INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

LIPOSOMAL FORMULATIONS of some drugs, most importantly doxorubicin (Doxil) and amphotericin B (Ambisome, Amphocil), are increasingly used in the treatment of cancer and other diseases (2). Although very efficient in improving the therapeutic efficacy of encapsulated agents, these liposomes can cause a poorly understood immediate hypersensitivity reaction in a relatively large number (up to 7%) of patients (1, 11, 14, 21, 28, 31, 39). The reaction usually develops at the start of infusion and includes symptoms of cardiopulmonary distress, such as dyspnea, tachypnea, hypo- and/or hypertension, chest pain, and back pain. Unlike in IgE-mediated (type I) allergy, however, the response to liposomes arises at the first exposure to the drug without prior sensitization, and the symptoms may lessen or disappear rather than increase upon rechallenge. Because of these and other unusual features, the reaction was recently called a "pseudoallergy" (1). Its mechanism, however, has not been clarified, to date.

Recently, we reported that intravenous injection of minute (milligram) amounts of large multilamellar vesicles (LMV) in pigs led within minutes to considerable hemodynamic disturbance, including a massive increase in pulmonary vascular resistance and pulmonary arterial pressure (PAP), with or without falls in cardiac output and systemic arterial pressure (34). We demonstrated that the reaction was mediated by thromboxane A₂, which, in turn, was released into the blood as a consequence of complement (C) activation. We suggested that the described porcine model could be used for quantitative studies on the above-mentioned hypersensitivity reactions to liposomes, tentatively called "C activation-related pseudoallergy," or "CARPA" (34).

The aims of the present study were to clarify the experimental conditions critical to the reaction and to define the mechanism of liposome-induced C activation in pigs. The specific questions were as follows. 1) How do the different vesicle properties and administration method influence the pulmonary reaction? 2) What are the roles of C3a, C5a, and C-activating antibodies? and 3) Is there correlation between in vitro antibody binding and C activation by liposomes and their in vivo pulmonary vasoactivity? Among the various liposome preparations, we studied liposome-encapsulated hemoglobin (LEH) to evaluate the safety of this potential red blood cell substitute (27) and liposomes containing 71 mol% cholesterol (Chol), which were previously used in our laboratory as efficient activators of C (3) and inducers of anticholesterol antibodies in animals (5, 32).

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Reagents. Dimyristoyl and distearoyl phosphatidylcholines (DMPC and DSPC), dimyristoyl phosphatidylglycerol (DMPG), and Chol were purchased from Avanti Polar Lipids (Alabaster, AL). Human C5a and rabbit anti-swine IgG-FITC conjugate were from Sigma Chemical (St. Louis, MO). Goat anti-swine IgM-alkaline phosphatase conjugate and FITC-conjugated F(ab)₂, directed against goat Ig, were from Kirkegaard and Perry Laboratories (Gaithersburg, MD) and Cappel (West Chester, PA), respectively. IgM-enriched intravenous Ig (IVIG, Pentaglobin) was obtained from Biotest Pharma (Dreieich, Germany), and IVIG containing Fc-depleted IgG (Gamma-Venin P) was from Centeon Pharma (Wien, Austria). Human C3a was obtained from Calbiochem (San Diego, CA). Lipopolysaccharide (LPS; Escherichia coli 0111:B4) was from Difco, Chicago, IL. Heat-sterilized, bis(3,5-dibromosalicyl)fumarate (diaspirin)-cross-linked human hemoglobin (Hb) was provided by the Blood Research Detachment of the Walter Reed Army Institute of Research. It was prepared as described earlier (41) and contained 12 g Hb/100 ml with <5% methhemoglobin.

Measurement of hemodynamic changes. Experiments were performed in accordance with the guidelines of the Committee on Animal Care of the Uniformed Services University of the Health Sciences. Yorkshire swine in the 25-40 kg range (female, except for 2 male pigs when indicated) were sedated with intramuscular ketamine (500 mg) and anesthetized with 1% halothane using an anesthesia machine. Other details of surgery, instrumentation and hemodynamic analysis were described previously (34). In the present study, we focused only on the changes of PAP, as it was shown previously to be the most consistent and reproducible measure of liposome-induced hemodynamic changes (34).

