Clinical, pathological and genetic features and follow-up of 110 patients with late-onset MADD: a single-center retrospective study

Abstract

To observe a long-term prognosis in late-onset multiple acyl-coenzyme-A dehydrogenation deficiency (MADD) patients and to determine whether riboflavin should be administrated in the long-term and high-dosage manner, we studied the clinical, pathological and genetic features of 110 patients with late-onset MADD in a single neuromuscular center. The plasma riboflavin levels and a long-term follow-up study were performed. We showed that fluctuating proximal muscle weakness, exercise intolerance and dramatic responsiveness to riboflavin treatment were essential clinical features for all 110 MADD patients. Among them, we identified 106 cases with ETFDH variants, 1 case with FLAD1 variants and 3 cases without causal variants. On muscle pathology, fibers with cracks, atypical ragged red fibers (aRRFs) and diffuse decrease of SDH activity were the distinctive features of these MADD patients. The plasma riboflavin levels before treatment were significantly decreased in these patients as compared to healthy controls. Among 48 MADD patients with a follow-up of 6.1 years on average, 31 patients were free of muscle weakness recurrence, while 17 patients had episodes of slight muscle weakness upon riboflavin withdrawal, but recovered after retaking a small-dose of riboflavin for a short-term. Multivariate Cox regression analysis showed vegetarian diet and masseter weakness were independent risk factors for muscle weakness recurrence. In conclusion, fibers with cracks, aRRFs and diffuse decreased SDH activity could distinguish MADD from other genotypes of lipid storage myopathy. For late-onset MADD, increased fatty acid oxidation and reduced riboflavin levels can induce episodes of muscle symptoms, which can be treated by short-term and small-dose of riboflavin therapy.

Introduction

Lipid storage myopathy (LSM) is a heterogeneous group of disorders of lipid metabolism characterized by impaired oxidation of fatty acids, leading to an accumulation of light microscopic lipid droplets in muscle fibers. LSM usually presents with muscle weakness or exercise intolerance accompanied by recurrent episodes of rhabdomyolysis triggered by fasting, exercise or infection (1). There are four main causes of LSM listed: primary carnitine deficiency (OMIM 212140), multiple acyl-coenzymeA dehydrogenation deficiency (MADD; OMIM 231680) and neutral lipid storage disease with myopathy (NLSDM; OMIM 610717) or with neutral lipid storage disease with ichthyosis (OMIM 25630).

MADD is an autosomal recessive inherited disorder, which is also known as glutaric aciduria type II. The typical biochemical findings are a combined elevation of short-, medium- and long-chain acyl-carnitines in blood and an increase in a variety of dicarboxylic acids (such as glutarate, adipate, suberate), ethylmalonic acid and 2-hydroxyglutaric acid in urine. The phenotype of MADD is classified into three types: neonatal onset form with (type 1) or without (type 2) congenital malformation and late-onset form (type 3) (2). In the late-onset type, metabolic disturbance in most patients is rather mild, patients usually present with muscle weakness, exercise intolerance, muscle pain and lipid storage in muscle fibers that can be corrected by riboflavin treatment. Most patients are shown to have ETFDH variants (3–5).

ETFDH encodes electron transfer flavoprotein-ubiquinone oxidoreductase (ETF-QO), which is a mitochondrial integral membrane protein, consisting of three structural domains: a flavin adenine dinucleotide (FAD) domain, a 4Fe4S cluster domain and a CoQ10 binding domain (6). The binding of FAD, a metabolite of riboflavin, might be able to correct the chaperone-mediated folding defects and bring in conformational stabilization of variant ETF-QO protein caused by ETFDH missense variants (7).

Late-onset MADD is the most common cause of LSM (5,8). In the published work from our research center, one of the biggest neuromuscular centers in China, a total of 91 cases of late-onset MADD were found out of 3400 cases of muscle biopsies until 2018 (9). We also proved three frequent mutations in another work published in 2014 (5): c.250G > A (most common in South China), c.770A > G and c.1227A > C, which were totally different from Caucasian. Further haplotype analysis suggested the possibility of founder effects of c.250G > A and c.770A > G.

To date, >700 cases with riboflavin responsive MADD have been reported (10,11); however, no sufficient follow-up data have been published. In this study, we present 110 patients with late-onset MADD. We investigate clinical, pathological and genetic features and study long-term disease outcomes and prognosis.

Results

Clinical and biochemical features

The onset age of 110 MADD patients varied from 5 to 73 years, with an average of 29.3, and the sex ratio was about 0.72:1(46:64, female/male). Proximal muscle weakness and exercise intolerance were documented in all patients. The most common clinical features were characterized as difficulty in keeping their head upright (83 cases, neck muscle extensor weakness, usually with paraspinal muscle atrophy) and fatigue of masticatory muscles during chewing (67 cases). Many patients suffered from muscle pain (49 cases), muscle atrophy (48 cases) or intermittent episodes of vomiting (24 cases). Thirty patients preferred fruits and vegetables to meat or eggs and 18 patients were documented with fatty liver. Almost all of the patients had fluctuating symptoms during the course of disease. Some environmental factors (such as cold, infection, pregnancy, stressing) can make the disease deteriorate. No patients showed significant symptoms in brain, heart or kidney. Laboratory examination revealed elevated plasma creatine kinase (CK) levels ranging from 102 to 11 022 U/L with an average of 1512.1 U/L (control range 26–178 U/L). Acylcarnitine and urine organic acid tests revealed increased levels of multiple medium- and long-chain acylcarnitines in serum and a variety of dicarboxylic acids (such as glutarate, adipate, suberate), ethylmalonic acid and 2-hydroxyglutaric acid in urine, which were consistent with MADD or GAII (61 out of 69). Electromyographic measures revealed myogenic (28 out of 53), neurogenic (14 out of 53) or normal patterns (11 out of 53). The clinical characteristics of the study population are summarized in Table 1.

