ABSTRACT
Polymyxin B (PMB) binds to and neutralizes endotoxin, but its systemic clinical utility is limited by neuro- and nephrotoxicity. PMX622 is a covalent conjugate of PMB and Dextran-70 designed to retain the ability of PMB to neutralize endotoxin and to retain the favorable colloidal, pharmacokinetic, and metabolic properties of Dextran-70. PMX622 has demonstrated efficacy in a number of animal models and effectively neutralized endotoxin in phase I clinical trials. Here, we systematically evaluated the pharmacodynamic properties of PMX622 in a murine model of endotoxin-induced lethality in galactosamine-sensitized mice. PMX622 completely and dose dependently inhibited lethality in this model. A stoichiometric relationship was found between the endotoxin challenge dose and the dose of PMX622 needed for protection. PMX622 neutralized endotoxin from four different genera of gram-negative bacteria but not Neisseria meningitidis. PMX622 was significantly less toxic than PMB in the mouse, suggesting that PMX622 has a better margin of safety than PMB. The timing of PMX622 administration relative to endotoxin was crucial. PMX622 was active for several hours prior to the endotoxin challenge; however, PMX622 did not protect mice if administered ≥15 min after endotoxin challenge. This suggests that PMX622 would best be clinically used prophylactically rather than therapeutically. These studies will be crucial in designing and interpreting human clinical trials assessing PMX622 efficacy.
Infection with gram-negative bacteria accounts for ∼50% of sepsis cases (36, 37). Endotoxemia has been correlated with morbidity in sepsis patients (8, 24, 32). Even when sepsis is caused by fungi, viral infection, or gram-positive bacteria, shock induces translocation of endotoxin from gram-negative bacteria in the intestine into the circulation due to a compromised intestinal tract; this endotoxin may contribute to the pathology (12, 31). Thus, it has been recognized that targeting the primary mediator, endotoxin, is an attractive therapeutic approach (24, 32). Furthermore, endotoxin may be an important factor later in the septic response. Other pathological responses also elicit endotoxin-translocation, and an anti-endotoxin therapy would be of benefit in these indications too.
Antisera or monoclonal antibodies against the endotoxin strain-specific O-chain can block sepsis caused only by that bacterial serotype (28). Clinical trials with monoclonal antibodies to cross-reactive epitopes have been disappointing (34, 35), perhaps because these antibodies are not broadly reactive and have low affinity for endotoxin (1, 30, 32). Bacterial permeability increasing protein (BPI), a protein released from activated neutrophils, also neutralizes endotoxin. A recombinant truncated form of BPI, rBPI21, has been assessed in clinical trials, and the results were favorable (22).
Polymyxin B (PMB), a naturally occurring cationic polypeptide, neutralizes endotoxin and has been used as a topical antibiotic for more than 40 years. PMB interacts with the endotoxin lipid A chain with a micromolar affinity (26), which is higher than anti-endotoxin antibodies. Because the lipid A moiety is conserved, PMB neutralizes most gram-negative bacteria and resistance to PMB is rare. However, neuro- and nephrotoxicity, as well as poor pharmacodynamic and pharmacokinetic properties, limit the systemic use of PMB. A PMB-nylon filter matrix has been clinically used for extracorporeal plasmapheresis to deplete endotoxemic plasma of endotoxin without exposing patients to the toxic effects of PMB (16, 19). However, a method to neutralize plasma endotoxin that does not require extracorporeal treatment would be advantageous.
PMB was chemically conjugated to Dextran-70 under controlled conditions to generate PMX622 (17). Dextran-70 was chosen because it is also used as plasma expander (Macrodex) in the treatment of sepsis (20) and its relatively high molecular weight limits PMB extravasation into the tissues. Other investigators have prepared PMB-soluble starch and PMB-immunoglobulin G conjugates, but these materials have not been prepared sufficiently characterized for clinical use (11). PMX622 is effective in an equine model of endotoxin-induced responses (25), a rabbit model of gram-negative peritonitis (7), a sheep model of cecal ligation and perforation (13), and a murine model of endotoxemia with viable Escherichia coli or Pseudomonas aeruginosa (6). However, because of the labor-intensive nature and complexity of these disease models, detailed and systematic studies to assess the pharmacodynamic properties of PMX622 could not be evaluated.
