In vivo quantitative mapping of human mitochondrial cardiac membrane potential: A feasibility study

Abstract

Purpose:

Alteration in mitochondrial membrane potential (ΔΨ_m) is an important feature of many pathologic processes, including heart failure, cardiotoxicity, ventricular arrhythmia, and myocardial hypertrophy. We present the first in vivo, noninvasive, assessment of regional ΔΨ_m in the myocardium of normal human subjects.

Methods:

Thirteen healthy subjects were imaged using [¹⁸F]-triphenylphosphonium ([¹⁸F]TPP+) on a PET/MR scanner. The imaging protocol consisted of a bolus injection of 300 MBq followed by a 120 min infusion of 0.6 MBq/min. A 60 min dynamic PET acquisition was started 1 hour after bolus injection. The extracellular space fraction (f_ECS) was simultaneously measured using MR T1-mapping images acquired at baseline and 15 minutes after gadolinium injection with correction for the subject’s hematocrit level. Serial venous blood samples were obtained to calculate the plasma tracer concentration. The tissue membrane potential (ΔΨ_T), a proxy of ΔΨ_m, was calculated from the myocardial tracer concentration at secular equilibrium, blood concentration, and fECS measurements using a model based on the Nernst equation.

Results:

In 13 healthy subjects, average tissue membrane potential (ΔΨ_T), representing the sum of cellular membrane potential (ΔΨ_c) and ΔΨ_m, was −160.7±3.7 mV, in excellent agreement with previous in vitro assessment.

Conclusion:

In vivo quantification of the mitochondrial function has the potential to provide new diagnostic and prognostic information for several cardiac diseases as well as allowing therapy monitoring. This feasibility study lays the foundation for further investigations to assess these potential roles.

ClinicalTrial Identifier: NCT03265431

Introduction

Myocardial mitochondrial dysfunction plays a key role in many pathologic processes, including heart failure [1], cardiotoxicity [2], ventricular arrhythmia [3], and reperfusion injury [4]. Alteration in mitochondrial membrane potential (ΔΨ_m) is a fundamental biomarker of mitochondrial and cellular dysfunction as under normal conditions, ΔΨ_m is maintained within narrow limits [5,6]. Currently, there is no technique enabling non-invasive measurement of ΔΨ_m. In vitro measurements of ΔΨ_m have been performed and validated for decades using fluorescent and radiolabeled lipophilic cation probes, including ³H-tetraphenylphosphonium (³H-TPP⁺) [7–9]. The in vitro assays rely on long-lived β emitters and are thus inappropriate for human in vivo investigations. Our group recently described the first successful method for in vivo mapping of the tissue membrane potential, ΔΨ_T, a proxy of ΔΨ_m, in pigs, using the fluorinated compound [¹⁸F]TPP⁺ [10]. Using PET/CT imaging, we were able to map the left ventricular ΔΨ_T in absolute units of millivolts (mV). The obtained results were consistent with previous in vitro bench-top methods and confirmed the basic feasibility of quantitative in vivo mapping of ΔΨ_T. Here, we extend this method to healthy volunteers and describe the first in vivo imaging of ΔΨ_T in humans.

Materials and Methods

Patient Population

Thirteen healthy subjects, 7 females and 6 males, aged between 41 and 74 years, without conditions or medication known to affect cellular membrane potential, such as diabetes, were studied. The study was conducted under the approval of the Partners Human Research Committee, the Institutional Review Board of Partners HealthCare.

Quantification of Membrane Potential

The method for in vivo quantification of ΔΨ_T has been described in detail previously [10]. Briefly, we divide the tissue distribution space of [¹⁸F]TPP⁺ into three compartments: the extracellular space comprised of the interstitial and plasma space, the cytosol and the mitochondria. Total activity in a voxel can be expressed as the sum of the activity in each compartment contained within that voxel. Therefore, at steady-state, the concentration of [¹⁸F]TPP⁺ measured by PET in a voxel can be written as:

$${\overline{C}}_{PET}=\left(1-f_{ECS}\right)\left(f_{mito}\cdot {\overline{C}}_{mito}+\left(1-f_{mito}\right)\cdot {\overline{C}}_{cyto}\right)+f_{ECS}\cdot {\overline{C}}_{ECS}$$ Eq. 1

