Dysphagia Pathology

In general, four patterns of swallowing impairment can be distinguished depending on the muscle groups affected (illustrated in Figure 2): Reduced pharyngeal contractility, cricopharyngeal dysfunction, reduced laryngeal elevation and esophageal hypomotility. In case of unclear dysphagia, knowledge of these mechanisms and the corresponding findings in instrumental dysphagia assessment can be helpful in the differential diagnosis [33].

**Figure 2.** Skeletal swallowing muscles and the associated dysphagia mechanisms: A: Reduced pharyngeal contractility; B: Cricopharyngeal dysfunction; C: Reduced laryngeal elevation; D: Esophageal hypomotility.

Reduced pharyngeal contractility can result in insufficient pharyngeal bolus clearance [139]. Consequently, myositis patients often show pharyngeal residue after swallowing [7,27,29,31,33,44, 58–61,65–67,69,71,73–80,83,84,86,87,91,93,96,100,101,103,104,106,111,112,115]. These can impact on swallowing safety and ultimately cause aspiration, which is a frequently reported finding [33,78,81,86]. Further findings indicating reduced pharyngeal contractility are absent or inadequate peristalsis or bolus propulsion [27,31,69,95,106,111], inadequate pharyngeal contraction [29,61,75,87,100,116, 126], nasal regurgitation due to velopharyngeal insufficiency [31,44,62,74,79,80,97,98,104,109–111,116], piecemeal deglutition when swallowing larger boluses [82,140] or reduced pharyngeal pressure in manometry [7,29,32].

Numerous authors reported a dysfunction of the upper esophageal sphincter (UES) due to cricopharyngeal impairment [7,23,24,27,29,31,32,38,58,61,63,66,68,71,72,77,78,80,82,86,87,89–94,96,97, 100–102,106,108]. Both, hypercontractility, e.g., a relaxation deficit of the UES [7,23,24,31,58,61,63,77, 78,80,82,86,87,89,93,94,96,97,100,102,106,108], and hypocontractility [24,38,82,85,90,98,114] have been described. This may be explained by the fact that muscle physiology is affected in a different way during the acute inflammatory phase compared to the chronic phase when fibrosis occurs [82]. Cricopharyngeal hypercontractility often leads to an opening or relaxation disorder of the UES, resulting in pronounced residue or pooling of saliva in the piriform sinus [7,27,31,33,65–67,69,73,74,76–78,80,86,87,101,103, 106,111,112,115], which is located directly above the UES. Typical findings in VFSS are a prominent cricopharyngeus muscle, also referred to as cricopharyngeal bar [23,32,61,68,86,91,93,97,99,106] and muscle propulsions or posterior indentations between C3 and C7 [7,71,96].

Another common finding in myositis is reduced laryngeal elevation [29,58,63,66,76,77,83,107,111]. This is probably caused by impaired contractility of the suprahyoid and longitudinal pharyngeal muscles [141]. Laryngeal elevation is a prerequisite for the UES to open [142] and reduced laryngeal elevation can lead to functional UES impairment. Therefore, the findings in dysphagia diagnostics can be similar to findings with primary UES disorder. Some studies suggest that the typical myositis-associated finding of residue in the piriform sinus may be primarily caused by reduced pharyngeal contractility of suprahyoid muscles rather than an actual dysfunction of the cricopharyngeus muscle itself [58,77]. In this context, one could speak of a pseudocricopharyngeal dysfunction due to reduced laryngeal elevation.

Various authors reported reduced or absent esophageal motility sometimes extending to the lower esophageal sphincter [29,38,114]. Most studies used manometry [24,29,38,59,85,90,98,105,114,117–119, 122,129,130] some also VFSS, barium swallow or scintigraphy [62,73,81,89,99,104,120,121,124,125,127] to detect esophageal impairment.

#### 3.1.3. Outcome

Dysphagia in patients with IIM not only affects quality of life [71], but is also associated with severe complications such as weight loss [30,138] or aspiration pneumonia [24,29,30,32,68,116,143–146]. Pneumonia after aspiration is particularly dangerous as this condition can be fatal [24,29,30,32,116, 143–145,147]. The rates of pneumonia/aspiration pneumonia in cohorts with dysphagic patients are reported between 6% and 36% [24,30,32,68,99,116], are four times more prevalent and, thus, significantly higher in dysphagic than in non-dysphagic patients [24]. Some studies report that aspiration pneumonia is the leading cause of death [29,30,32,147]. In fact, a survey-based study among physicians on patient cases with IBM suggests that dysphagia-associated complications may even be the only cause of premature mortality [148]. It is therefore not surprising that dysphagia is associated with increased mortality [34,144,145,149]. Conversely, the survival rate is associated with dysphagia recovery [143]. Nevertheless, some studies also reported no association between dysphagia and mortality [25,35,150]. Besides mortality, dysphagia is associated with a worse functional status or general condition of the disease [24,39,110,151–153] and represents a negative predictive factor for further disease progression [151].