Table 1

Summary of clinical features of 110 patients with late-onset MADD

Clinical features		Number
Sex	Male	64/110	(58.2%)
	Female	46/110	(41.8%)
Onset age (years)
	Onset age, mean (range)	29.3	(5–73)
	≤10	6/110	(5.5%)
	11–20	34/110	(30.9%)
	21–30	25/110	(22.7%)
	31–40	22/110	(20.0%)
	41–50	14/110	(12.7%)
	≥51	9/110	(8.2%)
Disease course (months)	Disease course, mean (range)	29.1	(0.5–252)
	≤6	46/110	(41.8%)
	7–12	26/110	(23.6%)
	13–24	10/110	(9.1%)
	25–60	12/110	(10.9%)
	≥61	16/110	(14.5%)
Muscle weakness		110/110	(100%)
	Proximal limbs	110/110	(100%)
	Neck	83/110	(75.5%)
	Masseter	67/110	(60.9%)
Exercise intolerance		110/110	(100%)
Muscle pain		49/110	(44.5%)
Muscle atrophy		48/110	(43.6%)
Extramuscular Symptoms
	Vomiting	24/110	(21.8%)
	Vegetarian diet	30/110	(27.3%)
	Fatty liver	18/110	(16.4%)
Electromyography
	Myogenic changes	28/53	(52.8%)
	Neurogenic changes	14/53	(26.4%)
	Normal	11/53	(20.8%)
CK (U/L), mean (range)		1512.1 (102–11 022)
Plasma acylcarnitine and urine organic acid	Increased serum multiple medium- and long-chain acylcarnitines or glutaric aciduria type II	61/69	(88.4%)

CK: normal value: 26–178 mmol/l.

Pathology features

We observed several pathology features of MADD that could be distinguished from other types of LSM, for example, NLSDM. Hematoxylin–eosin (H&E) staining revealed a variation in fiber size and a large quantity of vacuoles within muscle fibers in patients with each type of LSM. In MADD patients, most vacuoles were irregular with cracks on H&E (Fig. 1A, arrows), whereas in NLSDM patients, vacuoles could be rather round and large (Fig. 1D, arrows), further illustrations for ‘irregular vacuoles with cracks’ and ‘round and large vacuoles’ are shown in Supplementary Material, S1. In addition, necrotic fibers and inflammatory cell infiltration could be found in NLSDM. On modified Gomori trichrome (mGT) staining, fibers with subsarcolemmally red staining mimicking ragged red fibers (atypical RRFs) could be seen in MADD patients (Fig. 1B); however, in NLSDM patients, rimmed vacuoles were common (Fig. 1E) (12). On succinate dehydrogenase (SDH) staining, a diffused decrease in enzyme activity was found in 70 MADD patients (Fig. 1H). Nevertheless, lipid accumulation in muscle fibers disappeared and diffuse decrease of SDH activity was corrected after riboflavin treatment for 7 months in a case with late-onset MADD (Fig. 1G–L). By electron microscopy, the lipid droplets were nonmembranous and deposited parallel to the myofibril (Fig. 1M and N). Mitochondria with reduced matrix density and abnormal cristae were aggregation (Fig. 1O). Some fibers with cracks were stained red on mGT resembling RRFs and were darkly stained on nicotinamide adenine dinucleotide (NADH) and cytochrome c oxidase (COX), whereas they showed decreased activity on SDH staining (Fig. 1P–U). The immunohistochemistry studies did not show any abnormality.

Figure 1

Pathological features of MADD distinguished from NLSDM. Most vacuoles were irregular with cracks on H&E in MADD patients (A), whereas vacuoles could be rather round and large in NLSDM patients (D). Fibers with atypical RRFs could be seen in MADD patients on mGT staining (B), and rimmed vacuoles were common in NLSDM patients (E). A large quantity of lipid droplets within muscle fibers in MADD and NLSDM patients on ORO staining (C and F). A diffuse decrease in enzyme activity was found in MADD patients on SDH before treatment in a case with late-onset MADD (H), whereas it was normal after riboflavin treatment for 7 months (K); meanwhile, the lipid droplets disappeared after treatment (G versus J and I versus L). By electron microscopy, the lipid droplets are nonmembranous and parallel to the myofibril. Magnification ×5000 (M) and magnification ×15 000 (N). Abnormal mitochondria aggregation was seen (O). Serial sections on H&E, mGT, NADH, COX, SDH and ORO: some fibers with cracks (★) were dark stain on NADH and COX and red stain on mGT, whereas shows decreased activity on SDH staining (P–U).

The statistical difference was significant between late-onset MADD and NLSDM on muscle pathology regarding fibers with cracks or round vacuoles on H&E (P < 0.001, separately) and diffuse decrease of SDH activity (P < 0.001) (Table 2).

Genetic features and protein analysis

Genetic analysis revealed that 106 patients with late-onset MADD carried variants in ETFDH [Table 3, a part of mutations were reported previously by our team (5,13,14) or other teams (3,4,15–40)]; 101 patients were shown to have two variants, whereas in 5 patients only a single variant was identified. One patient carried FLAD1 gene compound heterozygous variants (c.1588C > T p.R530C and c.1589G > C p.R530P, from two different homologous chromosomes).

In three patients with typical late-onset MADD phenotype including good response to riboflavin treatment, LSM, increased multiple acyl-carnitines in serum and excessive urine organic acids, Sanger sequencing (both DNA and cDNA) of MADD candidate genes and next-generation sequencing (NGS) analysis of 1012 genes associated with metabolic or muscular disease did not reveal any responsible variants (Supplementary Material, S2). ETF-QO protein expression analysis from muscle samples was done on those three cases and showed a reduction in ETF-QO protein expression in patient 3 compared with four healthy controls. However, ETF-QO protein expression was normal in patients 1 and 2 (Fig. 2).

Riboflavin levels in plasma

Riboflavin is reported to be photodegraded into lumichrome and lumiflavin; thus, we measured each chemical to calculate the total amount (41,42). The total molecules of riboflavin, lumichrome and lumiflavin were calculated and listed in Figure 3. The plasma levels of total riboflavin in MADD patients before riboflavin treatment (N = 15, 11.51 nmol/L) were significantly lower than in healthy controls (N = 30, 64.38 nmol/L, P = 0.035). After riboflavin treatment (N = 14), the plasma levels of total riboflavin (152.85 nmol/L) were much higher in MADD patients than both healthy controls and patients before treatment, P < 0.001.

