2026 Volume 51 Issue 8 Pages 413-426
Aconite contains four highly toxic diester-diterpene alkaloids (DDAs): aconitine, mesaconitine, hypaconitine, and jesaconitine. The efficacy of charcoal hemoperfusion (CHP) in treating aconite poisoning has recently been reported in several East Asian cases. This study evaluated the potential efficacy of CHP using two simulation experiments and one clinical case. One simulation examined the saturated adsorption capacity of activated charcoal for aconitine, whereas the other assessed the time course of perfusate aconitine concentrations at the inlet and outlet of an activated-charcoal column. The clinical case involved a 61-year-old female with stage V chronic kidney disease who underwent CHP for an electrical storm due to aconite poisoning. Her initial plasma DDA concentration was markedly higher than those reported in previous cases. This elevated concentration raises the possibility of substantial ingestion, although the exact amount cannot be determined. CHP was initiated 2 hr after ingestion and continued for 7 hr. In the simulations, activated charcoal showed a saturated adsorption capacity of 110 mg of aconitine per gram, and the column removed most aconitine in a single pass. In the clinical case, plasma DDA concentrations rapidly decreased after 7 hr of CHP, with a clearance rate of approximately 70 mL/min at a blood-flow rate of 100 mL/min. The terminal elimination half-life of plasma DDAs was 109.5 hr, likely reflecting the patient’s underlying kidney disease. Taken together, these findings suggest that CHP may contribute to DDA elimination under severe renal impairment, although its broader clinical relevance remains uncertain.
Aconite contains four highly toxic diester-diterpene alkaloids (DDAs)—aconitine, mesaconitine, hypaconitine, and jesaconitine (Chan, 2009). These alkaloids have historically been used as poisons in many regions, including ancient Europe. They have also been used as traditional medicines in East Asia and India, where detoxified preparations were administered for analgesic, anti-inflammatory, and cardiovascular purposes (Chan et al., 2021). DDAs are potent cardiotoxins that impair sodium channel inactivation, yet their pharmacokinetics remain insufficiently characterized (Zhou et al., 2021). In this study, the term “DDAs” refers specifically to the four highly toxic aconitine-type DDAs (aconitine, mesaconitine, hypaconitine, and jesaconitine), excluding monoester-type derivatives and other non-aconitine-type diester alkaloids.
Aconite poisoning has been most commonly reported in East Asia and India due to improper pretreatment of herbal medicines, but cases have also been reported in other regions as the global use of herbal products has expanded. More than 100 species of aconite grow naturally in the Northern Hemisphere, and accidental ingestion mistaken for edible wild plants has been reported in several countries (Pullela et al., 2008).
Because aconitine has traditionally been regarded as highly lipophilic, extracorporeal treatments (ECTRs) have generally been considered ineffective (Tai et al., 1992). Given the high mortality rate of aconite poisoning (15%) (Coulson et al., 2017), identifying effective strategies to remove DDAs from the body remains an important clinical challenge. We investigated the potential efficacy of charcoal hemoperfusion (CHP) for aconite poisoning through two simulation experiments and a clinical case involving a patient with stage V chronic kidney disease (CKD)—a condition that can markedly reduce renal elimination of DDAs—who developed aconite poisoning (Oshima et al., 2024). In this case, plasma concentrations of DDAs were also measured and monitored, including during the CHP session, for over one week, providing additional information on DDA kinetics.
Two simulation experiments were conducted with an activated charcoal column (DHP-1, Kuraray, Osaka, Japan; Table 1). One experiment was conducted to investigate the saturated adsorption capacity of the activated charcoal for aconitine (Nakashima et al., 1979; Saini et al., 2023), while the other examined the time course of perfusate aconitine concentrations at the DHP-1 inlet and outlet in a CHP model (Fig. 1) (Carvallo et al., 1976; Nakashima et al., 1979). Aconitine was purchased from Sigma Chemical Co. (St. Louis, MO, USA). Because aconitine is poorly soluble in normal saline and stable in acidic solutions, 0.2 N HCl was used as a solvent (Dybing et al., 1951). The aconitine concentration was measured spectrophotometrically at 239 nm, with a lower limit of quantification (LLOQ) of 0.1 μg/mL (Karawya et al., 1976).
| Adsorbent | Petroleum bead-activated charcoal |
|---|---|
| Coating agent | Hydroxyethyl methacrylate polymer |
| Column material | Polypropylene |
| Column shape | Cylindrical, 55 mm (diameter) × 180 mm (length) |
| Column volume | 200 mL |
| Column weight | 480 g |
| Quantity of activated charcoal loaded | 100 g (dry weight) |
| Blood priming volume | 70 mL |
| Sterilization | Steam autoclaving (121°C, 20 min) |
Schematic illustration of a simulated hemoperfusion model.
Twenty milligrams (dry weight) of activated charcoal was suspended in 1 mL of 0.2 N HCl and designated as Solution A. Eight solutions were prepared and degassed for 1 min each using an aspirator. For Solution B, aconitine was dissolved in 10 mL of 0.2 N HCl to prepare eight concentrations: 10, 25, 50, 100, 250, 500, 900, and 1,800 μg/mL. Solution A was added to each Solution B concentration to create eight samples, each containing activated charcoal and a distinct aconitine concentration. The initial aconitine concentration was defined as the concentration in the sample immediately after adding Solution A to Solution B. The solvent volume of the sample was 11 mL, and the initial aconitine concentration was calculated as follows:
initial aconitine concentration = aconitine concentration of Solution B × 10/11 (μg/mL)
The samples were incubated in a rotary shaker at 37°C and 120 rpm for 120 min, after which the activated charcoal was filtered from each sample, and the final aconitine concentration in the filtrate was measured. Activated charcoal settles in the aconitine solution, so we used a shaker to maintain a uniform suspension and prevent rotor-related damage to the charcoal. A 120-min contact time was chosen to approximate the typical 3–4 hr of clinical use for a single DHP-1 column. As the weight of activated charcoal suspended in Solution A was 20 mg, the amount of adsorbed aconitine per mg of charcoal was calculated using the following equation:
amount of adsorbed aconitine per mg of charcoal = (initial aconitine concentration – final aconitine concentration) × 11/20 (μg/mg)
In a preliminary control experiment, 1 mL of 0.2 N HCl without activated charcoal was added to each of the eight concentrations of Solution B. The control samples were shaken under the same conditions at 37°C for 120 min, and the final aconitine concentrations were measured.