Preparation of liposomes and LEH. LMV and large multilamellar liposome-encapsulated Hb were prepared by thin-film hydration as follows. Chloroform solutions of the phospholipids and Chol were mixed in appropriate ratios in sterile, pyrogen-free round-bottom flasks, and a film was formed on the wall of the glass by removing the solvent in a rotary evaporator. This was followed by drying under vacuum for >1 h and hydration with PBS or Hb by vigorous vortex mixing. To remove unencapsulated Hb from LEH, liposomes were washed in PBS three times for 20 min at 10,000 g. LMV containing only DSPC were intermittently heated (at 65°C in a water bath) upon hydration to facilitate bilayer formation. Large unilamellar PBS and Hb-containing vesicles (LUV and LU-LEH) were prepared by subjecting the respective LMV to high-pressure shear stress in a small-volume microfluidizer (4 cycles at 60 psi; Microfluidics, Boston, MA). Because the washings and microfluidization were not associated with significant loss of lipid, the liposome doses injected and concentrations used for in vitro incubations were calculated on the basis of initial lipid input.

Analysis of vesicle size. The size distribution of microfluidized LUV was measured by quasi-elastic light scattering (QELS), using a Nicomp model 370 (Pacific Scientific, Silver Spring, MD) submicron particle sizer. This method was not sensitive in the supramicron range; therefore, LMV were analyzed in a Cell-DYN 3500 (Abbott Laboratories, Abbott Park, IL) flow cytometer in the platelet window [measuring particle volume distribution in the 2-30 femtoliter range, corresponding to spherical vesicles with diameter (d) 1.6 µm  d  3.9 µm]. We used the width/height ratio of the descending slope of the distribution curve (at 50% peak height) to quantitate the large liposome content (and, hence, polydispersity) of different LMV preparations.

Experimental protocol. Liposomes and LEH were administered in the jugular vein either as a bolus or by infusion. Bolus injections were 1- to 25-ml aliquots from the 5-40 mg lipid/ml (5-40 mM phospholipid) stock solutions, as specified in the legends to Figs. 1 and 2. They were injected within about 5-15 s and washed into the pulmonary circulation with 5-6 ml PBS. Infusion of 40 mg/ml liposomes (or LEH) at 2-8 ml/min, as specified in the legend to Fig. 3, and on Fig. 4 and Table 1, was performed with an infusion pump.

View larger version (26K): [in this window] [in a new window]   Fig. 1.   Effect of liposomes on pulmonary circulation in pigs. A: time course of pulmonary arterial pressure (PAP) changes in response to bolus injection (at time 0) of 5 mg large multilamellar vesicles (LMV) consisting of DMPC/DMPG/Chol 50:5:45 (mol%), suspended in 1 ml PBS. The two curves present typical responses to the same formulation in two animals. B: PAP responses to 5-mg boluses from different LMV, as shown in the key. Bars are means ± SE; no. of pigs is in parentheses. * Bars 2 and 3 significantly different from bars 1, 4, 6, and 7. ** Bar 6 significantly different from all other bars. *** Bar 7 significantly different from bar 2. #Bar numbers. PC, DMPC; PG, DMPG; Ch, Chol; SpC, DSPC. Statistical significance was determined by ANOVA followed by multiple comparisons (Student-Newman-Keuls test).