Follow-up data

Follow-up data are summarized and illustrated in Table 4 and Figure 4, respectively. The follow-up time of 48 MADD patients ranged from 1 to 15 years, depending on their diagnosis time, with an average of 6.1 years. Sixteen patients were placed on corticosteroids therapy before diagnosis, and 11 of them responded temporarily well to corticosteroids. All 48 MADD patients received riboflavin treatment after diagnosis, with an average dose of 68.3 (10–150) mg/day. The average time of riboflavin treatment was 10.4 (0.5, 84) months before drug withdrawal or the endpoint of follow-up. Each of the 48 patients were clinically cured but with some persistence in abnormal acylcarnitine species.

Among 48 MADD patients with closer follow-up for 6.1 years in average (a range of 1–15 years), 31 patients were free of muscle weakness recurrence: 13 patients took no further riboflavin administration till the endpoint of follow-up after an average of 2.2 months riboflavin treatment (the longest follow-up time was 13 years), 10 patients retook small-dose of riboflavin intermittently for prevention after an average of 4.5 months regular riboflavin treatment and 8 patients continued taking regular dose of riboflavin even though the treatment session showed a satisfactory result. The other 17 patients had episodes of slight muscle weakness after discontinued riboflavin treatment; however, after they restarted taking a small dose of riboflavin for a short term, the muscle weakness symptoms disappeared (Table 4).

For the 17 patients with muscle weakness recurrence after riboflavin discontinuation, the probable triggers were exhaustion (five cases), catching a cold (four cases), alcohol (two cases), pregnancy (one case) and unknown reason (five cases). The doses of riboflavin after restarting treatment were <30 mg/day (10 patients), 40–60 mg/day (2 patients) and 90 mg/day (5 patients). The frequency of muscle weakness recurrence was less than once a year (11 patients), 2–3 times/year (4 patients). Only two patients experienced muscle weakness recurrence 4–5 times/year. The treatment time of riboflavin after recurrence was <1 week a year (12 patients), 1–4 weeks a year (2 patients) or >5 weeks a year (3 patients) (Fig. 4).

The recurrence-free survival (RFS) is the length of time after primary treatment ends that the patient survives without any signs or symptoms of the disease. For the 48 patients, the median RFS was 47.3 months (0.5–148). The recurrence rate was 0% in both continuous riboflavin medication group (8 cases) and intermittent small dose of riboflavin administration group (10 cases). The cumulative probability of survival without recurrence was 52.0% in non-continuous medication group at 5 years (40 patients including 10 patients with intermittent medication and 17 patients with recurrence and 13 drug withdrawal patients, Fig. 5A), whereas it was 33.3% in drug withdrawal group at 5 years (30 patients including 17 patients with recurrence and 13 drug withdrawal patients, RFS 0.5–60 months, Fig. 5B).

Table 5 summarizes the univariate analysis of all the prognostic factors likely to affect the development of muscle weakness recurrence. The following factors were not related to an increased risk of muscle weakness recurrence: gender, age of onset, disease course, neck muscle weakness, muscle pain, vomiting, serum CK levels, ETF-QO CoQ10 domain mutation, ETF-QO truncated mutation, doses of riboflavin treatment, time of riboflavin treatment, decreased SDH activity and atypical RRFs on muscle pathology. Longer time from drug withdrawal to follow-up (≥1 year) increased the risk of muscle weakness recurrence but not significantly (P = 0.052). The risk of recurrence was significantly higher in patients with masseter weakness (P = 0.021) and vegetarian diet (P = 0.044). As compared with the riboflavin withdrawal group, the risk of recurrence was significantly lower in groups with intermittent riboflavin doses (P = 0.008) and continuous riboflavin doses (P = 0.004). Multivariate Cox regression analysis revealed two independent parameters associated with the risk of muscle weakness recurrence: masseter weakness [hazard ratio (HR) = 0.209, 95% confidence interval (CI): 0.058–0.753, P = 0.017] and vegetarian diet (HR = 0.279, 95%CI: 0.103–0.756, P = 0.012).

Table 2

Comparison on muscle pathology features between late-onset MADD and NLSDM

	Late-onset MADD	NLSDM*	P-value
Fibers with cracks on H&E	105/110	0/42	<0.001
Round vacuoles on H&E	0/110	42/42	<0.001
Atypical RRF on mGT	86/110	0/42	<0.001
Decreased SDH activity	70/110	0/42	<0.001
Rimmed vacuoles	0/110	21/42	<0.001

^*Reported previously (19).