CHP model analysisA 3,000 mL aconitine solution (20 μg/mL) was prepared in 0.2 N HCl. A CHP model (Fig. 1) was constructed by incorporating the DHP-1 and 3,000 mL reservoir into an extracorporeal blood circuit (JMS Co., Tokyo, Japan). In this model, the aconitine solution was circulated at a circuit flow rate of 100 mL/min for 180 min. Perfusate samples were collected from the DHP-1 inlet and outlet at 0, 20, 40, 60, 90, 120, and 180 min, and aconitine concentrations were measured. Before perfusion, the DHP-1 was primed with 1,000 mL of 0.2 N HCl. The DHP-1 column and extracorporeal circuit, filled with this solution, were flushed with the prepared aconitine solution under non-circulating flow to replace their contents. The reservoir was maintained at 37°C throughout the experiment. This experiment was repeated three times, each using a fresh extracorporeal circuit, DHP-1 column, reservoir, and aconitine solution. Clinically, aconitine exhibits toxicity at plasma concentrations in the ng/mL range (Mizugaki et al., 1998; Terui et al., 2008). However, the initial aconitine concentration in the perfusate was set at 20 µg/mL in this CHP model analysis because spectrophotometry was used for quantification, and the LLOQ of this method was 0.1 µg/mL.
Clinical case study of CHPThis study was approved by the Ethics Committee of Tottori University, Faculty of Medicine (approval number: 24A099). This case has been previously reported by Oshima et al. (Oshima et al., 2024). A 61-year-old female, who had been previously diagnosed with stage V CKD but had never been on dialysis, swallowed the fresh root of an aconite plant. Her estimated glomerular filtration rate on admission was 9.2 mL/min/1.73 m2. Prior to this emergency admission, her daily urine output was adequate, and no symptoms of renal failure, such as fluid retention or electrolyte imbalance, were observed. She developed an electrical storm (ES) 50 min after aconite ingestion. While the ES could be managed with intravenous boluses of 2 g magnesium sulfate (MgSO4), the underlying rhythm remained a sustained supraventricular tachycardia (SVT). Before recurrent pulseless ventricular tachycardia (VT) became sustained, 2 g of MgSO4 was administered intravenously over 5–30 min. Initially, she required MgSO4 every 10 min, but the interval became longer over time. Given this clinical course, CHP using a DHP-1 column was initiated 2 hr after ingestion. The right femoral vein was cannulated with an 11.5 Fr two-lumen catheter (Quinton-Mahurkar catheter, Quinton Instrument Company, Seattle, WA, USA) for vascular access. Additionally, the same perfusion system used in the CHP model analysis was employed in this case. CHP was administered at a circuit flow rate of 100 mL/min, with the first column used for 3 hr, followed by a 30-min pause, and the second column used for 4 hr. She returned to sinus rhythm 15 hr after ingestion. In total, 12 g of MgSO4 was administered intravenously during her 5-hr stay in the emergency department. The highest serum magnesium concentration in the emergency room was 4.8 mmol/L. Due to her CKD, hypermagnesemia persisted even on the next day and lasted for a week.
Plasma samples were collected at multiple time points during hospitalization, including during the CHP session, and stored at −80°C. Concentrations of the four DDAs were quantified by gas chromatography with selected ion monitoring (GC/SIM) (Mizugaki et al., 1988). Analytical reference standards of aconitine, mesaconitine, hypaconitine, and jesaconitine were kindly provided by Tsumura Co. (Tokyo, Japan). For analysis, 0.5 mL of plasma was acidified and extracted with ethanol, followed by liquid–liquid extraction and derivatization with BSTFA. The derivatized extracts were analyzed by GC/SIM (JEOL DX-303, DB-5 column) under standard GC–MS conditions (helium carrier gas, 70 eV). Ions at m/z 698, 684, 596, and 728 were monitored for aconitine, mesaconitine, hypaconitine, and jesaconitine, respectively. Quantification was performed using the external standard method. Calibration curves were linear over 30 pg–1 ng, with an LLOQ of 0.6 ng/mL.
Time course of the plasma concentrations of the four DDAsThe time course of the plasma concentrations of the four DDAs was monitored from admission for more than one week, including during the CHP session, until their levels fell below the LLOQ.
DDA extraction ratios of the DHP-1 columnPlasma samples were collected at the inlet and outlet of the column at three time points during the CHP session: 2 hr after the initiation of CHP for the first column and 3 and 4 hr for the second column. The plasma concentrations of the four DDAs were measured. The extraction ratios of each DDA in the DHP-1 column were calculated using the following equation:
extraction ratio = (Concinlet – Concoutlet)/Concinlet
To estimate the saturated adsorption capacity, the amount of aconitine adsorbed per mg of activated charcoal was plotted as a function of the final aconitine concentration (Fig. 2). The adsorbed amount of aconitine increased with increasing aconitine concentration and plateaued at approximately 110 μg/mg (Fig. 2). In the control samples prepared without activated charcoal, aconitine remained stable in 0.2 N HCl for 120 min and showed no adsorption onto the test tubes or other materials. Specifically, the amount of adsorbed aconitine at a final concentration of 1,440 μg/mL was 107.8 μg/mg. Thus, under strongly acidic conditions without competing adsorbates, one DHP-1 column packed with 100 g of activated charcoal is estimated to adsorb approximately 11 g of aconitine (Fig. 2).
Adsorption capacity of activated charcoal for aconitine in a simulation experiment. The amount of aconitine adsorbed per mg of activated charcoal is plotted as a function of the final aconitine concentration in each sample. The adsorbed amount increased with increasing final concentration and plateaued at approximately 110 μg/mg.