We used here a murine model of endotoxin-induced lethality to systematically measure efficacy, toxicity, the pharmacodynamic half-life, and other drug properties. Furthermore, PMX622 was shown to be safe and to neutralize an endotoxin challenge in humans in a phase I trial (23), but clear efficacy was not demonstrated in a phase II trial of septic patients who presented with at least one organ failure (unpublished data). Therefore, the detailed efficacy and pharmacodynamic studies described here are important in interpreting clinical trials with PMX622 and in designing new studies with PMX622, as well as in assessing other anti-endotoxin approaches in the clinical setting.
MATERIALS AND METHODS
Materials.PMB micronized sterile powder was obtained from Pfizer Specialty Chemicals (Pfizer, Inc., New York, N.Y.). Clinical-grade Dextran-70 (Macrodex) was purchased from Kabi Pharmacia AB (Uppsala, Sweden). PMB was conjugated to Dextran-70 by using previously published methods (17). The ratio was optimized to decrease toxicity and to improve pharmacodynamic and pharmacokinetic properties (18). For PMX622, the ratio of PMB conjugated to Dextran-70 was 38 mg of PMB/g of Dextran-70. All calculations for PMX622 dosage were based on the mass of the PMB component.
E. coli O111:B4, Salmonella enterica serovar Typhimurium, Klebsiella pneumoniae, and P. aeruginosa F-D type I endotoxin were purchased from List Biological Laboratories, Inc. (Campbell, Ca.). Endotoxin from Neisseria meningitidis was a gift from Chao-Ming Tsai (Center for Biologics Evaluation and Research, Food and Drug Administration, Bethesda, Md.). All endotoxin was dissolved at 1 mg/ml in isotonic pyrogen-free saline and probe pulse sonicated for 2 min (Branson Sonifier cell disrupter 200; Branson, Danbury, Conn.) to disperse any aggregated endotoxin. Aliquots were stored at −20°C and, just prior to use, endotoxin was thawed, diluted with isotonic saline, and probe sonicated for 30 s. d-(+)-Galactosamine hydrochloride (Sigma Chemicals, St. Louis, Mo.) was dissolved in isotonic pyrogen-free saline and mixed 1:1 with endotoxin.
Animals and animal care.Male C57BL/6J mice (6 to 8 weeks of age) were obtained from Jackson Laboratory (Bar Harbor, Maine) and were acclimated for one week before use. Mice were provided with food and water ad libitum and housed in a facility approved by the American Association for Accreditation of Laboratory Animal Care. Studies were carried out according to Animal Care and Use Committee guidelines and protocols approved by the Animal Care and Use Committee.
Endotoxin-induced lethality in mice.To determine the dose of endotoxin to use for further studies, endotoxin mixed 1:1 with d-(+)-galactosamine hydrochloride (500 mg/kg) in pyrogen-free 0.9% saline was administered to C57BL/6J male mice by intravenous (i.v.) tail injection. The sensitivity of mice to endotoxin can be enhanced more than 1,000-fold by coinjection with the liver-specific inhibitor, galactosamine (15). In this model, systemically released tumor necrosis factor alpha causes liver cell death and subsequent liver damage, resulting in lethality.
To determine the dose of endotoxin needed to kill 80 to 100% of the mice (LD80-100) for each type of bacterial endotoxin, lethality was assessed at six concentrations, with 10 mice per concentration (29).
