Differential diagnosis of vacuolar myopathies in the NGS era

Clinical examination

In the 32 index patients as well as in 36 additional family members a detailed neurological examination was carried out on several occasions by a neurologist with neuromuscular specialization (AF). In all 68 individuals (patients and family members) serum creatine kinase (CK) levels were measured repeatedly. Needle electromyography (EMG) investigation was carried out in all but two index patients and in four of the family members. In selected cases we performed nerve conduction studies. In 27 of the 32 index patients and in three of the family members muscle MRI of the lower extremities and of the pelvic girdle was performed using axial and coronal T1, T2 and STIR sequences.

Muscle biopsy

Open muscle biopsies were obtained for diagnostic purposes from mildly to moderately affected muscles based on the clinical features and MRI findings. In four cases, additional sural nerve biopsy was performed. Unfixed (cryostat), formaldehyde‐fixed (paraffin) and glutaraldehyde‐fixed (semi‐ and ultrathin) sections were stained using standard histological, enzyme histochemical and immunohistochemical methods (⁶⁵). Electron microscopy (EM) was performed in 28 out of 32 cases using a Philips CM10 transmission electron microscope; in four patients glutaraldehyde‐fixed muscle tissue for EM was not available. To validate the pathogenicity of detected genetic variants light and electron micrographs were revisited and in selected cases additional targeted immunohistochemistry and further EM was performed. A more detailed description of our neuropathological workflow is provided in Suppl. Methods.

Genetic testing

Written informed consent for genetic testing was obtained from all patients and family members involved in this study. Targeted gene analysis via Sanger sequencing was performed mainly at the Institute of Human Genetics, Würzburg. NGS‐based neuromuscular panel diagnostics was carried out mainly at the Centre for Genomics and Transcriptomics CeGaT (Tübingen, Germany). Custom‐design in‐solution hybridization (Agilent SureSelect, Agilent) was used to select panel genes. Over the course of this study, the panel was expanded several times ranging from 98 genes (version 2a) to 317 genes (version 4). Sequencing was performed on SOLiD s5500xl systems (Life Technologies) or Illumina HiSeq2500 and HiSeq4000 systems to an average read depth between 750× and 1400× per sample. SOLiD color‐space reads were aligned and variants called using lifescope (Life Technologies). HiSeq reads were trimmed (cutadapt 1.4.1 or Skewer 0.1.116) and aligned using bwa mem (0.7.2) against the hg19 reference genome. PCR duplicates were removed (samtools 0.1.18), and only uniquely aligned reads kept for variant calling (CeGaT proprietary software). Variants were called using samtools and VarScan (2.3.5 or 2.4.2). One case was diagnosed via NGS‐neuromuscular panel diagnostics at the Medical Genetics Centre, MGZ (Munich, Germany) that used Agilent SureSelectXT, IlluminaR sequencing technology. Patients and their relatives with non‐conclusive genetic workup were followed‐up for further examination and whole exome sequencing (WES). WES was performed at the Institute of Human Genetics, RWTH Aachen University Hospital. For whole WES enrichment, the Nextera Rapid Capture Exome (v1.2, Illumina, SanDiego, CA, USA) was applied. The exome library was sequenced on a NextSeq500Sequencer with 2x75 cycles on a high‐output flow cell. De‐multiplexing and fastq file generation was done using bcl2fastq2 (v.2.2, Ilumina, San Diego, CA, USA). The generation of the analyzed "bam,” “FastQ” and “vcf” files was carried out with an in‐house pipeline based on SeqMule (v.1.2.6.) (reference genome: hg19). The raw data were analyzed using the variant annotation software kggseq (v1, http://grass.cgs.hku.hk/limx/kggseq/). Variants with a minor allele frequency (MAF) of >0.75% in relevant databases (ie, EXAC, dbSNP, 1000G), synonymous variants and intronic variants that do not affect the canonical splice‐sites were not considered. Bioinformatics pathogenicity assessment (PolyPhen‐2, MutationTaster, SIFT and CADD) was performed, considering that this method only indicates a possible pathogenicity. Mean coverage depth was app. 75X. Segregation analysis was carried out in 29 of the family members.