View larger version (15K): [in this window] [in a new window]   Fig. 2.   The effect of vesicle size on liposome-induced pulmonary hypertension. A: dose-response curve for the pulmonary hypertensive effect of DMPC/DMPG/Chol 50:5:45 LUV boluses; data pooled from 6 pigs. Different small symbols indicate different animals. The large circles and squares represent the dose-response data reported earlier (34) for the corresponding LMV. Bolus volumes for LUV ranged between 1 and 25 ml according to the total amount of lipid injected from stock solutions containing either 5 or 40 mg/ml lipid. B: similar dose-response data for microfluidized liposome boluses consisting of DMPC/DMPG/Chol 26:5:71 (solid circles); DSPC/DMPG/Chol 50:5:45 (open squares), and DMPC/DMPG/Chol 50:5:45 with encapsulated hemoglobin (solid squares). Data are from a representative pig; repeat tests in 1-3 additional pigs resulted in essentially similar reactions. C: correlation between size and pulmonary hypertensive effect of LMV. The rises in PAP caused by different LMV (Fig. 1B) were plotted against the width-to-height ratio (at 50% height) of the upper, descending part of respective size distribution curves. The encircled numbers specify the preparation in Fig. 1B; circle 5 was not analyzed by flow cytometry. LUV, large unilamellar vesicles; DMPC, dimyristoyl phosphatidylcholine; DSPC, distearoyl phosphatidylcholine; and DMPG, dimyristoyl phosphatidylglycerol.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   Pulmonary effects of various liposome infusions. Bottom: pulmonary hypertensive effects of different microfluidized LUV (A and B) and of LMV (C), administered as bolus or infusion in individual pigs. The timing of injections is indicated by arrows, and the total amount of injected lipid is visualized by the bars (for bolus injection) and filled subcurve areas (for infusion) on the upper segment. Representative experiments as follows. In A the animal received 5- and 10-mg boluses from DMPC/DMPG/Chol LMV (50:5:45, d = <1.6-3.9 µm, solid black bars) to establish pulmonary reactivity, followed by bolus, and then infusion (at 4 ml/min) of the indicated amounts of DMPC/DMPG/Chol (50:5:45) LUV (d = 0.19 ± 0.1 µm, gray bar and gray subcurve area). After termination of the infusion, the animal received 20 mg DMPC/DMPG/Chol LMV (solid black bar) in bolus to rule out tachyphylaxis. In B, a pig was treated similarly, except that DMPC/DMPG/Chol (50:5:45) LUV (d = 0.19 ± 0.1 µm, gray bars) and DSPC/DMPG/Chol (50:5:45) LUV (d = 0.35 ± 0.25 µm, open bar) were also tested as bolus injections, in addition to the same LMV boluses as in A. This animal was infused with DMPC/DMPG/Chol (50:5:45) microfluidized liposome-encapsulated hemoglobin (LEH) (d = 0.45 ± 0.48 µm, gray subcurve area). In the experiment shown in C, after 3 pretest boluses from DMPC/DMPG/Chol LMV (solid black bars), the animal was treated with 10 mg/kg indomethacin iv, (open arrow) followed by bolus and then infusion of the same LMV preparation. Infusion in all 3 pigs was 4 ml/min from a 40 mg lipid/ml liposome stock. These and additional infusion experiments are quantified in Table 1.

View larger version (18K): [in this window] [in a new window]   Fig. 4.   The influence of infusion rate on the pulmonary response to microfluidized liposomes. DMPC/DMPG/Chol (50:5:45) microfluidized LEH (d = 0.45 ± 0.48 µm) was infused in a pig at rates and overall amounts shown in A. The corresponding changes in PAP are shown in B. A repeat experiment with vasoactive LUV (DSPC/DMPC/Chol, 50:5:45, d = 0.35 ± 0.25 µm) showed essentially similar correlation between infusion speed and PAP.

Measurement of liposome-induced complement consumption in vitro. Liposomes were incubated with undiluted pig serum for 5-30 min at 37°C with shaking, followed by centrifugal separation of vesicles (18,000 rpm for 3 min) and determination of the serum volume causing 50% hemolysis under standardized assay conditions (CH₅₀) per milliliter as described earlier (38).

Flow cytometric analysis of immunoglobulin binding to liposomes. Liposomes were incubated with undiluted pig serum (10 mg/ml) for 30 min at 37°C with gentle shaking. After centrifugal separation, vesicles were fixed in paraformaldehyde, washed, stained with FITC-conjugated antibodies directed against swine IgG and IgM, and analyzed with a FACSort flow cytometer as described previously (34).

Statistical methods. Data are presented as means ± SD, SE, or as individual observations. The efficacy of (liposome) treatments was established by comparisons with baseline using paired t-test; correlations between two variables were analyzed by linear regression, and differences between groups were examined using ANOVA followed by multiple comparisons with the Student-Newman-Keuls test.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Influence of lipid composition on the pulmonary hypertensive effect of LMV. Figure 1A demonstrates typical pulmonary responses to 5 mg DMPC/DMPG/Chol (50:5:45 mol%) LMV (defined as "standard" bolus in Ref. 34). As reported previously (34), this reaction was remarkably reproducible both in different animals and in one animal upon repetitive injection of the same dose, affording quantitative analysis of the contribution of individual liposome components and other experimental conditions.

