Evaluation of toxicokinetic interactions mediated by plasma protein binding during amoxapine intoxication

Abstract

Toxicity enhancement mediated by plasma protein binding during intoxication remains poorly understood. It is known that in mice, brain penetration of amoxapine (AMX) increases nonlinearly with increasing doses; therefore, this study investigated its potential to enhance toxicity via plasma protein binding. AMX was added to mouse or human plasma and adjusted to therapeutic, toxic, and lethal concentrations. The plasma protein-binding ratio and free AMX concentration were measured using ultrafiltration and equilibrium dialysis. Furthermore, the binding ratios of varying concentrations of chlorpromazine, which is often co-administered with AMX in cases of overdose, was analyzed in the presence of therapeutic AMX concentrations. The binding ratio of AMX exceeded 90%, thereby demonstrating a high binding rate; however, this ratio was lower in human plasma than in mice. A nonlinear increase in free AMX concentration was observed in mouse plasma, particularly at high concentrations. In contrast, free AMX concentration showed a linear increase in human plasma. Neither the therapeutic nor toxic concentration of chlorpromazine produced any visible effects on the plasma protein binding ratio or the free AMX concentration. These results suggest that protein binding of AMX is more readily saturated in mouse plasma than in human plasma. In addition, chlorpromazine inhibits AMX binding to α₁-acid glycoprotein; however, AMX may alternatively binds to albumin, which results in no apparent change in the total binding ratio. Further insights into the toxicokinetic interactions mediated by plasma protein binding are also needed for various toxic substances other than AMX.

INTRODUCTION

When considering the toxicity of drug overdoses, the pharmacodynamic effects during intoxication (toxicodynamics) are important. Particularly, substance-specific pharmacological effects in target organs are amplified and manifested, rendering them predictable to some extent. However, the toxicokinetics associated with absorption, distribution, metabolism, and excretion during drug overdose differ substantially from those observed at therapeutic doses, thereby complicating understanding of drug poisoning-induced organ toxicity. For example, during the absorption process, delayed absorption due to the formation of large drug aggregates within the gastrointestinal tract can lead to delayed-onset toxicity (Bosse and Matyunas, 1999). Furthermore, when drug aggregates do not form, the time to peak drug concentration during poisoning is generally delayed compared to that of therapeutic doses (Sue and Shannon, 1992). During the metabolic process, when drugs are administered in large doses, drug-metabolizing enzymes cannot process them sufficiently, which results in a nonlinear increase in blood concentration (Castanares-Zapatero et al., 2016; Fonck et al., 1998; de la Torre et al., 2000). During the excretion process, excretion rates vary depending on the urine pH. Acidic drugs, such as salicylic acid and phenobarbital, increase reabsorption from the renal tubules when urine pH decreases (Proudfoot et al., 2004). Furthermore, enterohepatic circulation prolongs blood concentrations of various drugs, including carbamazepine, thereby contributing to increased toxicity during intoxication (Sue and Shannon, 1992).

Drugs with larger distribution volumes (Vd) exhibit higher tissue penetration. For example, in cases of overdose involving tricyclic antidepressants or antipsychotics that possess an extremely large Vd, tissue concentrations may exceed blood concentrations. Factors determining the size of Vd include the tissue binding rate and plasma protein binding ratio (PBR) of the drug. The plasma proteins primarily involved in drug binding are albumin (ALB) and α₁-acid glycoprotein (AGP). Considering that unbound or free drugs exert pharmacological or toxicological effects on the target organ, changes in the PBR are important when considering toxicokinetics in drug distribution. In particular, an abnormal increase in the amount of free drug may enhance drug distribution to tissues and contribute to increased toxicity. Specifically, this could occur when two drugs are combined. For example, the free concentration of a drug may increase due to displacement of the binding sites of plasma proteins or saturation. At therapeutic doses, a drug may displace the plasma protein binding of another drug, leading to an unexpected increase in the free drug concentration and adverse events. For example, concomitant use of warfarin with other drugs increases the free warfarin concentration in plasma, leading to an increase in prothrombin time-international normalized ratio and subsequent bleeding events (Tilstone et al., 1977; Elias et al., 2024; Ansong et al., 2025). Furthermore, increasing the dose of valproic acid causes saturation of plasma protein binding, resulting in elevated free drug concentrations and possibly leading to adverse events (Wallenburg et al., 2017). Thus, although interactions mediated by plasma protein binding have been reported as phenomena occurring at therapeutic doses, to the best of the authors’ knowledge, no studies have described the mechanism of toxicity enhancement mediated by plasma protein binding during intoxication. Therefore, this study ultimately aimed to clarify the changes in the PBR and free drug concentrations within the toxic concentration range.

