Hereditary proximal spinal muscular atrophy (SMA) is an important genetic cause of infantile mortality and childhood disability. Degeneration of α-motorneurons in the ventral horns of the spinal cord is the most salient feature but other organs, in particular the heart, may also be affected as suggested by numerous case reports [1, 2].

SMA is caused by deficiency of the survival motor neuron (SMN) protein due to homozygous loss of function of the SMN1 gene. The human SMN locus contains a second, nearly identical, SMN copy (SMN2) that contains a critical point mutation in exon 7, resulting in exclusion of exon 7 from most SMN2 mRNA transcripts. SMN2 therefore produces residual levels of full length SMN2 mRNA and functional SMN protein [3–7]. SMN protein is ubiquitously expressed and is part of multiprotein complexes that probably have both general and motor neuron specific functions, including small nuclear ribonucleic protein (snRNP) assembly, pre-mRNA splicing, post-transcriptional gene regulation, axonal mRNA transport, ubiquitination homeostasis, maintenance and neuronal differentiation of embryonic stem cells and embryonic organ development [6, 8–13]. Variation in SMN2 copy numbers, which partly explains differences in SMN protein levels between patients, is the most important modifier of SMA severity. The severity spectrum encompasses prenatal SMA (type 0), infantile onset severe SMA (type 1), an intermediate form (SMA type 2), childhood onset SMA (type 3), and adult onset SMA (type 4). Higher copy numbers are associated with milder forms of SMA [5, 6].

The identification of non-neuromuscular complications of severe SMA, including disorders of the heart and cardiovascular system, may help to elucidate pathogenic pathways and are furthermore of increasing clinical importance since therapies that aim to attenuate or reverse SMN deficiency may be introduced soon.

To study the evidence for an association of SMA with cardiac pathology in more detail, we performed a systematic review of the available clinical and experimental literature.

We retrieved 3002 articles with our initial search. After selection, 72 articles met our predefined inclusion criteria, including 4 articles of which only the abstract was available [17–20]. These abstracts contained sufficient detailed information and were included for further analysis. We were unable to obtain full text or detailed abstracts of 15 possibly relevant articles. Twelve of these articles were identified in the original search, whilst the other 3 were found through the related articles search (Additional file 1: Table S4, [21–35]).

We identified a total of 264 published cases of SMA patients with cardiac pathology. Seven studies contained descriptions of patients with several SMA types, 28 studies of SMA type 1 only and 23 studies of SMA type 3 only. We found a total of 14 studies on cardiac pathology in SMA mouse models (Additional file 1: Tables S1–S3 and S5).

Cardiac pathology in patients with SMA type 1

We identified 77 patients with SMA type 1 (‘Werdnig-Hoffmann Disease’) and cardiac pathology [36–69]. Most studies used well defined clinical criteria for the diagnosis of SMA. Tests for homozygous SMN1 deletion were performed in 36 (47%) patients and confirmed in 31: five patients did not have a homozygous SMN1 deletion. It was not specified whether these patients had intragenic SMN1 point mutations, or a different neuromuscular disorder. The diagnosis in these 5 patients was based upon clinical characteristics combined with supportive information from muscle biopsies (n = 5), EMG (n = 4) and autopsy findings (n = 2) [44, 50, 54, 59].

Thirty-three out of 77 (43%) patients had electrocardiogram (ECG) abnormalities. ECGs of 15 patients (19%) showed severe symptomatic bradycardias, defined as a heart rate of less than 40 beats per minute. Baseline tremors were reported in the other 18 (23%) patients. These ECG baseline tremors were initially interpreted as abnormalities or even as being suggestive of cardiac pathology, but nowadays they are interpreted as artefacts due to peripheral muscle tremors. Two patients were diagnosed with a ‘late form of Werdnig-Hoffmann’ but it is unclear whether these two patients had SMA type 2 or 3 and were misclassified or had a relatively mild form of SMA type 1 (i.e. type 1c) with longer survival [70, 71]. The two children, aged 10 and 12, had echocardiographic abnormalities compatible with left ventricular hypertrophy (Additional file 1: Table S1, patients 8 and 9) [39].

The remaining 42 patients (55%) with SMA type 1 had structural cardiac defects. We used the IPCCC diagnostic classification system as outlined in Table 3 [14, 15].