General histopathological findings

Sixteen patients showed moderate (affecting <10%), 10 marked (10%–50%) and six severe (>50% of the muscle fibers in the sections examined) myopathic changes including myofibrillar disintegration, non‐subsarcolemmal nuclei, atrophy, basophilia and necrosis. By light microscopy, 16 biopsies exhibited numerous (>10) (rimmed) vacuoles, 10 biopsies had light (¹, ², ³) to moderate (³, ⁴, ⁵, ⁶, ⁷, ⁸, ⁹, ¹⁰) vacuolar defects in the sections examined; in six patients a considerable vacuolar myopathy characterized by large abnormal autophagic vacuoles was evident on the ultrastructural level only (Table ​(Table3).3). In 18 cases, there was additional neurogenic muscle fiber atrophy. Sural nerve biopsy revealed a predominantly axonal neuropathy with minor demyelinating component in three patients and a predominantly demyelinating neuropathy in one patient.

Vacuolar myopathies with identified pathogenic mutation

GNE myopathy (Hereditary inclusion body myopathy)

P15 who was of Turkish descent had difficulty in climbing stairs at age 23 and was wheelchair bound 13 years later. She developed high‐grade paraparesis and plegia of hip flexion and of foot dorsi‐ and plantarflexion, while quadriceps function was largely spared. Muscle biopsy showed severe vacuolar myopathy with rimmed vacuoles often extending along several sarcomeres (Figure ​(Figure1A).1A). Sanger sequencing revealed the homozygous GNE c.1760T>C, p.(Ile587Thr) mutation of known pathogenicity (²⁰).

Open in a separate window

Figure 1

Characteristic morphological findings in vacuolar myopathies. A. Prominent vacuoles containing granular osmiophilic material (arrows) in GNE myopathy (P14). Semithin section, toluidine blue. Scale bar = 40 µm. B,C. Large osmiophilic membranous and granular inclusions indicative of altered autophagy combined with granulofilamentous material and other myofibrillar breakdown products (asterisks) in desminopathy (P7). EM. Scale bar in B = 3 µm, in C = 1.5 µm. D. Filamentous bundles characteristic for ZASPopathy in P10. EM. Scale bar = 3 µm. E. Large autophagic vacuole containing pleomorphic granular and membranous material in a case of Bethlem myopathy (P11) caused by COL6A2 mutation. EM. Scale bar = 2 µm. F. Prominent endomysial fibrosis with concentric accumulation of collagen and of fibroblasts around individual muscle fibers in P11. Paraffin section, Coll VI immunohistochemistry (brown). Scale bar = 30 µm.

TRIM32 (Sarcotubular myopathy)

P1.1 suffered from weakness of the leg and back muscles starting at age 27 and developed plegia of hip flexion and knee extension over 16 years. His older brother P1.2 who showed similar symptoms at a later age was wheelchair bound at the age of 53. Initial biopsy analysis yielded marked dystrophic features. Linkage analysis of the family including the healthy parents and three additional healthy brothers suggested TRIM32 as a candidate gene. Panel NGS of DNA of P1.1 revealed the heterozygous c.467T>C, p.(Leu156Pro) TRIM32 variant (rs145907585) of unknown significance. This is a rare variant with a minor allele frequency <0.01% that is rated as pathogenic by Mutation Taster and Polyphen 2 and is located within a highly conserved region (GERP 5.36). By quantitative RTqPCR, an additional heterozygous deletion of exon 2, the only coding exon of TRIM32, was detected. Subsequent analysis of new semithin sections of the biopsies of P1.1 and P1.2 revealed rimmed vacuoles and many vacuoles located in rows in markedly affected muscle fibers (Figure 2A,B). In electron micrographs vacuoles were associated with autophagic membranous and tubular structures and focal myofibrillar disintegration (Figure 2C,D), a pattern typical for sarcotubular myopathy (⁴⁶, ⁶²). Gene panel NGS also yielded the heterozygous MYH7 variant c.4258C>T, p(Arg1420Trp) which has been described previously in patients with a cardiomyopathy (⁹⁷). However, the limb‐girdle phenotype, the AR mode of inheritance in this family and the lack of morphological alterations typical for MYH7 mutations such as myosin storage/hyaline bodies and cores/minicores or fiber type disproportion (²⁵, ⁹⁰) argued against the diagnosis of MYH7‐related myopathy. Moreover, subsequent analysis of DNA of the affected brother P1.2 showed the same TRIM32 genotype as found in P1.1. The five unaffected family members mentioned above underwent segregation analysis: The father carried the heterozygous p.(Leu156Pro) TRIM32 variant, while the mother and another brother harbored the TRIM32 deletion, confirming that one normal copy of TRIM32 is sufficient for proper gene function. Two other brothers harboured neither the exon 2 deletion nor the p.(Leu156Pro) mutation. We conclude that the combination of the p.(Leu156Pro) TRIM32 variant and the deletion of exon 2 cause LGMD2H (LGMDR8).