Figure 3 illustrates the time course of aconitine concentrations in the perfusate from the DHP-1 inlet and outlet. Before charcoal perfusion, the mean aconitine concentration at the inlet was 19.2 ± 0.3 μg/mL. The inlet concentration decreased rapidly after initiating charcoal perfusion and dropped below the LLOQ after 180 min. In particular, the inlet concentration fell to 50% of its initial value within 20 min. Even at high inlet concentrations, the outlet concentration consistently remained below the LLOQ. Therefore, under non-physiological conditions—specifically, a strongly acidic, protein-free perfusate lacking competing adsorbates—CHP can be considered as adsorbing most of the aconitine during a single pass through the column.
Adsorption efficacy of aconitine in a simulated hemoperfusion model (n = 3). Black circles and white squares indicate the perfusate aconitine concentrations at the inlet and outlet, respectively. Even when inlet aconitine concentrations were high, outlet concentrations remained consistently below the lower limit of quantification.
Jesaconitine was not detected throughout the study. Notably, the first sample was taken 1.5 hr after ingestion, which was 30 min before initiating CHP. The highest total DDA concentration was 37.9 ng/mL in the first sample collection. After 7 hr of CHP, the total DDA concentration rapidly decreased to 8.4 ng/mL. The total DDA concentration remained elevated at 5.5 ng/mL at 20 hr after ingestion. Thereafter, the concentrations of the three DDAs gradually declined but remained above the LLOQ until day 6 after ingestion. Fig. 4 illustrates the logarithm (base 10) of the total DDA concentration in plasma relative to the time after ingestion, using a semi-logarithmic graph. The total DDA concentration rapidly decreased up to 20 hr after ingestion (rapid phase), followed by a slow linear decline to below the LLOQ 7 days after ingestion (terminal phase). Notably, the patient had returned to sinus rhythm during this rapid phase (at 15 hr after ingestion). The terminal phase exhibited a clear log-linear decline, allowing reliable estimation of the plasma elimination rate constant (Kel) and elimination half-life. Importantly, the slope of the regression line in the terminal phase was used to calculate the plasma elimination rate constant (Kel) and the plasma elimination half-life using the following formula:
Slope = –Kel/ln(10)
elimination half-life = ln(2)/Kel
In the terminal phase, the plasma Kel was 0.0063/hr, and the plasma elimination half-life was 109.5 hr.
Semi-logarithmic plot of the total plasma concentrations of the three diester-diterpene alkaloids (DDAs) versus time after ingestion. The plot shows a rapid phase in which concentrations declined steeply up to approximately 20 hr after ingestion, followed by a terminal phase characterized by a slow, linear decline. Seven concentration–time points in the terminal phase were used to derive the regression equation (y = −0.0027x + 0.8384, R2 = 0.73), and the straight line depicts this regression. Charcoal hemoperfusion was performed from 2 to 9.5 hr after ingestion, overlapping with the rapid phase of steep concentration decline.
Table 2 and Fig. 5 present the same inlet and outlet plasma samples obtained during CHP, with Table 2 showing the extraction ratios and Fig. 5 displaying the corresponding raw concentrations. The 2, 3, and 4-hr extraction ratios of the DHP-1 column were 63.4%, 69.8%, and 35.9% for aconitine; 61.5%, 57.2%, and 49.9% for mesaconitine; and 81.5%, 82.9%, and 72.2% for hypaconitine, respectively (Table 2). Additionally, the extraction ratios of the three DDAs were approximately 70% (with a clearance rate of 70 mL/min at a flow rate of 100 mL/min) but slightly decreased 4 hr after perfusion.
| Alkaloid | 1C-2 hr | 2C-3 hr | 2C-4 hr |
|---|---|---|---|
| Aconitine | 63.4 | 69.8 | 35.9 |
| Mesaconitine | 61.5 | 57.2 | 49.9 |
| Hypaconitine | 81.5 | 82.9 | 72.2 |
Charcoal hemoperfusion was performed for 3 hr with the first column (1C), followed by 4 hr with the second column (2C). Extraction ratios were calculated at 2 hr during 1C, and at 3 hr and 4 hr during 2C. Ratios are expressed as percentages.
Plasma concentrations of the three diester-diterpene alkaloids (DDAs) at the inlet and outlet of the DHP-1 column during charcoal hemoperfusion (CHP). Plasma samples were obtained at three time points during CHP: 2 hr after initiation of the first column (4 hr after ingestion) and 3 and 4 hr after initiation of the second column (8.5 and 9.5 hr after ingestion). Inlet concentrations at 4 and 8.5 hr showed only modest changes, consistent with the plateau observed in venous plasma during this period, whereas all three DDAs exhibited a clear decrease at 9.5 hr. Outlet concentrations were consistently lower than inlet concentrations, reflecting ongoing extraction by the DHP-1 column. The 1.5-hr sample was collected before CHP initiation; therefore, no outlet sample was available at that time.
Plasma samples were collected at the column inlet and outlet at three time points during CHP. The inlet samples taken 2 hr after initiating CHP (4 hr after ingestion) and 3 hr after initiating CHP (8.5 hr after ingestion) showed only a modest difference in the plasma concentrations of the three DDAs. This finding is consistent with the minimal change in venous plasma concentrations between 4 and 8.5 hr after ingestion. In contrast, during the final hour of CHP (corresponding to 9.5 hr after ingestion), the plasma concentrations of all three DDAs clearly decreased (Fig. 5).
Although aconitine has long been used in both medicinal and non-medicinal settings, the clinical toxicology of its DDAs remains incompletely understood. Clinical manifestations may occur at very small doses, with reports describing toxicity from as little as 0.2 mg of the parent alkaloid (Gao et al., 2022). Following large ingestions, life-threatening ventricular tachyarrhythmias (VTAs) can develop within hours and may rapidly become fatal (Lawson et al., 2025). These VTAs are frequently refractory to antiarrhythmic drugs and electrical cardioversion/defibrillation, making them the leading cause of death in aconitine poisoning. This clinical profile underscores the need for therapeutic strategies that can promptly reduce circulating DDA levels early in the course of poisoning. Despite its low oral bioavailability (1.3–8.2%) (Tazawa et al., 2003; Tang et al., 2012), aconitine is rapidly absorbed and can exert lethal toxicity at very low doses, making it difficult to manage both as a therapeutic agent and as a poison.