Table 3

In total, 106 late-onset MADD patients with ETFDH variations

No.	Mutation 1	Protein domain1	Mutation 2	Protein domain2	MADD	LSM
1	c.1227A > C p.L409F★▲	FAD	c.872 T > G p.V291G★	CoQ10	Yes	Yes
2	c.1227A > C p.L409F★▲	FAD	Transcript↓	CoQ10	No	Yes
3	c.770A > G p.Y257C★▲	FAD	c.770A > G p.Y257C★▲	FAD	Yes	Yes
4	c.1227A > C p.L409F★▲	FAD	c.1691-3C > G p.?  ★▲		Yes	Yes
5	c.1227A > C p.L409F★▲	FAD	c. 389A > T p.D130V★▲	FAD	Yes	Yes
6	c.770A > G p.Y257C★▲	FAD	c.872 T > G p.V291G★	CoQ10	Yes	Yes
7	c.770A > G p.Y257C★▲	FAD	c.973del144 p.V325_Q372del★		Yes	Yes
8	c.770A > G p.Y257C★▲	FAD	c.1084G > A p.G362R★▲	CoQ10	Yes	Yes
9	c.1227A > C p.L409F★▲	FAD	c.1281_1282delAA p.I428Rfs*6		ND	Yes
10	c.1211 T > C p.M404T★	FAD	Transcript↓		Yes	Yes
11	c.770A > G p.Y257C★▲	FAD	c. -75A > G Transcript↓★		Yes	Yes
12	c.250G > A p.A84T★▲	FAD	c.524G > A p.R175H▲	FAD	ND	Yes
13	c. 389A > T p.D130V★▲	FAD	c.891G > A p.W297fs		ND	Yes
14	c. 236C > G p.A79G★	FAD	c.1281_1282delAA p.I428Rfs*6★		ND	Yes
15	c. 389A > T p.D130V★▲	FAD	c. 835 T > C p.W279R★	CoQ10	ND	Yes
16	c. 251C > T p.A84V★	FAD	c.1691-3C > G p.?  ★▲		ND	Yes
17	c.1227A > C p.L409F★▲	FAD	c.35-23_35-20delTAAT		Yes	Yes
18	c.770A > G p.Y257C★▲	FAD	c.1281_1282delAA p.I428Rfs*6★		Yes	Yes
19	c.389A > T p.D130V★▲	FAD	c. 835 T > C p.W279R★	CoQ10	ND	Yes
20	c.389A > T p.D130V★▲	FAD	c.1205C > T p.T402I★	FAD	Yes	Yes
21	c.770A > G p.Y257C★▲	FAD	c.775A > T p.K259 fs		Yes	Yes
22	c.236C > G p.A79G ★	FAD	c.872 T > G p.V291G★	CoQ10	ND	Yes
23	c.770A > G p.Y257C★▲	FAD	ND		Yes	Yes
24	c.568 C > G p.L190V	FAD	ND		ND	Yes
25	c.380 T > G p.L127C★	FAD	c. 835 T > C p.W279R★	CoQ10	ND	Yes
26	c.236C > G p.A79G★	FAD	c.872 T > G p.V291G★	CoQ10	Yes	Yes
27	c.1211 T > C p.M404T★▲	FAD	c.1450 T > C p.W484R★▲	CoQ10	Yes	Yes
28	c.770A > G p.Y257C★▲	FAD	c. 389A > T p.D130V★▲	FAD	ND	Yes
29	c.389A > T p.D130V★▲	FAD	c. 389A > T p.D130V★▲	FAD	ND	Yes
30	c.1227A > C p.L409F★▲	FAD	c.1395 T > G p.Y465fs★▲		No	Yes
31	c.1227A > C p.L409F★▲	FAD	c.740G > A p.G247D▲	FAD	Yes	Yes
32	c.389A > T p.D130V★▲	FAD	c.389A > T p.D130V★▲	FAD	ND	Yes
33	c.601G > C p.A201P	FAD	c.1327 T > C p.W443R	FAD	Yes	Yes
34	c.998A > G p.Y333C▲	CoQ10	c.1395dupT p.G466Wfs*24▲		ND	Yes
35	c.1205C > T p.T402I★	FAD	c.1205C > T p.T402I★	FAD	Yes	Yes
36	c.229G > A p.G77S▲	FAD	c.1211 T > C p.M404T★▲	FAD	ND	Yes
37	c.389A > T p.D130V★▲	FAD	c.1281_1282delAA p.I428Rfs*6★		Yes	Yes
38	c.1205C > T p.T402I★	FAD	c.805C > G p.Q269E	CoQ10	Yes	Yes
39	c.770A > G p.Y257C★▲	FAD	c.872 T > G p.V291G★	CoQ10	Yes	Yes
40	c.250G > A p.A84T★▲	FAD	c.1227A > C p.L409F★▲	FAD	No	Yes
41	c.389A > T p.D130V★▲	FAD	c.920C > G p.S307C▲	CoQ10	Yes	Yes
42	- c.829_830insG p.E277Gfs *6		ND		Yes	Yes
43	c.770A > G p.Y257C★▲	FAD	c.1227A > C p.L409F★▲	FAD	Yes	Yes
44	c.770A > G p.Y257C★▲	FAD	c.770A > G p.Y257C★▲	FAD	Yes	Yes
45	c.380 T > G p.L127R★	FAD	c.872 T > G p.V291G★	CoQ10	No	Yes
46	c.770A > G p.Y257C★▲	FAD	c.1157G > T p.G386V▲	FAD	Yes	Yes
47	c.250G > A p.A84T★▲	FAD	c.250G > A p.A84T★▲	FAD	Yes	Yes
48	c.389A > T p.D130V★▲	FAD	c.1395 T > G p.Y465fs★▲		Yes	Yes
49	c.1227A > C p.L409F★▲	FAD	c.872 T > G p.V291G★	CoQ10	Yes	Yes
50	c.1227A > C p.L409F★▲	FAD	c.393G > C p.W131C★	FAD	Yes	Yes
51	c.380 T > G p.L127R★	FAD	c.1531G > A p.D511N★▲	CoQ10	Yes	Yes
52	c.1436G > C p.R479T★	CoQ10	c.1395 T > G p.Y465fs★▲		Yes	Yes
53	c.715G > A p.A239T★▲	FAD	c.1395 T > G p.Y465fs★▲		No	Yes
54	c.1227A > C p.L409F★▲	FAD	c.872 T > G p.V291G★	CoQ10	ND	Yes
55	c.770A > G p.Y257C★▲	FAD	c.770A > G p.Y257C★▲	FAD	ND	Yes
56	c.242 T > C p.L81P★	FAD	c.487-9 T > C p.?		Yes	Yes
57	c.1227A > C p.L409F★▲	FAD	c. 389A > T p.D130V★▲	FAD	ND	Yes
58	c. 1295 T > A p.V432G★	FAD	c.C1528T p.P510S★	CoQ10	ND	Yes
59	c.770A > G p.Y257C★▲	FAD	c.1531G > A p.D511N★▲	CoQ10	ND	Yes
60	c. 389A > T p.D130V★▲	FAD	c.1395 T > G p.Y465fs★▲		ND	Yes
61	c.1227A > C p.L409F★▲	FAD	c.250G > A p.A84T★▲	FAD	ND	Yes
62	c.770A > G p.Y257C★▲	FAD	c.872 T > G p.V291G★	CoQ10	Yes	Yes
63	c.770A > G p.Y257C★▲	FAD	c.1372_1375delTGCC p.C458Tfs*9★		ND	Yes
64	c.256C > T p.R86C★	FAD	c.175 + 2 T > C p.?  ★		ND	Yes
65	c.770A > G p.Y257C★▲	FAD	c.1099A > G p.N367D★	CoQ10	ND	Yes
66	c.250G > A p.A84T★▲	FAD	c. 528G > C p.L176F★	FAD	ND	Yes
67	c.770A > G p.Y257C★▲	FAD	c.1027 T > C p.W343R★▲	CoQ10	ND	Yes
68	c. 242 T > C p.L81P★	FAD	c.606 + 7A > G p.?  ★		ND	Yes
69	c.389A > T p.D130V★▲	FAD	c.1395 T > G p.Y465fs★▲		ND	Yes
70	c.389A > T p.D130V★▲	FAD	c. 503A > G p.N168S★	CoQ10	ND	Yes
71	c.770A > G p.Y257C★▲	FAD	c. 1448C > T p.P483L▲	CoQ10	ND	Yes
72	c.1227A > C p.L409F★▲	FAD	c. 503A > G p.N168S★	CoQ10	ND	Yes
73	c.1227A > C p.L409F★▲	FAD	c. 692 T > C p.F231S★	FAD	ND	Yes
74	c.770A > G p.Y257C★▲	FAD	c.1395 T > G p.Y465fs★▲		ND	Yes
75	c.1227A > C p.L409F★▲	FAD	c. 524G > T p.R175L▲	FAD	ND	Yes
76	c.770A > G p.Y257C★▲	FAD	c. 1099A > G p.N367D★	CoQ10	ND	Yes
77	c.770A > G p.Y257C★▲	FAD	c. 1378G > T p.G460fs★		ND	Yes
78	c.770A > G p.Y257C★▲	FAD	c. 1026 G > T p.R342S★	CoQ10	ND	Yes
79	c.389A > T p.D130V★▲		c.361insT p.P121L fs*4		ND	Yes
80	c. 353 G > T p.C118F★	FAD	c. 1657 T > C p.Y553H★▲	4Fe4S	ND	Yes
81	c.1227A > C p.L409F★▲	FAD	c.210delTGT p.V71del		ND	Yes
82	c.405 + 1G > A p.?  ★		ND		Yes	Yes
83	c.1227A > C p.L409F★▲	FAD	c.998A > G p.Y333C▲	CoQ10	Yes	Yes
84	c.1205C > T p.T402I★	FAD	c.1411A > G p.T471A	CoQ10	Yes	Yes
85	c.389A > T p.D130V★▲	FAD	c.872 T > G p.V291G★	CoQ10	Yes	Yes
86	c. 835 T > C p.W279R★	UQ	ND		Yes	Yes
87	c.295C > T p.R99C▲	FAD	c.826 T > G p.V291G	CoQ10	Yes	Yes
88	c.1227A > C p.L409F★▲	FAD	c. 353G > T p.C118F★	FAD	Yes	Yes
89	c.152G > A p.R51Q★	FAD	c.152G > A p.R51Q★	FAD	No	Yes
90	c.770A > G p.Y257C★▲	FAD	c.872 T > G p.V291G★	CoQ10	Yes	Yes
91	c.1227A > C p.L409F★▲	FAD	c.1084G > A p.G362R★▲	CoQ10	Yes	Yes
92	c.250G > A p.A84T★▲	FAD	c. 389A > T p.D130V★▲	FAD	Yes	Yes
93	c.1227A > C p.L409F★▲	FAD	c.1227A > C p.L409F★▲	FAD	Yes	Yes
94	c.3G > C p.?  ★		c.1454C > G p.T485S★	CoQ10	Yes	Yes
95	c.250G > A p.A84T★▲	FAD	c. 528G > C p.L176F★571	FAD	Yes	Yes
96	c.250G > A p.A84T★▲	FAD	c.389A > T p.D130V★▲	FAD	Yes	Yes
97	c.3G > C p.?  ★		c.1454C > G p.T485S★	CoQ10	Yes	Yes
98	c.1227A > C p.L409F★▲	FAD	c.1211 T > C p.M404T★▲	FAD	Yes	Yes
99	c.770A > G p.Y257C★▲	FAD	Exon6 del		Yes	Yes
100	c.770A > G p.Y257C★▲	FAD	c.1448C > T p.P483L▲	CoQ10	Yes	Yes
101	c.1531G > A p.D511N★▲	CoQ10	c.1395 T > G p.Y465fs★▲		No	Yes
102	c.389A > T p.D130V★▲	FAD	c.1211 T > C p.M404T★▲	FAD	Yes	Yes
103	c.770A > G p.Y257C★▲	FAD	c.1531G > A p.D511N★▲	CoQ10	Yes	Yes
104	c.770A > G p.Y257C★▲	FAD	c.250G > A p.A84T★▲	FAD	Yes	Yes
105	c.251C > T p.A84V★	FAD	c.1531G > A p.D511N★▲	CoQ10	Yes	Yes
106	c.770A > G p.Y257C★▲	FAD	c.1211 T > C p.M404T★▲	FAD	No	Yes