Twenty-one of the 42 patients (50%) had a single structural abnormality of the heart. This included 11 (26%) patients with an atrial septal defect (ASD) and 5 (12%) with an isolated ventricular septal defect (VSD). The other 21 patients (50%) had multiple structural cardiac abnormalities. The combination of an ASD (IPCCC group 4) and VSD (IPCCC group 6) was reported relatively frequently (n = 5; 12%). Hypoplastic left heart syndrome was also found in 5 (12%) patients. Additional file 1: Table S1 includes details on all included patients. Figure 2 illustrates the reported structural cardiac pathology in SMA type 1.

We next assessed the relationship between SMN2 copy numbers and the severity of cardiac pathology. SMN2 copy numbers were available for only 13 of the 77 patients with SMA type 1 (17%). Ten patients had 1 SMN2 copy and 3 had 2 SMN2 copies (Table 4) [46, 53, 56, 63, 65–68].

Reference	Patient no.	SMN2 copy number	Cardiac pathology	IPCCC diagnostic group(s)
Devriendt (1996) [46]	1	1 copy	Small VSD	Group 6
Macleod (1999) [53]	2	1 copy χ	ASD	Group 4
	3	1 copy χ	ASD and mitral hypoplasia	Groups 4 and 5
García-Cabezas (2004) [56]	4	1 copy	ASD (secundum type)	Group 4
Rüdnik-Schöneborn (2008) [63]	5	1 copy	Large ASD (sinus venosus type), multiple VSDs and PDA	Groups 4, 6 and 8
	6	1 copy	Common atrium (i.e. a very large ASD) and PDA	Groups 4 and 8
	7	1 copy	Large ASD (secundum type) and VSD	Groups 4 and 6
	8	2 copies	Small VSD and PDA	Groups 6 and 8
Lumaka (2009) [65]	9	2 copies	ASD (secundum type)	Group 4
Rüdnik-Schöneborn (2010) [66]	10	1 copy	Large ASD, mild pulmonary and mild aortic stenosis	Groups 4 and 7
	11	1 copy	AVSD	Group 5
Parra (2012) [67]	12	1 copy	HLHS	Group 6
Ekici (2012) [68]	13	2 copies	Dextrocardia and Tetralogy of Fallot	Groups 1 and 2

The number of SMN2 copies of 13 patients with SMA type 1

Abbreviations: IPCCC International Paediatric and Congenital Cardiac Code; VSD ventricular septal defect; ASD atrial septal defect; PDA patent ductus arteriosus; AVSD atrioventricular septal defect; HLHS hypoplastic left heart syndrome. χ = SMN2 copy numbers were calculated based on the SMN:MPZ ratio provided by the authors in the original publication [53]

Vulnerability to SMN deficiency may not be confined to motor neurons. Cardiovascular abnormalities are among the most frequently reported non-neuromuscular complications in SMA [2]. In this systematic review, we identified 264 published possible cases of SMA with cardiac abnormalities and 14 studies reporting cardiac involvement in SMA mouse models. Structural cardiac pathology was almost exclusively reported in patients with SMA type 1, while acquired cardiac pathology, including arrhythmias and conduction abnormalities, were reported more frequently in less severely affected patients. Detailed classification of the reported abnormalities suggests convergence to specific pathologies in patients with SMA that may be linked to downstream effects of SMN deficiency. We did not identify large controlled studies that indicate the presence of cardiac pathology in SMA, preventing a definite conclusion as to whether the incidence of cardiac abnormalities is increased in SMA.

Structural cardiac abnormalities in SMA type 1 were almost exclusively defects of atrial and ventricular septa and/or defects of the cardiac outflow tract. Ventricular septal defects (VSD), pulmonary stenosis, a patent ductus arteriosus (PDA), and atrial septal defects (ASD) are, however, the most common structural cardiac abnormalities in newborns, with a reported incidence of approximately 1% [105–110]. Low SMN protein levels may increase the odds of abnormal cardiac development. This hypothesis is supported by several observations: interventricular septum abnormalities were also observed in animal models of severe SMA, and abnormal embryonic cardiogenesis induced by low SMN protein levels was identified as a possible underlying cause in one study [94]. Moreover, there was an over-representation of patients with SMA type 1 and cardiac defects who had only one SMN2 copy, which is associated with the lowest residual SMN protein levels that are compatible with life at birth [5]. The association between the lowest SMN2 copy number and occurrence of non-neuromuscular pathology, including cardiac abnormalities, has been suggested previously [63].