Several case reports from East Asia have described improved clinical outcomes in patients with aconite poisoning treated with CHP (Lin et al., 2004). Given the approximately 15% mortality associated with aconite poisoning (Coulson et al., 2017), these observations highlight the need to identify effective strategies for removing DDAs from the body.
The role of ECTRs warrants reevaluation. The use of ECTR in aconitine poisoning has traditionally been considered limited, largely because aconitine has long been assumed to be highly lipophilic (Tai et al., 1992). However, an autopsy case demonstrated extremely low DDA concentrations in adipose tissue (Ito et al., 2000). The patient died approximately four hours after ingestion, and although distribution into low-blood-flow tissues such as adipose tissue may not have reached equilibrium by that time, the measured concentrations were strikingly low even with this possibility in mind. For example, while skeletal muscle (greater psoas muscle)—also a low-blood-flow tissue—contained 13.1 ng/g of jesaconitine, adipose tissue contained only 3.4 ng/g, a disparity too large to be explained solely by perfusion differences. For reference, jesaconitine concentrations in the liver were approximately 240 ng/g in the same case, indicating that the extremely low adipose concentration cannot be attributed to low overall exposure. This inconsistency between the historically assumed lipophilicity of aconitine and the very low in vivo adipose concentrations prompted us to reexamine the physicochemical properties of aconitine-type DDAs and to reconsider the potential effectiveness of ECTR.
Although primary experimental measurements of the logP and pKa values of aconitine-type DDAs are not available, existing estimates provide useful indicators of their lipophilicity. The reported logP values—0.3 for aconitine, −0.1 for mesaconitine, and 0.9 for hypaconitine (Wei et al., 2024)—are not experimentally determined, yet they indicate that even the non-ionized forms have very low intrinsic lipophilicity. Similarly, the commonly cited pKa values for these DDAs (8.55–8.70), although lacking confirmed primary measurements, are widely referenced in analytical chemistry and pharmacology reviews and suggest that aconitine-type DDAs exist almost entirely in their ionized forms at physiological pH. Taken together, these physicochemical properties imply that the logD under physiological conditions—and therefore the effective in vivo lipophilicity—is likely far lower than traditionally assumed. As a result, membrane permeability is expected to be limited, restricting intracellular entry, and aconitine-type DDAs are predicted to distribute predominantly within extracellular fluid (ECF) compartments.
The physicochemical properties of aconitine-type DDAs—specifically their low logD and high pKa—suggest that early drug disposition is dominated by distribution, with elimination contributing only secondarily, giving rise to a three-phase distribution–elimination model. This distribution-driven process is expected to proceed through three sequential phases—Phase 1 (initial distribution), Phase 2 (re-equilibration), and Phase 3 (late distribution–elimination)—as schematically illustrated in Fig. 6. In Phase 1, rapid movement from plasma into the ECF of high-blood-flow organs produces a steep early decline in plasma concentrations, during which renal elimination contributes only minimally. In Phase 2, the markedly uneven distribution established early gradually shifts toward equilibrium, with passive back-flux from the ECF of high-blood-flow organs to plasma persisting throughout this period. Renal elimination begins to contribute more meaningfully; however, the decline in plasma concentrations remains predominantly driven by distribution processes, particularly slow redistribution from plasma into the ECF of low-blood-flow tissues. A characteristic feature of Phase 2 is the continued slow decline in plasma concentrations despite ongoing back-flux, reflecting this redistribution-dominant state. In Phase 3, slow redistribution into the ECF of low-blood-flow tissues persists as back-flux from high-blood-flow organs diminishes, and the plasma decline becomes elimination-dominant, with renal elimination becoming the principal driver of further concentration reduction. Across all phases, aconitine-type DDAs are expected to remain largely confined to ECF compartments, with minimal intracellular or adipose accumulation.
Schematic representation of the proposed three-phase distribution–elimination model. Phase 1 (initial distribution phase) involves rapid movement of the drug from plasma into the extracellular fluid (ECF) of high-blood-flow organs, producing a steep plasma decline and an uneven early distribution. Renal elimination contributes only minimally during this period. Phase 2 (re-equilibration phase) is characterized by passive back-flux from the ECF of high-blood-flow organs to plasma and slow redistribution into the ECF of low-blood-flow tissues, resulting in a distribution-dominant slow plasma decline. Renal elimination becomes more influential in this phase; however, the decline remains predominantly driven by distribution processes, particularly slow redistribution into low-blood-flow tissues. Phase 3 (late distribution–elimination phase) reflects continued slow redistribution, diminishing back-flux from high-blood-flow organs, and a shift toward elimination-dominant plasma decline, with renal elimination becoming the principal driver of further concentration reduction. These phases occur sequentially: Phase 1 within the first several hours after ingestion, Phase 2 over the next several hours, and Phase 3 thereafter. These time ranges are approximate and represent typical rather than strict phase boundaries.
The physicochemical prediction that aconitine-type DDAs follow our proposed three-phase distribution–elimination model is supported by the mouse study conducted by Wada et al. (2005). In that study, aconitine rapidly distributed into high-blood-flow organs such as the liver, kidneys, and heart, and concentrations in these organs declined quickly thereafter. This temporal pattern is consistent with the early phases of our proposed model, in which aconitine first undergoes rapid distribution from plasma into the ECF of high-blood-flow organs, followed by slower passive back-flux from the ECF of high-blood-flow organs to plasma (Supplementary Fig. S1). These findings support the assumption that aconitine-type DDAs distribute predominantly within ECF compartments.