^★Mutations were reported previously by our team.

^▲Mutations were reported previously by other teams.

Patients 1–48 had complete data of followed up.

ND: not detected.

Discussion

In this study, we reported the clinical, pathological and genetic features along with long-term follow-up of 110 patients with late-onset MADD in our single research center. Interestingly, it was observed that each patient with late-onset MADD was relieved by a single-dose riboflavin treatment and most of them did not need continuous high-dose riboflavin medication. We proved that ETFDH gene variants are the main, but not the only genetic basis for those patients. Furthermore, some pathological features, such as fibers with cracks or round vacuoles on H&E and decreased SDH activity, could differentiate MADD patients from other types of LSM before genetic analysis. Moreover, our studies revealed that reduced riboflavin levels in serum may play an important role in the pathogenesis of late-onset MADD.

It is well-known that late-onset MADD is well responsive to riboflavin, and the response rate is as much as 98.4% (8); however, the dose and course of treatment are on debate, and no systematic evaluation has been reported to date. The common view supports high dose and long-term riboflavin treatment, usually 100–400 mg/day (3). In this study, one patient only took riboflavin 10 mg/day for 2 months, then he was cured and stopped the medication. Although during the following 9 years he had three attacks of muscle weakness, for each recurrence, he took riboflavin only 5 mg/day for <1 week and he was back to normal strength again. In the present group of MADD patients, the average dose of riboflavin is 68.3 mg/day, which is much <100–400 mg/day reported in previous publications (3,43). As shown in this study, small-dose medication of riboflavin may work well for late-onset MADD patients. Further studies are needed to clarify whether different mutations require different treatments (44).

