Muscle imaging in patients with tubular aggregate myopathy caused by mutations in STIM1.

Tubular aggregate myopathy is a genetically heterogeneous disease characterized by tubular aggregates as the hallmark on muscle biopsy. Mutations in STIM1 have recently been identified as one genetic cause in a number of tubular aggregate myopathy cases. To characterize the pattern of muscle involvement in this disease, upper and lower girdles and lower limbs were imaged in five patients with mutations in STIM1, and the scans were compared with two patients with tubular aggregate myopathy not caused by mutations in STIM1. A common pattern of involvement was found in STIM1-mutated patients, although with variable extent and severity of lesions. In the upper girdle, the subscapularis muscle was invariably affected. In the lower limbs, all the patients showed a consistent involvement of the flexor hallucis longus, which is very rarely affected in other muscle diseases, and a diffuse involvement of thigh and posterior leg with sparing of gracilis, tibialis anterior and, to a lesser extent, short head of biceps femoris. Mutations in STIM1 are associated with a homogeneous involvement on imaging despite variable clinical features. Muscle imaging can be useful in identifying STIM1-mutated patients especially among other forms of tubular aggregate myopathy.

Introduction

Tubular aggregate myopathy (TAM) is a genetically heterogeneous, rare nosological entity characterized by tubular aggregates as the main pathological hallmark on muscle biopsy [1]. Tubular aggregates are regular arrays of tubules derived from the sarcoplasmic reticulum [2] and can also occasionally be found as an accompanying feature in other hereditary myopathies [3] or congenital myasthenic syndromes [4].

“Pure” TAM is usually sporadic or follows an autosomal dominant mode of inheritance, and is clinically characterized by myalgias, cramps and muscle stiffness, with or without weakness with a predominantly proximal distribution [3].

Recently, mutations in two genes encoding for interacting proteins involved in sarcoplasmic reticulum calcium storage sensing and intracellular influx, i.e. STIM1 [5] and ORAI1, the latter in complex phenotypes with miosis [6] or hypocalcemia [7], have been identified as molecular causes of TAM. However, some patients with TAM do not harbor mutations in any of these genes.

Muscle imaging through MRI is a strong tool to globally assess muscle involvement in neuromuscular patients in an accurate and non-invasive way, providing details that cannot be inferred from clinical examination alone [8].

The aim of our study was to investigate muscle involvement of STIM1-mutated patients on imaging, to define the MRI pattern and to compare it with that of TAM not caused by mutations in this gene.

Patients and methods

Muscle imaging data of seven patients with TAM from four different centers (Rome Bambino Gesù Children's Hospital, Rome Catholic University, Angers and Copenhagen) were available for retrospective analysis.

For all the patients, lower limb MRI had been performed in 1.5-T MR scanners as previously described [9]. MRI of the upper girdle [10] was available for four out of five STIM1-mutated patients. Due to claustrophobia, one patient (P4) could not undergo upper girdle MRI and was studied by means of computed tomography (CT) obtaining axial slices from the skull to the lumbar spines.

T1-weighted MR images and the CT scan were evaluated by one investigator (GT) and were judged qualitatively for the degree of fatty replacement of the single muscles and overall pattern of involvement.

All the scans were obtained either for diagnostic purposes or in the routine follow-up of the patients, and the local Ethics committees of the different Centers authorized their anonymized publication. All the involved subjects gave their informed consent to the study.

Results

Patients

Five of seven TAM patients (P1–P5) harbored different mutations in the EF hands domain of STIM1. Clinical and molecular data of these five patients have already been reported [11], [12]. Briefly, P1 (male, p.N80T mutation, age at imaging 36 years) complained about post-exercise myalgia since childhood and had very mild lower limb proximal weakness with onset in adulthood [12].

P2 (female, p.I115F mutation, age at imaging 6 years) and P3 (female, p.H109R mutation, age at imaging 16 years) showed a phenotype compatible with a congenital myopathy, with childhood onset and no significant cramps or myalgias [11].