Amoxapine (AMX) was previously examined in vivo using a mouse model, where administering AMX at toxic doses resulted in brain concentrations exceeding five times the dose-increase ratio of the therapeutic dose. These results suggested that AMX markedly increased brain penetration within the toxic range (Inoue et al., 2022). The current study hypothesized that this phenomenon was caused by a nonlinear increase in the free AMX concentration due to the saturation of plasma protein binding, and investigated its potential to enhance toxicity through plasma protein binding. AMX is a second-generation tricyclic antidepressant that is widely used to treat depression. Although AMX is rarely used as a monotherapy, the drug may be used in combination for treatment-resistant depression that cannot be controlled with existing first-line antidepressants, such as selective serotonin reuptake inhibitors or serotonin-norepinephrine reuptake inhibitors. Additionally, 7-hydroxyamoxapine, a metabolite of AMX, possesses dopamine antagonistic activity and is sometimes used as a monotherapy for mood disorders and psychotic depression accompanied by delusions or hallucinations (Cohen et al., 1982; Anton and Burch, 1993). However, in cases of overdose, AMX often exhibits toxicity to the central nervous system (CNS), which is rarely seen with other tricyclic antidepressants, such as imipramine or amitriptyline, and may cause convulsive seizures in refractory cases. As many as 36.4% of patients in AMX overdose cases experienced seizures, with a mortality rate of 15.2% (Litovitz and Troutman, 1983). The CNS toxicity observed with AMX is largely attributable to the potential of the drug for CNS toxicity and its high brain penetration rate. However, the mechanisms remain unclear. The increased brain transfer at the toxic dose observed in this mouse model may explain previously reported clinical cases. However, to the best of the authors’ knowledge, no studies have indicated that plasma protein binding contributes to AMX-induced CNS toxicity. Therefore, in this study, AMX was used to investigate changes in plasma protein binding during intoxication in human and mouse plasma. Additionally, to evaluate drug interactions mediated by the displacement of plasma protein binding during intoxication, combination testing was conducted with chlorpromazine (CPZ), which has been associated with sporadic fatalities when used concurrently with AMX (National Research Institute of Police Science, 2019-2022). The simultaneous ingestion of multiple drugs is frequently observed (Hashimoto et al., 2022). Therefore, to investigate the potential mechanism of toxicity enhancement due to displacement of plasma protein binding by concomitant drugs, we conducted the combination experiments with AMX and CPZ as an example. The study results provide a firm basis for understanding the enhancement of toxicity mediated by plasma protein binding.

MATERIALS AND METHODS

Materials

AMX, aspirin (ASA), and acetazolamide (AZ) were purchased from the Tokyo Chemical Industry (Tokyo, Japan). CPZ was purchased from Wako Pure Chemical Industries (Osaka, Japan). Diazepam-D5 and phenobarbital-D5 in methanol were purchased from Sigma-Aldrich (St. Louis, MO, USA). Human plasma was purchased from Sigma-Aldrich. Mouse plasma was supplied by our department. Amicon Ultra-0.5 centrifugal filters (10 kDa MWCO) were purchased from Merck Millipore (Darmstadt, Germany). Rapid equilibrium dialysis (RED) devices (8 K MWCO) were purchased from Thermo Fisher Scientific (Rockford, IL, USA). All other reagents used were of the highest grade and commercially available.

Stability evaluation of AMX in human plasma

AMX was added to human plasma and adjusted to 1.5 μM. After immediate adjustment or incubation at 37°C for 3, 6, or 24 hr, the samples were extracted using the QuEChERS method (Inoue et al., 2022). Plasma (10 µL) was added to 290 µL of distilled water, 100 mg of Bond Elut QuEChERS (Agilent, Santa Clara, CA, USA), and 300 µL of acetonitrile containing the internal standard (IS) and mixed well using a vortex mixer. The mixture was centrifuged for 10 min at 15,000 × g at room temperature. The supernatant was then collected and subjected to liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis.

Evaluation of the total recovery rate, plasma protein binding ratio (PBR), and free concentration of AMX by ultrafiltration

AMX was added to mouse or human plasma at the following concentrations: 1.5 µM (therapeutic range), 9 µM (toxic range), and 15 µM (lethal range) (Schulz et al., 2020). These concentrations were derived from the case in humans, yet toxic and lethal concentrations may be analogous in the murine model (Inoue et al., 2022). The samples were then incubated at 37°C for 1 hr. Thereafter, 500 µL of AMX-containing mouse or human plasma was applied to a filter device (Amicon Ultra-0.5 centrifugal filters, 10 kDa MWCO), which was pretreated with 0.05% Tween 20, centrifuged at 20°C and 14,000 × g for 10 min, and the filtrate was then collected. The filter device was inverted, centrifuged at 1,000 × g for 2 min at 20°C, and the concentrate was collected. Subsequently, AMX was extracted from mouse or human plasma as well as from their filtrates and concentrates using the QuEChERS method. Thereafter, the extracts were subjected to LC-MS/MS analysis. The concentrates were diluted with blank mouse or human plasma, as necessary. The total recovery rate and the PBR were calculated as follows:

Total recovery rate (%) = 100 × (Wc×Cc+Wf×Cf) / Wo×Co

PBR (%) = 100× (1 - (Cf / Co)

Wc = total weight of concentrate before assay, Wo = weight of original starting material, Wf = weight of filtrate, Cc = concentrate concentration, Co = original starting material concentration, Cf = filtrate concentration

Evaluation of PBR and free concentration changes using equilibrium dialysis

AMX was added and adjusted to concentrations of 1.5, 5, 10, 15, and 30 µM for mouse plasmas or 1.5, 15, and 60 µM for human plasma. In the combination experiment with CPZ, human plasma containing either a therapeutic 0.75 µM CPZ concentration or a toxic 2.5 µM CPZ concentration (Schulz et al., 2020), as well as 1.5 µM AMX, was used. In the combination experiment with AZ and ASA, 35 µM AZ (therapeutic concentration) and 2150 µM ASA (toxic concentration) (Sweeney et al., 1986) were added to human plasma.