We showed that neither dose nor time of riboflavin treatment increased the risk of muscle weakness recurrence according to statistical analysis. Moreover, the risk of recurrence was equally low in both the intermittent medication group and the continuous treatment group. Therefore, if the patient couldn’t insist on taking medicine, short term and small dose of riboflavin therapy could reduce the risk of recurrence as well. Masseter weakness and vegetarian diet were independent risk factors of muscle weakness recurrence, and both masseter weakness and vegetarian diet may increase the risk of malnutrition and riboflavin deficiency. Therefore, patients with those manifestations should be more cautious and take steps such as intermittent and small dose of riboflavin treatment to prevent recurrence. As this was an observation study limited in a single research center, prospective studies from multicenters are needed to support this conclusion.

Why a great number of late-onset MADD patients do not need long-term riboflavin treatment? Traditional hypotheses on the therapeutic mechanisms of riboflavin treatment suggested a common thread, where riboflavin treatment increases mitochondrial FAD content and thereby compensates for an impaired binding of FAD with mutated flavoproteins. Efficient binding of FAD is crucial for the catalytic activities of flavoproteins and for their correct folding and stability, which provides a basis for understanding high dose and long-term riboflavin treatment (7,45). Our latest findings suggested that FAD homeostasis disturbance was a potential mechanism of MADD disease onset in tissue with ETFDH variants and riboflavin deficiency (9). An increase in FAD-binding flavoproteins after riboflavin supplementation could release more FAD to mitochondrial matrix during degradation, keeping a larger circulating FAD pool even after discontinuation of riboflavin treatment. This may partially explain why short-term riboflavin supplementation brought a long period of resolution of symptoms in our study. Further studies are needed to support this hypothesis.

Although both MADD and NLSDM are associated with excessive lipid droplets in muscle fibers, their muscle pathology differs. In MADD patients, most vacuoles are irregular with cracks and mainly present in type I fibers. Also, MADD patients show atypical RRFs on mGT staining along with diffuse decreased SDH activity. If we observe excessive irregular lipid droplets with cracks and atypical RRFs along with diffuse decreased SDH activity, combined with typical clinical manifestations and biochemical profiles, riboflavin therapy should be given immediately. On the contrary, if vacuoles are rather round and big, seen also in type II fibers in great numbers, and rimmed vacuoles are found on mGT staining, then NLSDM is the most likely diagnosis. In MADD, irregular vacuoles and cracks could be caused by intermyofibrilar lipid droplets deposition, as evidenced by electron microscopy, whereas in NLSDM lipid droplets appear bigger because of the fact that smaller ones are degraded by lipophagy, an alternative to conventional cytosolic lipase-driven lipid droplets breakdown (46).

In three patients with typical clinical picture of late-onset riboflavin responsive MADD, we have screened classic MADD genes (ETFA, ETFB and ETFDH, both DNA and cDNA), riboflavin transporter genes (RFT1, RFT2, RFT3 and MFT), FAD synthesis-related genes (RFK, FLAD1) acyl-CoA dehydrogenases genes (short chain, medium chain, long chain and very long chain), other LSM genes (PNPLA2, ABHD5, SLC22A5 and CPT2) as well as another 1000 genes related to muscular diseases, with no positive results. Analysis of ETF-QO protein expression revealed a reduction in muscle from patient 3 suggesting that there might be some defect in protein translation, folding or stability because DNA and RNA sequence studies for ETFDH gene were normal. The two other patients had normal ETF-QO protein expression comparable to controls. A few single cases of myopathic patients with MADD-like acylcarnitines associated with mtDNA variants have been reported (47,48). As such, we cannot exclude that the three patients may have variants in mtDNA, although we failed to find any fibers with COX deficiency in muscle pathology and ruled out nDNA affecting respiratory chain function by our NGS analysis.

Figure 2

ETF-QO protein expression in three patients with no genetic findings. ETF-QO protein showed a reduced expression in patient 3, whereas normal expression was seen in patients 1 and 2. The figure is a representative immunoblot from four experiments.

Figure 3

Riboflavin levels in plasma in MADD patients before and after riboflavin treatment. The plasma levels of total riboflavin in MADD patients before riboflavin treatment were significantly lower than healthy controls (P = 0.035), whereas it was much higher in those after riboflavin treatment (P < 0.001). (A) Column chart for riboflavin levels in plasma. (B) Scatter diagram for riboflavin levels in plasma.

Table 4

Summary of follow-up data for 48 patients with late onset MADD

Features		Number
Follow-up (years), mean (range)		6.1	(1–15)
Corticosteroids before diagnosis		16/48	(32%)
	Well response	11/16	(68.8%)
Riboflavin treatment after diagnosis
Dose (mg/day), mean (range)		68.3	(10–150)
	≤30 mg/day	15/48	(31.3%)
	40–60 mg/day	12/48	(25.0%)
	75–90 mg/day	17/48	(35.4%)
	120–150 mg/day	4/48	(8.3%)
Course (months), mean (range)		10.4	(0.5–84)
	≤1 month	11/48	(22.9%)
	1–3 months	19/48	(39.6%)
	4–12 months	10/48	(20.8%)
	Continuous (until the endpoint of follow-up)	8/48	(16.7%)
Outcome
	Cured	48/48	(100%)
Riboflavin administration after cured
	Discontinuous	40/48	(83.3%)
	Continuous	8/48	(16.7%)
Muscle weakness recurrence
	Yes	17/48	(35.4%)
	No	31/48	(64.6%)
No recurrence in 31 cases
No further riboflavin treatment		13/31	(41.9%)
Small dose and intermittent riboflavin treatment for prevention		10/31	(32.3%)
Continuous riboflavin treatment although with a satisfied remission		8/31	(25.8%)
Muscle weakness recurrence resulting in   retaking riboflavin in 17 patients
Frequency	≤Once a year	11/17	(64.7%)
	2–3 times per year	4/17	(23.5%)
	4–5 times per year	2/17	(11.8%)
Dose
	≤30 mg/day	10/17	(58.8%)
	40–60 mg/day	2/17	(11.8%)
	90 mg/day	5/17	(29.4%)
Time
	≤1 week/year	12/17	(70.6%)
	1–4 week/year	2/17	(11.8%)
	≥5 week/year	3/17	(17.6%)
Causes of weakness
	Exhaustion	5/17	(29.4%)
	Cold	4/17	(23.5%)
	Alcohol	2/17	(11.8%)
	Pregnancy	1/17	(5.9%)
	No causes	5/17	(29.4%)