Influence of vesicle size. Another unsolved question regarding the pulmonary hypertensive effect of liposomes was the influence of size and associated homogeneity of vesicles. As shown in Fig. 2A, microfluidized liposomes displaying the lowest mean diameter and narrowest size distribution (DMPC/DMPG/Chol LUV, d = 0.19 ± 0.10 µm) had either no or minimal pulmonary hypertensive effect (25% increase in PAP) up to 1 g injected lipid. This (lack of) activity fundamentally differed from that of chemically identical LMV containing large, highly heterogeneous vesicles (d in the 1.6-3.9 µm range), the hypertensive effect of which was >40% above 2 mg lipid, and reached its plateau (at 100% increase of PAP) at doses 20 mg lipid (34).

5

Influence of administration method. Next we addressed the question whether slow infusion vs. rapid bolus administration has an impact on the vasoactivity of liposomes. As shown in Fig. 3, A-C, infusion of the smallest, highly homogeneous LUV caused no changes in PAP (Fig. 3A), whereas intermediate size LUV (Fig. 3B) and LMV (Fig. 3C) led to significant pulmonary hypertension. Thus the pattern of pulmonary response to liposome infusion was the same as seen with bolus injection; i.e., the hypertensive effect correlated with vesicle size above a certain threshold. Figure 3, A-C, and Table 1 also reveal the pigs' cardiovascular tolerance of various liposomes; we could infuse up to 10 g (0.2 g/kg) DMPC/DMPG/Chol LUV without detectable rise in PAP and up to 20 g intermediate size LUV or LEH with only transient, reversible rises in PAP. Because infusion of LMV led to irreversible circulatory collapse and death within minutes, the question of interest with these large liposomes was how effective indomethacin, the most potent inhibitor of standard LMV bolus-induced hemodynamic changes (34), was in the case of LMV infusion. As shown in Fig. 3C and Table 1, indomethacin allowed the infusion of up to 4.5 g LMV without changes in PAP; however, its protective effect vanished after about 30 min, and a steep rise in PAP caused the death of animals.

Influence of infusion speed. The extent of pulmonary hypertension caused by infusion of intermediate size LUV (e.g., that containing Hb or DSPC) depended on the rate of infusion. This effect is illustrated in Fig. 4 for the case of microfluidized LEH, where increasing the infusion rate from 4 ml/min to 6 and 8 ml/min caused additional 20-30% elevations in PAP over the values observed at the lower speed.

Correlation between liposome-induced complement activation in vitro and pulmonary hypertension. Table 2 shows that short-term (10 min) incubation of different liposomes with pig serum resulted in C consumption that was proportional with the pulmonary hypertensive effects of vesicles. However, after 30-min incubation, or at a higher serum liposome level (8 vs. 2 mg/ml), C activation was advanced with all preparations, and the proportionality with pulmonary effects tended to disappear. Accordingly, linear regression analysis of PAP (% of baseline) plotted against %C consumption gave statistically significant linear correlation at 10 min (R² = 0.71, P < 0.05), but not at 30 min. These data are consistent with a causal relationship between C activation and pulmonary vasoactivity of liposomes, showing at the same time that the parameters of in vitro C assay are critical in demonstrating such correlation.

Pulmonary effects of human C3a, C5a, and immunoglobulins. Additional yet unresolved questions regarding the involvement of C in the pulmonary reaction to liposomes relate to the roles of C5a vs. C3a and to the exact pathway of C activation. We injected pigs intravenously with human C3a, C5a, and two human Ig preparations: 1) an IgM-enriched mixture of immunoglobulins that reacts with porcine endothelial cells causing massive C activation (Pentaglobin) (9, 29) and 2) an IgG F(ab)₂ product (Gamma-Venin P) that lacks the C activating Fc portion of Ig. Table 3 shows that C3a, C5a, and Pentaglobin caused significant rises in PAP on a time course that was essentially identical (within 1-2 min) with that of the liposome-induced reactions, whereas Gamma-Venin P caused no pulmonary reaction even at a 3,000- to 5,000-fold higher dose. These observations provide evidence that 1) both C3a and C5a can induce pulmonary hypertension, although C5a has significantly greater efficacy, and that 2) the reaction to liposomes has the same basic features as an anaphylactic shock induced by xenoreactive antibodies via classic pathway activation of C. 