The procedure was performed according to the user guide for the Pierce RED Device Single-Use Plate with Inserts (8 K MWCO). The RED device consisted of the following two chambers: a sample chamber and a buffer chamber. The sample chamber was loaded with 300 µL of plasma, and the buffer chamber was loaded with 550 µL of phosphate-buffered saline (PBS). The device was sealed with tape, placed on a rotator, and incubated overnight at 37°C. Thereafter, 50 µL of the sample was recovered from the sample and buffer chamber. To standardize the solvent matrix of the samples, 50 µL of PBS or plasma was added to each, bringing the total volume to 100 µL. Distilled water (200 µL), 300 µL of acetonitrile, 100 mg of Agilent Bond Elut QuEChERS, and 20 ng of diazepam-D5 were then added to form an IS. The resultant mixture was mixed thoroughly and centrifuged at 15,000 × g for 10 min at 20°C. The supernatant was collected and analyzed using LC-MS/MS. Depending on the drug measured, the following IS was added: 20 ng diazepam-D5 for AMX, CPZ, and AZ, and 100 ng of phenobarbital-D5 for ASA. The PBR was calculated as follows:

PBR (%) = 100 - (Concentration in the buffer chamber / Concentration sample chamber) × 100%

LC-MS/MS analysis

LC-MS/MS was performed using an LC-40ADXR and LCMS-8045 (Shimadzu, Kyoto, Japan). Chromatographic separation was achieved on a Phenomenex Kinetex XB-C18 column (2.1 mm I.D. × 100 mm., 2.6 µm; Phenomenex, Torrance, CA, USA) with an equivalent Phenomenex Security Ultra C18 guard column (2.1 mm ID). The column temperature was set at 40°C. The injection volume was 1 µL or 5 µL, depending on the concentration of the sample. The mobile phases were (A) 10 mmol/L ammonium formate and 0.1% formic acid and (B) methanol containing 10 mmol/L ammonium formate and 0.1% formic acid. The initial elute condition was set at 5% B, changed linearly to 95% B in 7.5 min, held for 2.5 min, immediately changed back to 5% B, and held for 5 min. The flow rate was set to 0.3 mL/min. After electrospray ionization, the samples were analyzed in multiple reaction-monitoring modes. AMX, CPZ, AZ, and diazepam-D5 were detected in positive mode, whereas ASA and phenobarbital-D5 were detected in negative mode. The flow rates of the nebulizer, drying, and heating gases were set to 3, 10, and 10 L/min. The temperatures of the interface, desolvation line, and heat block were set at 300, 250, and 400°C, respectively. The instrumental conditions for each drug are listed in Table 1.

Table 1. Liquid chromatography–tandem mass spectrometry (LC-MS/MS) conditions.

RESULTS

To investigate the binding pattern of AMX with plasma proteins, fundamental aspects, such as AMX stability and the total recovery rate in plasma, were examined. Time-dependent changes in AMX during incubation at 37°C in human plasma are shown in Fig. 1A. The AMX concentration in human plasma measured using LC-MS/MS showed little change for up to 6 hr but decreased to 91% after 24 hr. Therefore, AMX remained stable in human plasma for at least 6 hr.

Fig. 1

Assessing the feasibility of plasma protein binding analysis of AMX by ultrafiltration. (A) Changes in the retention rate of AMX in human plasma incubated at 37°C were measured using LC-MS/MS at 0, 3, 6, and 24 hr. Values represent the mean ± SD (n = 3). (B and C) Total recovery rate of AMX was measured via ultrafiltration using mouse (B) and human (C) plasma spiked with therapeutic (1.5 µM), toxic (9 µM), and lethal (15 µM) concentrations of the drug. Prior to ultrafiltration, the filter device was pretreated with 0.05% Tween 20 to prevent adsorption. AMX in the filtrate and the residue was measured using LC-MS/MS. Values represent the mean; B, n = 2 (1.5 µM and 9 µM) or n = 3 (15 µM) or mean ± SD (C, n = 3). AMX, amoxapine; LC-MS/MS, liquid chromatography–tandem mass spectrometry; SD, standard deviation.

Subsequently, the total recovery of AMX was examined using mouse and human plasma spiked with therapeutic to lethal AMX concentrations (Fig. 1B and C). Ultrafiltration was employed to investigate whether highly reproducible experiments could be achieved, given the high hydrophobicity and strong adsorption properties of AMX. Agents characterized by high PBR, such as AMX, have been observed to adsorb to plastic laboratory equipment. To prevent such adsorption, a pretreatment of the filter device with 0.05% Tween 20 prior to ultrafiltration was employed. When mouse plasma was used, the total recovery rate was approximately 80% at the therapeutic concentration (1.5 µM), but this rate was stable at over 90% at the toxic (9 µM) and lethal (15 µM) concentrations. In human plasma, the total recovery rate showed a slight decrease at toxic and lethal concentrations compared to that at the therapeutic concentration but generally remained favorable at 87–99%. Based on these findings, AMX was efficiently recovered by ultrafiltration, regardless of whether mouse or human plasma was used. Furthermore, PBR could be measured with good reproducibility.

The PBR and free AMX concentration measurements in mouse and human plasma using ultrafiltration are shown in Fig. 2. The PBR of AMX in mouse plasma showed a slight decrease proportional to the AMX concentration (Fig. 2A). Free AMX concentrations increased by 6.8-fold and 17-fold at the toxic concentration (9 µM) and the lethal concentration (15 µM), respectively, compared to that at the therapeutic concentration (1.5 µM). This exceeded the concentration increase ratio and exhibited nonlinearity particularly at high concentrations (Fig. 2B). In contrast, the PBR of AMX in human plasma showed no substantial change, even when the AMX concentration was increased (Fig. 2C). Free AMX concentrations increased in a concentration-dependent manner at the toxic concentration (9 µM) and lethal concentration (15 µM), reaching 6-fold and 11-fold increases, respectively, compared to that at the therapeutic concentration (1.5 µM), thereby demonstrating linearity (Fig. 2D).

