Research ArticleNanomedicine

Liposome-supported peritoneal dialysis for detoxification of drugs and endogenous metabolites

See allHide authors and affiliations

Science Translational Medicine  15 Oct 2014:
Vol. 6, Issue 258, pp. 258ra141
DOI: 10.1126/scitranslmed.3009135


Peritoneal dialysis confers therapeutic advantages in patients with renal insufficiency and has proven beneficial in other indications, such as removal of excess metabolites or overdosed drugs. However, it is used in only about 10% of the dialyzed population worldwide, partly owing to the lower clearance rate compared with hemodialysis. We have developed a dialysis medium based on liposomes with a transmembrane pH gradient (basic or acidic aqueous core) that could improve the efficacy of peritoneal dialysis, specifically for the removal of excess metabolites or overdosed drugs. These scavenging vesicles are able to extract ionizable drugs and toxic metabolites into the peritoneal space and can be easily withdrawn from the body at the end of dialysis. This approach was used to successfully remove ammonia from rats with a greater extraction efficiency than traditional peritoneal dialysis, and may therefore prove useful in the treatment of severe hyperammonemia. Liposomal dialysis was also used to concentrate exogenous compounds in the rat peritoneal cavity, allowing for sequestration of several drugs that are frequently involved in overdose in people. In particular, liposomal dialysis counteracted the hypotensive action of the cardiovascular drug verapamil more efficiently than did control dialysis in a rat model of drug overdose. These findings highlight the versatility and advantage of this liposome-based approach for emergency dialysis.


Renal replacement therapy is indicated as a life-supporting measure in acute kidney injury and chronic kidney disease as well as in nonrenal applications, such as for the treatment of intoxications and several metabolic disorders. The primary renal replacement modalities include intermittent hemodialysis (HD), peritoneal dialysis (PD), and continuous dialysis therapies (hemofiltration and hemodiafiltration), with an incontestable clinical predominance of HD. Although the proportion of patients receiving PD steadily decreased between 1995 and 2008, recent clinical studies suggesting comparable prognosis between HD and PD, as well as improved technology and governmental support, have fostered a renewed medical interest for PD; for instance, incidence increased by 300% over the past 5 years (13).

PD offers flexibility and autonomy to the patients, and there are indeed instances where PD would represent a valuable alternative to HD, such as in emergency situations (severe acute intoxications), when specialized HD facilities are not available, or when the physiological or health status of the patient makes HD a difficult and risky procedure (neonates or hemodynamic instabilities). PD uses the peritoneum as a natural semipermeable membrane to continuously draw waste products and excess fluids from the patient’s blood into the dialysate. As opposed to thin and highly porous filters used in HD, the peritoneum is composed of mesothelial cells as well as connective and interstitial tissues, constraining metabolic waste products to travel a considerable distance to the peritoneal cavity. For this reason, the solute clearance capacity of PD is low, especially when compared to the newer continuous renal replacement therapies implemented in intensive care units (4).

To circumvent this problem and enhance the effectiveness of PD, new dialysis fluids have been introduced in the clinic. A relatively recent breakthrough is icodextrin, a biocompatible glucose polymer that acts as an osmotic agent and has been associated with beneficial effects, such as increased ultrafiltration and better volume balance (5). However, the clearances obtained with this fluid are contingent on a limited colloidal osmotic gradient and remain inferior to what can be achieved with extracorporeal dialysis. This limitation compels patients to long or continuous dialysis exchanges for the removal of metabolic waste products. To date, no other technological advances have addressed this issue sufficiently. In this work, we report a liposome-based dialysis medium with a transmembrane pH gradient (vesicles with basic or acidic aqueous core) that increases clearance efficiency of low–molecular weight ionizable solutes and that could be used to treat various intoxications. This system is unique in that it concentrates the toxic compounds in the liposomal core and therefore generates a virtual sink in the PD fluid (Fig. 1A).

Fig. 1. Liposome-supported peritoneal dialysis.

(A) Concept of LSPD: Weak bases, including drugs (“D”) or ammonia (NH3), diffuse from the blood to the peritoneal space and into the acidic interior of the transmembrane pH gradient liposomes. They become protonated (DH+ and NH4+, respectively) in the liposome’s aqueous core, where they remain trapped because the diffusion of the protonated species through the phospholipid membrane is hindered, thus shifting the tissue distribution of these metabolites and exogenous chemicals toward the peritoneal space. This concept can also be applied to the extraction of weak acids using liposomes containing an alkaline buffer (not shown). Left panel image from Keystone (, adapted with permission. (B) Liposome drainage from the peritoneal space to the blood. After intraperitoneal administration of BODIPY-labeled small (250 nm) and large (850 nm) liposomes (60 ml/kg), the dye fluorescence was measured in plasma aliquots to obtain the liposomal lipid concentration. Data are means + or − SD (n = 3 rats per group). (C) Photographs of rat plasma, sampled 4 hours after the onset of PD (60 ml/kg). The plasma samples were obtained from rats dialyzed with icodextrin solutions containing or not containing liposomes; the more opaque the plasma, the greater the presence of liposomes in the blood.