Open in a separate window

Figure 2

Rimmed vacuoles in sarcotubular myopathy and FSHD2. A. Central rimmed vacuoles (white arrows) in P1.2 with compound heterozygous TRIM32 mutations. Cryostat section, Gömöri trichrome. Scale bar = 30 µm. B. Numerous smaller empty vacuoles arranged in rows typical for sarcotubular myopathy in P1.1, the brother of P1.2 who carried the same compound heterozygous TRIM32 mutations as P1.2. Semithin section, toluidine blue; scale bar = 30 µm. C,D. By EM, the small vacuoles in the biopsy of P1.1. At least partially correspond to dilations of the sarcoplasmic reticulum. Scale bar in C = 4 µm, in D = 2 µm. E. Muscle biopsy of the FSHD2 patient P6 showing subsarcolemmal rimmed vacuoles (arrow). Cryostat section, H&E. Scale bar = 20 µm. F. Subsarcolemmal rimmed vacuoles in the same biopsy as in (E). Cryostat section, Gömöri trichrome. Scale bar = 20 µm. G. The autophagic vacuoles in P6 contain pTDP‐43‐immunoreactive granular material (arrows). Paraffin section. Scale bar = 30 µm.

FKRP

P2 and P3 initially presented with severe proximal paresis of the legs. Proximal arm weakness was severe in P2 while P3 had relatively well‐preserved arm function. The sister of P2 had died of respiratory muscle failure at the age of 39 caused by an undiagnosed muscle disease. P2 developed wheelchair dependency and respiratory muscle weakness 14 years after disease onset. P3 received an implantable cardioverter‐defibrillator at age 37 caused by an episode of ventricular tachycardia and suspected cardiomyopathy. In the muscle biopsy of P2, subsarcolemmal proliferations of enlarged mitochondria projected deeply into many muscle fibers and were associated with marked proliferations of the T‐system and focal myofibrillar degradation (not shown). Autophagic vacuoles containing membranous, myelin‐like material were frequently seen immediately below the sarcolemma. Similar predominantly subsarcolemmal autophagic vacuoles were also found in the biopsy of P3 and were associated with cytoplasmic and nuclear accumulations of 12–18 nm tubulofilaments (Figure 3A,B). Because of the limb‐girdle distribution of muscle weakness and the AR inheritance pattern in both families, targeted Sanger sequencing of FKRP was performed, revealing the most frequent FKRP mutation c.826C>A, p.(Leu276Ile) in P2 in a homozygous state (⁷). P3 was compound heterozygous for this mutation combined with the c.1486T>A, p.(*496Argext*21) mutation that abolishes a stop codon and leads to an extension of the protein by 21 amino acids (¹⁹). In line with the genetic findings, alpha‐dystroglycan immunohistochemistry yielded absent reactivity; the immunoblot showed a markedly reduced intensity of the alpha‐dystroglycan band and an additional band of reduced molecular weight (not shown).