The reported intravenous volume of distribution (Vd) of aconitine in rats is based on only two studies, which yield markedly divergent values of 1.7 ± 0.4 L/kg (Tazawa et al., 2003) and 7.7 ± 1.1 L/kg (Chen et al., 2009). Autopsy findings showing that aconitine does not accumulate in adipose tissue (Ito et al., 2000), together with consistently observed features—low intrinsic lipophilicity, a predominantly ionized state, and rapid back-flux from the ECF of high-blood-flow organs—indicate that aconitine-type DDAs have a moderate to somewhat small Vd. The value of 1.7 L/kg aligns most closely with these physicochemical and distributional characteristics and therefore provides direct data-based support for a moderate to slightly reduced Vd. This value is also consistent with the presumed distribution process, in which rapid initial distribution into circulating blood and high-blood-flow organs is followed by slow, blood-flow-dependent redistribution into the ECF of low-blood-flow tissues. Furthermore, a Vd of 1.7 L/kg lies near the commonly cited 1.0–2.0 L/kg threshold used when evaluating the applicability of ECTRs (Ghannoum et al., 2014), indicating that a non-negligible intravascular fraction remains.
A closer examination of the study by Wada et al. (2005) further supports the interpretation that aconitine distributes predominantly within ECF compartments. In that study, mice were orally administered approximately half of the median lethal dose, and the highest measured concentrations of aconitine were observed at the earliest sampling point (15 min) in both plasma and high-blood-flow organs such as the liver and kidneys (measured in tissue homogenates). The observed maximum concentrations (Cmax) in these organs were approximately 100-fold higher than those in plasma. Moreover, aconitine concentrations in both plasma and high-blood-flow organs became undetectable within 24 hr. This temporal pattern indicates that the presence of aconitine in high-blood-flow organs is brief and transient, and is most reasonably interpreted as reflecting ECF-restricted distribution rather than intracellular accumulation.
When interpreting the organ concentration data reported by Wada et al. (2005), it is important to recognize that organ homogenates contain multiple compartments. These homogenates include ECF, intracellular fluid, residual blood, and interstitial fluid, and in the case of aconitine, the high measured concentrations are considered to reflect primarily the ECF fraction together with membrane-bound and tissue-bound forms. If substantial intracellular penetration or accumulation were present, such rapid disappearance from organs would not occur. Direct comparisons between organ homogenate and plasma concentrations therefore have methodological limitations. Accordingly, the observed organ concentrations are most appropriately interpreted as reflecting aconitine distributed predominantly within ECF compartments. Because drugs that accumulate intracellularly typically decline much more slowly, the rapid disappearance observed in Wada et al.’s study is most consistent with ECF-restricted distribution.
In the study by Wada et al. (2005), the first plasma and organ samples were both obtained at 15 min; however, the true time to maximum concentration was very likely to have occurred earlier in plasma, and the true plasma Cmax was probably higher than the value measured at the 15-min sampling point. Consequently, the organ-to-plasma ratio calculated at 15 min (approximately 100-fold) is likely an overestimation of the true Cmax ratio. This overestimation reflects not only the sampling-time mismatch but also the characteristic distribution of aconitine, which localizes predominantly within the ECF of high-blood-flow organs. Although the detailed mechanisms underlying this phenomenon fall outside the scope of the present discussion, it is clear that the appearance of high organ concentrations does not, in itself, imply intracellular accumulation.
As outlined in our proposed three-phase distribution–elimination model, aconitine-type DDAs are predicted to undergo rapid initial distribution from plasma into the ECF of high-blood-flow organs, followed by slow redistribution into the large ECF volumes of low-blood-flow tissues together with passive return from high-blood-flow organs to plasma. This distribution pattern is also reflected in the delayed peak of urinary excretion. For typical ECF-restricted drugs (Vd approximately 0.2–0.3 L/kg), distribution is largely confined to the small ECF volumes of high-blood-flow organs; consequently, back-flux to plasma occurs rapidly, and urinary excretion usually peaks within 1–3 hr. In contrast, aconitine-type DDAs are predicted to distribute into the substantially larger ECF volumes of low-blood-flow tissues, resulting in delayed back-flux due to a capacity effect, with further delay arising from a flow-limited effect caused by slow blood flow in these tissues. Consistent with this prediction, human case reports describe markedly delayed urinary excretion peaks at approximately 15 hr or even 20–35 hr (Yoshioka et al., 1996; Terui et al., 2008), supporting the interpretation that aconitine-type DDAs remain within the ECF of low-blood-flow tissues for prolonged periods.
Understanding the metabolic pathways of DDAs is important; however, their in vivo fate remains only partially elucidated. Decoction hydrolyzes the ester bonds of DDAs, converting them into the less toxic monoester-type diterpenoid alkaloids (MDAs) (Zhou et al., 2021). These MDAs exhibit markedly lower cardiotoxicity and overall toxicity than their parent DDAs, reflecting the loss of one ester moiety essential for high potency. In vivo, DDAs also undergo hydrolysis by carboxylesterases (CE), producing MDAs such as benzoylaconine (BAC) from aconitine. Numerous oxidative pathways mediated by cytochrome P450 enzymes have been reported, although the extent to which they contribute in vivo during acute poisoning remains unclear (Zhang et al., 2015). The toxicological properties of P450-derived metabolites are also unknown, and data on their renal elimination are virtually nonexistent. Autopsy studies have demonstrated extremely low MDA/DDA ratios in plasma, urine, and liver (Niitsu et al., 2013), and mouse data indicate that BAC formation after acute dosing is minimal, with aconitine persisting substantially longer and reaching much higher concentrations than CE-derived metabolites in all measured tissues, including liver, kidneys, heart, and plasma (Wada et al., 2005). Taken together, these findings suggest that hepatic metabolism is unlikely to play a major role in DDA clearance during acute poisoning, and that renal excretion is the most plausible primary elimination pathway. However, its quantitative contribution has not yet been fully defined, and a contribution from P450-mediated metabolism cannot be excluded. These findings underscore the need for clinical confirmation.
