ACAD9 treatment with bezafibrate and nicotinamide riboside temporarily stabilizes cardiomyopathy and lactic acidosis

Abstract

Pathogenic ACAD9 variants cause complex I deficiency. Patients presenting in infancy unresponsive to riboflavin have high mortality. A six-month-old infant presented with riboflavin unresponsive lactic acidosis and life-threatening cardiomyopathy. Treatment with high dose bezafibrate and nicotinamide riboside resulted in marked clinical improvement including reduced lactate and NT-pro-brain type natriuretic peptide levels, with stabilized echocardiographic measures. After a long stable period, the child succumbed from cardiac failure with infection at 10.5 months. Therapy was well tolerated. Peak bezafibrate levels exceeded its EC₅₀. The clinical improvement with this treatment illustrates its potential, but weak PPAR agonist activity of bezafibrate limited its efficacy.

1. Introduction

The oxidative phosphorylation system provides energy from reducing equivalents generated by fuel oxidation, and consists of the respiratory chain complexes I, II, III and IV to generate the inner membrane electrochemical proton gradient, and complex V for generation of ATP. Primary mitochondrial disorders are inborn errors of metabolism affecting this mitochondrial mechanism to generate energy. They are highly genetically heterogenous and have variable symptoms (1,2). Primary mitochondrial disorders which present with cardiomyopathy have few therapeutic options (3,4). Deficiency of complex I of the respiratory electron transfer chain (ETC) is similarly heterogenous. It presents with a variety of clinical manifestations that are mostly neurological (5). One of the more common causes of isolated complex I deficiency consists of pathogenic variants in the ACAD9 gene (6–9). The ACAD9 protein interacts with NDUFAF1, ECSIT, and TMEM126B as an assembly factor to add the ND2-containing subcomplex to the nascent assembled complex (10–12). ACAD9 emerged as a homolog of acyl-CoA dehydrogenase enzymes and retains this activity for oxidation of fatty acids with chain lengths of C12-C22 (13,14). In this enzymatic function, it uses flavin adenine dinucleotide (FAD) to transfer electrons to electron transfer flavoprotein (ETF). However, the pathophysiology is considered to be primarily related to defective complex I assembly resulting in deficient complex I activity with a limited contribution of the fatty acid oxidation function (15,16).

Patients with biallelic pathogenic variants in ACAD9 present with cardiomyopathy, lactic acidosis, and developmental delays (6–9,15–20). Riboflavin, the vitamin precursor of FAD, provides therapeutic value to many patients (9). The cardiomyopathy is usually hypertrophic, with dilated cardiomyopathy and pulmonary hypertension rarely observed (19). Heart failure and cardiovascular collapse are common causes of death (20). While presentation is mostly infantile, neonatal and even prenatal presentations have been reported (21,22). Milder presentation with optic atrophy, which may be riboflavin responsive, is rarely reported (23). Patients treated with cardiac transplantation have later developed late onset neurological, muscular, renal, and ophthalmological problems [20]. Patients that present before 1 year of age have a poor prognosis with a 70% mortality by 2 years of age, whereas patients that present after age 1 year have a much lower mortality 10% (9). Many patients have marked riboflavin-responsiveness with recommended dosing of riboflavin at 20 mg/kg/day (9). In vivo, this results in greatly improved survival (10). Mechanistically, responsiveness corresponds to an increase in oxygen consumption and, in some cases, a mild increase in ACAD9 protein levels (9). For those not responsive to riboflavin, few therapeutic options have been reported. A single case report showed improvement with a combination treatment of sodium pyruvate, a β-blocker, and coenzyme Q (24).

PPAR agonists have been used in various metabolic disorders (25). Bezafibrate is a pan-PPAR agonist with effect on both PPAR-α and PPAR-δ. After binding with a ligand on its C-terminal ligand binding domain, PPAR receptors are translocated to the nucleus to form a complex with the RXR receptor, and when activated, bind to transcription activation sites (26). PPAR hormone response elements are present in the gene promoters of all fatty acid oxidation enzymes and also increase the transcription of PGC-1α. PGC-1α expression results in increased transcription of the mitochondrial biogenesis factors NRF1, NRF2 and ERR-α, resulting in increased expression of all the respiratory chain enzyme complexes. ACAD9 has a putative NRF1 response element in its promotor region (13). In fibroblasts exposed to increasing bezafibrate doses, increased expression of respiratory chain enzyme activities in fibroblasts was observed at doses of 100 μM with an optimal effect at 300 μM (27). A modest positive effect was shown for complex I-deficient cell lines. In the ACAD9 in vitro studies, it was also shown that bezafibrate treatment at 400 μM increased oxygen consumption in 13 of 17 fibroblast cell lines, and increased ACAD9 protein levels in 6 of 7 ACAD9 fibroblasts cell lines (9).