Correlation between pulmonary vasoactivity and antibody binding to liposomes. To investigate further the possibility that classic pathway activation of C is a major underlying cause of pulmonary reaction, we measured the binding of natural antibodies to different liposomes in vitro. The fluorescence-activated cell sorting (FACS) analysis shown in Fig. 5 indicated abundant binding of IgG and IgM to the highly vasoactive LMV, whereas the chemically identical, vaso-inactive LUV showed no or minimal binding. In further correlation with pulmonary vasoactivity, liposomes with 71% Chol bound significantly more Ig than those containing 45% Chol. These findings, taken together with analogous correlations between in vitro C consumption and liposome size and Chol content (Table 2), imply that the binding of naturally occurring IgG and IgM antibodies to liposomes may be rate limiting to both C activation and subsequent pulmonary changes. Consequently, natural antibody-mediated classic pathway C activation may be the predominant ultimate cause of liposome-induced pulmonary changes.

View larger version (19K): [in this window] [in a new window]   Fig. 5.   Binding of porcine IgG and IgM to liposomes in vitro. Following incubation with PBS or pig serum, liposomes were fixed, washed, incubated with antibodies against porcine IgG or IgM, and counterstained with FITC-conjugated F(ab)2. For each liposome sample, 10,000 data points were acquired. Vesicles stained with an unrelated polyclonal antibody showed no greater fluorescence than those incubated with PBS (data not shown). The PBS control, i.e., vesicles incubated with PBS instead of pig serum and stained with anti-IgG or anti-IgM, were identically negative for all varieties of liposomes (shown in the first column for 45% Chol liposomes). The presented data were repeated in an independent experiment with essentially identical results. The PBS control, i.e., vesicles incubated with PBS and stained with anti-IgG or anti-IgM, indicated no antibody binding to any of the studied preparations (shown for 45% Chol LMV).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Role of C in liposome-induced pseudoallergy. Since the first description of liposome-induced C activation in 1969 (20), numerous studies have analyzed this phenomenon (reviewed in Ref. 33). The profound hemodynamic actions of anaphylatoxins were also recognized some 30 years ago (6), yet, to our knowledge, the possible causal connection between C activation and acute hemodynamic side effects of liposomal drugs in patients has not been proposed and investigated in a systematic fashion prior to a recent study from our laboratories (34). In the latest report on Doxil-induced hypersensitivity reactions (31), for example, the authors recognized the similarity of clinical symptoms to those caused by hemodialysis, which is known to be mediated by C5a (22). Also, they described transient neutropenia and increased leukocyte adherence, i.e., classic hallmarks of C activation (22), only in patients who displayed hypersensitivity symptoms. Nevertheless, C activation was not considered as an underlying cause (31). One reason for a common oversight of this mechanism may lie in the traditional perception of the C system as being associated with immunology (host defense) and not with pharmacology (adverse drug reactions) or physiology (circulatory changes).

Influence of liposome lipids. The fact that electrically neutral LMV composed of only DMPC and DSPC were capable of inducing pulmonary hypertension was unexpected, as earlier in vitro studies showed C activation only by vesicles supplemented with charged and/or antigenic (glyco)lipids or other ligands (33). Thus the present study provides evidence for the first time that C-mediated vasoactivity may be an intrinsic property of even the simplest monocomponent phospholipid vesicles, provided they exceed a threshold size, as discussed below. The observation that DMPG augmented the pulmonary reaction to liposomes is in keeping with numerous reports on C activation by anionic liposomes (8, 25, 33). This may occur both through the classic and the alternative pathways by binding immunoglobulins and C3, respectively (8, 25, 33). The effect of immunoglobulins may, however, overshadow activation via the alternative pathway (24).

Effect of liposome size. In theory, antibody binding and C activation by liposomes should be proportional with the overall surface area of vesicles directly exposed to plasma, which (considering equal amounts of lipid) is smaller with LMV than with small unilamellar vesicles. Our findings therefore that LMV had significantly greater C-activating and pulmonary hypertensive effects than their smaller counterparts, and that liposomes with 0.2 µm caused no hemodynamic changes, were counter-intuitive and unexpected. It is conceivable that natural antibodies have increased binding affinity to liposomes with flat surface, due to steric proximity or favorable positioning of epitopes. A further factor could be the spatial arrangement of surface-bound antibodies, as the activation of complement protein 1 (C1) requires multiple IgG molecules to be aligned in a special parallel configuration (22). Finally, it is not excluded that other plasma proteins (albumin, C reactive protein, C1q) bound on the surface of LMV accelerate C activation in a cooperative fashion.

