A New Approach Methodology (NAM) for the Prediction of (Nor)Ibogaine-Induced Cardiotoxicity in Humans

Fig. S2: Effects of repeated addition of 0.05% (v/v) acetonitrile (squares) and 0.1% (v/v) DMSO (circles) on the FPDc relative to baseline conditions in the vehicle control well set at 100% Vehicle control addition 0 on the X-axis represents the response of the baseline control set at 100%. 1-7 represent the 1 to 7 addition of vehicle controls corresponding to the 1 to 7 addition. Each data point represents the mean ± SD of three independent experiments.


A New Approach Methodology (NAM) for the Prediction of (Nor)Ibogaine-Induced Cardiotoxicity in Humans
Supplementary Data 1 Fig. S1: Irregular waveforms of field potential observed in human induced pluripotent stem cell-derived cardiomyocytes using the multiple-electrode array (a) Arrhythmia-type waveform induced by 1 µM ibogaine; (b) arrhythmia-type waveform induced by 3 µM noribogaine. Waveforms present in (a) and (b) were not used for defining the in vitro concentration-response curves for FPDc effects.

1% (v/v) DMSO (circles) on the FPDc relative to baseline conditions in the vehicle control well set at 100%
Vehicle control addition 0 on the X-axis represents the response of the baseline control set at 100%. 1-7 represent the 1 st to 7 th addition of vehicle controls corresponding to the 1 st to 7 th addition. Each data point represents the mean ± SD of three independent experiments.

Fig. S3: Concentration-response curves for cardiotoxicity in hiPSC-CMs of the reference compounds (a) dofetilide and (b) isoproterenol
The response of the baseline at 0.1% (v/v) DMSO was set at 100%. Data represent the mean of results obtained from three independent experiments, each containing six well replicates. Each data point represents the mean ± SD. Statistically significant changes compared to the solvent control are marked with * , p < 0.05; * , p < 0.01; ** and p < 0.001, *** .

Fig. S4: Concentration-dependent formation of (a) noribogaine from ibogaine and (b) noribogaine glucuronide from noribogaine using in vitro incubations with human liver microsomes
Data represent the mean of three independent experiments. Each data point represents the mean ± SD. a Internet-purchased ibogaine with unknown purity; b Baseline was assumed to be 405 ms and 411 ms for male and female, respectively, given that no baseline information was reported (Wedam et al., 2007); c a dose of 3500 mg ibogaine was corrected for the reported purity of 15%.