Fig. 2

Concentration-dependent changes in PBR (A and C) and free concentrations (B and D) of AMX in mouse (A and B) and human (C and D) plasma measured using ultrafiltration. Prior to ultrafiltration, the filter device was pretreated with 0.05% Tween 20 to prevent adsorption. After ultrafiltration of plasma containing AMX at therapeutic (1.5 µM), toxic (9 µM), and lethal (15 µM) concentrations, the AMX in the filtrate and the residue was measured using LC-MS/MS. Values represent the mean ± SD (n = 3). PBR, protein binding ratio; AMX, amoxapine; LC-MS/MS, liquid chromatography–tandem mass spectrometry; SD, standard deviation.

Changes in PBR and free AMX concentrations in the mouse plasma after equilibrium dialysis are shown in Fig. 3A and B, respectively. The PBR of AMX in the mouse plasma decreased in a concentration-dependent manner, similar to the results obtained via ultrafiltration (Fig. 3A). Free concentrations also increased beyond the dose-increase ratio at toxic concentrations (5 µM, 10 µM) and lethal concentrations (15 µM, 30 µM) compared to those at the therapeutic concentration (1.5 µM), showing nonlinearity at 13-fold, 34-fold, 57-fold, and 122-fold increases, respectively (Fig. 3B). In contrast, the PBR of AMX in human plasma differed from that in mouse plasma, showing no considerable change even when increased beyond the lethal concentrations (15 µM and 60 µM, Fig. 3C). Free concentrations increased in a concentration-dependent manner at lethal concentrations (15 µM and 60 µM) compared to that at the therapeutic concentration (1.5 µM), by 11-fold and 44-fold, respectively, thereby demonstrating linearity (Fig. 3D).

Fig. 3

Concentration-dependent changes in PBR (A and C) and free concentrations (B and D) of AMX in mouse (A and B) and human (C and D) plasma measured using equilibrium dialysis. AMX was added to mouse plasma at therapeutic (1.5 µM), toxic (5 μM and 10 μM), and lethal (15 μM and 30 μM) concentrations, or to human plasma at therapeutic (1.5 μM) and lethal (15 μM and 60 μM) concentrations. After equilibrium dialysis, total and free AMX concentrations were measured using LC-MS/MS. Values represent the mean ± SD (n = 3). PBR, protein binding ratio; AMX, amoxapine; LC-MS/MS, liquid chromatography–tandem mass spectrometry; SD, standard deviation.

To conduct the co-administration experiment with CPZ, which is often observed in combination with AMX in overdose cases (National Research Institute of Police Science, 2019-2022), we first verified whether a combined effect of ASA and AZ, which was previously reported (Sweeney et al., 1986), could be confirmed in our experimental system using equilibrium dialysis. Changes in PBR and free concentration upon addition of therapeutic AZ concentration (35 µM) and toxic ASA concentration (2150 µM) to human plasma are shown in Fig. 4A and B, respectively. The addition of toxic concentrations of ASA reduced the PBR of AZ from 91.4% to 56.1% (Fig. 4A). Free concentration of AZ also increased approximately fivefold, from 3 µM to 15.4 µM after addition of ASA (Fig. 4B). These results were in good agreement with the previous report (Sweeney et al., 1986), confirming that the combined effect can be detected in our experimental system. We then proceeded to examine the interaction between AMX and CPZ in protein binding, given that CPZ is frequently co-administered in cases of AMX overdose (National Research Institute of Police Science, 2019-2022). In this experiment, we considered that protein binding of AMX at the therapeutic concentration was more susceptible to the influence of other agents than at the toxic concentration. Therefore, changes in PBR and free AMX concentrations were examined when therapeutic AMX concentrations (1.5 µM) and therapeutic (0.75 µM) or toxic (2.5 µM) CPZ concentrations were added to human plasma. The PBR of AMX consistently remained around 90% and showed no concentration-dependent decrease, even when CPZ concentration was increased and added at therapeutic (0.75 µM) and toxic (2.5 µM) levels (Fig. 4C). Furthermore, no CPZ concentration-dependent increase in free AMX concentration was observed (Fig. 4D).

Fig. 4

Drug interactions mediated by human plasma proteins at toxic concentrations. PBR (A) and free concentrations (B) of AZ (35 µM) were measured in the absence or presence of a toxic concentration of ASA (2150 µM) in human plasma using equilibrium dialysis. PBR (C) and free concentrations (D) of AMX (1.5 µM) were measured in the absence or presence of therapeutic (0.75 µM) or toxic (2.5 µM) concentration of CPZ in human plasma using equilibrium dialysis. Total and free AZ and AMX concentrations were measured using LC-MS/MS. Values represent the mean ± SD (n = 3). AZ, acetazolamide; PBR, protein binding ratio; ASA, aspirin; AMX, amoxapine; CPZ, chlorpromazine; LC-MS/MS, liquid chromatography–tandem mass spectrometry; SD, standard deviation.

DISCUSSION

Plasma protein binding is important when considering the tissue distribution of drugs following an overdose. However, few studies have observed nonlinear protein binding using drugs at toxic concentrations or investigated interactions with other drugs. This study focused on AMX, which exhibits CNS toxicity during poisoning, with the aim of elucidating the mechanism of toxicity manifestation mediated by plasma protein binding during drug overdose.