The plasma concentration–time profile of DDAs in this case exhibited a biphasic pattern consisting of a rapid phase followed by a terminal phase characterized by prolonged persistence. The first total DDA concentration measured on day 1 was 37.9 ng/mL, clearly higher than values reported in previous cases; even if the first blood sample was obtained earlier than in many prior reports, this finding indicates substantial ingestion. However, the prolonged detectability of DDAs for six days cannot be explained by ingestion amount alone and is best attributed to reduced renal excretion associated with CKD. Whereas all three DDAs remained quantifiable in plasma for six days in this case, DDAs in patients without renal impairment typically become undetectable in plasma within 1-2 days (Mizugaki et al., 1998; Terui et al., 2008). Terui et al. reported that the urinary excretion rate of unchanged DDAs peaked 20–35 hr after ingestion and that increases in urinary excretion paralleled declines in plasma concentrations. Mizugaki et al. likewise reported urinary detection for up to six days, supporting the interpretation that renal excretion is the principal elimination pathway in acute poisoning. Taken together, these findings indicate that the final elimination of DDAs is predominantly renal.
To interpret the prolonged terminal phase observed in this case, it is essential to clarify the pharmacokinetic structure underlying the biphasic decline; thus, we first outline the characteristics of the initial distribution phase and the terminal phase. No clinical factors suggesting impaired hepatic metabolic capacity were identified, nor were there findings indicative of abnormally delayed redistribution from low-blood-flow tissues. Accordingly, the biphasic decline in plasma DDA concentrations can be attributed to the fundamental pharmacokinetic structure of these compounds—an initial distribution phase followed by a terminal phase predominantly governed by reduced renal elimination. During the initial distribution phase, DDAs rapidly distribute from plasma into the ECF of high-blood-flow organs, and the contribution of renal clearance is inherently minimal. In contrast, renal dysfunction markedly limits elimination during the terminal phase, resulting in the prolonged persistence of DDAs in the circulation.
To describe how CHP influenced the rapid phase, we outline the mechanisms shaping the concentration–time profile during this period (Fig. 5). The initial decline observed during the first 4 hr after ingestion (i.e., the first 2 hr of CHP) is driven predominantly by rapid distribution from plasma into the ECF of high-blood-flow organs, with CHP contributing only modestly to the overall rate of decline during this period. The subsequent plateau between 4 and 8.5 hr after ingestion can be interpreted as a transient balance between back-flux from the ECF of high-blood-flow organs to plasma and concurrent CHP-mediated extraction. The renewed decline after approximately 8.5 hr suggests that back-flux had diminished and that CHP-mediated extraction had become predominant. These temporal features are summarized schematically in Supplementary Figure S2. Moreover, in the absence of CHP, the true terminal phase would likely have begun at a higher concentration than the measured values suggest, indicating that the time required to reach the LLOQ would have been longer than observed.
The effectiveness of CHP in this case is best explained by the persistence of high circulating DDA concentrations and the initiation of treatment at an appropriately early time point. Because the ingested dose was extremely large, a substantial circulating fraction remained available for extracorporeal removal when CHP was started 2 hr after ingestion. CHP was continued for 7 hr, from 2 to 9.5 hr after ingestion—a period during which redistribution to low-blood-flow tissues would normally be the predominant contributor to the decline in plasma concentrations, with renal elimination contributing to a lesser extent, while back-flux from high-blood-flow organs would partially counteract these processes. In patients with preserved renal function, renal elimination would accelerate the decline during this period; however, in this case, markedly impaired renal function greatly limited elimination. As a result, despite ongoing redistribution to low-blood-flow tissues, a large circulating fraction of the drug persisted. Thus, CHP was clinically meaningful because it removed a substantial circulating DDA burden that persisted as a result of the extremely large ingested dose together with markedly impaired renal elimination.
Building on the patient-specific insights described above, these considerations can be generalized by examining how Vd-based pharmacokinetic assumptions influence the predicted time window during which CHP can meaningfully remove circulating drug. Static models estimate that only 2–3% of the total body burden of DDAs resides intravascularly, based on a plasma volume of 0.04 L/kg and a lower reported Vd of 1.7 L/kg. This underlies the conventional view that toxins with a Vd greater than 1–2 L/kg are poor candidates for ECTR. However, Vd is a terminal parameter derived after distribution has completed and plasma–tissue equilibrium has been reached; it does not represent the intravascular fraction during the initial distribution phase. As a fundamental pharmacokinetic principle, the initial distribution phase is universally the period in which plasma concentrations—and therefore the accessible intravascular fraction—are at their absolute highest across the entire concentration–time profile. Plasma concentrations decline more slowly during the subsequent re-equilibration phase, as back-flux from high-blood-flow organs persists and redistribution to low-blood-flow tissues proceeds gradually. Renal elimination becomes increasingly relevant only after the re-equilibration phase has progressed, and its contribution is substantially delayed in patients with impaired renal function. These considerations support the theoretical rationale for CHP not only during the initial distribution phase but also during the re-equilibration phase, particularly in patients with reduced renal clearance. Even in patients without pre-existing renal disease, aconite-induced circulatory failure can acutely reduce renal perfusion and prolong the elimination half-life of DDAs, as suggested in a PCPS-treated case (Fujita et al., 2007).
The theoretical basis for CHP lies in the adsorption properties of activated charcoal and the molecular structure of DDAs. Activated charcoal efficiently adsorbs hydrophobic compounds through its hydrophobic pores and has high adsorption capacity for molecules in the 100–3000 Dalton (Da) range. DDAs (approximately 600–700 Da) fall within this range, and adsorption is further promoted by the aromatic rings within their diterpene skeletons. Although DDAs exist largely in ionized forms at physiological pH, activated charcoal can also adsorb ionized species through its acidic surface functional groups (Radovic et al., 2001).
In the rat study by Tang et al. (2012), the protein binding rate of aconitine was reported as 24–32%, corresponding to a free fraction of 68–76%. Because protein-bound aconitine forms complexes with plasma proteins—effectively increasing its apparent molecular size and preventing adsorption to activated charcoal—only the free fraction is adsorbable in vivo. Thus, the 68–76% free fraction represents the theoretical upper limit of aconitine adsorption in vivo. In our acidic, protein-free model, most aconitine was extracted in a single pass, supporting the ability of activated charcoal columns to adsorb the free fraction to a high degree. This theoretical upper limit was consistent with the extraction ratios of the three DDAs observed in this case (approximately 70%).