Where ${\overline{C}}_{\text{mito }},{\overline{C}}_{\text{cyto }}$ and ${\overline{C}}_{ECS}$, represent the steady-state concentration of [¹⁸F]TPP⁺ in the mitochondria, cytosol, and extracellular space respectively, 𝑓_𝐸𝐶𝑆 and 𝑓_{𝑚𝑖𝑡𝑜} represent the extracellular space fraction and mitochondrial fraction respectively. At equilibrium, the Nernst equation can be used to relate the concentration of tracer on each side of a membrane to its electric potential ΔΨ:

$$\frac{{\overline{c}}_{in}}{{\overline{c}}_{out}}=e^{-\beta \text{ΔΨ}}$$ Eq. 2

Where $\beta =\frac{zF}{RT}$ is the ratio of known physical parameters: z is the valence, F denotes Faraday’s constant, R is the universal gas constant and T is the temperature in Kelvin. Noting that ${\overline{C}}_{ECS}={\overline{C}}_p$ at steady state, division of Eq. 1 by ${\overline{C}}_p$ yields the following expression:

$$V_T=\frac{{\overline{c}}_{PET}}{{\overline{c}}_p}=\left(1-f_{ECS}\right)\left(f_{\text{mito }}\cdot e^{-\beta \left(\text{Δ}{\text{Ψ}}_m+\text{Δ}{\text{Ψ}}_c\right)}+\left(1-f_{mito}\right)\cdot e^{-\beta \text{Δ}{\text{Ψ}}_c}\right)+f_{ECS}$$ Eq. 3

Where V_T, ΔΨ_𝑚, and ΔΨ_𝑐 represent the volume of distribution of the tracer, and the mitochondrial and cellular membrane potential respectively. Because V_T is >> 1, Eq. 3 can be approximated by

$$V_T\approx \left(1-f_{ECS}\right)\cdot f_{mito}\cdot e^{-\beta \left(\text{Δ}{\text{Ψ}}_T\right)}$$ Eq. 4

Where ΔΨ_𝑇 is the total tissue membrane potential, defined as ΔΨ_𝑚 + ΔΨ_𝑐. The f_mito parameter was assumed to be constant at 0.25 [11].

[¹⁸F]TPP⁺ synthesis

[¹⁸F]TPP⁺ was synthesized and purified on a GE Tracerlab FXN synthesis unit using [¹⁸F]fluoride was produced onsite by a GE PETtrace cyclotron bombarding >98% enriched ¹⁸O-water. The [¹⁸F]TPP⁺ was sterile filtered into a sterile vial and tested to ensure it met all FDA and USP specifications for a PET sterile radiopharmaceutical prior to injection. The radiation dose exposure from [¹⁸F]TPP⁺ is estimated at 0.0178 mSv/MBq, corresponding to an effective dose of 6.6 mSv for a dose of 370 mBq.[12]

PET/MR imaging

A bolus of 300 MBq of [¹⁸F]TPP⁺ was injected intravenously using PTFE tubing immediately followed by an infusion of 0.6 MBq/min over 2 hours, so that the magnitude of the bolus (K_bol) corresponds to 500 min of infusate. This bolus with infusion protocol allows the system to reach secular tracer equilibrium in the myocardium and blood. Approximately 60 minutes after the bolus injection, a 60 minutes PET acquisition in list mode was performed centered on the thorax. During PET acquisition, MR sequences were acquired before and after administration of gadolinium (Dotarem 0.5 mM/mL, 0.1 mM/kg) to measure the myocardium extracellular space fraction. Modified Look-Locker Inversion recovery (MOLLI) sequences were used to acquire T1 maps before and 15 minutes after contrast administration [13]. Dynamic PET images were reconstructed using an iterative algorithm (OSEM 3 iterations, 21 subsets) with 344×344×127 pixels and 1 min frame length. Attenuation correction was performed using a segmentation method based on Dixon sequence with model-based bone estimation [14].

Extracellular Space Fraction Quantification

Region-of-interests (ROI) were drawn over the myocardium and left ventricular cavity on T1 images. The extracellular space (ECS) fraction was calculated using the following formula:

$$f_{ECS}=\frac{\left(\frac{1}{T_{1myo,post}}-\frac{1}{T_{1myo,pre}}\right)}{(\frac{1}{T_{\text{1 }blood,post}}-\frac{1}{T_{\text{1 }blood,pre})}}\times (100-Hct)$$ Eq. 5

where 𝑇_{1 𝑚𝑦𝑜,𝑝𝑟𝑒} and 𝑇_{1 𝑚𝑦𝑜,𝑝𝑜𝑠𝑡} each denote the 𝑇₁ of myocardium measured before and after gadolinium (Gd) injection, respectively, 𝑇_{1 𝑏𝑙𝑜𝑜𝑑,𝑝𝑟𝑒} and 𝑇_{1 𝑏𝑙𝑜𝑜𝑑,𝑝𝑜𝑠𝑡} each denote the 𝑇₁ of blood measured before and after Gd injection, respectively, and 𝐻𝑐𝑡 denotes hematocrit in percentage measured from blood samples.