The effect of PPAR coactivators is based in part on tissue distribution. PPAR-α is primarily expressed in liver, heart and to a lesser extent in skeletal muscle (26), whereas PPAR-δ is primarily expressed in liver, kidney, and muscle, and PPAR-γ is found mainly in liver, lymphocytes and monocytes. Bezafibrate is a dual PPAR agonist activating both the PPAR-α and the PPAR-δ coactivators but has limited potency (28). It upregulates expression of fatty acid oxidation enzymes and is approved in multiple countries for its lipid lowering effect in severe hypertriglyceridemia and mixed dyslipidemia in individuals not tolerating statins (29–31). The dosing in adults is 300 mg 2x/day for the sustained-release preparation. Peak concentration at this dosing occurred at 1.5 hours at 25 μM, with mostly urinary excretion (30). Primary side effects include intestinal symptoms, dizziness, and pruritus. Laboratory changes include elevation of transaminases and increase in creatine kinase. At high doses, the primary risk is for rhabdomyolysis, with blood dyscrasias an uncommon side effect. Bezafibrate treatment improved the function in some mitochondrial myopathy mouse models, but not in all (32–34). In clinical trials of long-chain fatty acid oxidation defects, treatment with bezafibrate was well tolerated with mixed clinical effects (35–38). Treatment with bezafibrate 600 – 1200 mg/day in adult human patients with the mitochondrial DNA mutation m.3243A>G was well tolerated (39).

Nicotinamide adenine dinucleotide (NAD) coenzymes are the carriers of high energy electrons in fuel oxidation and are used in anabolic processes, but also participate in cell signaling (40–42). Levels of NAD coenzymes are decreased in many diseases and conditions including mitochondrial myopathy (43) and heart failure (44). NAD repletion with vitamin precursors of NAD has ameliorated a variety of these conditions. In the mouse model of heart failure, nicotinamide riboside, a recently discovered NAD precursor vitamin (45), but not nicotinamide was capable of repleting cardiac NAD owing to depressed expression of the nicotinamide salvage gene NAMPT and highly induced expression of the NR salvage gene NMRK2 (43). In human primary mitochondrial myopathy, nicotinic acid improved muscle function (43). In fibroblasts of human mitochondrial patients, nicotinic acid restored the changes to the transcriptional network (46). In a mouse model of mitochondrial DNA depletion, nicotinamide riboside restored the NAD status and improved mitochondrial biogenesis and function (47). The gain in mitochondrial biogenesis has been attributed to deacetylation of PGC-1α possibly through activation of Sirt1 (41). Nicotinamide riboside chloride has been well characterized for human safety in healthy adults with typical adult dosing of 1000 mg/day (48–51).

We present here the case history of a child with mutations in ACAD9 with severe life-threatening cardiomyopathy, who was treated with a combination of high dose bezafibrate and nicotinamide riboside chloride.

2. Methods

The patient was treated under an emergency IND# 166404 from the FDA with consent on an IRB-approved study COMIRB#23–0447. He was also consented on an IRB-approved study COMIRB#16–0146 for the biochemical studies.

Fibroblasts were grown from a skin biopsy and cultured in MEM-α with 10% fetal bovine serum, 10% non-essential amino acids and antibiotics and antimycotics until 80% confluency. The respiratory chain enzyme activities were assayed spectrophotometrically as described (52,53). The complexes were further assessed by blue native PAGE with in-gel activity staining as described (52–54). The assembly of complex I was traced on a non-denaturing gel following western blotting and detection with an antibody against NDUFS2 as described (55–57). The level of ACAD9 protein was measured in a fibroblast homogenate by SDS-PAGE followed by western blotting and detection by an anti-ACAD9 antibody, a kind gift of Dr. J. Vockley, University of Pittsburgh. The abundance of the respiratory chain enzyme complexes (RCA) was determined by proteomics studies as described in Supplemental Methods (58,59).