Liposomes are hollow, spherical, self-closed structures formed by concentric lipid bilayers surrounding an aqueous core. Their size can be adjusted from tens of nanometers to several micrometers. With more than 10 U.S. Food and Drug Administration–approved treatments on the market and many more in clinical phases, liposomal formulations are one of the most successful nanomedicines (6, 7). Liposomes with a transmembrane pH gradient have been used in drug formulation for almost three decades to actively encapsulate therapeutic compounds with high yields (8, 9). In the case of weak bases, this “remote-loading” technique involves the diffusion of uncharged molecules into the vesicle’s acidic core, where they become protonated and subsequently accumulate owing to limited release (Fig. 1A).

Recently, we showed that intravenously injected nanosized liposomes with a transmembrane pH gradient can efficiently reverse drug overdose effects by sequestering the drug in the bloodstream, making it less available to the peripheral tissues (10, 11). Although the preliminary data were promising, the intravenous route of administration was not ideal because of the inherent difficulty of physically removing the encapsulated drug from the body and the risks associated with complement-activated pseudoallergic reactions (observed in several patients treated with chemotherapeutic liposomal formulations) (12). In this regard, PD represents a safer administration mode, because blood exposure to liposomes can be reduced to low levels, even when high amounts of lipids are administered. Moreover, at the end of the dialysis procedure, the physical withdrawal of the toxin-loaded vesicles from the body may eventually spare the liver from extensive metabolic overload.

Here, we assessed in rats the performance of liposome-supported PD (LSPD). Liposomes with an acidic internal buffer were supplemented to an icodextrin PD solution and tested in two different preclinical applications: the extraction of a potentially toxic endogenous metabolite, ammonia, and the extraction of several of the most commonly overdosed prescription drugs, including a calcium channel blocker, verapamil. Removing excessive metabolic waste like ammonia would be valuable to treat hyperammonemia, especially in newborns for whom HD or continuous renal replacement therapy is problematic, partly owing to limited venous access. By removing overdosed drugs, LSPD can counterbalance the negative cardiovascular sequelae and shorten the recovery time compared with standard PD, as demonstrated in this study for verapamil overdose in rats. Both of these examples show the broad clinical potential for liposome-based dialysis therapies in routine and in emergency medicine.


Characterization of LSPD

The first aim was to determine the conditions for preparing a stable, long-lasting PD suspension containing liposomes. The dialysis medium contained 7.5% (w/v) of icodextrin because this glucose polymer promotes sustained peritoneal ultrafiltration (5) and increases solute clearance (13, 14). The choice of the bilayer composition was dictated by previous optimizations (11); formulations containing dipalmitoylphosphatidylcholine (DPPC) and 45 mole percent (mol %) cholesterol are stable under physiological conditions and simultaneously allow rapid drug uptake. Poly(ethylene glycol) (PEG)–conjugated lipids were added to the lipid bilayer at low concentration (1 mol %) to prevent aggregation of the liposomes in the peritoneal medium. Figure 1B shows the concentration of small (250 nm) and large (850 nm) fluorescently labeled liposomes in the blood after intraperitoneal administration to rats. Small liposomes entered the systemic circulation rapidly (<2 hours), whereas the larger vesicles retained longer in the peritoneal cavity. Indeed, only about 0.2% of the total injected dose of large vesicles could be found in the plasma after 4 hours of dialysis, an amount too low to increase the turbidity of the plasma (Fig. 1C). In contrast, at this time point, 2.3% of the small vesicles could be found in the plasma, giving rise to a slightly opalescent plasma (Fig. 1C). Accordingly, the large vesicles were selected for the subsequent PD experiments.

Extraction of endogenous metabolites by PD

When infused into the peritoneal cavity, the liposomes displayed an increase in the internal pH from 3.2 to 5.0 between 3 min and 12 hours after intraperitoneal administration (Fig. 2A). This phenomenon could not be principally ascribed to a loss of the encapsulated buffer, because this particular lipid composition is not leaky in plasma (11). Among the potential endogenous weak bases, ammonia (NH3) was the agent that was most likely to diffuse into the liposomes and neutralize the internal acidic buffer. It has a low molecular weight and can easily partition in lipid membranes (15). Indeed, in an in vitro diffusion system (fig. S1A), the large transmembrane pH gradient liposomes extracted 95% of ammonia added in less than 8 hours (Fig. 2B). When serum was added to the in vitro setting to mimic physiological conditions, the uptake exceeded the total amount of added ammonia, indicating that the ammonia naturally present in the serum was also taken up by the liposomes.

Fig. 2. In vitro and in vivo ammonia (base) extraction and in vitro acidic metabolite uptake.

