**1. Introduction**

Idiopathic inflammatory myopathies (IIM) are a heterogeneous group of autoimmune diseases in which inflammation of the striated skeletal muscles leads to myalgia and weakness. As distinct subgroups, they include dermatomyositis (DM), inclusion body myositis (IBM) and polymyositis (PM), defined by clinical, serological and histological criteria. In DM, both muscles and skin tissues are affected. IBM owes its name to the histological findings of protein aggregates in muscle cells. In PM there is no skin involvement and an inflammation of the muscle tissue occurs without evidence of inclusion bodies in muscle biopsy. Besides these major groups, there are also overlap syndromes in

which symptoms of other rheumatological diseases occur in combination with muscle impairment. In recent years, the role of autoantibodies has been increasingly recognized in both research and diagnostics. Specific autoantibodies are hypothesized to be involved in the pathophysiology of inflammation and thus are associated with distinct disease entities, e.g., the Jo-1 antibody is highly specific for the antisynthetase syndrome.

Swallowing is a complex neuromuscular process that requires the precise motor coordination of the oropharynx, larynx and esophagus [1,2]. While smooth muscles are located in the lower and middle part of the esophagus, the upper part and the oropharynx consist of striated skeletal muscle tissue [2], which is typically affected by inflammation in IIM. It is therefore not surprising that myositis can cause dysphagia via inflammatory involvement of the swallowing muscles. In fact, dysphagia is part of the current American College of Rheumatology/European League Against Rheumatism (ACR/EULAR) diagnostic criteria as an item indicating IIM in patients with symptoms of myalgia [3]. Instrumental assessments, e.g., flexible endoscopic evaluation of swallowing (FEES) or videofluoroscopy (VFSS) are considered the diagnostic gold standard [4,5].

The study data available on dysphagia in IIM are heterogeneous with partly conflicting results, e.g., the reported prevalence rates range from 0% [6] to 100% [7]. Similarly, heterogeneous study results can be found in the instrumental characterization of dysphagia, its consequences or therapeutic implications. The aim of this systematic review was therefore to summarize and analyze the existing evidence on epidemiology, pathophysiology, outcome and therapeutic effects and to estimate pooled prevalence rates in a meta-analysis.

## **2. Methods**

### *2.1. Review*

2.1.1. Inclusion and Exclusion Criteria in the Review

Studies had to meet the following inclusion criteria:

	- a. Epidemiology or prevalence of dysphagia in a population with a minimum of five subjects;
	- b. Pathophysiology of dysphagia;
	- c. Outcome of a patient cohort with dysphagia;
	- d. Therapeutic effects on dysphagia or swallowing.

Articles were excluded if:


#### 2.1.2. Search Strategy

To identify studies, MEDLINE was searched for all relevant articles on dysphagia and myositis from inception to January 2020 (last update in January 2020). The following PubMed search algorithm was used:

("deglutition disorders"(MeSH Terms) OR ("deglutition"(All Fields) AND "disorders"(All Fields)) OR "deglutition disorders"(All Fields) OR "dysphagia"(All Fields)) AND (("myositis"(MeSH Terms) OR "myositis"(All Fields)) OR ("polymyositis"(MeSH Terms) OR "polymyositis"(All Fields)) OR ("dermatomyositis"(MeSH Terms) OR "dermatomyositis"(All Fields)) OR ("myositis, inclusion body"(MeSH Terms) OR ("myositis"(All Fields) AND "inclusion"(All Fields) AND "body"(All Fields)) OR "inclusion body myositis"(All Fields) OR ("inclusion"(All Fields) AND "body"(All Fields) AND "myositis"(All Fields))) OR ("Antisynthetase syndrome"(Supplementary Concept) OR "Antisynthetase syndrome"(All Fields] OR "antisynthetase syndrome"(All Fields)))".

Furthermore, reference lists of published articles were screened for additional studies.