#### 3.1.4. Therapy

#### Immunomodulatory Therapy

There are several articles reporting positive therapeutic effects of immunomodulatory medication on symptoms and/or on findings of objective swallowing evaluations. These include intravenous methylprednisolone pulse therapy [19,33,70,81,87,134,154–159], methotrexate [30, 76,89,124,136,154,159–161], long-term prednisone/prednisolone [29,30,33,59,64,70,74,76,81,83,84,87, 89,101,103,111,118,122,124,134,136,143,154,158,159,161–167], azathioprine [29,30,33,83,84,87,155,158, 163], intravenous immunoglobulin (IVIG) [24,30,44,65,69,70,73,76,78,81,85,87,114,117,118,121,136, 155,161,168–176], subcutaneous immunoglobulin [173,177,178], hydroxychloroquine [30,118, 154], tacrolimus [162,172], cyclophosphamide [83,170,172], mycophenolate mofetil [118,179], cyclosporine [123] and rituximab [155]. If dysphagia does not respond to medical therapy, it may be helpful to switch to another group of medication, e.g., from steroids to IVIG [24]. The effects on swallowing function were reported in all forms of IIM including IBM [176].

#### Therapy of Malignancy

IIM is associated with malignant diseases and can occur as paraneoplastic syndrome. Therefore, treatment of malignancy can improve muscle symptoms. This effect is also described for dysphagia [44]. Both tumor resection [180–182] and chemotherapy [182–184] can improve or relieve impaired deglutition.

#### Non-Pharmacological Interventional Therapy

Non-pharmacological interventional therapies are symptomatic strategies without a modulatory effect on the disease course, aiming to improve swallowing physiology. Preliminary data suggest that the pneumonia rate can be reduced by interventional therapy if aspiration is reduced [68]. To date, all non-pharmacological interventional procedures attempt to relieve or eliminate the symptoms of cricopharyngeal dysfunction. Three different procedures have been reported:

Injection of botulinum toxin A in the cricopharyngeus muscle: This procedure can reduce the pressure in the UES [61,86] which may result in both symptom relief [61,68] and improvement in objective swallowing diagnostics [68,72]. The effect of this treatment usually lasts for a few months, hence repetitive treatments are necessary [72]. Some authors also reported no improvement [29].

Cricopharyngeal dilatation: This procedure is usually performed endoscopically via a balloon catheter. A clinical improvement of symptoms [29,30,32,80,102] as well as improvement in objective dysphagia diagnostics [78,80] have been described. Here, too, the effect may not be permanent, so that repetitive treatments may become necessary [32,80,102].

Cricopharyngeal myotomy: This is a non-reversible intervention with a surgical sectioning of the cricopharyngeus muscle. It can lead to an improvement of symptoms [24,29–32,86,91,96,97,100, 102,108,110] and an improvement in objective swallowing diagnostics [91,96,97,100]. In some cases, improvement of symptoms without corresponding improvement in VFSS were reported [29]. Other articles reported improvement in swallowing diagnostics without benefits being perceived by the patients [91].

In the absence of interventional trials with clinically meaningful endpoints, the available studies do not allow for a direct comparison between these treatment options and related treatmentspecific recommendations.

#### Behavioral Therapy

In myositis patients, various behavioral swallowing therapies such as diet modifications, compensatory techniques and exercises are used [29]. Unfortunately, there is little evidence for these techniques as there are few studies investigating behavioral therapy in IIM. In individual cases, it was reported that the Mendelson maneuver (pressing the back of the tongue against the palate

when swallowing) has helped to maintain oral food intake without aspiration pneumonia or weight loss [29]. In addition, a case report suggests that isometric tongue strengthening has contributed to the maintenance of posterior tongue pressure [60].

### *3.2. Meta-Analysis*

A total of 109 studies representing 10,382 subjects were included in the meta-analysis of the total patient cohort with IIM. The overall estimate of prevalence of dysphagia was 36%. In patients with IBM, a particularly high prevalence of 56% was estimated. No significant differences in prevalence were found between PM and DM (prevalence and CIs are visualized in Figure 3).