Patients P4 (female, p.F108I mutation, age at imaging 40 years) and P5 (male, p.F108I mutation, age at imaging 47 years) are siblings and had a later onset with a clinical diagnosis of limb-girdle muscular dystrophy [12].

P6 and P7 have TAM not associated with mutations in STIM1. P6 (male, age at imaging 69 years) was investigated for hyperCKemia at age 57. Family history was negative for neuromuscular disorders. Muscle biopsy showed the presence of myopathic changes and tubular aggregate formations, without any other significant abnormality. Both STIM1 and ORAI1 were sequenced with normal results. P7 (male, age at imaging 44) has TAM caused by mutations in the gene encoding for the muscle subunit of phosphoglycerate mutase (PGAM2) [13]. Since age 18, he has complained of contractures and stiffness during exercise. PGAM2 was also screened in P6 by direct sequencing with negative results.

MRI and CT scans

Lower limb imaging in STIM1-mutated patients

The STIM1-mutated patients with predominant muscle weakness (P2–P5) showed, at the level of the pelvis, fatty replacement of obturator and gluteal muscles without prominent differences between gluteus minimus, medius and maximus in early disease stages. In more advanced stages, gluteus minimus and medius appeared more affected than the maximus (Fig. 1). In the thigh, the sartorius was involved early. Diffuse involvement of anterior and posterior thigh followed, with relative sparing of adductor longus, gracilis and short head of the biceps femoris even in late stages. In the lower leg, we found a consistent involvement of the posterior compartment (gastrocnemii and soleus) and peroneal muscles, followed by tibialis posterior later on, with selective sparing of tibialis anterior and extensor hallucis and digitorum longus. Notably, flexor hallucis longus was affected in all of these patients (Fig. 2). The STIM1-mutated patient with predominant myalgias (P1) showed a milder but similar radiological picture (Fig. 1). Vastus lateralis, adductors, semitendinosus and semimembranosus were mildly affected. Leg muscles were spared with the exception of gastrocnemius medialis and, strikingly, the distal part of flexor hallucis longus (Fig. 2).

Fig. 1

Overview of lower limb imaging findings of the seven TAM patients. The five STIM1-mutated patients with different mutations are ordered by disease severity and cover a wide radiological spectrum. Some remarkable findings were sartorius involvement in early disease (arrowheads, P2–P3) and sparing of tibialis anterior in later stages (short arrows, P4–P5). At variance, non-STIM1 patients were characterized by different features, such as gluteus minimus (asterisk) and monolateral gastrocnemius medialis involvement (long arrow) (P6), or mild gluteus maximus involvement (triangle, P7).

Fig. 2

Distal lower leg axial MRI slices of the TAM patients Flexor hallucis longus was bilaterally affected in all STIM1-mutated patients (P1–P5) but not in non-STIM1 patients (P6 as an example) (short arrows).

Lower limb imaging in other TAM

At variance with STIM1 TAM, the non-STIM1 TAM patients displayed either bilateral gluteus minimus and asymmetric gastrocnemius medialis fatty replacement (P6) or only initial glutei involvement with sparing of thigh and leg muscles (P7).

Upper girdle imaging in STIM1-mutated patients

In the upper girdle (Fig. 3), P1 showed only initial subscapularis involvement. In the mildest patient with predominant muscle weakness (P2), subscapularis and teres major were mildly affected. P3 showed subscapularis and neck extensors involvement; spinati, teres major, latissimus dorsi and pectoralis major were affected to a lesser extent. P4 and P5 displayed a more diffuse involvement, with subscapularis being consistently the most severely affected, followed by teres major and latissimus dorsi. In all of the patients, masticator muscles (masseter and pterygoids) were spared and trapezius completely or relatively spared. Fatty replacement of the tongue could be noticed in patients with more severe disease (P3–P5).