The pharmacokinetics of bezafibrate were determined in a time course following a morning dose. Bezafibrate levels were measured by stable isotope dilution tandem mass spectrometry as described in Supplemental Methods. The pharmacokinetics of NAD metabolites was determined in whole blood collected into EDTA tubes, and 100 μL blood added to 1 μL 5 μM immucilin H, and stored at −80°C until analysis by liquid chromatography-tandem mass spectrometry as described (58) with the modifications as outlined in Supplemental Methods. After addition of internal standards and deproteinization and reconstitution in buffer, samples were analyzed by reverse phase liquid chromatography tandem mass spectrometry. Metabolites were reverse-quantified in a sample of ¹³C-cultivated Pichia pastoris extract purchased from Cambridge Isotope Labs (ISO1). The pharmacodynamics of the intervention was determined by following the level of ACAD9 and of the abundance of complex I in peripheral blood mononuclear cells (PBMCs) isolated from 3 mL EDTA blood using SepMate (STEMcell Technologies) isolation tubes according to manufacturer’s instructions and quantitative proteomics performed as described in detail in Supplemental Methods (59,60). The biomarkers growth differentiation factor 15 (GDF15) and fibroblast growth factor 21 (FGF21) were determined by enzyme-linked immunosorbent assay (ELISA) as described (61).

3. Case Report

The six-month-old boy was followed closely by his pediatrician for difficulty feeding and poor weight gain. At age 5 months, he presented with mild metabolic acidosis (bicarbonate 16 mEq/L, anion gap 19). Two and a half weeks later, he was admitted for worsened lactic acidosis to Children’s Hospital Colorado with bicarbonate 9, anion gap of 26, lactate 11.59 mM, and pyruvate 0.42 mM and lactate /pyruvate ratio 27.6 (normal <20). He presented with severely elevated alanine (1262 μM, normal 143–439) with normal levels of other amino acids, increased hydroxylated medium-chain acylcarnitines, mild β-hydroxybutyrate 0.85 mM, but normal liver function tests and ammonia. Qualitative urine organic acids revealed elevated lactate and ketones.

Upon admission at age 5 months 3 weeks, his initial lactate varied between 4 to 8 mM (Figure 1A). An echocardiogram showed a hypertrophic cardiomyopathy with fair contractility with left ventricle mass 143.4 g/Ht^2.7 (95^th %ile= 80g) (62); left ventricle ejection fraction (LV EF) 65%, left ventricle fractional shortening (LV FS) 34% (−1.42 SD) (Figure 2), very elevated NT-pro-brain type natriuretic peptide (NT pro-BNP) 21,700 pg/mL (Figure 1B). The next day decreased cardiac function to LV EF 46.9% and LV FS 23% (−6.19 SD) was observed, and he had a first cardiovascular collapse with altered mental status, poor perfusion, lactate to 18.54 mM, pH 6.7, and was intubated, ventilated, sedated and paralyzed. He recovered overnight, with down-trending lactate and he was extubated 3 days later.

Figure 1: Lactate and NT-pro-basic natriuretic peptide levels in a patient with ACAD9 disorder.

A. Lactate levels in mmol/L in a patient with ACAD9 disorder. The onset of treatment is shown and the time of an adenoviral gastroenteritis and of a parainfluenza respiratory infection shown. B. Levels of NT-pro-basic natriuretic peptide in pg/mL in this patient with ACAD9 disorder. The bar indicates the first hospital admission with the area in red prior to treatment and in green on treatment with bezafibrate and nicotinamide riboside.

Figure 2: Basic parameters of the cardiac function on serial echocardiogram evaluations.

The ejection fraction (A) and the fractional shortening (B) are shown. The ventricular mass in gram/body surface areâ^2.7 is shown (C). The bar indicates the first hospital admission with the area in red prior to treatment and in green on treatment with bezafibrate and nicotinamide riboside.

Suspecting a mitochondrial disease, rapid whole genome sequencing revealed biallelic variants in ACAD9: c.453+2T>A, paternal, pathogenic (affecting a consensus splice site); and c.1525G>A p.Gly509Ser, maternal, likely pathogenic (highly conserved amino acid, very rare in gnomAD, not previously reported, located in the C-terminal acyl-CoA dehydrogenase domain). The ACAD9 mitochondrial disorder was treated with coenzyme Q and riboflavin staring at 20 mg/kg/day in three doses.