Open in a separate window

Figure 3

Ultrastructural pathology of FKRP, CAV3 and DYSF mutation cases. A. Prominent autophagic vacuoles combined with intranuclear tubulofilamentous inclusions (white arrow) in LGMD2i (LGMDR9) caused by compound heterozygous FKRP mutations (P3). Scale bar = 3 µm. B. P3: Membranous autophagic material accumulating within a myonucleus (arrows) and in the perinuclear sarcoplasm. Scale bar = 3 µm. Inset: High power view of sarcoplasmic tubulofilaments. Scale bar = 0.5 µm. C,D. Subsarcolemmal vacuoles and proliferations of membranes in a case with CAV 3 mutation (P15). Scale bar in C = 1 µm and D = 2 µm. E. Similar subsarcolemmal membranous proliferations in LGMD2B caused by compound heterozygous DYSF mutations (P5.2). Scale bar = 2 µm. F. Exocytosis of membranous autophagic material in the dysferlinopathy patient P5.1, the sister of P5.2. Scale bar = 2 µm. G,H. Large extracellular electron‐dense globules with abundant amorphous extracellular material surrounded by surplus basal lamina in the biopsy of P5.1. Scale bar in G = 10 µm, in H = 1.5 µm.

Glycogen storage diseases

P12 presented with mild proximal arm and leg paresis and respiratory muscle weakness requiring nocturnal BiPAP therapy at the age of 52. Biopsy revealed many PAS‐positive vacuoles in numerous muscle fibers suggestive of GSD. By EM (Figure ​(Figure4A),4A), the extensive glycogen deposits were associated with autophagic vacuoles and abnormal lipofuscin as well as considerable mitochondrial pathology, consistent with GSD II (²⁴, ⁵¹). Alpha‐1,4‐glucosidase activity was reduced to 0.9 µmol/l/h (normal >3.3 µmol/l/h). Sanger sequencing detected the intronic mutation c.‐32‐13T>G, which is present in most late‐onset GSD II cases and gives rise to alternatively spliced transcripts with deletion of the first coding exon (³⁴), combined with the c.1796C>A, p.(Ser599Tyr) missense mutation of known pathogenicity (⁶⁹).

Open in a separate window

Figure 4 1

Ultrastructure of GSD II, GSD XV, chloroquine myopathy and cases without identifiable gene defect. A. Intermyofibrillar deposits of excess glycogen in GSD II (Pompe disease; P12). Scale bar = 3 µm. B. Muscle biopsy of P13 with novel compound heterozygous GYG1 (GSD XV) mutations showing excess glycogen with distinct round clusters of less osmiophilic fine granules, possibly representing partially degraded glycogen (arrows). Scale bar = 2 µm. C. P13: Autophagic vacuoles (arrows) associated with the glycogen deposits. Scale bar = 2.5 µm. D. P13: Large intermyofibrillar autophagic vacuole. Scale bar = 8 µm. E. Autophagic vacuoles combined with extensive accumulations of lipofuscin and curvilinear material (arrow) typical for chloroquine myopathy (P16). Scale bar = 2 µm. F. Arrows marking the membranous borders of exocytosed material between the plasma membrane and the basal lamina in P21. However, no mutation in LAMP‐2 or any other myopathy gene was found in this patient by WES. Scale bar = 2 µm. G. Distinct OPMD‐like tubulofilaments within a myonucleus in P27, who did not harbor a detectable PABPN1 repeat expansion or any other mutation in a myopathy gene. Scale bar = 0.5 µm. H. Deposition of granulofilamentous material in P19 suggested the diagnosis of a desminopathy/myofibrillar myopathy, but no defect in any of the relevant known genes was detected by WES. Scale bar = 0.5 µm.

P13 presented with severe foot drop and mild hip flexion paresis starting at age 50. Muscle biopsy revealed numerous large autophagic vacuoles, cytoplasmic and intranuclear accumulation of tubulofilamentous material, myonuclear degeneration, mitochondrial alterations and focal myofibrillar breakdown. In several muscle fibers, large abnormal, partially membrane‐bound accumulations of glycogen granules were associated with the autophagic vacuoles. Within the glycogen accumulations, round clusters of finely granular material which was less osmiophilic than normal glycogen were frequent. Some of these clusters were fully or partially engulfed by a membrane (Figure 4B–D). NGS panel detected the previously undescribed compound heterozygous mutations c.164_165delTT, p.(Phe55*) in exon 3 and c.646C>T, p.(Arg216*) in exon 6 of GYG1. The clinically unaffected adult son and the healthy maternal aunt of the patient were heterozygous for p.(Arg216*), in line with a compound heterozygous disease process in the index patient.