Fig. 3

Upper girdle imaging in STIM1-mutated patients. Bilateral subscapularis involvement, ranging from mild to end-stage, is present in all the subjects (P1–P5, long arrows). Teres major involvement is consistently found in the most severely affected patients (shown in P3 and P5, arrowheads). All the patients showed complete or relative trapezius sparing (more evident on coronal sections displayed on the right hand side, asterisk), as well as sparing of pterygoid muscles (P3, triangle).

Discussion

Despite the differences in the clinical presentations (age of onset, presence/absence of myalgias, overall severity), muscle imaging showed a common pattern of involvement in all STIM1-mutated patients, although with variable extent and severity of lesions. In the lower limbs, anterior and posterior thigh muscles, especially sartorius in the patients with childhood onset, together with calf and peroneal muscles were the earliest and most affected. In all the patients we found a consistent involvement of flexor hallucis longus, which is very rarely affected in other muscle diseases, with sparing of gracilis, tibialis anterior and, to a lesser extent, short head of the biceps femoris. This MRI pattern is similar but not identical to that reported in a heterogeneous cohort of non-genetically defined TAM patients [14], which could reflect a high prevalence of STIM1 mutations in that cohort. However, in those patients the authors did not describe a prominent involvement of flexor hallucis. In contrast, our non-STIM1 TAM patients had a different type of involvement, with no overlap even with the mildest of the STIM1-mutated patients.

In the initial stages, lower limb muscle imaging findings of STIM1-mutated patients may share features (calf and peroneal muscle fatty replacement in the lower leg, and diffuse thigh involvement with adductor longus and gracilis sparing) with RYR1-related myopathies, especially with autosomal recessive forms that have a less clear-cut imaging signature [15]. These similarities in imaging pattern could reflect shared molecular pathogenetic processes between the two diseases that may involve perturbed intracellular calcium homeostasis, RYR1 being a sarcoplasmic reticulum protein with a role in calcium entrance in the cytoplasm to allow muscle contraction [16].

In late stages, the pattern of STIM1-mutated patients has overlaps with forms of muscular dystrophies such as the alpha-dystroglycanopathies [17] or calpainopathy [18], but tibialis posterior and flexor hallucis are often relatively spared in these diseases.

Flexor hallucis involvement has been described in TPM2-related myopathies, but their overall pattern of muscle involvement is very different from STIM1 with predominantly distal distribution of fatty replacement involving both anterior and posterior lower leg compartments and masticators [19], [20].

Upper girdle MRI can be helpful to distinguish those diseases that overlap with STIM1 on lower limb imaging. The preferential involvement of subscapularis with sparing of trapezius is useful to differentiate STIM1 from calpainopathy, since in the latter subscapularis and trapezius are usually affected to a similar extent [10]. Upper girdle imaging in RYR1-related myopathies has been described only in few patients, in whom the authors reported a mild diffuse involvement [21], [22]. At variance with STIM1, neck extensors and masticators were the most affected muscles, while subscapularis was relatively spared. Further data are needed to confirm whether upper girdle imaging can be useful to distinguish between the two diseases.

A limitation of our study is the small number of patients, although significant considering the rarity of this condition. To this regard, particularly the findings about non-STIM1 mutated patients should be carefully interpreted given the very limited number of subjects. Muscle imaging features of patients harboring dominant mutations in ORAI1 have been reported only in one family so far [7]: in these three individuals, the described pattern could resemble that of the most severe patients in our cohort, although with a heavier involvement of paraspinal and trapezius muscles. However, more detailed and systematic analysis of muscle involvement in ORAI1-related TAM is required to allow an exact comparison with forms caused by STIM1 mutations.

In conclusion, mutations in STIM1 are a common cause of TAM and cause a homogeneous involvement on imaging despite different clinical features. The detection of a compatible imaging pattern, although not completely specific, can be helpful to distinguish STIM1-mutated patients from other forms of TAM.