Text S1
Development of the PBK models A PBK model consisting of multiple organ compartments was developed to describe the ADME of ibogaine and its metabolite noribogaine upon oral administration. Noribogaine has also been reported to cause prolongation effects on the QTc interval in human (Glue et al., 2016). Therefore, an oral administration route was included in the submodel of noribogaine, which enables modeling of noribogaine kinetics and prediction of its cardiotoxicity upon oral administration. Human physiological parameters reported in Brown et al. (1997) were used in the PBK model (Tab. S2).
For the absorption parameters, the ka values of ibogaine and noribogaine were extrapolated from in vitroderived Papp values obtained in the present study as described in the "in vitro intestinal transport studies" section. Due to limited pharmacokinetic data of both ibogaine and noribogaine, the experimental fractions absorbed (Fa) were not available. However, many studies demonstrated a positive correlation between Papp values and human Fa and indicated that Fa values can be estimated to be 1 when the Papp value is higher than 10 -5 (cm/s) (Lozoya-Agullo et al., 2015;Lüpfert and Reichel, 2005;Skolnik et al., 2010). Considering the relatively high Papp values measured for ibogaine and noribogaine (see Results), the Fa values for both compounds were assumed to be 1.
To describe how ibogaine and noribogaine distribute in organs and the systemic blood circulation upon absorption, tissue:blood partition coefficients (P) of ibogaine and noribogaine were obtained by converting tissue:plasma partition coefficients using the corresponding blood/plasma ratio (BPr) as previously described (Shi, 2020). The tissue:plasma partition coefficients were predicted using the QIVIVE tool (version 1.0) from Wageningen Food Safety Research (WFSR, 2020), in which the algorithm of Berezhkovskiy (2004) was applied for ibogaine and the algorithm of Rodgers and Rowland (2006) was used for noribogaine given it generally shows better prediction for zwitterions (Graham et al., 2012;Utsey et al., 2020). Other input parameters included acid-base properties (pKa), lipophilicity (logP) and fraction unbound in plasma (fu,p). The logP and pKa values were predicted using Chemicalize (ChemAxon, Hungary). The logP and pKa of ibogaine were 3.53 and 8.97, respectively. The log P and pKa of noribogaine were 3.0 and 8.87 (basic) and 9.66 (acidic). The fu,p values were determined using pooled human plasma in the present study. A BPr value of 2.5 for noribogaine in human was reported by Mash et al. (2016) while no published BPr value was available for ibogaine. Given that the concentration was reported to be higher in the blood compared to plasma also for ibogaine (Alper, 2001;Maciulaitis et al., 2008), the BPr value of ibogaine was assumed to be the same as that for noribogaine. Tissue:blood partition coefficients for ibogaine and noribogaine are summarized in Table S3.
Based on in vitro metabolism and in vivo pharmacokinetic studies, liver was considered the major organ for the metabolism of ibogaine and noribogaine (Glue et al., 2015b(Glue et al., , 2016Obach et al., 1998). The kinetic parameters obtained in the current study were used to define the conversion of ibogaine to noribogaine and the glucuronidation of noribogaine by applying Michaelis-Menten kinetics. To extrapolate the in vitro Vmax to an in vivo Vmax, a total microsomal protein per gram of liver (MPL) value of 32 mg/g was applied in the PBK model (Barter et al., 2007). The in vivo Km was assumed to be similar to the in vitro Km.
Hepatic metabolism was reported to be the major elimination route for ibogaine (Mash et al., 2016). For noribogaine, Glue et al. (2015a) found that only a small amount of the administered dose (1.4-3.9%) was detected in urine as noribogaine and its glucuronide after a single oral dose of noribogaine in human, indicating the negligible contribution of urinary excretion to the elimination of noribogaine. For this reason, renal excretion was not considered in the PBK model of noribogaine. Given the higher molecular weight of ibogaine and noribogaine than the cut-off value of 275 Da for biliary excretion in human, the compounds could be excreted via bile instead of via urine (Haddad and Nong, 2020). This is supported by the fact that both ibogaine and noribogaine were detected in human bile (Kontrimavičiūtė et al., 2006;Maciulaitis et al., 2008) and were excreted via the gastrointestinal tract (Alper, 2001) and were present in the feces in rat (Jeffcoat et al., 1993). Therefore, biliary excretion was assumed to be the major elimination route for ibogaine and noribogaine and was included in the PBK model. The biliary excretion rate constant (kb) of noribogaine was obtained by the curve fitting option in Berkeley Madonna (version 8.3.18, UC Berkeley, CA, USA), in which the predicted blood maximum concentration (Cmax) of noribogaine was fitted to the Cmax of noribogaine in the blood that was reported in clinical studies (Glue et al., 2015a(Glue et al., ,b, 2016. The averaged fitted kb for noribogaine was 0.575 (/h). Due to the limited pharmacokinetic data on ibogaine and little influence of biliary excretion on ibogaine blood kinetics (see the results of the sensitivity analysis), the same kb value was assumed for ibogaine. Kinetic model calculations and curve fitting were performed with Berkeley Madonna, applying Rosenbrock's algorithms for solving stiff systems. Model equations are shown in Text S2.  ; the kb of noribogaine was assumed to be same for ibogaine kbnor=0.575 ; biliary excretion rate constant (/h) of noribogaine was obtained by fitting CVBnor to reported in vivo data (Glue et al., 2016;Glue et al., 2015a;Glue et al., 2015b).  ; Concentration of noribogaine in heart venous blood (ug/l) BPribo=2.5 ;blood to plasma ratio of ibogaine, assumed to be same as noribogaine BPrnor=2.5 ;blood to plasma ratio of noribogaine (Mash et al. 2016) fupibo=0.04 ;fraction unbound in plasma of ibogaine obtained from the current study fupnor=0.26