Neutralization of endotoxin-induced lethality by PMB or PMX622.PMB or PMX622 was administered to mice by i.v. tail injection at various doses, followed immediately by an i.v. tail injection with an otherwise-lethal dose (LD80-100) of endotoxin and galactosamine. Survival was assessed at 72 h. The dose of PMB or PMX622 that protected 50% (PD50) of the mice from endotoxin-induced lethality was calculated by the method of Reed and Muench (29). For some studies, PMB or PMX622 was premixed for 1 h with an LD80-100 dose of endotoxin and then mixed with galactosamine and administered by i.v. tail injection. In the time course experiments, PMX622 was administered i.v. at various times before or after an i.v. endotoxin challenge.
Acute toxicity of PMB or PMX622.A single bolus of increasing doses of either PMB or PMX622 was administered by i.v. tail injection. Lethality was assessed after 7 days. The dose of PMB or PMX622 that was lethal to 50% of the mice (LD50 of PMB or PMX622) was calculated by the Reed-Muench method (29).
RESULTS
Systemic administration of PMX622 prevents endotoxin-induced lethality in galactosamine-sensitized mice.The efficacy of PMX622 in vivo was assessed in a model of endotoxin-induced lethality in galactosamine-sensitized mice. The i.v. administration of either PMB or PMX622 immediately prior to an i.v. endotoxin challenge resulted in a dose-dependent increase in survival. As shown in Fig. 1 and as summarized in Table 1, an LD80-100 of E. coli endotoxin (2 μg/kg) in the absence of PMX622 or PMB resulted in 0% survival at 72 h (average of 14 experiments). Fifty percent of the animals survived (PD50) an LD80-100 endotoxin challenge dose when treated with 287 × 107 μg of PMB/kg or 520 ± 180 μg of PMX622/kg (differences are not statistically significantly different; see Fig. 1). Complete protection was observed at a dose of 2 mg of PMX622/kg. As described in Materials and Methods, all calculations for the PMX622 dosage were based on the PMB component, so these numbers can be directly compared. In other studies, PMB and PMX622 were shown to neutralize endotoxin with similar potency in a number of in vitro assays (for example, neutralization of lipopolysaccharide in vitro requires a dose of 14.5 μg/kg for PMB versus a dose of 20 μg/kg for PMX622 [unpublished data]). Thus, conjugation of PMB does not result in a loss of endotoxin-neutralizing activity. Dextran-70 showed no activity and HA-1A, a human immunoglobulin M anti-endotoxin monoclonal antibody (29) was not efficacious in this in vivo model (data not shown).
Dose-dependent protection by PMX622 or PMB after lethal E. coli endotoxin challenge in galactosamine-sensitized mice. PMX622, PMB, or Dextran-70, followed immediately by a lethal dose (LD80-100) of endotoxin, was administered by i.v. tail injection. Each datum point represents the average of six to eight experiments ± the standard error of the mean (SEM). Within each experiment, 10 animals were used per group.
PMX622 protects mice against a lethal challenge of endotoxin from four gram-negative bacterial strains but not from challenge with N. meningitidis
Quantitative relationship between the dose of PMX622 needed for protection and the dose of the endotoxin challenge.The ability of PMX622 to prevent lethality due to endotoxin doses that varied over 2.5 orders of magnitude was assessed. Administration of the lowest dose of E. coli endotoxin (0.05 μg/kg) resulted in 50% survival in the absence of PMX622 (not shown). As shown in Fig. 2A, at this dose of endotoxin (i.e., 100 μg/kg) PMX622 completely prevented lethality. In contrast, challenge with endotoxin at 2 or 20 μg/kg resulted in only ≤20% survival in the absence of PMX622 (not shown). As shown in Fig. 2A, at these higher doses of endotoxin, more PMX622 was needed to prevent lethality. For example, 100 μg of PMX622/kg had no effect on the lethality induced by 2 or 20 μg of lipopolysaccharide/kg. At 1 mg/kg, PMX622 prevented lethality at all tested doses of endotoxin. Figure 2B shows the relationship between endotoxin dose and the concentration of PMX622 needed for protection by comparing the PD50 for PMX622 at different endotoxin doses. In summary, the dose of PMX622 necessary to neutralize endotoxin is proportional to the dose of endotoxin.