AMX has an extremely high PBR, and often exhibits high adsorption on plastic laboratory materials. Therefore, prior to this study, experimental systems were investigated to accurately determine the PBR. The following three primary methods are used to determine the PBR: ultrafiltration, equilibrium dialysis, and ultracentrifugation. Owing to throughput considerations and equipment availability, ultracentrifugation was excluded from the investigation. Ultrafiltration is a simple method that can be performed inexpensively; however, the total recovery rate drops substantially for highly adsorbable drugs. As expected, when the AMX PBR was examined using ultrafiltration, a pronounced decrease in the total recovery rate was observed. However, the addition of 0.05% Tween suppressed adsorption and substantially improved recovery (Fig. 1B and C). Therefore, ultrafiltration was performed using 0.05% Tween just before centrifugation in subsequent experiments. Ultrafiltration enables the observation of total recovery rates that are difficult to achieve with equilibrium dialysis. In contrast, equilibrium dialysis, which requires expensive equipment, has high throughput and low adsorption. Given that AMX, a highly adsorbent drug, was the target compound, ultrafiltration and equilibrium dialysis were performed simultaneously to compare and evaluate the results.

The PBR of AMX in mouse plasma was calculated using ultrafiltration and equilibrium dialysis. At the therapeutic concentration, both methods showed a high binding rate of approximately 98%, and the binding rate decreased as the AMX concentration increased (Fig. 2A and Fig. 3A). These results suggested that AMX exhibited saturated binding to mouse plasma proteins at concentrations exceeding the toxic range. Moreover, a nonlinear increase in free AMX concentration was observed, particularly at high concentrations (Fig. 2B and Fig. 3B). Therefore, the toxic concentration of AMX in mouse plasma induces saturation of plasma protein binding, leading to a nonlinear increase in free AMX concentration. These results indicate that the phenomenon previously demonstrated in vivo in mouse models, where brain AMX concentrations increased dramatically within the toxic range, can be partially explained by the nonlinear increase in free AMX concentrations in plasma. However, the nonlinear increase in the free AMX concentration was not substantial, and it is difficult to conclude that the saturation of plasma protein binding was the primary factor responsible for the abnormal increase in brain concentration. When the same experiment was performed using human plasma, the PBR was 92% and 90% with ultrafiltration and equilibrium dialysis, respectively, at the therapeutic concentration, which is in agreement with previously reported data (DRUGBANK 5.1.8). However, increasing the AMX concentration resulted in only a slight change in the PBR with both ultrafiltration (Fig. 2C) and equilibrium dialysis (Fig. 3C). Furthermore, free AMX concentration showed a consistent linear increase at the therapeutic and toxic concentrations (Fig. 2D, Fig. 3D). The observed differences between mice and humans may be attributed to species-specific variations in plasma proteins. Among plasma proteins, AMX exhibits a high binding affinity for AGP (DRUGBANK 5.1.8, Ferry et al., 1986). Increased AGP concentrations correlate with reduced brain transfer rates of tricyclic antidepressants (Holladay et al., 1996). Blood AGP concentrations are 0.4–1.0 mg/mL in humans, whereas they are lower in rats (0.1–0.2 mg/mL) and mice (0.2–0.4 mg/mL) (Israili and Dayton, 2001; Ceciliani and Pocacqua, 2007). Therefore, in mouse plasma, a low AGP concentration results in a low absolute number of available binding sites, rendering the saturation of AMX binding to AGP more readily achievable. Furthermore, the binding affinity of certain drugs to human AGP is markedly higher than that to mouse or rat AGP (Huang and Ung, 2013). Based on these findings, the interspecies differences in AMX binding observed in this study were presumed to result from variations in plasma protein concentrations and binding characteristics. Consequently, plasma protein binding is thought to be saturated more readily in mouse plasma than in human plasma.

The linear increase in free AMX concentration accompanying the elevated AMX concentrations observed in human plasma is known as linear protein binding. This is observed in drugs with a low PBR or those with a large number of available plasma protein-binding sites (Deitchman et al., 2018). However, this phenomenon is generally reported within the therapeutic range, and to the best of the authors’ knowledge, no studies have observed linear protein binding within the toxic range. The fact that AMX exhibits linear binding with human plasma proteins suggests that AMX possesses numerous binding sites on plasma proteins within the toxic range. In AMX poisoning, the binding to AGP may first become saturated. Unbound AMX may bind to ALB, which possesses abundant binding sites, with lower affinity. Consequently, even when the overall AMX concentration increased, the free AMX concentration was expected to increase linearly as a result of maintaining a high PBR. Given that PBR is lower in the plasma of humans than in mice, free drug concentrations were higher at the same plasma concentration (Fig. 3B and D). Consequently, the amount of drug transferred to the brain is high, even within the therapeutic range, and it is highly likely that an abnormal increase in brain drug concentration occurs with increasing doses. Table 2 shows cases of CNS toxicity due to AMX overdose. Refractory seizures commonly coexist with metabolic acidosis, and cases in which these complications developed were predominantly those involving large doses of 2000 mg or more. The binding characteristics between human plasma proteins and AMX in this study may partially explain the phenomenon reported in clinical settings, where serious CNS toxicity, such as refractory seizures, occurs in proportion to increased AMX intake.

Table 2. Cases of seizures caused by amoxapine overdose.