The predominant circulating DDA varies among poisoning cases, including reports in which jesaconitine is highest (Matsuo et al., 2023). In this case, the three C19 diester diterpenoid alkaloids showed broadly similar extracorporeal extraction behavior. Although our in vitro adsorption experiment was conducted using aconitine alone, this finding suggests that CHP may have therapeutic applicability beyond aconitine, although caution is required when extrapolating to cases with different predominant DDAs.
Aconitine-type DDAs bind to voltage-gated sodium channels from the extracellular side, inhibit inactivation, and thereby induce sustained depolarization leading to toxicity. The principal site of toxicity is the myocardium, particularly the myocardial interstitial ECF, which represents the relevant extracellular compartment. Toxicity is determined by the concentration of freely diffusible, non-membrane-bound molecules within this space, a key consideration when defining the therapeutic target of CHP. CHP can directly reduce only the protein-unbound fraction in plasma (step 1). Lowering this fraction induces back-flux from the ECF to plasma (step 2), which in turn decreases the ECF unbound concentration—the determinant of toxicity (step 3). As the ECF concentration declines, aconitine-type DDAs bound to sodium channels are expected to dissociate progressively from the membrane.
Toxicity is determined by the concentration within the myocardial interstitial ECF, and the clinical course therefore depends on how this concentration changes over time. In the mouse study by Wada et al. (2005), myocardial aconitine concentrations reached a Cmax approximately eightfold higher than plasma, with the highest measured value observed at the earliest common sampling point (15 min). Thereafter, myocardial concentrations declined largely in parallel with plasma and became undetectable within 24 hr. Based on the pharmacokinetic principle that drug concentrations in high-blood-flow organs rapidly equilibrate with plasma, the myocardial interstitial concentration in humans is likewise expected to decline in parallel with plasma. Consistent with this expectation, human survival cases typically show a progressive attenuation of the initially frequent VTAs, including ventricular fibrillation (VF), as well as SVT, with sinus rhythm generally returning within 24 hr (Terui et al., 2008). In this case, the early phase was marked by recurrent VT and VF consistent with ES, followed by a period dominated by SVT with bundle-branch block, making differentiation from VT difficult (Oshima et al., 2024). These arrhythmias gradually subsided, and sinus rhythm was restored at approximately 15 hr. In our concentration–time profile, the rapid phase corresponded to 0–20 hr. The clinical period during which arrhythmias attenuated overlapped with the early portion of this rapid phase and extended into the subsequent re-equilibration phase. This temporal relationship suggests that the clinical severity and overall arrhythmia pattern parallel the decline in myocardial interstitial concentration.
A practical clinical indicator is required to assess the early toxicokinetic phase. Because plasma DDA concentrations can be measured only at a small number of specialized facilities, real-time monitoring is not feasible in acute clinical settings. Clinicians therefore should, in practice, rely on the persistence of severe VTAs as a surrogate indicator of persistently elevated DDA concentrations in both the myocardial interstitial ECF and plasma, helping to guide the timing of CHP. Accordingly, CHP may be considered in cases of substantial ingestion where severe VTAs remain refractory despite supportive care, whereas performing CHP after these arrhythmias have resolved offers little pharmacokinetic or clinical benefit.
If ES can be adequately controlled, patients without severe comorbidities generally return to sinus rhythm within approximately 24 hr (Terui et al., 2008), and aggressive reduction of plasma DDA concentrations by CHP is not always required. In contrast, when ES cannot be controlled, patients often present with profound hemodynamic instability, and mechanical circulatory support is typically initiated before CHP becomes feasible. Reports of ECMO-assisted CHP exist (Ren et al., 2021), but discussion of such cases would broaden the scope of this study and is therefore not pursued further. In cases of massive ingestion, or when renal impairment delays elimination, plasma and myocardial interstitial DDA concentrations may rise steeply. When ES persists yet hemodynamics can still be maintained, early CHP on the day of ingestion may offer meaningful therapeutic benefit.
Before considering the clinical implications further, it is important to acknowledge the limitations of our simulation model. First, the experiment was conducted under markedly non-physiological chemical conditions. A strongly acidic environment (0.2 N HCl) was required to maintain aconitine solubility, resulting in the compound being almost completely ionized. Under such conditions, adsorption is dominated by electrostatic attraction between protonated aconitine and acidic surface functional groups on activated charcoal. In vivo, however, aconitine exists in a more moderate acid-base environment, and a small but meaningful uncharged fraction is present at physiological pH. These neutral species are more readily adsorbed through hydrophobic and π–π interactions, suggesting that the strongly acidic conditions used here may underestimate the adsorption capacity of activated charcoal under physiological conditions. Second, the perfusate lacked key biological matrix components such as plasma proteins and competing adsorbates. The absence of protein binding and competitive adsorption simplifies the system and prevents reproduction of the complex distributional and competitive processes that occur in vivo. Third, the aconitine concentrations used in the experiment (10–1800 µg/mL) were far higher than clinically relevant toxic concentrations (ng/mL), a requirement imposed by the 0.1 µg/mL LLOQ of the spectrophotometric assay. Overall, although this simulation demonstrates that ionized aconitine can be adsorbed by activated charcoal under strongly acidic conditions, it does not replicate the complex plasma environment present in vivo and therefore limits direct extrapolation to clinical settings.
Several uncertainties remain regarding the pharmacokinetics of DDAs and the role of CHP. Key parameters—including protein binding, Vd, and tissue distribution characteristics—are based primarily on limited animal data. The only experimentally reported protein-binding estimate (24–32%) originates from a single rat study (Tang et al., 2012), and the two available intravenous Vd values (1.7 and 7.7 L/kg) likewise derive solely from rat experiments (Tazawa et al., 2003; Chen et al., 2009), underscoring the need for human data.