Results

Characteristics of the studied population is presented in Table 1. For all 13 subjects, secular equilibrium was achieved after ≤ 90 minutes with relatively constant myocardial and blood tracer concentrations (Figure 1b). For one subject, motion introduced significant artifacts in the PET images and only the motion free portion of the acquisition was used for the analysis. Overall, the average f_ECS was 0.293 ± 0.025 and ranged from 0.261 to 0.339. Average V_T was 73.3 ± 12.0 and ranged from 60.4 to 106.7. Average ΔΨ_T was −160.7 ± 3.7 mV and ranged from −170.4 to −156.2 mV (Figure 1c). Values of ΔΨ_T across the standard 17 left ventricular segments were homogeneous (Figure 1d, Table 2).

Table 1:

Population characteristics.

Female, N (%)	7 (53.8%)
Age (y), mean ± SD (range)	56 ± 10 (41 − 74)
Ethnicity, N (%)
Caucasian	11 (84.6%)
African American	1 (7.7%)
Hispanic	0 (0.0%)
Asian	1 (7.7%)
BMI (kg/m2), mean ± SD (range)	25.4 ± 3.5 (18.8 − 31.5)
Hypertension, N (%)
No	11 (84.6%)
Yes	2 (15.4%)
Diabetes, N (%)
No	13 (100.0%)
Yes	0 (0.0%)
Smoking, N (%)
Never	12 (92.3%)
Current	0 (0.0%)
Remote (>1 year)	1 (7.7)%
Dyslipidemia, N (%)
No	13 (100.0%)
Yes	0 (0.0%)
Heart Rate (min−1)	66 ± 14 (46 − 96)
Systolic blood pressure (mmHg)	127 ± 20 (101 − 173)
Diastolic blood pressure (mmHg)	78 ± 7 (64 − 87)
Rate pressure product (mmHg/min)	8297 ± 1949 (5757 − 12,032)

Figure 1:

A, Average extracellular fraction (f_ECS) of the 13 study subjects. B, Time activity curves (TAC) of a representative subject showing that tracer concentrations in the blood and in the myocardium are at secular equilibrium between 90 and 120 minutes post tracer injection. C, Average tissue membrane potential (ΔΨ_T) of the 13 study subjects. D, ΔΨ_T of the 17 left ventricular segments of the 13 study subjects.

Table 2:

Average tissue membrane potential (ΔΨ_T) for the 17 left ventricular segments

Segments	ΔΨT (mV)
1	−160.4 ± 3.7
2	−159.0 ± 4.3
3	−161.6 ± 4.1
4	−163.5 ± 4.0
5	−161.6 ± 5.2
6	−161.1 ± 4.4
7	−160.4 ± 4.5
8	−159.9 ± 4.2
9	−162.1 ± 3.9
10	−164.5 ± 4.0
11	−163.4 ± 4.9
12	−162.5 ± 4.5
13	−161.2 ± 4.4
14	−158.9 ± 4.2
15	−161.2 ± 4.7
16	−162.1 ± 5.2
17	−154.0 ± 4.4
Left Ventricle	−160.7 ± 3.7

With this method, high quality parametric images of ΔΨ_T can be generated, allowing for regional assessment of ΔΨ_T (Figure 2). In our sample, no significant correlations between age and f_ECS (R²=0.06579, p=0.40) or between age and ΔΨ_T (R² = 0.0340, p=0.55) was observed (Figure 3). In addition, there was no significant difference between male and female in f_ECS (0.284±0.025 vs 0.301±0.024, p=0.23) and ΔΨ_T (−162.4±4.4 vs −159.2±2.2 mV, p=0.11).

Figure 2:

Parametric image of the volume of distribution (top) and tissue membrane potential (bottom). SA, short axis. HLA, horizontal long axis. VLA, vertical long axis.

Figure 3:

Extracellular fraction (f_ECS) and tissue membrane potential (ΔΨ_T) versus age (left) and according to sex (right).

Discussion

Quantification of ΔΨ_m has been performed in vitro by measuring the concentration of different lipophilic cations in mitochondria. These molecules cross the cellular and mitochondrial phospholipid membranes without significant interaction and their concentration on either side of a membrane at equilibrium abide by the Nernst equation. In this study, we demonstrated for the first time the feasibility of in vivo quantification of ΔΨ_T, a proxy of ΔΨ_m, in humans. The method presented in this work relies on the exact same principle on which in vitro techniques are founded but uses PET imaging to measure tissue concentration of the lipophilic cation [¹⁸F]TPP⁺. The total tissue membrane potential, ΔΨ_T, is often used instead of ΔΨ_m for in vitro studies, including explanted hearts, as isolation of the different cellular compartment is not easily feasible [9,15].