In our large cohort of patients with late-onset MADD, we learned three facts: in the background of lipid metabolic defects because of ETFDH gene variants, the significantly decreased plasma levels of riboflavin and the possible triggers (such as exhaustion, cold, excessive alcohol or pregnancy) are essential to muscle weakness onset, which were also proved in our mice study (9). Therefore, ETFDH deficiency, riboflavin deficiency and environmental/catabolic stress might contribute to the pathogenesis of MADD synergistically (32).

In conclusion, we presented 110 patients with late-onset MADD, which is the largest single-center cohort of MADD patients to date. On muscle pathology, fibers with cracks, atypical RRFs and diffuse decreased SDH activity could differentiate MADD from other genotypes of LSM. Follow-up data on 48 patients showed that for late-onset MADD, increased fatty acid oxidation and reduced riboflavin levels can induce episodes of muscle symptoms, which can be treated by short-term and small-dose medication of single riboflavin therapy. ETFDH gene variants, low riboflavin levels in plasma and increased fatty acid oxidation by environmental stress may synergistically act on the pathogenesis of MADD.

Materials and Methods

Participants

We performed retrospective study on patients with a diagnosis of late-onset MADD obtained at Research Institute of Neuromuscular and Neurodegenerative Diseases, Qilu Hospital, Shandong University between 1995 and 2019. From 4265 muscle biopsies with myopathy, a total of 110 patients with late-onset MADD were identified, in which 48 patients were followed up at least for 1 year, and 59 patients were previously published only with genetic data by our previous work (5,13,49,50).

The diagnosis of late-onset MADD was made on the basis of clinical manifestation, excessive lipid droplets on muscle pathology, increased multiple acyl-carnitines in serum and genetic investigations (13). Muscle pathology and genetic studies were carried out for each of 110 patients, whereas 69 of them had acyl-carnitine profile and urine organic acid analysis. However, for three patients without gene mutations, the diagnosis of MADD was made according to good response to riboflavin, muscle pathology, increased multiple acyl-carnitines in serum and excessive urine organic acids. Muscle strength was assessed using the Medical Research Council grading scale. Patients were followed up through outpatient service and telephone consultation once a year. A standardized template was used for data collection during clinic visits. Ethics approval was provided by the Medical Ethics Committee of Qilu Hospital. All the participants granted written informed consent.

This study is real world and retrospective. MADD is a rare disease and those 110 patients in this study were collected during the last two decades. In the process of understanding this disease, we tried different initial treatment doses of riboflavin to determine the best therapeutic protocol for late-onset MADD, and to our surprise, all doses of riboflavin worked well. For this reason, the initial treatment doses of riboflavin varied. Although we suggested continuous riboflavin treatment, most patients discontinued riboflavin intake after they were clinical cured and might retake riboflavin all by their own decision when they got recurrence of muscle weakness. Therefore, the treatment time for riboflavin varied as well.

The statistical analyses were performed by SPSS for windows (Version 25.0) on a personal computer. RFS was evaluated using the Kaplan–Meier method. Cox regression analyses were used to identify the factors independently associated with muscle weakness recurrence. The categorical variables were studied using Chi-square analysis. The numerical variables were studied using t-test. Two-sided P-values were computed, and P < 0.05 was considered statistically significant.

Muscle pathology

Open muscle biopsy in biceps brachii was carried out in all patients under local anesthesia. Muscle specimens were frozen in isopentane that was precooled in liquid nitrogen and stored at −80°C. For histological examination, serial frozen sections (8 mM) were stained with H&E, oil red O (ORO), SDH, mGT, NADH-tetrazolium reductase (NADH-TR), COX and periodic acid Schiff. For immunohistochemistry studies, serial frozen sections (5 mM) were incubated with dystrophin, dysferlin, α-sarcoglycan, β-sarcoglycan, γ-sarcoglycan, δ-sarcoglycan and caveolin-3 antibodies.

For electron microscopic analysis, muscle tissue was fixed as described previously (51).

Figure 4

Riboflavin treatment for 48 patients with late onset MADD. (A) The doses of riboflavin treatment for 48 patients. (B) The time of riboflavin treatment for 40 patients before first discontinuation. (C) The probable causes of muscle weakness in 17 patients with recurrence after discontinuation. (D,E) The doses of retaking riboflavin in 17 patients with recurrence. (F) The frequency of muscle weakness in 17 patients with recurrence. (G,H) The time of retaking riboflavin medication in 17 patients with recurrence.

Molecular studies and western blot analysis

Genomic DNA-based PCR amplification and sequence analysis of LSM-related genes such as ETFA, ETFB, ETFDH, PNPLA2, FLAD1, MFT, ACADM and ACAD9 were performed. Genomic DNA was extracted from frozen muscle biopsy specimen with a genomic DNA extract kit (Tiangen, PR China). Control DNA from unaffected Chinese individuals was obtained from peripheral blood lymphocytes. PCR amplifications of all coding regions and intron–exon boundaries were carried out under standard conditions, and the PCR products were subjected to direct bidirectional sequencing. The initiating ATG codon is numbered as bp 1–3, and the initiating methionine is numbered as amino acid 1.

For three patients, we could not find any pathogenic variants in the LSM-related genes by Sanger sequencing. In those patients, the NGS using the target area capture technology and high-throughput sequencing technology was applied (RunningGene Inc., Beijing, China). Meantime, the RNA was extracted with TRIzol Reagents (Invitrogen) from frozen muscle samples derived from the three patients and reverse transcription was performed with a RT-PCR kit (Tiangen, PR China), followed by cDNA sequence and expression analysis of ETFA, ETFB and ETFDH.