Vacuolar myopathy associated with chloroquine therapy

P16 presented with left hand weakness, severe foot drop, marked bilateral pes cavus and claw toes. He reported problems with his feet since childhood; there had been similar complaints of ambulation by his father, sister and nephew. CK levels were normal. Panel NGS for CMT and myopathy genes yielded variants of unknown significance in KIF1B and DYSF (Table S1). Muscle biopsy revealed severe vacuolar myopathy with numerous rimmed vacuoles combined with extensive accumulations of lipofuscin and of abundant curvilinear material (Figure ​(Figure4E).4E). Prominent build‐up of lipofuscin and of curvilinear material are typical for chloroquine myopathy (⁶⁴). When asked explicitly, the patient then reported that he had taken chloroquine for several years to treat systemic lupus erythematosus.

Biopsy findings expand the phenotype of known mutations

In the second group, markedly disturbed autophagy was linked to known pathogenic mutations in genes not regularly associated with autophagy alterations. Such changes were prominent in P2 and P3 with LGMD2i (LGMDR9) caused by the p.(*496Argext*21)/p.(Leu276Ile) and p.(Leu276Ile)/ p.(Leu276Ile) FKRP mutations (Figure 3A,B). They are consistent with the known function of FKRP in the glycosylation of sarcolemmal proteins and correlate with previous findings by others in an FKRP mouse model (¹¹) and in two FKRP patients homozygous for p.(Leu276Ile) (⁷³). The severe mitochondrial changes in P3 have not been described before in LGMD2i (LGMDR9).

Dysferlin is involved in membrane repair and induces sarcolemma resealing following damage (¹⁰). Dysferlin mutations lead to marked sarcolemmal defects combined with characteristic extracellular electron‐dense globules detectable by EM (⁸⁴), as confirmed in P5.1 and 5.2. Associated with the membrane defects, we also confirmed subsarcolemmal autophagic vacuoles and extracellular deposits of membranous cell debris, consistent with the hypothesis that dysferlin mutations interfere with normal autophagy (²², ⁸⁴).

Caveolin‐3 is a major component of caveolae (⁵⁹). P15 and his affected mother carried the heterozygous p.(Ala93Thr) CAV3 mutation previously associated with AD rippling muscle disease, RMD (⁴⁵), but did not show muscle rippling. EM analysis revealed a marked reduction of caveolae, large subsarcolemmal vacuoles and focal proliferations of the sarcoplasmic reticulum, findings typical for caveolinopathy (⁵, ⁶, ⁵⁹). Polyneuropathy has been described by us before in RMD caused by the c.80G>A, p.(Arg27Gln) CAV3 mutation; in functional assays this mutation diminished signaling of the nerve growth factor receptor TrkA, potentially accounting for the manifestation of neuropathy (⁵). Hence, the present cases expand the spectrum of CAV3 mutations leading to neuromyopathy (⁴⁵, ¹⁰²).

Collagen VI contributes to the anchoring of the muscle fiber basal lamina in the extracellular matrix (⁴⁷). Defective autophagy has been observed in a COL6A2 myopathy mouse model, along with decreased levels of beclin‐1 and Bnip3 in human Bethlem myopathy patients (³⁰). The marked vacuolar changes detectable in P11 harboring the known pathogenic AD COL6A2 mutation p.(Asp621Asn) (⁷⁶) support the notion that altered autophagy is a relevant pathomechanism of collagen VI‐related myopathy. Similar to the prominent core/rod pathology suggestive of altered proteostasis reported by us in a patient with COL6A3 mutation (¹³), the exact sequence of events leading to these changes remains unclear.

Previously, we have described five unrelated patients with genetically confirmed FSHD1 whose biopsies showed vacuolar myopathy with rimmed vacuoles (⁷⁴). To our knowledge, we here report the first case of genetically confirmed FSHD2 associated with markedly vacuolar myopathy. Again, the pathomechanism leading to the vacuolar features is unclear.