Disturbances of cardiac rhythm were a second abnormality reported across the spectrum of SMA severity, i.e. in SMA types 1–3. Leaving out baseline tremors, which are to be considered an artefact caused by the characteristic peripheral tremor in patients with SMA, impulse initiation disorders were the most common cardiac rhythm abnormalities. Taking into account the very low reported incidence of for example atrial flutter or atrial fibrillation in patients under the age of 50 years [111, 112], impulse initiation disorders occurred at a strikingly young age in patients with SMA (atrial flutter, n = 2, ages 24 and 49 years [77, 80]; atrial fibrillation, n = 4, reported ages ranging from 29 to 35 years [17, 19, 88]). This may suggest a developmental origin associated with SMN deficiency. In theory, both dysfunction of either the cardiac electrical conduction system or the ANS, which influences cardiac rhythm in vivo, may underlie cardiac arrhythmias [113]. Significant abnormalities of the cardiac ANS were also found in SMA mouse models [93, 95].

Myocardial fibrosis was reported in 8 patients and may contribute to arrhythmias in SMA [19, 73, 76, 89]. Fibrosis of the myocardium was also a frequent finding in both severe and intermediate SMA mouse models, in which arrhythmias were virtually omnipresent. Bradycardia was reported most often, due to delays in the cardiac electrical conduction system, causing various types of atrioventricular and bundle branch blocks. It should be noted that myocardial fibrosis is a hallmark of normal aging [114], and the limited number of patients precludes a definite conclusion whether impulse conduction disorders in SMA are caused by presenile cardiac fibrosis secondary to SMN deficiency.

There are several other possible explanations how SMN deficiency causes cardiac abnormalities, including specific mRNA-splicing defects that could interfere with normal cardiac development [11, 115]. Low SMN levels have already been shown to influence embryonic organ development in animal models, including cardiogenesis [2, 6, 116]. Furthermore, very low levels of SMN protein may predispose to dysfunction of specific cell types other than alpha-motor neurons, that are involved in cardiogenesis [117–119]. A potential candidate cell type is the neural crest cell (NCC), as a subset of NCCs migrate and differentiate into cardiac neural crest cells (cNCCs) that are involved in development of the musculoconnective tissue (tunica media) of the great vessels, cardiac outflow tract septa (dividing the conotruncus into the aorta and pulmonary trunk) and, to some extent, septation of the atria and ventricles [120–126]. SMN protein deficiency may alter the function of downstream signalling pathways that are important for the migratory process of NCCs [123]. Furthermore, although the cardiac electrical conduction system itself originates from cardiomyocytes [127], the cardiac ANS that contributes to arrhythmias, develops from NCCs [113].

Several limitations of this systematic review need to be addressed. First, we cannot exclude the possibility of publication bias towards cases with particular findings or with severe forms of SMA and heart disease. Published cases may, therefore, not be representative of all cardiac pathology in SMA and single cases could have been missed if they were not represented in the databases used. However, given the relatively large number of patients included, it is unlikely that these cases would have substantially influenced our overall findings. Furthermore, publications and reports differed significantly in clinical detail and the time of diagnosis. Many studies were published before genetic testing for homozygous SMN1 deletion became widely available (i.e.: cases before 1995) which leaves open the possibility of inclusion of disorders other than SMA, in particular for SMA types 2 and 3. The cases of patients with SMA and heart disease included in our work were published between the late 1960s [73] and 2015 [91]. During this period, significant modifications of diagnostic criteria and classifications of SMA types occurred [128, 129]. Although these changes are largely irrelevant with regard to observing a cardiac abnormality in a patient with SMA, we had to assume the correct diagnosis of SMA (in the absence of genetic confirmation of the diagnosis) and severity in some patients. With a view to addressing these issues, at least in part, we reviewed all available clinical data of included cases (Additional file 1: Tables S1–S3) in an attempt to maximise diagnostic accuracy. Finally, considerable differences in the diagnostic methodology for cardiac evaluation, ranging from a limited number of diagnostic tools to assess cardiac pathology, to a more comprehensive combination of ECG, radiographs, echocardiography, or autopsy, clearly results in differences in quality of observations between studies.

On the basis of the data available, if present, structural abnormalities of the heart are predominantly expected in the more severely affected SMA patients (i.e. SMA type 1), disturbances of the cardiac rhythm in the more mildly affected patients (i.e. SMA types 2 and 3). Future studies would benefit greatly from a controlled, standardized, uniform, and comprehensive protocol for cardiac work-up of genetically confirmed cases of SMA.