Stoichiometric protection by PMX622 from increasing concentrations of E. coli endotoxin in mice. (A) PMX622, followed immediately by an LD20-75 (0.05 μg/kg), an LD80-100 (2 μg/kg), or an LD100 (20 μg/kg) challenge of endotoxin, was administered by i.v. tail injection. Each PMX622 concentration was tested in three or four individual experiments for a given endotoxin experiment; each experiment contained 10 animals per group. The error bars represent the average ± the SEM. (B) PD50 values calculated from Fig. 2A.
PMX622 neutralizes endotoxin from different genera of gram-negative bacteria but not from Neisseria meningitidis.The ability of PMX622 to protect mice against endotoxin from five clinically relevant gram-negative bacteria was assessed. For these studies, mice were challenged i.v. with an LD80-100 of each type of endotoxin (i.e., equally lethal doses). For each endotoxin, an LD80-100 was determined. The LD80-100 varied from 0.5 to 10.0 μg/kg, depending on the molecular weight of the endotoxin species. As shown in Fig. 3 and as summarized in Table 1 (column 2), PMX622 protected mice against four of these bacterial endotoxins (E. coli, S. enterica serovar Typhimurium, K. pneumoniae, and P. aeruginosa F-D type I) but not against N. meningitidis endotoxin-induced lethality. Only 20% survival was observed with an LD80-100 challenge of N. meningitidis endotoxin and, even at the highest doses of PMX622 tested, i.e., 10 mg/kg, there was no protection from this challenge.
Protection by PMX622 against lethal challenge with four gram-negative bacteria strains. An LD80-100 dose of endotoxin and galactosamine was administered to mice by i.v. tail injection, followed immediately by i.v. administration of PMX622. Each curve represents the mean ± the SEM of three or four individual experiments. For each experiment, each group (PMX622 dose) contained 10 mice.
Despite the 20-fold difference in the quantity of different endotoxins administered to achieve an LD80-100, the PD50 of PMX622 did not differ for different endotoxins, as shown in Table 1. Therefore, the potency of PMX622 was equal against biologically equivalent challenges of diverse endotoxins, even if the challenge doses differed greatly in amount.
Premixing PMX622 with endotoxin increases potency 5,000-fold and irreversibly neutralizes endotoxin.To determine whether PMX622 neutralization of endotoxin was reversible, an otherwise-lethal challenge of E. coli endotoxin was premixed in vitro with increasing concentrations of PMB or PMX622 prior to i.v. injection in mice. As shown in Fig. 4, premixing with endotoxin resulted in a 5,000-fold increase in PMX622 potency and a 1,000-fold increase in PMB potency. Premixing endotoxin with Dextran-70 did not prevent lethality (data not shown). These studies show that if large amounts of endotoxin are neutralized in vitro, it is not biologically active when injected into the mouse.
Endotoxin neutralized in vitro does not elicit lethality when injected in vivo. Mice (10 per group) were injected by i.v. tail injection with either PMX622 or PMB premixed with an LD80-100 challenge of E. coli endotoxin as described in Materials and Methods. This was compared to survival in mice injected by i.v. tail injection with an LD80-100 challenge of endotoxin, followed immediately by PMX622.
Relationship between the time of PMX622 administration and the endotoxin challenge.These experiments were designed to address two questions. First, the pharmacodynamic half-life of PMX622 was determined by injecting PMX622 at various periods before endotoxin. Thus, if PMX622 is eliminated very rapidly, it will not be present at the time of endotoxin challenge. As shown in Fig. 5, administration of PMX622 4 h prior to endotoxin resulted in 30% survival. Survival progressively increased as the time of PMX622 administration approached the time of endotoxin administration, suggesting a relevant pharmacodynamic half-life of PMX622 in mice of ca. 2 h.