Biopsy confirms the pathogenicity of novel mutations detected by NGS

The third group included cases with novel mutations that were classified as pathogenic based on the clinical findings including segregation analysis and on the biopsy results. GYG1 mutations cause AR GSD XV (³³, ⁵⁴, ⁶¹). The novel p.(Phe55*) and p.(Arg216*) GYG1 mutations in P13 each lead to premature termination of the coding sequence. The disease phenotype was only expressed by the index patient, but not by the heterozygous family members, thus confirming that one functional copy of the GYG1 gene is sufficient. Glycogenin‐1 is thought to be a primer of glycogen synthesis. As reported already in GSD XV cases and in glycogenin‐1 KO mice (⁹¹), glycogen accumulated in the muscle fibers of P13. Of note, the signs of altered autophagy were particularly prominent (Figure ​(Figure4D).4D). Accumulation of finely granular material, probably of smaller glycogen granules, was a further distinguishing feature of our case (Figure 4B,C). Interestingly, smaller than usual (partially degraded?) glycogen granules have already been shown to occur in GSD II muscle fibers (²¹).

TRIM32 is an E3 ubiquitin ligase located in the Z‐band (⁵³). All LGMD2H (LGMDR8) mutations reported so far are located within the NHL domain of TRIM32 which is critical for the recognition of protein targets for ubiquitination (²⁷, ⁵³). Two brothers with clinical LGMD phenotype (P1.1. and P1.2) showed a deletion of exon 2, the only coding exon of TRIM32, and the sequence variant p.(Leu156Pro), resulting in a change of leucine to the structurally unrelated proline in a highly conserved region, but not within the NHL domain. These results indicate that mutation in TRIM32 outside of the NHL domain can cause sarcotubular myopathy/LGMD2H/LGMDR8. Notably, our initial histopathology of the muscle biopsy of P1.1 had not shown the typical vacuolar, sarcotubular expansions of the sarcoplasmic reticulum combined with myofibrillar disintegration (⁴⁶, ⁷⁹). However, subsequent semithin sections and EM examination of archival material established the diagnosis of sarcotubular myopathy, and thus confirmed the functional relevance of the p.(Leu156Pro) mutation. This focal occurrence of the sarcotubular alterations confirms previous observations by others (⁶²). Our morphological workup did not yield any evidence for pathogenicity of the detected MYH7 sequence variant, in line with the limb‐girdle phenotype of the index patient. Thus, highlighted by this case, repeated and extended morphological examination may eventually settle an uncertain genetic diagnosis.

Non‐inflammatory vacuolar myopathies with negative WES

In the fourth group consisting of 15 patients including the chloroquine myopathy patient, no pathogenic gene defects were detected by broad genetic screening and NGS panel analysis. In 12 of these cases, additional WES was inconclusive. Trio WES analysis would be a more powerful and promising approach to identify causal mutations, however, especially in patients with late‐onset disease, parental samples are often not available. Several of these patients showed characteristic pathology such as OPMD‐like intranuclear filamentous inclusions, prominent exocytosis or granulofilamentous deposits. It remains unclear whether such cases can at some point be assigned to mutations that are not within the sequenced coding regions or to copy number variations or other structural alterations of the genome. Alternatively, at least some of them might represent sporadic degenerative myopathies of non‐genetic origin which share characteristic features of autophagy impairment. This is reminiscent of neurodegenerative disorders such as Alzheimer´s disease, which are also age‐related and characterized by abnormal autophagy, comprising early‐onset, often genetic and late‐onset, often non‐genetic forms.

We thank Mrs. A. Knischewski, Mrs. C. Krude, Mrs. H. Mader and Dr. I. Katona for technical assistance and support. Dr. I. S. Zechel is a member of the European Reference Network for Rare Neuromuscular Diseases (ERN‐Euro NMD). J. Weis is a member of the German Muscular Dystrophy, Motor Neuron Disease and Charcot‐Marie‐Tooth Disease Networks (MD‐Net, NMD‐Net and CMT‐Net, respectively). This study was funded by the German Ministry of Science and Education (BMBF‐MND‐Net; Funds 360644), the German Research Foundation (INST 948/41‐1 FUGG, 2018‐2019), the German Myopathy Society (DGM) and the Interdisciplinary Centre for Clinical Research, RWTH Aachen University (IZKF Aachen, N7‐4) to J. Weis.