(A) Change in the internal pH of acidic liposomes after intraperitoneal injection. The vesicles contained the membrane-impermeant dye LysoSensor Yellow/Blue dextran that undergoes a pH-dependent fluorescence emission (Fem) at λem = 540 nm and an isosbestic point at λem = 485 nm (fig. S3A). Dashed lines indicate pH thresholds determined by calibration (fig. S3B). Data are means ± SD (n = 3 different animals). (B) In vitro ammonia uptake by acidic liposomes in dialysis fluid (icodextrin) and in 50% serum (50% isotonic 20 mM Hepes buffer) at 37°C. The red dashed line represents 100% uptake of the ammonia added to the system. The in vitro setup used is illustrated in fig. S1A. Data are means ± SD (n = 6 independent experiments per condition). (C) Plasma and peritoneal (dialysate) concentrations of ammonia during PD and LSPD (60 ml/kg). Data are means ± SD (n = 5 to 6 rats per group). **P ≤ 0.01, comparing dialysate (LSPD) to both plasma LSPD and dialysate (PD), two-sided Mann-Whitney test. (D) Comparison of in vitro ammonia extraction (PD setup in fig. S1B, n = 3) with in vivo animal data in (C) (n = 5). Data are means ± SD. (E) Total ammonia removed from HD, PD, and LSPD in vitro setups after 3 hours of simulation. Data are means + SD (n = 3 to 6 independent experiments per dialysis method). P values were determined by one-way analysis of variance (ANOVA) and multiple pairwise comparison (Tukey test). (F) In vitro uptake kinetics of propionic and isovaleric acids by basic liposomes (containing an internal buffer of calcium acetate, pH 10) in isotonic 20 mM Hepes buffer, pH 7.4, at 37°C. The red dashed line represents 100% uptake of acidic molecules added to the system. Data are means ± SD (n = 6 independent experiments per group).

In vivo PD experiments in rats confirmed the ability of the liposomes to take up endogenous ammonia (Fig. 2C). A 3-hour LSPD treatment sequestered ammonia into the peritoneal space at concentrations more than 20-fold the normal plasma levels for rodents. The extraction process was also fast: after 30 min of dialysis, the peritoneal fluid was 7.5-fold more concentrated in ammonia than plasma. Conversely, the control PD medium without liposomes matched the plasma levels of ammonia without any accumulation effect. We sought to reproduce the in vivo uptake kinetics of LSPD and PD in vitro using custom-made dialysis setups, mimicking different dialysis modalities (fig. S1, B and C). The LSPD extraction efficiency for ammonia after 3 hours of dialysis approached that of simulated (in vitro) HD (3.6 versus 5.5 μmol of ammonia, respectively) (Fig. 2, D and E).

Moreover, by only adjusting the liposome internal pH buffer (calcium acetate buffer, pH 10) to allow the entrapment of acidic compounds, endogenous metabolites other than ammonia could be taken up. For example, the LSPD system was able to extract propionic and isovaleric acids in vitro at an efficiency approaching 50%, which is probably related in part to the lower pH gradient of the liposomes containing the basic medium (Fig. 2F). These acids can build up in the body in the rare congenital metabolic disorders propionic and isovaleric acidemias, respectively.

Treatment of drug-overdosed rats with LSPD

LSPD was then tested in a rodent model of overdose on verapamil, a calcium channel blocker for which no specific antidote is available. The ability of large transmembrane pH gradient liposomes to capture verapamil in vitro was first verified in the absence and presence of serum. In both conditions, 90% of the drug was sequestered in the vesicles within 8 hours (Fig. 3A). At a lipid dose of 600 mg/kg, the liposomes concentrated orally administered verapamil in the peritoneal space of rats (Fig. 3B). With LSPD, the drug was continuously removed from the systemic circulation, with an efficacy that surpassed that of standard PD. After 3 and 11 hours of LSPD, the extracted drug concentrations exceeded those obtained with the icodextrin control solution (PD) by 30- and 80-fold, respectively (Fig. 3B). This control solution (PD) yielded a maximal peritoneal verapamil concentration of 0.12 μg/ml after 1 hour of dialysis (Fig. 3B), which was below the drug’s plasma level (0.52 μg/ml) at the time of sampling (Fig. 3C).

Fig. 3. Liposomal capture of verapamil in vitro and in vivo.

(A) In vitro verapamil uptake by acidic liposomes in Hepes-buffered icodextrin and in 50% serum (50% isotonic 20 mM Hepes buffer) at 37°C. The red dashed line represents 100% uptake of verapamil added to the system. Data are means ± SD (n = 3 or 5 independent experiments per group). (B) Verapamil concentration in dialysate during icodextrin PD and LSPD. Verapamil was administered by oral gavage at t = 0 hours (50 mg/kg), followed by the intraperitoneal injection of dialysis fluid 1 hour later (60 ml/kg). Data are means + or − SD (n = 6 rats per group). **P ≤ 0.01, two-sided Mann-Whitney test. (C and D) Plasma verapamil concentrations (left panels) and area under the plasma versus time curves (AUC) (right panels). (C) Animals were given verapamil orally (50 mg/kg) and then treated with large (850 nm) or small (250 nm) liposomes or PD solution (icodextrin) intraperitoneally (60 ml/kg) 1 hour later. Data are means + or − SD (n = 6 rats per group). (D) Small, long-circulating liposomes (135 nm, 250 mg/kg) or control saline solution was injected intravenously in the tail vein 1 hour after intoxication. Data are means + or − SD (n = 4 to 13 rats per time point). Numbers of replicates are not identical at all time points. See supplementary file [3009135data.xlsx] for detailed sampling times on individual animals. P values in (C) and (D) were determined by Kruskal-Wallis one-way ANOVA on ranks and multiple pairwise comparison (Dunn’s test) and by two-sided Mann-Whitney test, respectively.