**Figure 3.** Pooled estimated prevalence of dysphagia: prevalence in % (y-axis): the blue and orange bars represent the 95% confidence interval; IIM: Idiopathic inflammatory myopathy, PM: Polymyositis, DM: Dermatomyositis, IBM: Inclusion body myositis, low bias risk studies: cohort of studies with low risk of bias.

Only six studies were classified as "low bias risk". In those studies, all with gold-standard instrumental assessments of dysphagia, the prevalence estimate was 82% and thus clearly higher compared to the total cohort. The estimate of dysphagia prevalence in non-cancer-associated IIM was 26% and 52% in cancer-associated IIM. In patients with NXP2-negative IIM, the estimated prevalence was 33% and 56% in patients with NXP2 antibodies. The CIs in these two comparative analyses did not overlap, so that a significant difference between patients with and without malignancy and NXP2-antibodies can be assumed. The forest plot for studies on malignancy is illustrated in Figure 4 and for studies on NXP2-antibodies in Figure 5. All other comparisons between patients with and without specific antibodies did not reveal significant differences in prevalence. Therefore, of the risk factors presented in Section 3.1.1.2, only malignancy and NXP2 antibodies could be confirmed in our meta-analysis. The estimate of the pooled prevalence, the 95% CI, the number of included studies, the number of included subjects, I<sup>2</sup> as measure for heterogeneity, the *p*-value of the Egger's test as measure for publication bias and the percentage of studies with low bias risk for all analyses are shown in Table 1. The included studies with prevalence and CI in forest and funnel plots for all analyses are shown in the Supplementary Materials S5.


**Table 1.** The estimate of the pooled prevalence, the 95% confidence interval (CI), the number of included studies and subjects, I<sup>2</sup> as measure for heterogeneity, *p*-value of the Egger's test and percentage of studies with low risk of bias for all meta-analyses.

**Figure 4.** Forest plot for malignancy: Forest plot of the studies comparing prevalence in cancer and non-cancer-associated IIM: x-axis shows the prevalence in %.

**Figure 5.** Forest plot for NXP2: Forest plot of the studies comparing prevalence in NXP2-positive and -negative IIM: x-axis shows the prevalence in %.

#### **4. Discussion**

Dysphagia is a frequent complication in IIM with an estimated pooled prevalence of 36% and a peak prevalence of 56% in IBM. Due to the worse outcome associated with dysphagia and the fact that standard immunomodulatory therapy as well as interventional treatment options can improve swallowing impairment, we propose to systematically evaluate swallowing function in patients with IIM and, if present, to include dysphagia as a therapeutic target. The association with malignancy and NXP2 antibodies may have diagnostic relevance in two ways: On the one hand, dysphagia should be considered early on in patients with these risk factors and therefore initiate instrumental swallowing assessment for detailed analysis. On the other hand, in patients with proven dysphagia it might be particularly relevant to carefully look for the presence of an associated malignancy, as dysphagia was shown to be associated with malignant comorbidities [40].

The fact that specific antibodies are associated with an increased risk of swallowing impairment could be an indication that specific pathophysiologic mechanisms might be prone to the oropharynx or the esophagus. The NXP2 antibody associated with dysphagia in this study is particularly common in patients with juvenile dermatomyositis [185]. In the studies on NXP2 antibodies included in our meta-analysis, there was one study in which only juvenile IIM was investigated [49], and another study in which juvenile IIM patients were included in addition to adult patients [47]. The remaining three studies were conducted in adult patients. In addition, the antibody is associated with calcinosis and in adult patients possibly also with malignancy [185]. Thus, an association with dysphagia may also seem possible by association with malignancy which, in turn, is associated with dysphagia. A connection between dysphagia and calcinosis also seems possible, although we did not find a supporting mechanistic explanation for this connection in the literature. However, other antibodies such as TIF-1y, for which in this study no increased prevalence of dysphagia could be proven, are also associated with malignancy [185] (although individual studies associate TIF-1y to dysphagia). Furthermore, in one of the studies on NXP2 antibodies from our meta-analysis, no association with malignancy and calcinosis in adult patients was found [186]. A higher prevalence of dysphagia is also observed in malignancy with compared to malignancy without active IIM [43]. This suggests that dysphagia is not due to an unspecific general deterioration caused by the malignant disease alone. Specific paraneoplastic immune-mediated mechanisms might therefore contribute to swallowing

dysfunction. Further, the reported cases of isolated dysphagia (Section 3.1.1.1) might, similarly to orbital myositis [187], represent a distinct inflammatory entity.