Understanding the changes in the PBR during intoxication requires consideration of the interactions caused by the displacement of protein-binding sites when two drugs are administered simultaneously. Table 3 shows cases in which adverse events occurred owing to interactions via the displacement of ALB binding. The most frequently reported drug was warfarin, and all cases reported its displacement on ALB. To date, no cases of AGP involvement have been reported. A common feature of these cases is that most drugs, including concomitant medications, exhibit high binding affinity, with PBR values exceeding 90%. However, all these previously reported interactions occurred in cases where therapeutic doses were administered. Sweeney et al. (1986) reported two cases in which patients receiving AZ developed symptoms of AZ toxicity, including confusion, difficulty walking, and incontinence, and fell into a coma after concomitant administration of ASA. To demonstrate the drug interaction observed in these cases, plasma samples containing therapeutic AZ (8.0 µg/mL, 35 µM) were prepared in vitro. When a toxic ASA concentration (386 µg/mL, 2150 µM) was added to these samples, the free fraction of AZ increased from 3.3% to 30% compared to that of samples without ASA using an ultrafiltration method (Sweeney et al., 1986). This experiment was replicated using equilibrium dialysis in the present study and the PBR of AZ similarly decreased from 91.4% to 56.1% and the free AZ concentration increases from 3 µM to 15.4 µM, thereby confirming the previous results (Fig. 4A and B). Therefore, it was confirmed that displacement of binding occurred even during intoxication when drugs with high PBRs were co-administered.

Table 3. Cases where interactions due to albumin (ALB) binding displacement were suspected with therapeutic doses.

In Japan, fatal AMX overdoses are caused by multiple medications (National Research Institute of Police Science, 2019-2022), and the effects of concomitant drugs should be considered. In these cases, although a toxicodynamic contribution from concomitant medications to the increase in toxicity is entirely plausible, the toxicokinetic involvement remains poorly understood. Accordingly, this study analyzed the PBR of varying concentrations of CPZ, which is often co-administered with AMX in cases of overdose, in the presence of therapeutic AMX concentrations. However, neither the therapeutic (0.75 µM) nor the toxic (2.5 µM) CPZ concentrations produced visible effects on PBR or free AMX concentrations (Fig. 4 C and D). CPZ binds well to AGP in a variant-nonselective manner; therefore, it should antagonize AMX at the AGP binding site (Jolliet-Riant et al., 1998). However, no displacement reaction of AMX binding to the plasma protein was observed with concomitant incubation, thereby indicating that the mechanism of increased toxicity with the combined drugs cannot be explained by plasma protein binding. One possible cause is alternative binding to other plasma proteins. Motoya et al. (2006) conducted experiments on plasma protein substitution by adding the anti-HIV drug nelfinavir, which binds to AGP and ALB in either AGP or ALB solutions, followed by the addition of ritonavir, which is highly specific to AGP, or salicylic acid, which is highly specific to ALB. This markedly increased the free concentration of nelfinavir in both solutions. However, when the same experiment was conducted using plasma, no pronounced increase in the free nelfinavir levels was observed. Even if one binding site is displaced in plasma, binding to other binding sites serves as a substitute. Similarly, in the case of AMX, CPZ inhibits AMX binding to AGP, but AMX alternatively binds to ALB, resulting in no apparent change in the PBR.

In conclusion, the experimental hypothesis that enhanced CNS toxicity during AMX overdose involves a nonlinear increase in the free AMX form due to plasma protein binding saturation was partially confirmed in mouse plasma, but not in humans. The enhancement of CNS toxicity by AMX in humans cannot be fully explained by the plasma protein-binding interactions examined in this study. Therefore, further investigation of multiple aspects, such as the presence of AMX-binding proteins within the brain, is necessary. Conducting the experiments performed in this study with AMX and other toxic substances will contribute to further insights into the interactions mediated by plasma protein binding.

Funding

No funding was provided for the work.

Conflict of interest

The authors declare that there is no conflict of interest.

Data availability statement

The data in this study are included in the article/supplementary materials. Contact the corresponding author directly to request the underlying data.

Author contribution statement

Conceptualization: Akifumi Okamoto, Satoshi Numazawa

Data curation: Akifumi Okamoto, Yoshitaka Yamazaki, Kenji Momo, Natsumi Hattori-Usami

Formal analysis: Akifumi Okamoto, Yoshitaka Yamazaki, Satoshi Numazawa

Funding acquisition: Asuka Kaizaki-Mitsumoto, Satoshi Numazawa

Investigation: Akifumi Okamoto, Yoshitaka Yamazaki, Asuka Kaizaki-Mitsumoto, Kenji Momo, Natsumi Hattori-Usami

Methodology: Yoshitaka Yamazaki, Asuka Kaizaki-Mitsumoto, Satoshi Numazawa, Natsumi Hattori-Usami

Project administration: Asuka Kaizaki-Mitsumoto, Satoshi Numazawa

Resources: Asuka Kaizaki-Mitsumoto, Satoshi Numazawa

Software: Not applicable

Supervision: Asuka Kaizaki-Mitsumoto, Satoshi Numazawa, Kenji Momo

Validation: Yoshitaka Yamazaki, Asuka Kaizaki-Mitsumoto

Visualization: Akifumi Okamoto, Asuka Kaizaki-Mitsumoto, Satoshi Numazawa

Writing – original draft: Akifumi Okamoto, Asuka Kaizaki-Mitsumoto, Satoshi Numazawa

Writing – review & editing: Akifumi Okamoto, Asuka Kaizaki-Mitsumoto, Satoshi Numazawa