In terms of the pharmacokinetic profile, the large liposomes used in LSPD seemed to minimally affect the AUC of verapamil (apparent systemic exposure to the drug) (Fig. 3C). After 6 hours of continuous LSPD, the verapamil plasma level was higher in the LSPD group than in the icodextrin (PD) control. We suspected that after some time, the diffusion of a fraction of liposomes (with encapsulated verapamil) from the dialysis fluid into the blood would artificially increase the total plasma drug levels. To verify this hypothesis, the verapamil pharmacokinetics after PD with small liposomes (these liposomes reached the systemic circulation more rapidly and in higher amounts) and the injection of liposomes directly in the tail vein were characterized (Fig. 3, C and D). It can be seen that as the amount of liposomes in the systemic circulation increased, the total measured plasma concentrations of verapamil rose. However, the drug was trapped in the vesicle core and would therefore be pharmacologically inactive as long as it remained in the liposomes. Indeed, the physical entrapment of drugs in the liposomes directly in the bloodstream and subsequent change in the biodistribution kinetics was shown previously to decrease the pharmacological activity of calcium channel blockers (10, 11).

Besides verapamil, LSPD was able to extract in rats structurally different basic and acidic drugs that are often associated with intoxications (16). These included propranolol (β-blocker), amitriptyline (tricyclic antidepressant), haloperidol (antipsychotic), and phenobarbital (anticonvulsant) (Fig. 4, A and B). Depending on the drug’s structure, uptake efficacies with LSPD surpassed that of icodextrin PD by 1.5- to 130-fold (Fig. 4A). The plateau of dialysate drug concentration reached after 2 to 4 hours (Fig. 4B) might be due to a saturation of the uptake process. The transmembrane pH gradient liposomes were also shown to entrap salbutamol and clenbuterol (Fig. 4C), which are β2-adrenergic agonist anti-asthmatic agents that are occasionally implicated in human intoxications, especially when used as an illicit substance for muscle performance enhancement or weight loss (17). Uptake of salbutamol by the liposomes was significantly lower than that of clenbuterol (Fig. 4C), probably because the latter has a higher logP (2.61 versus 0.02) and possesses two protonable amines, which would improve retention in the acidic liposomal core (18). These drugs have also been associated with food poisoning because they are sometimes illegally used by the meat industry to increase the muscle/fat ratio of livestock (19). In this context, LSPD has the potential to improve the detection of these doping compounds. Following its intravenous administration, clenbuterol was no longer detected in the plasma after only 8 hours whereas it could be found in PD fluids of LSPD-treated rats after 13 and 15 hours (Fig. 4D).

Fig. 4. Extraction of verapamil, propranolol, amitriptyline, haloperidol, phenobarbital, and clenbuterol by PD and LSPD.

(A) Concentrations of drugs in dialysate after 3 hours of dialysis. Drugs were administered by oral gavage at t = 0 hours (30 mg/kg, except for verapamil: 50 mg/kg), followed by the intraperitoneal injection of the dialysis fluid 1 hour later (60 ml/kg). For phenobarbital, liposomes with an internal calcium acetate buffer pH 10 were used. N.D., not detected. P values were determined by a two-sided Mann-Whitney test. Data are means + SD (n = 4 to 6 rats per group). (B) Extraction kinetics of drugs in dialysate during PD and LSPD. Data are means ± SD (n = 4 to 6 rats per drug). The symbol “x” indicates that amitriptyline concentration was not detected for PD. (C) In vitro uptake kinetics of clenbuterol and salbutamol by transmembrane pH gradient liposomes in 50% serum (50% isotonic 20 mM Hepes buffer) at 37°C. The red dashed line represents 100% uptake of drugs added to the system. Data are means ± SD (n = 3 independent experiments). The area under the uptake versus time curve (AUC0–8 h) are compared for the different drugs on the right panel of the figure. AUC0–8 h are computed over the course of 8 hours from kinetics in (C) and Fig. 3A. P values were determined by one-way ANOVA and multiple pairwise comparison (Tukey test). Data are means ± SD (n = 3 independent experiments, except for verapamil, where n = 5). (D) Pharmacokinetics of clenbuterol after intravenous injection (2 mg/kg). LSPD (30 ml/kg) was initiated 10 hours after the drug was administered, and the PD fluid was withdrawn after 3 or 5 hours. The limits of quantification and detection (LOQ and LOD, respectively) are displayed as horizontal red dashed lines. The n values above time points 4 and 6 show how many animals were above the LOQ and therefore were averaged. For the 8- and 10-hour time points, open triangles were replaced by “x” because clenbuterol concentrations were below LOQ for all of the animals. Data are means ± SD (plasma pharmacokinetics, n = 3 to 11 rats per time point). Numbers of replicates are not identical at all time points. See supplementary file [3009135data.xlsx] for detailed sampling times on individual animals. The blue circles on the right represent clenbuterol concentration in dialysate after 3 or 5 hours of LSPD.

