GM-CSF and G-CSF, and the seminal paper describing the activity of purified human G-CSF referred to it as a pluripotent factor (22), possibly in error because of assaying it on impure cell preparations. Nevertheless, from the early days, experiments where G-CSF was used as a single activity showed that although it was a modest CSF, it was highly selective in its actions on neutrophilic progenitor cells (9,63). As it turned out, the modesty of its in vitro actions was misleading, but its selectivity was probably not (for review see Ref. 29). The dominant clinical effect of G-CSF action is neutrophilia, though minor or sporadic effects on other blood cells have been reported. Most notably, G-CSF is well documented to increase monocyte proliferation (64,65), which may be linked also to reports of increased osteoclast-mediated bone turnover (66,67). These data illustrate that increased bone turnover, at least in rodents, results from expanded osteoclast activity after treatment (68). Whether this is related to the profound effects of G-CSF on monocyte production kinetics awaits definition of the relationship between these monocytes and osteoclast development.

Humans injected with G-CSF can expect a neutrophil response within one to two days (69-71). However, this is not the case after cancer chemotherapy where G-CSF is normally used to treat neutropenia, because the marrow is often not capable of responding on that timescale (69). This PD response is driven by a rapid absorption of typically SC administered G-CSF, wherein peak concentrations are noted within two to eight hours. The elimination half-life after either SC or IV administration is two to four hours depending on dose and neutrophil count (61,72). As G-CSF is administered daily, the neutrophil count increases, and in parallel, the clearance time of G-CSF is shortened; a relationship that was correlated even in early studies with receptor number on neutrophils (73). As noted above, this appears to be a very similar mechanism to that suggested for the M-CSF PK/PD relationship, that is, the cellular response to a cytokine in turn selectively clears that very cytokine, while in parallel a less saturable pathway (renal clearance) accounts for the balance of the elimination.

In an extension of this very satisfying model, a novel form of G-CSF was engineered specifically to evade the nonselective clearance pathway, yielding a new drug tailored to effect a neutrophil response that could only be cleared by those very neutrophils once they accumulate to a sufficient level (74-78). This form (pegfilgrastim) was designed for use in patients undergoing cancer chemotherapy and in whom support for neutrophil production was required. The underlying hypothesis in designing a form of G-CSF that would not be cleared by the kidney yet would remain sensitive to neutrophil-mediated clearance was that a degree of self-regulation would be an intrinsic feature of the molecule. This was proven to be correct first in animal and then in clinical studies. During neutropenia, the drug has an extended half-life; upon neutrophil recovery clearance is reactivated (75). Thus, for the first time, a drug that offered "automated" control of neutrophil counts was developed. This exciting mechanism of action has led to the broad uptake of pegfilgrastim in medical practice, but has yet to be applied to other therapeutics.


EPO is widely used in the treatment of anemia since it is the central regulator of erythropoiesis. The major quantitative site of EPO production is the kidney, so patients with declining renal function were the first and are still the most obvious candidates for EPO therapy (79). Use in anemia associated with cancer treatment is also common. Although controversial, a number of other experimental uses have emerged since EPO was approved for use in 1989 (80), including stroke, nerve crush injury, heart failure, myocardial infarction, immunomodulation, and for improving cognitive function. It remains unclear how these latter effects work in the absence of EPO receptor on many of the target tissues (see Ref. 81 for a critique of methods used to claim otherwise).

Confining our discussion to the effects of EPO on erythropoiesis, it must be borne in mind how highly dynamic is the process of red blood cell production. A normal 70-kg human produces on the order of 2.5 x 1011 erythrocytes per day, and this rate of production is maintained by a basal EPO level of around 10 to 20 mU/mL (82,83). Pharmacological administration of EPO at a dose intended to sustain a three times per week dosing cycle (150 U/kg) or a weekly treatment cycle (40,000 U/kg) leads to a Cmax of 150 or 850 mU/mL, respectively (84). Reticulocytes are released earlier than normal from the bone marrow and reside for a disproportionately longer fraction of their life span in the blood following EPO therapy. Despite this being the first PD readout of EPO administration, the more important result is a change in hemoglobin concentration. In the same study (84), the reticulocyte shift could be clearly seen in the blood by five days and a readily discernable change in hemoglobin by day 8—the two dosing regimens being approximately the same despite the 30% dose increment with the weekly regimen. This inefficiency is suggested to be driven by the non-linearity of EPO PK, which seems to lean toward reduced clearance at higher doses. In this case, it is likely that the similar PD response was driven by the accumulated time above the concentration threshold required for pharmacological action, which was similar between the two regimens.

A model was expounded in the early 1990s (85,86) that still yields a satisfactory explanation of the relationship between EPO exposure and response. Furthermore, this model has to date proven satisfactory to explain the PD response to all erythropoiesis-stimulating agents (ESAs). The model states, in essence, that the time between administrations during which the ESA serum level exceeds the threshold for response is the sole driver of efficacy. Of course, the details of the model parameters change with intrinsic potency of the ESA, dose, and clearance parameters, but the model remains the same across all ESAs. The implication is that all ESAs perform similarly when matched for the time above this threshold level. Inefficiency does become a factor as the interval between injections gets longer—explaining the 30% dose penalty with EPO administered once versus three times per week, as shown in the above study. Longer-acting analogs of EPO specifically engineered to improve half-life [darbepoetin alfa (87)] and pegylated EPO [e.g., PEG-EPO/3 (88)] are not hampered by this inefficiency until after a longer interval and are, therefore, able to sustain a desired clinical outcome for up to three or four weeks between injections. It remains to be seen how dosing of a non-EPO-based ESA may be approved by regulatory authorities (89), but initial observations suggest adherence to the same PK/PD model.

0 0

Post a comment