At age 6 months 1 week, he had another acute clinical deterioration associated with a lactic acidosis crisis, pulseless cardiac arrest occurred requiring reintubation, which afterwards recovered with stabilized lactate levels while intubated and sedated, and the riboflavin dose was increased to 50 mg/kg/day. Given rising lactates at minor stressors, sedation and gabapentin were started, and further chronic alkalinization added. Outside of acute events, lactate levels remained around 3–5 mM. Three and four days later, he had two more acute episodes of profound systolic heart failure, in addition to multiple episodes of acute severe lactic acidosis, triggered by minor handling like a diaper change. The next two days, he continued experiencing multiple episodes of instability with pallor, diaphoresis, needing increasing bicarbonate and sedation, and also another pulseless cardiac arrest and started to accumulate fluids. He remained intubated, sedated, and ventilated. Cardiac evaluation showed mildly reduced cardiac contractility that exacerbated in acute crises with rapidly worsening lactic acidosis, and severe cardiac dysfunction. Given critical condition and apparent lack of responsiveness to riboflavin, new treatments were pursued. Calcium pyruvate and nicotinamide riboside chloride were started at half dose, and the next day increased to target dosing: calcium pyruvate at 900 mg/kg/day and nicotinamide riboside chloride at 300 mg 2x/day via nasogastric tube. Cardiac support was escalated with milrinone, increasing diuretics and high PEEP pressure on a ventilator, as the LV EF reduced to 18%. Investigations had not shown any other system involvement with normal ophthalmologic exam, normal renal function, and normal EEG.

Thus, at age 6 months 1 week, the child’s cardiac condition was critical and deteriorating with multiple cardiac crises and arrests. Resuscitation efforts were requested discontinued by the family, and it was anticipated that his demise was likely within the next 48 hours. In this critical situation given the rationale outlined and after obtaining emergency authorization approval from the FDA, treatment with bezafibrate was started in the evening at 100 mg 2x/day (dosing weight was 7 kg), while nicotinamide riboside chloride was continued and pyruvate discontinued. The next days, the lactate remained low and dropped to the normal range. The child remained stable, and sedation was released 4 days later with decrease in ventilatory support, followed by extubation. Liver function tests and creatine kinase levels remained normal with no evidence of nausea or gastric discomfort. Cardiac contractility improved with LV EF plateauing around 45%, and LV FS around 24%, and the NT-pro-BNP levels decreased (Figure 2). The next week, the bezafibrate dose was gradually increased to 2x/day 150 mg, then 200 mg, then 300 mg (86 mg/kg/day), while biochemically monitoring for liver or muscle toxicity, which remained normal. He remained stable while weaned off cardiac pressor support with only carvedilol remaining. He acted as a normal infant with eyes opening, and with tracking and grasping toys. Brain MRI at 7 months 1 week appeared normal with no elevated lactate peak on MRS. The lactate levels before treatment median 5.305 (IQR 3.92–8.91, range 2.0–18.54) were significantly higher than after treatment median 3.03 (IQR 2.15–3.91, range 1.24–5.19), p<0.001 (Figure 1C). His cardiac status stabilized at moderate cardiac weakness: fractional shortening 20–25%, ejection fraction (34–47%), and NT-pro-BNP levels 2540–4440. Although there was a trend for better EF and FS, this was not statistically significant (MWU p=0.73 for EF and p=0.53 for FS). The cardiac hypertrophy persisted without statistical or clinically meaningful differences (Supplemental Figure 1). He displayed much more energy with increased movements without tachycardia or shortness of breath. Additional metabolic supportive medications used included: riboflavin at 20 mg/kg/day, nicotinamide riboside chloride at 300 mg 2x/day, D,L-3-hydroxybutyrate (mixed salt) at 900 mg/kg/day) and coenzyme Q10 100 mg 2x/day (26.7 mg/kg/day). He had transient vomiting, suspected to be related to cow’s milk intolerance and resolved with hypoallergenic formula. He received continuous nasogastric drip feeding, and had swallowing dysfunction. He was slowly weaned from opiates and discharged in stable condition on 4/3/2023 at age 7.5 months.

Over the next months, his growth was normal. He was making developmental progress but had marked skeletal muscle weakness. At age 8.5 months, on the Alberta Infant Motor Scale (AIMS) he had a score of 7 (equivalent to the 5th percentile, age equivalent of a 3.25 months old). On the Bayley Scales of Infant and Toddler Development, Third Edition, at age 8.5 months, he had a poor gross motor score (age equivalent of 20 days), but good fine motor skills (age equivalent of 6 months). He had not started regular physical or occupational therapy following his prolonged hospitalization.