Last, we evaluated the effect of a 3-hour treatment with LSPD on the pharmacological activity of a high dose of verapamil (Fig. 5A). Oral administration of verapamil (50 mg/kg) induced a 20% drop in mean arterial pressure (MAP) (Fig. 5, B and C). The group that was treated with icodextrin dialysis (PD) for 3 hours showed no difference in recovery time from the untreated group and did not alleviate hypotension (Fig. 5, B to D). With PD, the blood pressure did return to a preintoxication value, but only after more than 21 hours. By contrast, treatment with LSPD for 3 hours had a more rapid beneficial effect on the hemodynamics of the animals, significantly reducing their MAP recovery time compared to PD by ~70% within 6.5 hours (Fig. 5, C and D).

Fig. 5. Hemodynamic effect of LSPD in rats intoxicated with verapamil.

(A) Experimental design. Baseline blood pressure measurements were taken 24 hours before verapamil oral administration (50 mg/kg). Dialysate (60 ml/kg) was given 1 hour later, for 3 hours, at which time the fluid was removed. Animals were observed for 24 hours total. (B) Normalized MAP, calculated as a percentage change of basal MAP, of the control groups. Untreated animals (n = 7) received only verapamil (50 mg/kg); data are means – SD. Non-intoxicated rats (n = 3) received LSPD (60 ml/kg), but no verapamil; data are means + SD. (C) Impact of LSPD (n = 9 rats) and PD (n = 6 rats) on the hypotensive effect of verapamil. Data are means + SD (LSPD) or means – SD (PD). (D) Recovery times for reaching basal blood pressure after verapamil administration and dialysis. P values were determined by Kruskal-Wallis one-way ANOVA on ranks and multiple pairwise comparison (Dunn’s test). (E) Effect of LSPD on hemodynamic parameters. AUC0–24 h computed from normalized curves such as those in (B) and (C). ABP, amplitude blood pressure (DBP – SBP); DBP, diastolic blood pressure; DNP, dichroic notch pressure; dP/dt+, maximal rate of pressure increase; HR, heart rate; SBP, systolic blood pressure. Data are means – SD (n = 6 rats for PD, 9 for LSPD). P values were determined by Mann-Whitney test.

Figure 5E illustrates the AUC over 24 hours (AUC0–24 h) for various blood pressure parameters for the control (icodextrin PD) and LSPD groups. AUC0–24 h values were computed to measure the impact of the treatment during the whole observation time, not only at specific time points. Significant improvements were achieved for LSPD-treated animals in most of the pressure-related signals, including systolic, diastolic, and dichroic notch pressures, with the greatest difference observed for the diastolic blood pressure [2.8-fold reduction of the AUC versus the icodextrin (PD) group] (Fig. 5E). The positive change in amplitude blood pressure did not reach statistical significance. The heart rate was barely affected by the drug at the selected dose; therefore, the effect of LSPD on this parameter could not be measured.


PD has lately attracted renewed interest within the medical community, because more widespread implementation in patients with kidney failure and in emergency medicine could relieve the pressure on congested health care systems (3, 4, 20, 21). To expand the attractiveness of PD and make it a valuable alternative to HD (at least in some indications), there is an urgent need to develop more effective PD fluids. Here, we demonstrated that supplementing an icodextrin PD fluid with transmembrane pH gradient liposomes represents a simple yet efficient means to improve the extraction process in specific intoxications. The method is well suited for a wide range of small ionizable compounds, but could not be adapted to neutral molecules (for example, urea), to compounds that have a high molecular weight (for example, heparin and bacterial toxins), or to compounds that are too polar (for example, phosphate ions) owing to limited permeation through the phospholipid bilayer.

Developing new PD solutions requires an understanding not only of the complex transport mechanisms of toxic solutes of interest but also of the absorption of the dialysis solution itself. Since its first description in the mid-1980s, icodextrin has been investigated in numerous studies aimed at characterizing its biological fate. Owing to its large molecular weight (about 17 kD), this polydispersed glucose polymer is retained for a longer period in the peritoneal cavity than conventional glucose-based solutions. Unlike glucose and other small molecules, which diffuse rapidly across the peritoneal capillary endothelium, icodextrin’s absorption primarily occurs through convective fluid movement via the dense lymphatics of the peritoneum (22). This lymphatic drainage is virtually the only route for the absorption of macromolecules. Similar to icodextrin, small liposomes (50 to 100 nm) cannot diffuse across the blood capillary endothelium and therefore must travel through the lymphatics to reach the systemic circulation (half-life of about 30 min) (23, 24). When injected intraperitoneally in low volume and concentration, the peritoneal residence time of small liposomes was shown to increase with their size. A recent study indicated that an optimal liposome size for long retention is about 1 μm (25). Our data suggest that during PD, the diffusion of 250-nm liposomes into the blood is further delayed, possibly owing to the important convective transport of peritoneal fluids through the lymphatic vessels. We observed that 850-nm liposomes were detectable in plasma only after 4 hours of dialysis, which was 3 hours later than what had been reported for small-volume intraperitoneal injections (25).