Figure 5

RFS analysis by the Kaplan–Meier method. (A) The recurrence rate was 0% in continuous riboflavin medication group (8 cases), and the cumulative probability of survival without recurrence was 52.0% in noncontinuous medication group at 5 years (10 intermittent administration patients, 17 patients with recurrence and 13 drug withdrawal patients). (B) The recurrence rate was 0% in both continuous riboflavin medication group (8 cases) and intermittent riboflavin administration group (10 cases), and the cumulative probability of survival without recurrence was 33.3% in drug withdrawal group at 5 years (17 patients with recurrence and 13 drug withdrawal patients.

Western blot analysis of protein extracted from muscle homogenate was performed with a polyclonal antibody raised toward ETF-QO (Protein Tech Group, USA) and a monoclonal antibody raised toward a-tubulin (Santa Cruz Biotechnology, USA). The procedures for protein extraction, measurements of protein concentration and western blot analysis have been described elsewhere (52).

Isotope dilution method for determination of riboflavin, lumiflavin and lumichrome in plasma using liquid chromatography–tandem mass spectrometry

Riboflavin and lumichrome were obtained from Sigma-Aldrich (USA), and lumiflavin was obtained from Toronto Research Chemicals Inc. (Canada). Riboflavin-dioxopyrimidine-¹³C4,¹⁵N2 and lumichrome-D8, to be used as internal standard, were obtained from Toronto Research Chemicals Inc. (Canada). Plasma samples were obtained and stored at −80°C. The samples and operations were kept out of light.

Table 5

Results of univariate Cox regression analysis of the prognostic factors potentially associated with recurrences

Prognostic factors		Cases	Recurrences	HR (95% CI)	P-value
Gender
	Female	17	9	1.0
	Male	31	8	0.309 (0.089–1.076)	0.060
Age of onset
	<30 years	29	10	1.0
	≥30 years	19	7	0.902 (0.270–3.015)	0.867
Disease course
	<8 months	22	8	1.0
	≥8 months	26	9	1.079 (0.330–3.535)	0.900
Neck muscle weakness
	Yes	38	15	1.0
	No	10	2	2.609 (0.486–14.004)	0.252
Masseter weakness
	Yes	29	14	1.0
	No	19	3	4.978 (1.189–20.845)	0.021
Muscle pain
	Yes	24	11	1.0
	No	24	6	2.538 (0.746–8.633)	0.131
Vegetarian diet
	Yes	19	10	1.0
	No	29	7	3.492 (1.012–12.051)	0.044
Vomiting
	Yes	8	3	1.0
	No	40	14	1.114 (0.231–5.366)	0.893
Serum CK (U/L)
	<1000	28	9	1.0
	≥1000	20	8	0.711 (0.215–2.349)	0.575
CoQ10 domain mutation on ETF-QO protein
	Yes	13	3	1.0
	No	35	14	0.450 (0.105–1.931)	0.276
Truncated mutation on ETF-QO protein
	Yes	15	5	1.0
	No	33	12	0.875 (0.242–3.168)	0.839
Doses for riboflavin treatment
	<60 mg/day	23	8	1.0
	≥60 mg/day	25	9	0.948 (0.290–3.100)	0.930
Time of riboflavin treatment
	<3 months	22	7	1.0
	≥3 months	26	10	0.747 (0.226–2.468)	0.632
Riboflavin withdrawal
	Yes	30	17
	Intermittent and small doses	10	0		0.008
	Continuous medication	8	0		0.004
Time from drug withdrawal to follow-up
	<1 year	8	0
	≥1 year	40	17		0.052
Decreased SDH activity on muscle pathology
	Yes	31	11	1.0
	No	17	6	1.320 (0.368–4.731)	0.670
Atypical RRFs on mGT on muscle pathology
	Yes	38	12	1.0
	No	10	5	0.462 (0.112–1.901)	0.278

A stock solution with 1.0 mg/ml concentration and work solutions with 0.1; 1.0; 2.0; 5.0; 10.0; 50.0; 100.0; 200.0 ng/ml concentrations of riboflavin, lumichrome and lumiflavin were prepared in ultrapure water, for standard curve. A riboflavin-dioxopyrimidine-¹³C4,¹⁵N2 stock solution with 1.0 mg/ml concentration was prepared in 5 mM potassium hydroxide. A lumichrome-D8 stock solution with 1.0 mg/ml concentration was prepared in dimethyl sulfoxide (DMSO). The mixture internal standard work solution of riboflavin-dioxopyrimidine-¹³C4,¹⁵N2 and lumichrome-D8 at 10.0 ng/ml was prepared in methanol-0.1% formic acid. All stock and working solutions were stored at −20°C. The concentration of the lower and upper limit of calibration standards was proposed on the basis of a reference value for vitamin B2 in plasma proposed in literature (53,54).

The content of chemicals was determined using a Sciex API 6500 Plus liquid chromatography-tandem mass spectrometry (LC–MS/MS) system. About 50 μl of calibrator, quality controls (the low-, medium- and high-quality controls at three different concentrations) and plasma samples were spiked with 200 μl work solution of internal standard (riboflavin-dioxopyrimidine-¹³C4,¹⁵N2, lumichrome-D8). The mixture was stirred vigorously for 10 min using a vortex apparatus and centrifuged for 5 min at 14 000 rpm. The clear supernatant was transferred into a vial and 100 μl of water-0.1% formic acid was added. Then stirred for 5 min and centrifuged for 2 min at 14 000 rpm. Finally, 5.0 μl of solution were injected in the chromatographic system (AB SCIEX 6500+ LC–MS/MS).

Acknowledgements

We thank the neurologists that have followed up the patients in this study.

Conflict of Interest statement. None declared.

Funding

National Natural Science Foundation of China (82071412), the Youth Program of National Natural Science Foundation of China (81701058), the policy supported projects of collaborative innovation and achievement transformation in universities and research institutes of Jinan (2019GXRC050), the Taishan Scholars Program of Shandong Province; Qingdao Key Health Discipline Development Fund.