At age 8 months 3 weeks, he had an adenovirus gastroenteritis with vomiting and diarrhea, which resulted in decompensation with hypernatremia, lactic metabolic acidosis (max lactate > 20 mM) and cardiac decompensation (worst cardiac function: EF 27.5%, FS 17.5%, cardiac mass stable (168 g/ht^2.7), NT-pro-BNP 58,100). He was hospitalized and supported with fluids, bicarbonate, pressors, and BiPAP, and was started on enalapril. He had intermittent premature ventricular contractions, which resolved. His ketone treatment was discontinued and he was started on low dose triheptanoin, while other medications were continued. His decompensation resolved over the next days with resolution of the lactic acidosis (lactate levels 1.01–4.59 mM), improved cardiac status (NT-pro-BNP at 8010), and he was discharged at age 9 months 2 weeks. After discharge, his lactate remained stable and his NT-pro-BNP was at 4440 pg/mL.

At age 10 mo, he was admitted with a parainfluenza viral infection with fever and cough, and cardiopulmonary exacerbation requiring non-invasive high-pressure ventilation and cardiac pressor support with milrinone and adrenaline. He transiently stabilized, but remained critical with exacerbations, increasingly unresponsive to interventions, and died after two weeks from cardiovascular failure.

From a safety perspective, he had normal transaminase levels with minimal elevation only during adenoviral infection, and normal creatine kinase levels. His hemoglobin levels (8.4–13.2) had been low normal, but he also had numerous blood draws causing iatrogenic anemia. His platelet counts were normal to mildly elevated (patient range 296–573, normal 150–500). He had persistent neutropenia. Prior to treatment, absolute neutrophil numbers in 10³/μL were mean 2.32±1.20, median 2.08 (IQR 1.21–3.51); after treatment this worsened to mean 1.13±0.55, median 0.90 (IQR 0.735 – 1.365), p<0.001.

4. Results

4.1. Diagnostic studies:

In fibroblasts, respiratory chain enzyme activities (complexes I, II, III, II-III combined, IV and citrate synthase) were normal including for complex I (61.4 nmol/min/mg protein, normal 49.3–131.1, 69% of average normal, Z-score −1.1 SD) (Supplemental Table 1). The blue native PAGE with in-gel activity staining showed normal activities including of complex I activity (Figure 3A). However, the complex I assembly assay showed decreased holo-complex with added intermediate at 460 kDa (Figure 3B). The amount of ACAD9 protein in fibroblasts was markedly less abundant relative to controls, but not absent (Figure 3C). This pattern was previously observed in fibroblasts from a patient with riboflavin-responsive ACAD9 disorder, whereas the results in riboflavin non-responsive ACAD9 deficient cells had shown decreased complex I enzyme activity spectrophotometrically and on blue native PAGE. A trough riboflavin plasma level in our treated patient was 102 μg/L, with normal levels in human plasma 1–19 μg/L (Mayo medical laboratories). On quantitative proteomics analysis of the cellular proteins in the fibroblasts, the volcano plot of mitochondrial proteins showed markedly reduced ACAD9 protein, and reduction in complex I subunits (Figure 3D). On relative complex abundance (RCA) analysis, the abundance of complex I was reduced to 73% of controls after correction for mitochondrial abundance, whereas other respiratory chain enzyme complexes were unchanged (Figure 3E). Interestingly, the amount of the mitoribosomal proteins (large subunit LSU and small subunit SSU) was increased.

Figure 3: Analysis of mitochondrial function in fibroblasts.

(A) The activity and assembly of the respiratory chain enzyme complexes (I, II, IV, and V) were analyzed by blue-native PAGE followed by in-gel activity staining. The activities of all enzyme complexes including complex I were normal with normal assembly of complex V noted. (B) The assembly of complex I is followed on non-denaturing gel after western blotting and identification with an antibody against NDUFS2. Normal fibroblasts show a large amount of fully assembled complex I at 1000 KDa with a small band at 230 kDa. The cells of the patient with ACAD9 show reduced amount of the holocomplex, and an added band of incompletely assembled complex I at 460 kDa. Huh-7cells treated with chloramphenicol are shown as a positive control. (C) Western blot analysis of the ACAD9 protein using SDS-PAGE was performed in fibroblast homogenates followed by detection using enhanced chemiluminescence (ECL). The amount of patient ACAD9 protein was 52% of the average control amount (n=3). Citrate synthase (Abcam: ab129095) was used as the loading control. D. Proteomics analysis after correction of mitochondrial abundance shows a large significant decrease in ACAD9 abundance. The subunits of complex I (red) are decreased. E. Quantification of the abundance of the mitochondrial complexes shows a decrease in complex I, but not the other complexes. The p-value and the relative % of the complex in the patient compared to the controls are provided in sequential rows above the figure. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001, ns = non-significant. CI = complex I, CII = complex II, CIII = complex III, CIV = complex IV, CV = complex V, mtLSU = large subunit of the mitoribosome, mtSSU = small subunit of the mitoribosome.