The long residence time of the large (850 μm) transmembrane pH gradient liposomes in the peritoneal cavity and their ability to actively trap ionizable compounds in their aqueous core make them a translational platform for the removal of several toxic substances. LSPD was found to concentrate high amounts of ammonia in the peritoneal space. Ammonia is a neurotoxic waste that is harmful when present in elevated plasma concentrations. Congenital hyperammonemia leads to a high mortality rate (up to 85%) and is often the consequence of rare metabolic diseases, such as urea cycle disorders, organic acidemias, or fatty acid oxidation defects (26). The cornerstone of a successful therapy is the emergency reduction of the serum ammonia level (27). Along with low-protein dietary control and the use of urea cycle supplements and ammonia scavengers, PD plays a role in the first-line management of severe hyperammonemia (2729). PD is even more important in neonates, whose limited vascular access makes HD challenging and who have a peritoneal membrane surface area twice that of an adult when scaled to body weight (30). With ammonia extraction efficiency surpassing that of the control icodextrin solution by at least 20-fold after 3 hours of dialysis, LSPD presents a promising option for pediatric patients. Apart from ammonia, LSPD could be tailored for the removal of other ionizable toxic metabolites, such as isovaleric acid and isopropionic acid shown in our study, which can cause secondary hyperammonemia when present in excess (acidemias) (27).

To mimic a clinical scenario where dialysis is crucial, we tested LSPD for treating drug overdose, a condition for which efficient treatments to counteract potentially severe adverse effects are lacking (31). Apart from a few specific antidotes (for example, N-acetylcysteine for acetaminophen or naloxone for opioids) and recent experimental strategies to counteract anticoagulant agents (32, 33), minimal progress has been made in the field of antidote medicine in the past decade. For treating an increasing number of intoxicated patients (about 2.5 million in the United States per year) (16), emergency clinicians primarily rely on nonspecific supportive measures, such as gastric lavage or administration of activated charcoal applied immediately after overdose, which are moderately effective (34). Here, we used LSPD to remove from the body different drugs that were administered orally at relatively high doses. In particular, we showed that the rapid extraction of verapamil from intoxicated rats counteracted the adverse cardiovascular effects (hypotension) of the drug. Verapamil is a first-generation calcium channel blocker, which when taken in excess can generate lethal combinations of impaired conduction, dysrhythmia, vasodilatation, and negative inotropy (35). LSPD continuously removed verapamil from the treated animals to such an extent that the extracted drug concentration in the peritoneal cavity exceeded by 80-fold that of the icodextrin PD control group after 11 hours of treatment. The most striking outcome was the rapidity with which the treatment counteracted the hypotensive effect of the drug. A 3-hour LSPD session was sufficient to completely restore the basal blood pressure level within 6.5 hours, reducing the recovery time by 70%. This recovery time with PD was 10.5 hours faster than what was obtained in a previous study in which PEGylated liposomes were administered intravenously in rats (11). Such drastic improvement in recovery time could be attributed to physical removal of liposomes from the peritoneal cavity after the dialysis session, thus reducing risk of leakage of encapsulated drug into the bloodstream.

In conclusion, although recent years have witnessed the emergence of a variety of parenteral nanoparticulate systems for sequestering toxins (36, 37), drugs (38), radionuclides (39), and alcohol (40), the intravenous route of administration generally chosen for these approaches remains risky, owing to potential adverse reactions associated with the injection of high doses of practically nonremovable materials in already weakened patients (12). In this respect, LSPD may represent a valuable and simple alternative to intravenous detoxification strategies with applications that extend beyond those of neutralizing exogenous toxic compounds. The possibility of withdrawing the PD fluids from the body at the end of the treatment, combined with the established safety profile of liposomes, positions this simple platform among the biodetoxifying systems with the highest potential for translational emergency medicine. As compared to other detoxifying colloidal systems described in the literature, which are generally specific to one or a few compounds (36, 40, 41), LSPD is characterized by a relatively wide extraction spectrum, minimizing the need to optimize the formulation for each new toxic agent. The uptake capacity of the transmembrane pH gradient liposomes is also 20-fold superior to that of intravenous lipid emulsions (11), such as Intralipid, that are sometimes used off-label in emergency units to treat drug poisoning (42). Finally, the liposomes presented in this work are composed of biocompatible materials that are already found in commercial formulations (DSPE-PEG, cholesterol) or in advanced clinical trials (DPPC) (7), which in principle make them more suitable for use in people than detoxification systems based on nonbiodegradable polymers (36, 43), silica-based nanocapsules that can be hemolytic (44), or nanoparticles containing potentially antigenic exogenous proteins (40).

Future work for translation should aim to scale up the fabrication process of the PD medium under good manufacturing practice conditions, monitor its long-term stability, and characterize its safety profile. An important challenge will be the implementation of a sterilization procedure that will not affect the physical and chemical stabilities of the PD medium. Moreover, it would be interesting to investigate whether icodextrin could be substituted by dextrose as osmotic agent in the dialysis procedure. Although icodextrin is characterized by longer dwell times (5), it is also more expensive and may not be necessary if LSPD is carried out for relatively short periods of time (2 to 4 hours).