In aconitine poisoning, the severity of toxicity in humans is largely determined by the ingested dose, and absorption varies substantially depending on the formulation, such as crude herbal materials, wild plants, extracts, or processed products (Chan, 2009). In contrast, clinical data describing the time course of distribution and metabolism are scarce, and existing reports—including those quantifying blood and urinary metabolites—are insufficient to evaluate interindividual variability. Faster elimination has been observed in patients with preserved renal function and without circulatory failure (Fujita et al., 2007), whereas impaired renal function represents a clinically relevant source of interindividual variability and may lead to delayed elimination and a prolonged half-life, as seen in this case. Although CE and P450 exhibit interindividual variability, their relative contribution to DDA inactivation and the toxicological relevance of P450-derived metabolites remain unclear. Accordingly, patient-specific differences beyond renal function and circulatory status cannot be meaningfully assessed at present.
Several additional uncertainties relate to the physicochemical properties and elimination pathways of DDAs, as well as future research requirements. Given the scarcity of human pharmacokinetic data, available estimates of logP and pKa (Wei et al., 2024) suggest that DDAs are more hydrophilic and more extensively ionized at physiological pH than traditionally assumed, implying a smaller effective Vd and greater dependence on renal elimination. Nevertheless, these estimates remain indirect, and important aspects of the elimination pathways and distributional behavior are still not well defined. Future studies should evaluate adsorption under physiological pH, in matrices containing plasma proteins, and at clinically relevant concentrations. When plasma is used as the perfusate in future models, however, some degree of acidification may still be required to maintain uniform aconitine solubility. Because aconitine poisoning is rare, prospective studies are essentially infeasible, and progress will depend on high-quality case accumulation and multicenter collaboration. Existing reports from China suggest that CHP may improve survival or shorten the duration of arrhythmias, but its pharmacokinetic effectiveness has not been systematically evaluated (Lin et al., 2004; Ke et al., 2024).
This study provides three complementary findings. First, activated charcoal possesses a high saturated adsorption capacity for aconitine. Second, under simplified conditions, a CHP column can adsorb most of the aconitine during a single pass. Third, in this case, approximately 70% of plasma DDAs were extracted in a single pass, corresponding to an effective clearance of about 70 mL/min at a blood-flow rate of 100 mL/min and confirming the adsorption capacity of the column. Although this does not imply whole-body removal, plasma DDA concentrations decreased markedly during the 7-hr CHP session, falling from 37.9 ng/mL in the early post-ingestion period to 8.4 ng/mL at the end of the session, reflecting the combined effects of the initial distribution and re-equilibration phases as well as CHP-mediated extraction. In actual clinical use, adsorption efficiency declines after approximately 4 hr because of protein and cellular debris accumulation on the column surface. In this case as well, a clear reduction in the extraction ratio was observed at the 4-hr mark, indicating that this practical limitation also manifests clinically. For this reason, in poisoning cases requiring prolonged treatment, replacement of the CHP column every 4 hr is generally recommended.
In summary, the pharmacokinetic behavior observed in this case suggests that CHP can reduce the plasma unbound fraction of DDAs, thereby promoting the decline in myocardial interstitial concentrations that drive toxicity. This mechanism aligns with the temporal resolution of VTAs in this case, which occurred during the rapid phase of concentration decline. The present findings provide rare pharmacokinetic evidence that CHP may facilitate the reduction of circulating aconitine-type DDAs. In cases involving massive ingestion or impaired renal elimination, early reduction of plasma concentrations by CHP may contribute to improved clinical outcomes. Although additional case accumulation and clinical data are needed, these observations offer preliminary pharmacokinetic support for the potential therapeutic value of early CHP in severe aconitine poisoning. However, because this report describes a single case, the generalizability of these findings is inherently limited.
We are grateful to Prof. Dr. Michinao Mizugaki of the Department of Pharmaceutical Sciences, Tohoku University Hospital, for measuring the four DDA concentrations in our clinical case. We are also grateful to Mr. Motomu Nada for preparing the figures and to Ms. Kyoko Nakada for her assistance with reference collection. We also thank Dr. Ichiro Hisatome and Dr. Naoto Burioka of Yonago Medical Center for their valuable advice during the preparation of this manuscript.
FundingThis research was funded by a joint research effort conducted by the authors and Kuraray Medical Incorporated.
Conflict of interestThis research was supported by a joint research effort conducted by the authors and Kuraray Medical Incorporated. However, the company had no control over the writing or publication of this work. Co-author Shuhei Nakaji was formerly an employee of the company.
Data availabilityThe data in this study are included in the article. Additional data are available from the corresponding author upon reasonable request.
Author contributionsConceptualization: Akira Tanaka
Methodology: Shuhei Nakaji
Validation: Naoki Moriyama, Aki Aoki, Ryo Endo
Formal analysis: Yoshiaki Oshima and Shuhei Nakaji
Investigation: Yoshiaki Oshima, Yukari Minami, Masaharu Fukuki, Shuhei Nakaji
Resources: Shuhei Nakaji
Data curation: Masato Nakasone, Shuhei Nakaji
Writing—original draft preparation: Yoshiaki Oshima, Yukari Minami, Masaharu Fukuki, Naoki Moriyama, Aki Aoki
Writing—review and editing: Akihiro Otsuki, Ryo Endo, Masato Nakasone, Yoshinori Kitagawa, Hiroyuki Minato
Visualization: Yoshinori Kitagawa, Hiroyuki Minato
Supervision: Akihiro Otsuki
Project administration: Yoshiaki Oshima
Funding acquisition: Akira Tanaka
Ethical approval and consent to participateThe study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Ethics Committee of Tottori University, Faculty of Medicine (number: 24A099; approval date: October 16, 2024). Because this clinical case study was conducted retrospectively using data already in clinical use and anonymity was ensured, the Ethics Committee waived the requirement for written informed consent and used an opt-out method by posting notices on the hospital’s website. This decision was made in accordance with the Ethical Guidelines for Medical and Health Research Involving Human Subjects issued by the Japanese Ministry of Health, Labor, and Welfare.
Patient consent for publicationBecause this clinical case study was conducted retrospectively using data already in clinical use and anonymity was ensured, the Ethics Committee waived the requirement for written informed consent for publication. We acknowledge that although the patient was successfully treated for aconite poisoning, she passed away six months later due to another illness.