Role of endotoxin. In our previous study (33) we have shown that the liposome-induced hemodynamic response in pigs can be completely blocked by the specific C inhibitor, soluble C receptor type 1 (sCR1), which strongly argues against a major direct role of endotoxin in vasoactivity. Endotoxin can activate C; nevertheless, the possibility that the trace amounts coinjected with liposomes played a major role in C activation could also be ruled out on the following basis. First, in control experiments, we observed significant rise of PAP after intravenous injection of 50 µg/kg E. coli LPS, but not 2.5 µg/kg, i.e., only at five to six orders of magnitude higher LPS doses than the maximal amount coinjected with liposomes (data not shown). Consistent with these data, the reported dose where S. enteritidis caused massive hemodynamic changes in pigs was in the 100-250 µg/kg range (16). Second, there was no correlation between LPS contamination and PAP changes, with the purest empty LMV preparation causing maximal rises in PAP. Finally, previous studies in rats provided evidence that LPS contamination was not responsible for C activation by liposomes and LEH in vitro (37).

Mechanism of liposome-induced C activation. At present our hypothesis regarding the mechanisms of C activation by LMV in pig blood and its acceleration by 71% membrane Chol is as follows. The reaction is essentially due to the binding by LMV of naturally occurring, C-activating anti-lipid antibodies that are present in the blood of most mammalian species including pigs (4, 40) and which include anti-phospholipid and anti-Chol antibodies. As mentioned, it is known from the literature that antibody binding to liposomes is accelerated by negative charges on the membrane, and we have shown here that 71 mol% Chol further increases this binding (vs. 45%). One possible explanation for this difference is that the excess Chol in 71% liposomes may become exposed on the membrane surface as patches of microcrystals, becoming increasingly accessible to anti-Chol antibodies. There are several lines of evidence supporting this theory: 1) phosphatidylcholine dispersions enriched in Chol to levels of 2-3 mol Chol per mole phosphatidylcholine are metastable, with formation of Chol monohydrate crystals (10); 2) negatively charged liposomes containing 71% Chol were shown to bind IgM with associated formation of giant, multimicron structures in plasma (15); 3) the recognition of Chol by anti-Chol antibodies in biological membranes and in lipoproteins critically depends on the steric position of -OH epitopes (4), and 4) crystalline Chol is known to efficiently bind anti-Chol antibodies and to activate C in human blood (19, 30). It seems worthwhile to point out in this context that C activation by microcrystalline Chol may be an important pathogenic factor in precipitating acute thrombotic or thromboembolic events in arteriosclerosis, at sites of ulcerated arteriosclerotic plaques (17-19, 30).

Pulmonary effects and long-term consequences of liposome and LEH infusion. In addition to composition and size, the third critical factor in liposome-induced pulmonary hypertension was the method and speed of intravenous administration. In particular, administration of intermediary size (0.2 µm  d  1.1 µm) liposomes by infusion attenuated and delayed the pulmonary reaction relative to that observed with bolus injection, and the speed of infusion was also critical with these vesicles. These observations can most easily be rationalized by blood anaphylatoxin levels being rate limiting to pulmonary vasoactivity. As is known, the steady-state levels of functional anaphylatoxins are set by the relative rates of production from plasma C3 and C5 and clearance by cellular receptors and plasma carboxypeptidases (22). With clearance most likely uninfluenced by the method of administration, anaphylatoxins level in the blood will depend not only on the potency of vesicles for C activation but also on the speed by which liposomes enter in blood. This mechanism provides explanation for the reversal of hypersensitivity reaction to liposomes in some patients by slowing down the rate of infusion (14).

Role of anaphylatoxins and pathway of liposome-induced C activation. The observation that human C3a and C5a mimicked the liposome-induced vascular reaction is consistent with the partial efficacy of murine anti-C5a antibody in our previous study (34). It suggests that inhibition of the formation or the action of C5a may not be sufficient to completely suppress the vasoactivity of liposomes. The observed difference between the efficacies of C3a and C5a is in keeping with the relative efficacies of these anaphylatoxins in other spasmogeneity assays (23).

Clinical and theoretical implications. By highlighting the C mechanism, the present study helps in solving the hypersensitivity riddle observed with Doxil and other liposomal drugs. We demonstrate that small unilamellar, highly homogeneous liposomes, like those supposedly present in liposomal drugs, are not prone to activate C. This suggests that the clinical formulations of reaction-causing liposomal drugs may have unique yet unclarified features that render them C activators in vivo that may need to be explored.