4.2. Biomarkers:

The evolution of the lactate levels is shown in Figure 1A, and of NT-pro-BNP levels in Figure 1B. Key cardiac function parameters on echocardiogram are shown in Figure 2.

During admission and before the start of the treatment, GDF15 and FGF21 were both elevated. GDF15 levels were elevated to 1745 and 2233 ng/L (normal 98th percentile cutoff 874 ng/L), and decreased after treatment with combined bezafibrate and nicotinamide riboside chloride to 1085, 1282, 936, 1552 ng/L, and exacerbated during the final admission to 7402 ng/L. Before treatment, the FGF21 levels were elevated to 3616 and 5465 ng/L (normal 98th percentile cutoff 347 ng/L). After treatment, values were 4037, 1901, 2659, and 5977 ng/L, but they spiked during the final admission to 207,846 ng/L.

4.3. Pharmacokinetics and pharmacodynamics:

The pharmacokinetic study of bezafibrate showed a peak concentration of 104 μM with a half-life of 2.4 hours (Table 1). In whole blood at the time points where samples were available, the NAD metabolome was considerably expanded compared to reference values (Table 2). Specifically, the level of nicotinamide was increased by 23%, NR was increased 35 fold, while levels of NAD⁺ and NADP⁺ coenzymes were increased 3.7 and 2.9 fold respectively. We evaluated the molecular effect of this therapy using PBMCs and quantitative proteomics. While ACAD9 protein was detected well (quantified from 4–9 unique peptides) in 4 of 5 controls, it was poorly detected (quantified from 1 or 2 unique peptides in a minority of replicates) in the patient, such that no estimate on the relative reduction in protein abundance could be calculated. Analysis using SDS PAGE and western blot analysis also failed in this preparation. However, complex I protein abundance as determined from proteomics in the PBMCs was decreased compared to controls before treatment as expected, and remained low throughout treatment (Figure 4A). We also observed an overall increase in mitochondrial biogenesis reflected by an increase in the abundance of all detectable mitochondrial proteins (Figure 4B). The abundance of complex I proteins increased relative to the patient’s baseline on days 7 and 12, but this did not persist. The increase was in line with the increases seen in other mitochondrial proteins, excluding a selective effect on complex I biogenesis. (Figure 4). A persistent increase over time was not observed in this cell type.

Table 1:

Pharmacokinetics studies of bezafibrate

Time (min)	Bezafibrate (μm)
0	3.4
30	103.6
60	79.3
90	74.1
120	56.7
180	31.2
240	19.8
360	9.2

Table 2:

NAD metabolites

Metabolite	90 min	120 min	180 min	Reference values
Nicotinamide (NAM)	12.1	11.1	13.6	10.0 (7.5–11.9)
Nicotinamide riboside (NR)	5.32	4.56	7.01	0.16 (0.13–0.18)
Nicotinamide mononucleotide	1.14	0.97	1.55	1.1 (0.8–1.46)
NAD+	72.8	63.9	96.3	20.9 (15–28.2)
NADP+	19.7	17.7	20.9	6.7 (5.2–9.4)
1-methylnicotinamide	3.95	3.77	5.68	NA
Nicotinic acid riboside	0.0898	0.0762	0.0899	NA
Nicotinic acid mononucleotide	2.99	2.60	3.67	NA
Total NAD metabolome	118	105	149	NA

The NAD metabolome analytes in μM are provided in whole blood obtained after administration of nicotinamide riboside chloride 300 mg at the indicated times after administration, compared to published human adult reference values, shown as mean (range) (50). NA = not available

Figure 4: Proteomics analyses of peripheral blood mononuclear cells during treatment with bezafibrate and nicotinamide riboside.

(A) Volcano plot from quantitative proteomics data depicting reduced levels of Complex I subunits (blue) in mitochondrial proteins (gray). (B) Relative abundance of mitochondrial Complex I proteins shows decreased levels on Day 0 compared to controls. Following treatment, the abundance of Complex I increases over time compared to Day 0, except on Day 62. (C) Relative abundance of mitochondrial proteins shows increased level in the overall level of mitochondrial proteins up to Day 62 when compared to Day 0.