Ethics approval and consent to participate

Not applicable

Patient consent for publication

Not applicable

REFERENCES

•  Abad-Santos,  F.,  Carcas,  A.J., F-Capitán, C. and Frias, J. (1995): Retroperitoneal haematoma in a patient treated with acenocoumarol, phenytoin and paroxetine. Clin. Lab. Haematol., 17, 195-197.
•  Alvarez,  P.A.,  Egozcue,  J.,  Sleiman,  J.,  Moretti,  L.,  Di Girolamo,  G. and  Keller,  G.A. (2012): Severe neutropenia in a renal transplant patient suggesting an interaction between mycophenolate and fenofibrate. Curr. Drug Saf., 7, 24-29.
•  Ansong,  S.,  Doad,  J.,  Igweonu-Nwakile,  E.O. and  Okafor,  C. (2025): Drug-Drug Interaction of Warfarin and Doxycycline Leading to a Large Rectus Sheath Hematoma. Cureus, 17, e81312.
•  Anton,  R.F. Jr. and  Burch,  E.A. Jr. (1993): Response of psychotic depression subtypes to pharmacotherapy. J. Affect. Disord., 28, 125-131.
•  Bird,  J. and  Carmona,  C. (2008): Probable interaction between warfarin and torsemide. Ann. Pharmacother., 42, 1893-1898.
•  Bock,  J.L.,  Cummings,  K.C. and  Jatlow,  P.I. (1982): Amoxapine overdose: a case report. Am. J. Psychiatry, 139, 1619-1620.
•  Bosse,  G.M. and  Matyunas,  N.J. (1999): Delayed toxidromes. J. Emerg. Med., 17, 679-690.
•  Browne,  J.L.,  Tsuang,  M.T. and  Perry,  P.J. (1982): Amoxapine neurotoxicity: a case report with long-term follow-up. Drug Intell. Clin. Pharm., 16, 404-407.
•  Castanares-Zapatero,  D.,  Gillard,  N.,  Capron,  A.,  Haufroid,  V. and  Hantson,  P. (2016): Reversible cardiac dysfunction after venlafaxine overdose and possible influence of genotype and metabolism. Forensic Sci. Int., 266, e48-e51.
•  Ceciliani,  F. and  Pocacqua,  V. (2007): The acute phase protein alpha1-acid glycoprotein: a model for altered glycosylation during diseases. Curr. Protein Pept. Sci., 8, 91-108.
•  Cohen,  B.M.,  Harris,  P.Q.,  Altesman,  R.I. and  Cole,  J.O. (1982): Amoxapine: neuroleptic as well as antidepressant? Am. J. Psychiatry, 139, 1165-1167.
•  Deitchman,  A.N.,  Singh,  R.S. and  Derendorf,  H. (2018): Nonlinear Protein Binding: Not What You Think. J. Pharm. Sci., 107, 1754-1760.
•  de la Torre,  R.,  Farré,  M.,  Ortuño,  J.,  Mas,  M.,  Brenneisen,  R.,  Roset,  P.N.,  Segura,  J. and  Camí,  J. (2000): Non-linear pharmacokinetics of MDMA (‘ecstasy’) in humans. Br. J. Clin. Pharmacol., 49, 104-109.
• DRUGBANK 5.1.8: https://go.drugbank.com/
•  Elias,  J.H.,  Olney,  L.W.,  Frederick,  P.M.,  Yokote,  M. and  Herur,  J. (2024): Supratherapeutic International Normalized Ratio in a Patient on Warfarin Following Initiation of Brexpiprazole Pharmacotherapy: A Case Report. Cureus, 16, e70457.
•  Ferry,  D.G.,  Caplan,  N.B. and  Cubeddu,  L.X. (1986): Interaction between antidepressants and alpha 1-adrenergic receptor antagonists on the binding to alpha 1-acid glycoprotein. J. Pharm. Sci., 75, 146-149.
•  Fonck,  K.,  Haenebalcke,  C.,  Hemeryck,  A.,  Belpaire,  F.,  Jordaens,  L.,  Calle,  P. and  Buylaert,  W. (1998): ECG changes and plasma concentrations of propafenone and its metabolites in a case of severe poisoning. J. Toxicol. Clin. Toxicol., 36, 247-251.
•  Goldberg,  M.J. and  Spector,  R. (1982): Amoxapine overdose: report of two patients with severe neurologic damage. Ann. Intern. Med., 96, 463-464.
•  Hasan,  S.A. (2007): Interaction of doxycycline and warfarin: an enhanced anticoagulant effect. Cornea, 26, 742-743.
•  Hashimoto,  T.,  Kaneda,  Y.,  Ozaki,  A.,  Hori,  A. and  Tsuchiya,  T. (2022): Eleven-Year Trend of Drug and Chemical Substance Overdose at a Local Emergency Hospital in Japan. Cureus, 14, e32475.
•  Holladay,  J.W.,  Dewey,  M.J. and  Yoo,  S.D. (1996): Steady-state kinetics of imipramine in transgenic mice with elevated serum AAG levels. Pharm. Res., 13, 1313-1316.
•  Huang,  Z. and  Ung,  T. (2013): Effect of alpha-1-acid glycoprotein binding on pharmacokinetics and pharmacodynamics. Curr. Drug Metab., 14, 226-238.
•  Ide,  A. (2006): Case of amoxapine-induced refractory status epilepticus which was successfully treated with propofol. Chudoku Kenkyu, 19, 407-408.
•  Inoue,  Y.,  Kaizaki-Mitsumoto,  A. and  Numazawa,  S. (2022): Toxicokinetic evaluation during intoxication of psychotropic drugs using brain microdialysis in mice. J. Toxicol. Sci., 47, 99-108.