Study design

Male Sprague-Dawley rats were used to investigate the ability of LSPD to concentrate ammonia and different drugs in the peritoneal space and accelerate the recovery of altered cardiovascular parameters after the oral gavage of a high dose of verapamil. The oral dose of administered verapamil was chosen to induce a drop in basal blood pressure, which was at least of 20% but not producing any signs of physical distress. The animals were randomly assigned to treatment groups, and the studies were not blinded. The sample size was not based on statistical power considerations. The number of animals in each group was considered sufficient to achieve the study objectives, that is, establish the superiority of LSPD versus PD at extracting ammonia and the tested drugs, and in counteracting the pharmacological effect of verapamil on blood pressure. Sample sizes for each experimental measurement are provided in the figure legend. Data were analyzed using parametric or nonparametric statistical methods for in vitro and in vivo experiments, respectively (see Statistical analysis). In one experiment (Fig. 4D), an outlying value was rejected because of an improper intravenous injection of clenbuterol, which led to an excessive concentration of drug in the dialysate.

Preparation of transmembrane pH gradient liposomes

Liposomes were composed of DPPC (Lipoid); 1,2-distearoyl-sn-glycero-3-phosphoethanol-amine-N-[methoxy(PEG)-2000] (DSPE-PEG) (Lipoid); and cholesterol (Sigma-Aldrich). They were prepared by the film hydration/extrusion method as described previously (11). Unless otherwise stated, DPPC, cholesterol, and DSPE-PEG were co-dissolved at 54:45:1 mol % in chloroform. The organic solvent was subsequently evaporated, and the lipid film was kept under vacuum overnight. The dried film was hydrated with isotonic citrate (250 mM, pH 2, for the entrapment of bases) or calcium acetate buffers (250 mM, pH 10, for the entrapment of acids) unless otherwise stated, and the solution was extruded through double-stacked 400-nm or single 2-μm pore size polycarbonate membranes to achieve about 250- or 850-nm liposomes, respectively, as assessed by dynamic light scattering (DelsaNano C; Beckman Coulter). The pH gradient was established by exchanging the vesicles’ external buffer with a solution of physiological pH and osmolarity, either by size exclusion chromatography (PD MidiTrap G-25; GE Healthcare) in Hepes-buffered saline (20 mM Hepes, 140 mM NaCl, pH 7.4) or by overnight dialysis against normal saline (Float-A-Lyzer 1000 kD; Spectrum Labs) for the in vitro or in vivo studies, respectively.

Determination of liposome concentration in plasma

The cholesteryl-conjugated fluorescent dye BODIPY (FL-C12, Life Technologies) was dissolved in chloroform and incorporated into the lipid mixture at a 0.05 mol % ratio. Solutions of BODIPY-labeled liposomes were diluted in fetal bovine serum (Invitrogen) to concentrations that ranged from 0.25 to 8 mM of total lipids and cholesterol content, and their fluorescence emission (λex = 470 nm, λem = 520 nm) was acquired (Infinite 200 PRO plate reader; Tecan) to construct a calibration curve. The plasma samples from rats dialyzed with icodextrin, icodextrin supplemented with small liposomes, and icodextrin supplemented with BODIPY-containing liposomes (small or large) were measured using the same conditions (see below for sampling protocol). Fluorescence intensities were subsequently compared with the calibration curve to determine the plasma liposomal lipid concentration. Rat plasma volumes were estimated to be 4.22 ml per 100 g of body weight (45) and used to determine the portion of the injected liposomes present in the plasma after 4 hours.

In vivo characterization of intraperitoneally injected liposomes

All the animal experiments were performed in accordance with procedures and protocols approved by the cantonal veterinary authorities (Kantonales Veterinäramt Zürich, licenses 2009082 and 2012189). Adult male Sprague-Dawley rats (weighing about 300 g; Charles River Laboratories) were allowed at least 5 days to acclimate to the surroundings; they had access to food and water ad libitum, and they followed a 12-hour light/dark cycle. On the day of the experiment, freshly dialyzed transmembrane pH gradient liposomes [labeled or not with BODIPY, or LysoSensor Yellow/Blue dextran (L-22460), Life Technologies] were diluted with icodextrin (Extraneal 7.5%; Baxter) to obtain a final lipid concentration of 17 mM. The dialysis solutions were prewarmed to 37°C and slowly infused (60 ml/kg) in the peritoneal cavity of rats kept under isoflurane anesthesia (1.5 to 2.5% in 0.8 ml/min oxygen flow). The instillation was performed with a 20-gauge hypodermic needle. At the end of each experiment, the animals were euthanized either by cardiac puncture performed under deep isoflurane anesthesia (3%) or by carbon dioxide asphyxia followed by a thoracotomy.

Blood sampling. Blood aliquots were withdrawn from the tail veins of rats placed inside a metallic restrainer. To ease the process, the tail was dipped into a warmed water bath for 1 to 2 min before sampling (with a 25-gauge needle). Blood aliquots (maximum 200 μl) were collected in lithium heparin-coated tubes (Microtainer; Becton Dickinson), stored on ice, and centrifuged at 3000g for 10 min at 4°C, within the hour after sampling. The plasma liposome concentration was assayed by spectrofluorimetry, or the plasma was frozen and stored at –80°C for additional ammonia content determination (Supplementary Materials and Methods).