Efficacy with various times of PMX622 administration from endotoxin challenge. Mice were given an i.v. bolus tail injection of PMX622 at various time intervals prior to endotoxin challenge. E. coli endotoxin and galactosamine was administered intraperitoneally at t = 0. Each datum point represents the average ± the SEM of three experiments; each experiment with 10 mice per group.
Second, the window of time after administration of endotoxin in which PMX622 can prevent lethality was determined. Therapeutic administration of PMX622 (i.e., after endotoxin) provided significant reduction in mortality when PMX622 was administered 1 h after endotoxin challenge, but there was a rapid loss in efficacy after this time.
Conjugation of PMB to Dextran-70 reduces the toxicity.The clinical utility of a drug depends on the window between its efficacious dose and its toxic dose. One measure of toxicity is the direct lethality due to high doses of the drug, i.e., the acute lethality. As shown in Fig. 6, administration of high doses of either PMB or PMX622 i.v. as a single bolus to mice resulted in lethality within 72 h. PMB at 10 mg/kg resulted in 100% lethality with an LD50 of 8 mg/kg. In contrast, 100% lethality was not observed until 25 mg of PMX622/kg was administered. Therefore, the LD50 of PMX622 is 22 mg/kg, i.e., 2.75-fold less toxic than that of PMB. Equivalent doses of nonconjugated Dextran-70 were not lethal (data not shown). Native PMB mixed with Dextran-70 was as toxic as PMB alone (data not shown).
Acute toxicity due to high doses of PMX622 or PMB. Mice were given an i.v. bolus tail injection of increasing concentrations of PMB or PMX622. The dose of PMB and PMX622 was determined by the concentration of PMB. The dose of Dextran-70 was a dose equivalent to the Dextran-70 content of PMX622 (38 mg of PMB/g of Dextran-70) Survival was assessed at 72 h. Values represent the average of three experiments, each with 10 mice per group.
DISCUSSION
In the present study, we have shown that a novel anti-endotoxin therapeutic, PMX622, neutralizes endotoxin from a number of gram-negative bacteria in vivo in a model of endotoxin-induced lethality in galactosamine-sensitized mice. PMX622 is two to threefold less toxic than unconjugated PMB in this model and should have a bigger therapeutic window in vivo.
A single dose of PMX622 effectively prevented endotoxin-induced lethality in galactosamine-sensitized mice in a dose-dependent manner (Fig. 1A). Dextran-70 had no effect on survival, suggesting that PMX622 does not prevent lethality by plasma expansion. A stoichiometric relationship was found between the endotoxin dose and the dose of PMX622 needed for protection (Fig. 2B). Since endotoxin blood levels are much lower (nanograms/milliliter [10]) in human sepsis than the levels resulting from the bolus injections used in these studies (micrograms/milliliter), extrapolation suggests that low doses of PMX622 should be efficacious in humans.
In addition, an effective anti-endotoxin therapy must neutralize endotoxin from a variety of gram-negative bacteria that may be encountered in the clinical setting. PMX622 neutralized lipopolysaccharide from four of the most clinically relevant bacteria implicated in bacterial sepsis. PMX622, like PMB, does not neutralize Neisseria endotoxin (2, 3). The ability of PMX622 to neutralize the “biologically active” dose and not the quantity of endotoxin suggests that PMX622 neutralizes the pathogenic pharmacophore of endotoxin. These results demonstrate that PMX622 has a specific mechanism of action and behaves in a mechanistically predictable manner. In other studies, PMX622 was shown to bind to radiolabeled endotoxin (data not shown), further demonstrating that the in vivo activity is based on the predicted mechanism.