5. Discussion

This child presented with lactic acidosis and cardiomyopathy at age 6 months in critical condition clinically unresponsive to high dose riboflavin treatment in combination with coenzyme Q. This represents a common phenotype for ACAD9 deficiency. Riboflavin-unresponsive ACAD9 patients that present prior to 1 year of age have a very poor prognosis with high mortality in the first two years of life. The child here was critically ill with numerous critical decompensations including repeated cardiac arrests, which were triggered by even minimal caregiving acts, a situation that persisted despite multiple critical care therapeutic support measures for three weeks prior to intervention. This situation dramatically improved after start of the combined therapy of bezafibrate and nicotinamide riboside chloride, and was stable during the following three weeks of hospitalization. Comparing the three weeks before and after the start of the intervention showed clear statistical and clinically meaningful improvements in lactate, and of NT-pro-BNP. The latter progressively declined until 1 month after start of treatment. Despite the strongly reduced NT-pro-BNP with a trend for better EF and FS, this was not statistically significant, and cardiac hypertrophy persisted. The persistence of the life-threatening events before treatment, and the direct temporal association with a dramatic and persistent improvement following the initiation of treatment intervention makes a causal relationship likely. A fortuitous spontaneous clinical recovery cannot be excluded in a single case, but appears unlikely given the persistent course prior to this intervention. However, the impact long-term was less evident. He did remain with pronounced skeletal myopathy with severe muscle weakness. His NT-pro-BNP started to advance during follow-up, and the trend of the echocardiographic measures shows a clear linear decline. Ultimately, on the second infectious episode, he was unable to recover resulting in his death.

The clinical evolution suggests substantive clinical benefits, but were insufficient to improve the cardiac dysfunction over time. This response suggests that there is a pathway to clinical benefit via PGC-1α activation coupled to NAD⁺ boosting. We further hypothesize that a possible synergistic effect could exist whereby PPAR activation increases PGC-1α transcription, and NAD⁺ boosting activates PGC-1α. We suggest opportunities to improve upon this. Incomplete effectiveness by bezafibrate resulting from the weak agonist activity requiring high concentrations for therapeutic effect compared to the achievable pharmacokinetics could be improved by more effective PPAR activators. The increased cardiac mass suggests that increased mitochondrial biogenesis needs to be matched with repair of poorly formed complexes through increased mitochondrial quality control processes. Nicotinamide riboside improves mitochondrial maintenance (63,64) in mouse models, but it may have to be strengthened.

Bezafibrate increased PGC-1α and Tfam in mitochondrial cybrids, increased OXPHOS enzyme activities, ATP synthesis and oxygen consumption with less lactate production (27). It increased residual ACAD9 protein levels and the oxygen consumption in 6/7 ACAD9 cell lines in an experiment at a concentration of 400 μM (9). Surprisingly, there was no correlation between ACAD9 levels and complex I enzyme activity (9). Bezafibrate activates PPAR receptors and via PGC-1α is a known inducer of mitochondrial biogenesis. Bezafibrate is a weak pan-PPAR activator with a cell-based EC₅₀ of 50 and 60 μM for PPAR-α and PPAR-γ (65). It was shown to increase respiratory chain enzyme activities in various mitochondrial disorders at a concentration of 100 μM (27), but with optimal concentrations used in most experiments of 300 or 400 μM (9,27). These concentrations are well above peak concentration at usual dosing. In adults at 300 mg 2x/day dosing (approx. 10 mg/kg/day), the peak concentration of bezafibrate was only 25 μM (29). To achieve this higher concentration, we treated our patient with an 8 times higher dose at 80 mg/kg/day, and indeed achieved a peak concentration of 100 μM, but not a sustained elevation as would be present in the tissue culture experiments.

The heart and the liver have both PPAR-α and PPAR-δ receptors and combined stimulation would be optimal (66,67). The prominent involvement of PPAR-α makes pure PPAR-δ agonists (e.g. GW501516, bocidelpar, mavodelpar) less optimal compounds for use in this clinical condition where cardiomyopathy is such a defining feature. The weak binding activity of the pan-agonist bezafibrate makes it difficult to achieve effective therapeutic concentrations, and thus more active multiple PPAR agonists should be explored. Elafibranor (GFT505) is a dual PPARα/δ with EC₅₀ of 45 and 175 nM respectively (68); lanifibranor (IVA337) is a pan-PPAR agonist (69); and seladelpar showed the strongest induction for PGC-1α (70). All three have been used in clinical trials of liver diseases with a good safety profile in humans (71–73). Thus, these can be immediately clinical trial ready, an advantage compared to other PPAR agonists that are expected to become available later (74). These compounds should be evaluated in patient fibroblasts, and preferably also in a mouse model of a missense variant of ACAD9. For pertinent in vivo experiments, a novel mouse model with a missense variant with residual activity would be needed, as the current ACAD9 mouse model, which involves a complete null mutation, would be less appropriate to investigate the therapeutic impact of upregulation of ACAD9 residual activity (75).