Peritoneal fluid sampling. The rats were briefly anesthetized (using isoflurane inhalation, under similar conditions), and about 400 μl of dialysate was withdrawn through a sterile abdominal puncture with a 22-gauge perforated silicone catheter (Venflon; Becton Dickinson). The aliquots were analyzed by spectrofluorimetry or immediately frozen and kept at –80°C for additional ammonia content determination by colorimetry (46) (Supplementary Materials and Methods).

Pharmacokinetics and peritoneal uptake of drugs

After a 3-day acclimation period to the blood sampling procedure, adult male Sprague-Dawley rats were gavaged orally with verapamil (50 mg/kg, in 10% polysorbate 80), propranolol, amitriptyline, haloperidol, or phenobarbital (30 mg/kg, in 10% polysorbate 80), or injected via the tail vein with an isotonic clenbuterol solution (2 mg/kg, 1.5 mg/ml normal saline). All drugs were purchased from Sigma-Aldrich except phenobarbital (from Hänseler AG). One or 10 hours (for clenbuterol experiments only) after drug administration, the rats were briefly anesthetized (isoflurane, 1.5 to 2.5%), and an icodextrin solution with or without transmembrane pH gradient liposomes (17 mM) was infused intraperitoneally. At predetermined time intervals, the plasma and peritoneal fluids were sampled and frozen for the analysis of drug content (Supplementary Materials and Methods). At the end of the PD procedure, the animals were euthanized.

Assessment of LSPD efficacy in verapamil-overdosed rats

Transmitter implantation. For hemodynamic monitoring, the rats were implanted with a telemetric transducer (TL11M2-C50-PXT; Data Sciences International) capable of recording electrocardiography (ECG), arterial blood pressure, body temperature, and activity (body movements) signals. The procedure is described in Supplementary Materials and Methods.

Experimental setup. The rats were pair-housed in environmentally enriched cages equipped with receivers (RPC-1; Data Science International). Before each experiment, active and resting phase basal values were recorded for 24 hours. On the morning of the study, the animals were weighed and gavaged with a solution of verapamil (50 mg/kg) in 10% polysorbate 80, followed 1 hour later by the intraperitoneal infusion of icodextrin (60 ml/kg) or pH gradient liposomes (60 ml/kg, 600 mg/kg) under light isoflurane anesthesia. After a 3-hour dialysis session, the PD fluid was withdrawn using a perforated Venflon catheter, and the peritoneal cavity was rinsed with 10 ml of normal saline to ensure complete removal of the liposomes. The control experiments without drug or without PD treatment were performed similarly (Fig. 5B). To reduce the number of animals, we subjected each implanted rat to two consecutive, randomly distributed experiments before they were euthanized. Between each experiment, the rats were allowed a washout period of 1 week to ensure complete balancing of blood volume and clearance of the drug and detoxification systems.

Telemetric data collection and analysis. Arterial pressure, ECG, body temperature, and activity were continuously acquired by Dataquest A.R.T. 4.3 acquisition software (DSI). MAP, SBP, DBP, ABP, DNP, dP/dt+, and HR were derived from blood pressure waveforms and averaged every 5 min, using the ecgAUTO analysis software (Emka Technologies). All of the physiological parameters were normalized to active and passive basal values acquired during the night and light phases of the day before the experiment. The AUCs were calculated for 24-hour periods subsequent to drug administration. The recovery time after intoxication was set as the time when the cumulative AUC of MAP reached its first minimum (fig. S2).

Statistical analysis

We analyzed the data from in vivo experiments using nonparametric statistical methods. We assessed the statistical differences for two-group comparisons using a two-sided Mann-Whitney test (OriginPro 9; OriginLab Corp.). For multiple group comparisons (Fig. 5D), we used a Kruskal-Wallis one-way ANOVA on the rank test followed by a post hoc pairwise comparison with Dunn’s test (Prism 6; GraphPad Software). P ≤ 0.05 was considered to be statistically significant. The in vitro experiments were analyzed using Levene’s test of variance homogeneity and a one-way ANOVA followed by a Tukey post hoc test for pairwise comparisons.


Materials and Methods

Fig. S1. Schematic of the two- and three-compartment side-by-side diffusion systems.

Fig. S2. Calculation of blood pressure recovery time.

Fig. S3. Emission of LysoSensor Yellow/Blue dextran and its pH calibration.

Table S1. HPLC instrumentation and elution methods.

Numerical values of experimental data: Supplementary file [3009135data.xlsx]

Reference (47)


  1. Acknowledgments: We thank I. Acimovic for her contribution with the high-performance liquid chromatography (HPLC) analysis of clenbuterol samples and P. Luciani, D. Brambilla, and M. Ivarsson for their critical reading of the manuscript. Funding: This work was supported, in part, by the Swiss National Science Foundation (grant 31003A_124882). Author contributions: J.-C.L. and V.F. conceived the study, analyzed the data, and wrote the manuscript. V.F. performed all of the experiments, except the ammonia and clenbuterol/salbutamol in vitro uptake experiments, and the verapamil/amitriptyline/clenbuterol in vivo quantification, which were conducted by R.D.S. and M.R. Competing interests: On behalf of the authors, ETH Zürich filed a European patent based on the invention described in this study in 2012: Eur Pat Appl EP 12005796.3. Data and materials availability: Individual numerical values of the reported data are available in supplementary file [3009135data.xlsx].
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article