Systemic use of PMB in sepsis is limited by toxicity. In vitro studies have shown that PMB is not released from PMX622 in human serum (no measurable PMB released after 16 h at 37°C). The reduction in PMB toxicity on conjugation to Dextran-70 could occur by any, or all, of the following mechanisms. First, it is expected that the increased molecular weight of the drug reduces renal filtration, tubular uptake, and exposure of the nephrons and kidney parenchyma tissue to PMB and therefore reduces nephrotoxicity. Second, conjugation blocks the ability of PMB to insert into cellular membranes. PMB acts as an antibiotic because of the cationic detergent nature of the compound. PMX622 does not retain this antimicrobial activity. In studies not presented here, we have demonstrated that PMX622, unlike PMB, will not directly damage kidney tubule cells or result in mast cell degranulation. Third, other studies have demonstrated that conjugation of PMB to Dextran-70 increases the half-life and reduces drug extravasation into tissues (data not shown). This increases the intravascular concentration of PMB and should reduce the daily dose required for effective protection.
Importantly, PMX622 anti-endotoxin therapy is only effective if given before or within a short time frame after endotoxin challenge (Fig. 5). PMX622 was not effective if administered more than 1 h after the endotoxin. This is consistent with evidence suggesting that an endotoxin bolus elicits toxic levels of tumor necrosis factor alpha and interleukin-1 within 1 to 3 h in mice (9, 12). However, clinical sepsis is more complex than can be modeled in these studies. Patients may be subjected to a large bolus of endotoxin release from an infected site, several small boluses, or even a continual release of endotoxin depending on the patient. Clinical studies of endotoxin levels show that endotoxin is released intermittently over time. Experimental studies have shown that exposure of bacteria to bactericidal antibiotics can cause significant amounts of endotoxin release that can further exacerbate sepsis (10, 21, 33). At present, antibiotics are the mainstay of sepsis therapy. Therefore, although early initiation of PMX622 is preferable, a potential opportunity for therapeutic benefit may exist later in the syndrome. In a murine model of endotoxemia with viable E. coli or P. aeruginosa, antibiotics resulted in bacteria lysis and lethality, and administration of PMX622 prevented antibiotic-induced lethality (6). However, it is likely that later in the disease course, when cytokine levels are high and physiological impairment is observed, neutralization of endotoxin may no longer be sufficient to improvement in patient survival.
Although PMX622 had a pharmacodynamic half-life of ∼2 h, similar to the half-life measured by enzyme-linked immunosorbent assay. However, in sheep, the pharmacodynamic half-life of the material is ca. 2 h, which is less than the ∼5-h immunological half-life measured by enzyme-linked immunosorbent assay. In a phase I study, PMX622 neutralized a bolus injection of endotoxin in humans, demonstrating that PMX622 was sufficiently stable in human plasma (23).
Neutralization of endotoxin should be part of a clinically useful therapy for a number of other pathological conditions and in high-risk patients. Other procedures and conditions are known to cause significant increases in endotoxin levels in plasma; these conditions and procedures include burns (4), bowel surgery, bowel transplantation (27), pancreatitis, and coronary bypass surgery (5). PMX622 might be an effective prophylactic drug for use in these situations.
There are few compelling therapeutic strategies for sepsis. A recombinant form of activated protein C, Drotreocogin alfa (activated) (Xigris; Eli Lilly and Co., Indianapolis, Ind.), has been approved for sepsis (14). Administration of Xigris resulted in a 6% reduction in mortality, but with a 3.5% increase in severe bleeding and a cost of $6,800 per treatment. Therefore, other, less-expensive therapies that target different mechanisms might be of significant benefit either as monotherapy or in combination with antibiotics and other therapeutic approaches. Clinical trials with agents such as PMX622 can establish the significance of endotoxin release, in particular, late endotoxin release, in sepsis and in other pathologies in which endotoxin has been implicated.
ACKNOWLEDGMENTS
We thank Beat Weidman, Mary Ellen Digan, and Joseph Colacino for careful reading of the manuscript.
FOOTNOTES
- Received 5 November 2003.
- Returned for modification 26 January 2004.
- Accepted 4 April 2004.
- Copyright © 2004 American Society for Microbiology