As normal mitochondrial function depends on the maintenance of membrane potential in a narrow range, pathologies associated with mitochondrial dysfunction could potentially be studied with membrane potential tracers such as [¹⁸F]TPP⁺. Indeed, several pathologies have been associated with mitochondrial dysfunction, including cancers [18], diabetes [19], and cardiotoxicity [16]. The first attempt to measure ΔΨ_T in vivo has been reported over 30 years ago by Fukuda et al. [15] who studied the distribution of the lipophilic cation triphenylmethylphosphonium labeled with ¹¹C (¹¹C-TPMP).[15] Using primitive PET systems, they reported ΔΨ_T in dogs (−148.1±6.0 mV), rats (−146.7±3.8 mV), and mice (−139.3 ± 5.8 mV) hearts, with results comparable to those obtained for humans in this study (−160.7 ± 3.7 mV). However, the resolution of the PET scanner was very low and quantitation methods are questionable by today’s standards. More recently, Gurm et al reported an attempt to measure myocardial ΔΨ_m in vivo with [¹⁸F]TPP⁺ in a swine model of ischemic heart disease [17]. Their results were however discordant with previous in vitro and in vivo studies, with a reported average ΔΨ_m of −91 mV. This is likely related to two critical methodological issues: first, the Nernst equation was applied while tracer concentrations in the blood and myocardium were changing, whereas both the tracer and subject must be in steady state for concentrations to exhibit Nernstian behavior [18,19]. Second, they did not account for extracellular volume, which would lead to underestimation of ΔΨ_m. Our group reported the first successful technique to image and quantify ΔΨ_T in swine [10]. The ΔΨ_T values obtained in that study, which is based on similar methodology, were slightly different (−129 mV) than those obtained in the present study [10]. We hypothesize that two main factors could explain these differences. First, species variability could account for part of this difference, and second, the use of isoflurane for anesthesia in pigs could lead to partial mitochondrial depolarization [20,21].

Average ΔΨ_c can be estimated from typical ventricular action potential. Assuming a QT of 400 ms, using the average subjects’ heart rate of 65 bpm, a polarized (phase 4) ΔΨ_c of −90 mV, and an average depolarized (phase 0–3) ΔΨ_c of +30 mV (peak +50mV), the average ΔΨ_c is approximately −38 mV. Therefore, we can estimate ΔΨ_m (ΔΨ_m = ΔΨ_T-ΔΨ_c) to be −123 mV. This value is in excellent agreement with the measurements of −118 mV obtained in explanted intact perfused rat hearts using the Langendorff method and a physiological glucose infusion [9]. It is important to acknowledge that the mitochondrial environment significantly affects ΔΨ_m and it must be taken into consideration when comparing in vivo and in vitro measurements. Indeed, it has been shown that in situ assessment of ΔΨ_m typically yields smaller values compared to isolated mitochondrion (−180 to −190 mV) [22–24]. This can be accounted for by variable substrate availability which alters ΔΨ_m [9,25].

A main limitation of this study is the lack of an in vivo gold standard for the measurement of ΔΨ_m. However, this technique simply represents the extension of well-established chemistry principles and in vitro methods. Importantly, membrane potential measurements of this study agreed well with those of previous and comparable experiments, supporting the validity of the presented methodology. Another limitation is the fact that the cellular mitochondrial content (f_mito) was not quantified and assumed the same for all subjects. However, myocardial mitochondrial morphology and volume is known to be affected in several conditions such as diabetes [26,27] and heart failure [28]. Unfortunately, in vivo measurement of f_mito is not currently feasible. Using Eq. 1, we can estimate the variation induced by not accounting for f_mito: a change of 10% in f_mito will lead to a change of 2–3 mV in ΔΨ_T. In several pathologies of interest, it is expected that f_mito will decrease, leading to overestimation of depolarization (less negative ΔΨ_T). It should also be noted that f_mito enters Eq. 1 as constant multiplier, meaning that for normal human subjects systematic errors in f_mito do not affect the method’s ability to compare subjects or groups. Another limitation of this study is its small sample size. Nonetheless, this study demonstrated the feasibility of in vivo measurements of membrane potential in humans. Finally, 2 subjects had a medical history of hypertension and 3 subjects had a systolic blood pressure >130 mmHg at time of imaging, which could affect the measured ΔΨ_T.

Conclusion

We demonstrated for the first time the feasibility of noninvasive, in vivo, quantitative assessment of cardiac ΔΨ_T in humans, with very low variability of ΔΨ_T between subjects meaning that relatively small sample size would be sufficient for hypotheses testing. Furthermore, true quantitation provided by this method allows comparison between subjects and enables the investigation of several pathologies of interest such as diabetes and chemotherapy-induced cardiotoxicity, which are typically associated with diffuse myocardial involvement. ΔΨ_T has the potential to provide early diagnosis for various pathologies, including cardiotoxicity. In addition, given the new therapies targeting the mitochondria, ΔΨ_T could serve as a surrogate endpoint in clinical trials and be used for the assessment of response to therapy. Finally, the method is generalizable to other organs and tissue, and could be used for characterization of tumors.[29] More studies are required to assess these potential roles.