•  Israili,  Z.H. and  Dayton,  P.G. (2001): Human alpha-1-glycoprotein and its interactions with drugs. Drug Metab. Rev., 33, 161-235.
•  Jolliet-Riant,  P.,  Boukef,  M.F.,  Duché,  J.C.,  Simon,  N. and  Tillement,  J.P. (1998): The genetic variant A of human alpha 1-acid glycoprotein limits the blood to brain transfer of drugs it binds. Life Sci., 62, PL219-PL226.
•  Kulig,  K.,  Rumack,  B.H.,  Sullivan,  J.B. Jr.,  Brandt,  H.,  Spyker,  D.A.,  Duffy,  J.P. and  Shipe,  J.R. (1982): Amoxapine overdose. Coma and seizures without cardiotoxic effects. JAMA, 248, 1092-1094.
•  Litovitz,  T.L. and  Troutman,  W.G. (1983): Amoxapine overdose. Seizures and fatalities. JAMA, 250, 1069-1071.
•  Lopez,  L.M. and  Russell,  W.L. (1983): Amoxapine overdose: case report and pharmacokinetic profile. J. Toxicol. Clin. Toxicol., 20, 101-105.
•  Mancias,  P.,  Kramer,  L. and  Butler,  I.J. (1995): Amoxapine overdose in a young man: a transient mitochondrial abnormality? Pharmacotherapy, 15, 528-532.
•  Merigian,  K.S.,  Browning,  R.G. and  Leeper,  K.V. (1995): Successful treatment of amoxapine-induced refractory status epilepticus with propofol (diprivan). Acad. Emerg. Med., 2, 128-133.
•  Miles,  M.V.,  Greenwood,  R.S. and  Hussey,  B. (1990): Diagnostic pitfalls associated with amoxapine overdose: a case report. Am. J. Emerg. Med., 8, 335-337.
•  Miller,  M.T.,  Sleigh,  J.,  Rawlinson,  F.,  de Jongh,  R. and  Godwin,  Y. (1990): Amoxapine overdose: recovery after severe metabolic acidosis (pH 6.69) and status epilepticus. Anaesth. Intensive Care, 18, 246-248.
•  Motoya,  T.,  Thevanayagam,  L.N.,  Blaschke,  T.F.,  Au,  S.,  Stone,  J.A.,  Jayewardene,  A.L.,  Chi,  J. and  Aweeka,  F.T. (2006): Characterization of nelfinavir binding to plasma proteins and the lack of drug displacement interactions. HIV Med., 7, 122-128.
• National Research Institute of Police Science. (2019-2022): Annual case reports of drug and toxic poisoning in Japan, nos. 61-64 (in japanese). National Police Agency, Kashiwa.
•  Nosko,  M.G.,  McLean,  D.R. and  Chin,  W.D. (1988): Loss of brainstem and pupillary reflexes in amoxapine overdose: a case report. J. Toxicol. Clin. Toxicol., 26, 117-122.
•  Proudfoot,  A.T.,  Krenzelok,  E.P. and  Vale,  J.A. (2004): Position Paper on urine alkalinization. J. Toxicol. Clin. Toxicol., 42, 1-26.
• Rodrigues-Silva, N., Venâncio, Â. and Bouça, J. (2013): Risperidone, a risk factor for valproate-induced encephalopathy? Gen. Hosp. Psychiatry., 35, 452. e5-6.
•  Rogol,  A.D.,  Schoumacher,  R. and  Spyker,  D.A. (1984): Generalized convulsions as the presenting sign of amoxapine intoxication. Clin. Pediatr. (Phila.), 23, 235-237.
•  Ross,  S.R. and  Rodgers,  S.R. (1981): Physostigmine in amoxapine overdose. Am. J. Hosp. Pharm., 38, 1121-1122.
•  Schulz,  M.,  Schmoldt,  A.,  Andresen-Streichert,  H. and  Iwersen-Bergmann,  S. (2020): Revisited: therapeutic and toxic blood concentrations of more than 1100 drugs and other xenobiotics. Crit. Care, 24, 195.
•  Smith,  C.A. (1982): Amoxapine overdose. N. Z. Med. J., 95, 827.
•  Sue,  Y.J. and  Shannon,  M. (1992): Pharmacokinetics of drugs in overdose. Clin. Pharmacokinet., 23, 93-105.
•  Sweeney,  K.R.,  Chapron,  D.J.,  Brandt,  J.L.,  Gomolin,  I.H.,  Feig,  P.U. and  Kramer,  P.A. (1986): Toxic interaction between acetazolamide and salicylate: case reports and a pharmacokinetic explanation. Clin. Pharmacol. Ther., 40, 518-524.
•  Tasset,  J.J. and  Pesce,  A.J. (1984): Amoxapine in human overdose. J. Anal. Toxicol., 8, 124-128.
•  Tilstone,  W.J.,  Gray,  J.M.,  Nimmo-Smith,  R.H. and  Lawson,  D.H. (1977): Interaction between warfarin and sulphamethoxazole. Postgrad. Med. J., 53, 388-390.
•  Wallenburg,  E.,  Klok,  B.,  de Jong,  K.,  de Maat,  M.,  van Erp,  N.,  Stalpers-Konijnenburg,  S.,  Essink,  G. and  van Luin,  M. (2017): Monitoring Protein-Unbound Valproic Acid Serum Concentrations in Clinical Practice. Ther. Drug Monit., 39, 269-272.
•  Winek,  C.L.,  Wahba,  W.W. and  Rozin,  L. (1984): Amoxapine fatalities: three case studies. Forensic Sci. Int., 26, 33-38.