Boosters of the NAD⁺ concentration have been evaluated for therapeutic potential (42, 76). In whole blood, we showed that nicotinamide riboside greatly expanded the NAD metabolome, including levels of NAD⁺ and NADP⁺ coenzymes. Among the non-mutually exclusive mechanisms of actions are correction of a cardiac NAD⁺ deficit (43), activation of SIRT1 which could stimulate PGC-1α increasing mitochondrial biogenesis (27,41), and activation of SIRT7 by NAD⁺ which could stimulate NRF2 (77) thus mitochondrial mitogenesis. Past human studies in mitochondrial patients have focused on skeletal myopathy showing significant positive effects, but not on cardiomyopathy (42,46). Notably, nicotinamide riboside boosts NAD⁺ synthesis in tissues such as liver (48), skeletal muscle (50), and cardiac muscle (43). Boosting of NAD⁺, including by nicotinamide riboside, also had cardioprotective effects in mouse models of congestive cardiomyopathy and of myocardial ischemia/reperfusion injury (43,78). In both mouse and human, cardiomyopathy and heart failure are associated with depressed NAD⁺ levels, which is accompanied by increased expression of nicotinamide riboside kinase 2(NMRK2) (43). It will need to be studied if there is indeed a synergistic effect between upregulation of PGC-1α transcription using PPAR agonists, and its activation using NAD⁺ booster therapy, which can be evaluated in patient fibroblasts, and in animal models.

The therapy appeared to be safe. Despite the high dosing used for both bezafibrate and nicotinamide riboside, no severe adverse effects were encountered. Only the preexisting neutropenia worsened, which may relate to the known bone marrow effect of bezafibrate (29–31), but still this patient responded to acute infections with raised neutrophils. Hepatotoxicity listed for several fibrates in rodents is a species-specific difference, possibly mediated by the Let-7c microRNA, that is not observed in humans or other animal species (79,80). Acute toxicity studies were done in mice and rats with an LD₅₀ for the oral route as 722–759 mg/kg in mice and 1080 mg/kg in rats. In rabbits or beagle dogs, all animals survived the maximum dose of 2500 mg/kg and in Rhesus monkeys, the acute lethal dose was 1000–2000 mg/kg. Chronic toxic effects were noted at 500 mg/kg/day or 1000 mg/kg/day. No mutagenic effect was noted in multiple assays. Lower body weights were noted in neonatal rates at 200 mg/kg/day. There is no significant tumorigenic effect noted in rats or mice in three studies. For comparison purposes, our currently highest dose of 600 mg/day reflects 86 mg/kg/day, which was well below these toxicity levels. No serious adverse effects have been documented for NAD⁺ boosters in multiple clinical trials as summarized (43). An increase in mitochondrial stress response FGF21 and GDF15 was reported in a clinical study of MELAS patients (39), but was not observed in our patient, who had elevated levels prior to treatment, possibly in association with his cardiomyopathy.

Finally, surveying pharmacodynamic effects of these interventions is difficult. This study shows that it is possible to evaluate mitochondrial complexes longitudinally using PBMCs, and the profile in PBMCs was indicative of an ACAD9 related complex I deficiency. However, the cell types involved of mainly lymphocytes and monocytes only express PGC-γ effect (26). The effect suggested a temporary improvement of the mitochondrial mass and perhaps of the complex I abundance. More direct target tissue measurements of mitochondrial function and of complex I will have to be explored. It might be possible to measure complex I and overall mitochondrial abundance using small catheter based cardiac biopsies using proteomics, but this remains an invasive procedure with risks that cannot be easily repeated.

In summary, combined treatment with high dose bezafibrate to achieve peak concentrations that are above its EC₅₀ combined with high dose nicotinamide riboside resulted in temporary improvement in the clinical status, which may be useful for patients in critical condition. It did not prevent further long-term worsening of the cardiomyopathy. We outlined several opportunities for improvement over this first signal of therapeutic efficacy. Several improved dual or pan-PPAR agonists are available with human safety documented in clinical trials. If cellular and mouse studies provide efficacy, they will be readily translatable into human clinical trials. It is worth examining if NAD⁺ boosting by nicotinamide riboside would have a synergistic effect.

6.2. Data sharing statement:

The data that support the findings of this study are available from the corresponding author upon reasonable request.