Text S3
Method of sensitivity analysis A local parameter sensitivity analysis was conducted to estimate to what extent the model parameters can influence the model output, which refers to Cmax of ibogaine and noribogaine in the heart venous blood upon the oral administration of ibogaine or noribogaine. Furthermore, given that the in vivo cardiotoxicity of ibogaine is dependent on the unbound concentration of both ibogaine and noribogaine, the sensitivity analysis was also performed for the unbound toxic equivalence (TEQ) concentration (details see in "QIVIVE using PBK modeling-based reverse dosimetry" section). The sensitivity coefficient (SC) was calculated according to the Eq. S1: SC= (C'-C) (P'-P) x P C Eq. S1 where P and C represent the initial value of the model parameter and output, respectively. P' and C' stand for the model parameter and model output after a 1% increase in an individual model parameter value, respectively. Only parameters with an absolute SC > 0.1 are considered influential on the model output (Rietjens et al., 2011). The sensitivity analysis was carried out for a subject with a body weight of 70 kg (Brown et al., 1997) and for a single oral dose of 20 and 500 mg ibogaine, representing a safe and well tolerated dose for healthy people (Glue et al., 2015b) and a clinically relevant dose for the treatment of drug addiction (Maciulaitis et al., 2008), respectively. For the sensitivity analysis of the noribogaine model, a single oral dose of 20 mg and 200 mg was chosen, respectively representing a safe dose for healthy people and a dose level associated with prolonged QTc in human (Glue et al., 2016). Figure S5 shows the results of the sensitivity analysis presenting the influential model parameters for the prediction of Cmax of ibogaine and noribogaine in the heart venous blood and of the Cmax expressed in unbound ibogaine equivalents using a TEQ approach, upon exposure to an oral dose of ibogaine or noribogaine. For the oral administration of ibogaine (Fig. S5a), results reveal that Cmax of ibogaine in the heart venous blood is most sensitive to the body weight, fraction absorbed of ibogaine, fraction of liver, percentage of blood to liver and metabolic parameters for conversion of ibogaine to noribogaine (MPL, Vmaxc1 and Km1). When the oral dose increased from 20 mg to 500 mg, the normalized SC values of body weight, fraction of liver, absorption related parameters (Faibo and kaibo) and metabolic parameters (MPL and Vmaxc1) increased 2-to 3-fold while the normalized SC value of Km1 showed a 3.6-fold decrease.

Results of sensitivity analysis
As illustrated in Figure S5b, similar SC values were obtained for the prediction of the Cmax of noribogaine in the heart venous blood at two oral doses of 20 mg and 200 mg noribogaine. The predicted Cmax of noribogaine in the heart venous blood is most affected by the fraction absorbed of noribogaine and the body weight with the normalized SC values being 1. Parameters related to percentage of blood to tissues also influence the prediction especially the percentage of blood to liver, rapidly perfused tissue, slowly perfused tissue, and kidney with the normalized SC values above 0.5. Other model parameters show less influence with the normalized SC values ranging from 0.14 to 0.37 (Fig.  S5b). Figure S5c shows that the unbound TEQ concentration expressed in ibogaine equivalents is most sensitive to the percentage of blood flow to slowly perfused tissue, followed by the percentage of blood flow to liver and to rapidly perfused tissue with normalized SC values above 1. Besides, body weight, fraction absorbed of ibogaine, blood:plasma ratio of noribogaine, unbound fraction of noribogaine in plasma, and TEF of noribogaine show a high influence on the prediction with the normalized SC values being 1. Figure S5c also indicates that parameters related to percentage of blood flow to tissues (QSc, QLc, QRc, QKc, QFc and QHc) show a dose-dependent influence on the prediction with the normalized SC values being higher at 500 mg compared to those at 20 mg while the SC of other model parameters generally are not dose-dependent at the two doses of ibogaine. Dotted lines indicate the normalized SC with an absolute value higher than 0.1. BW, body weight; VLc, fraction of liver; VRs, fraction of rapidly perfused tissue; VSc, fraction of slowly perfused tissue; QLc, percentage of blood flow to liver; QKc, percentage of blood flow to kidney; QHc, percentage of blood flow to heart; QRc, percentage of blood flow to rapidly perfused tissue; QSc, percentage of blood flow to slowly perfused tissue; PSibo, partition coefficient slowly perfused tissue:blood of ibogaine; MPL, microsomal protein per gram of liver; Vmaxc1, unscaled maximum rate of ibogaine metabolism in liver; Km1, Michaelis-Menten constant for ibogaine metabolism in liver; PLnor, partition coefficient liver:blood of noribogaine; PSnor, partition coefficient slowly perfused tissue:blood of noribogaine; kaibo, absorption rate constant of ibogaine; kanor, absorption rate constant of noribogaine; Faibo, fraction absorbed of ibogaine; Fanor, fraction absorbed of noribogaine; Kbnor, biliary excretion constant of noribogaine; BPnor, blood to plasma ratio of noribogaine; fupnor, unbound fraction of noribogaine in human plasma; TEFnor, toxic equivalency factor of noribogaine.