A higher prevalence of dysphagia of 82% was estimated in the low bias risk studies with instrumental assessment. This finding corroborates previous studies showing that refined instrumental evaluation is more sensitive for detecting dysphagia than clinical testing [188,189]. Further, this suggests that oropharyngoesophageal dysfunction may also be present in patients who subjectively experience no swallowing complaints and, therefore, do not report symptoms of dysphagia [7,27,138]. Consistent with this, silent penetration and aspiration (clinically unapparent without symptoms, e.g., coughing or dyspnea) are reported in patients with IBM [71]. The reported prevalence rates vary widely which is also reflected by the strong heterogeneity of the overall cohort. There are four main explanations for these inconsistencies: (1) IIM is not a uniform disease but instead represents a heterogeneous group of diseases with different pathophysiologic mechanisms. Thus, there are presumably real differences in prevalence between different subgroups of the disease. If this is the case, heterogeneity in a meta-analysis should decrease when individual disease groups are analyzed separately; (2) Many different definitions of dysphagia were used, e.g., oropharyngeal vs. esophageal dysphagia. The prevalence rates of the different forms of dysphagia may differ; (3) Different forms of assessment of dysphagia were used, e.g., clinical (patient chart review, swallowing examination) vs. instrumental (FEES, VFSS, manometry, scintigraphy, real-time swallowing MRI). If this is a cause of different prevalence rates, heterogeneity in a meta-analysis should decrease when studies using a uniform assessment procedure are analyzed separately; (4) Dysphagia was determined at different points in the course of the disease (Section 3.1.1.1). Indeed, the heterogeneity partly decreased in the subgroup analysis of IBM, PM and DM and disappeared in the analysis of low bias risk studies with instrumental assessment. Therefore, the heterogeneity in the overall cohort seems to be due to both the different definitions and assessments of dysphagia and real differences in the investigated patient cohorts with differing pathophysiology. The low heterogeneity in most subgroup analysis with specific antibodies may indicate that in case of uniform pathophysiology prevalence rates converge.

Both the funnel plot and the Egger's test suggest that there was a publication bias in our overall cohort, i.e., studies with small sample sizes show higher prevalence rates than studies with large sample sizes. If the bias risk of individual studies is taken into account, an alternative conclusion emerges: Prospective studies with instrumental procedures generally had a smaller sample size, presumably due to the increased recruitment and data collection effort. However, they reported higher prevalence rates due to more sensitive and high-quality diagnostic procedures. In line with this explanatory approach, the funnel plot and the Egger's test no longer indicate a publication bias when studies with low bias risk are analyzed separately.

There are several limitations to this study that must be considered. First, in the overall cohort of the meta-analysis, only few studies had a low bias risk. Especially in the studies with a significant bias risk, different definitions of dysphagia were used and the classification as dysphagic and non-dysphagic was often based solely on clinical evaluation or symptoms. However, due to the lack of objective swallowing diagnostics, it is not possible to say with certainty whether oropharyngoesophageal dysfunction was actually present in these studies with significant bias risk. This has certainly contributed to the considerable heterogeneity and may have contributed to the publication bias. Second, in the meta-analysis of factors associated with increased risk of dysphagia, only studies comparing the prevalence in groups with and without the respective factors were included. However, several potential factors were reported where no such comparison was possible. Third, the majority of included studies were retrospective observational studies, some with small sample size or even only individual case reports. Thus, many conclusions are based on studies with low quality and evidence levels. This applies in particular to the therapy section, where not a single prospective randomized controlled trial could be included. Fourth, although studies at the same institutions with overlapping recruitment periods were excluded, it is possible that overlapping patient groups may also have occurred between registry studies and studies at individual institutions. Fifth, the review

as well as the assessment of the bias risk were conducted by only one observer, which may reduce reliability. Sixth, for the systematic review of this meta-analysis, only Medline was searched with Pubmed, so studies that are only listed in other databases may not have been found. Seventh, due to different reporting standards and partially missing information, no demographic data were pooled and included in the meta-analysis. Especially when comparing groups (e.g., patients with malignant disease and without malignant disease), the groups may differ not only in the prevalence of dysphagia but also in demographic characteristics. For the available demographic data of the studies included in the meta-analysis, we refer to Table S1 (column "cohort") in the Supplementary Materials.