#### *2.2. Meta-Analysis*

#### 2.2.1. Inclusion and Exclusion Criteria in the Meta-Analysis

All studies that reported the prevalence of dysphagia in a cohort of a minimum of five subjects were included in the meta-analysis (Document S1). Only studies that reported directly on a cohort were included (no survey data with estimates of prevalence among physicians). If both instrumental and clinical results were available, the results of the instrumental diagnostics were used. If studies at the same institution had recruited subjects during overlapping periods, only the study with lowest bias risk (Section 2.2.2) was included, or, in case of equal bias risk, the study with the larger sample was included. An equivalent procedure was applied to overlapping cohorts of registry studies or precursor cohorts of a registry. If studies at the same institution did not report an overlapping recruitment period, studies were excluded only if one of the studies stated that all available patients at the institution were included. If studies reported on an identical patient cohort with the same bias risk and sample size, the study that allowed for more subgroup analyses was included.

In addition to the total cohort of IIM, pooled prevalence for dysphagia was estimated in the PM, DM and IBM subgroup and in the subgroup of studies with low bias risk regarding study cohort and dysphagia assessment (Section 2.2.2). Also, the pooled prevalence was estimated for cancer associated myositis, and non-cancer associated myositis in all studies that compared these two groups. All studies on myositis associated/specific antibodies were reviewed to determine whether dysphagia was reported to be associated with (or with the absence of) a specific antibody. If two or more studies compared the prevalence in a population with one of these reported antibodies to a population without the respective antibody, pooled prevalence was again estimated in both of these groups. Studies in the subgroup analysis were only included, if the sample size of the subgroup contained a minimum of five subjects.

#### 2.2.2. Bias Risk in Individual Studies

In all studies included in the meta-analysis, the bias risk was assessed according to the two domains relevant for observational studies, "study participation" and "outcome measurement" of the "Quality in Prognosis Studies Tool" [13]. The domains were adapted to the topic of dysphagia, e.g., in the outcome measurement it was evaluated if studies relied on an instrumental gold-standard assessment including the pharyngeal phase of swallowing. The aim was to evaluate if the presence or absence of oropharyngoesophageal involvement had been assessed by an objective procedure and that dysphagia had not been determined by clinical examination or the presence of symptoms alone. The following criteria were evaluated:

Study participation criteria: (1) Study population represents the total population of IIM or one of its subgroups (DM, PM, IBM, JDM etc.) without additional clinical, demographic or diagnostic criteria, e.g., not only subjects with specific diagnostic procedure or additional clinical hallmark. Excluded from this were clinical criteria, which exclusively represented the contraindications of the instrumental diagnostics used. (2) Adequate description of recruitment: Either a defined period of time at a particular institution/region had to be specified, or it had to be evident that all available patients of an institution/region were included. (3) Adequate description of inclusion and exclusion criteria.

Outcome measures: (1) A clear definition of dysphagia or swallowing pathologies assessed is provided. (2) Dysphagia was assessed with an instrumental gold-standard procedure (flexible endoscopy of swallowing, VFSS, real-time MRI, scintigraphy) that includes the visualization of the pharyngeal phase of swallowing. (3) Identical method and setting of outcome measurement was applied for all study participants.

All points in this list had to have been fulfilled for a study to be classified as "low bias risk". If there was no indication of bias risk, the study was classified as "low bias risk", otherwise the study was classified as "significant bias risk".

#### 2.2.3. Statistical Analysis

A random effect model (restricted maximum likelihood) was chosen to estimate the pooled prevalence rates. The effect size and standard deviation was calculated with Microsoft Excel 16 using the following approach [14]: If no patient had dysphagia in a study population (0 events), a "continuity correction" of 0.5 was added to the event column as well as to the sample size column to enable inverse variance weighting [15]. The further analysis was calculated with the software JASP 0.11.1. The pooled prevalence, the 95% confidence interval (CI), I2 as a measure for heterogeneity and a funnel plot with the Egger's test as a measure for publication bias were calculated for each analyzed group. In the comparison of subjects with a parameter to subjects without the respective parameter (Section 2.2.1), prevalence rates were considered to be significantly different when the 95% CI did not overlap.

#### **3. Results**

Figure 1 illustrates the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram of the reviewed literature [16].
