Plasma exchange in adult idiopathic inflammatory myositis: a systematic review and meta-analysis

Abstract

Objectives

To systematically evaluate the efficacy and safety of plasma exchange (PE) in adult patients with idiopathic inflammatory myopathies (IIM).

Methods

A systematic review was conducted in accordance with PRISMA guidelines and registered on PROSPERO. MEDLINE, Embase and Web of Science were searched from inception to 17 January 2025. Studies involving adult patients with IIM treated with PE were included. Baseline demographics, clinical characteristics, treatment outcomes and adverse events were extracted.

Results

Thirty-five studies involving 473 patients were included, of whom 361 received PE. Most studies were observational, and PE was predominantly used as adjunctive or rescue therapy alongside immunosuppressive treatment. Across included studies, PE was reported to be associated with improvements in organ-specific outcomes, including muscle strength, dysphagia, pulmonary manifestations and biochemical markers such as creatine kinase, ferritin, KL-6 and myositis-specific autoantibody titres. A meta-analysis of seven comparative studies did not demonstrate a statistically significant mortality benefit associated with PE in patients with refractory or rapidly progressive interstitial lung disease (RR 0.41, 95% CI 0.16–1.06, I² = 70%). In immune-mediated necrotising myopathy, PE was frequently associated with clinical improvement and reductions in anti-HMGCR antibody titres. Reported adverse events were generally mild to moderate, with no procedure-related mortality.

Conclusion

PE has a biologically plausible but clinically selective role in adult IIM and is most commonly used as an adjunctive therapy in severe or refractory disease. Evidence is heterogeneous and largely observational, limiting causal inference. Prospective studies with standardised outcome measures are required to better define the role of PE in IIM.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10067-026-08037-7.

Introduction

Idiopathic inflammatory myopathies (IIM) are heterogeneous, acquired autoimmune disorders characterised by muscle inflammation and multisystem involvement [1]. The classification of these diseases is evolving, but current recognised subtypes include dermatomyositis (DM), polymyositis (PM), anti-synthetase syndrome (ASyS), overlap myositis (OM), inclusion body myositis (IBM) and immune-mediated necrotising myopathy (IMNM) [2].

The clinical manifestations of IIM are variable, but the majority present with progressive muscle weakness with or without other systemic features (cutaneous, joint, respiratory, cardiac), often along with histopathological findings of immune-mediated muscle damage. IIM has been associated with increased morbidity and mortality, with major contributing factors including malignancy, infections, profound effects of muscle weakness, cardiovascular and respiratory complications [3–7]. In population-based cohorts, mortality remains substantially higher than in the general population: estimates vary by subtype but 5-year mortality in pooled adult cohorts has been reported between 10 and 35%, with higher rates in patients complicated by interstitial lung disease (ILD) or malignancy [3, 5, 8, 9]. Rapidly progressive ILD (RP-ILD) associated with anti-melanoma differentiation-associated protein 5 (MDA5) DM in particular carries a poor short-term prognosis, and consequently treatment paradigms for these phenotypes include early aggressive combination immunosuppression [10].

Given the heterogeneity of disease manifestations and outcomes, assessment of disease activity and prognosis in IIM relies on a combination of clinical evaluation, organ-specific assessments, and supportive laboratory markers rather than a single definitive measure [11]. Circulating biomarkers such as ferritin, Krebs von den Lungen-6 (KL-6), and myositis-specific autoantibody (MSA) titres have been explored in specific IIM subtypes, but remain adjunctive tools and are not validated surrogates for clinical outcomes [12].

Current therapy for IIM includes glucocorticoids and immunosuppressive agents such as methotrexate (MTX), mycophenolate, azathioprine (AZA), rituximab (RTX), calcineurin inhibitors (CNI), cyclophosphamide (CYC) and intravenous immunoglobulin (IVIg), with treatment influenced by IIM subtype and individual patient factors. There has been increasing use of Janus kinase inhibitors (JAKi), supported thus far by a limited number of open-label trials [13].

Therapeutic plasma exchange (PE), often used synonymously with plasmapheresis in the literature, is an extracorporeal blood purification technique designed for the removal of large molecular weight substances from the plasma including autoantibodies, immune complexes and other lipoproteins [14]. The theoretical basis for its use in IIM is principally the rapid reduction of pathogenic autoantibodies and immune complexes, thereby reversing or halting the pathological processes mediated by these molecules [15]. PE may also modulate immune function through effects on lymphocyte proliferation and function and alterations in the immune system including changes in B and T cell numbers and activation, increased T suppressor function and alteration in T-helper cell type 1/2 (Th1/Th2) ratio [15]. Despite advances in immunosuppressive and targeted therapies, a subset of patients with severe or rapidly progressive IIM continue to experience high early morbidity and mortality. There remain clinical scenarios, such as refractory disease, life-threatening RP-ILD or severe organ compromise, where rapid reduction of circulating pathogenic mediators is desirable and extracorporeal approaches continue to be considered. In this context, PE is used as adjunctive or rescue therapy in selected patients and is supported by consensus statements, narrative reviews and apheresis guidelines, including those of the American Society for Apheresis (ASFA) [16–18]. However, the available evidence remains fragmented and largely observational. To date, no comprehensive synthesis has critically evaluated the efficacy, safety and clinical context of PE across adult IIM subtypes.

The aim of this systematic review and meta-analysis is to synthesise the available evidence on the efficacy and safety of PE in adult IIM, to describe how PE has been used across different IIM phenotypes and to highlight gaps in the literature that require prospective evaluation.

Methods

This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [19]. The review was prospectively registered on PROSPERO on 31 st December 2022 (CRD42022381560). The databases MEDLINE, Embase, Web of Science and Cochrane Database of Systematic Reviews (Wiley) were searched from inception until 17th January 2025. Full details of the search strategy, selection criteria and data extraction are available in Online Resource 1.

Quality assessment was performed using the National Institutes of Health (NIH) quality assessment tools and the Joanna Briggs Institute (JBI) critical appraisal tools [20, 21]. Data were synthesised qualitatively, with descriptive analyses used to present study characteristics. Because of the limited number of participants and lack of standardisation across studies included in our review, a statistical comparison was not attempted for the majority of outcomes measured. A meta-analysis was undertaken when deemed appropriate, using a random-effects model due to the variation observed among studies. Dichotomous data were analysed using risk ratios with 95% confidence intervals. For assessment of statistical heterogeneity, I² statistic was utilised.

Eligibility criteria and definitions

We included studies of adult patients (≥ 18 years) with clinician-diagnosed IIM of the following subtypes (DM, PM, ASyS, IMNM and overlap myositis). We recorded the diagnostic or classification criteria (including the Bohan and Peter criteria [22, 23], European League Against Rheumatism (EULAR)/American College of Rheumatology Classification Criteria (ACR) [24], Solomon [25], Connors [26] and Sontheimer [27]) used in each study and classified subtypes according to the authors’ report.

We defined RP-ILD as new or rapidly worsening respiratory symptoms and, where available, corresponding progression of interstitial changes on chest imaging over weeks–months (authors’ definitions were recorded and harmonised where possible). Refractory disease was defined as inadequate clinical response to at least one adequate trial of glucocorticoid and one or more steroid-sparing agent as stated in the source publication; if the study did not provide an explicit definition, we recorded the indication reported (e.g. ‘treatment failure’, ‘refractory’) and listed the study as ‘unclear’ for the pre-specified refractory definition.

Outcomes and pre-specification

Pre-specified primary outcomes were (1) mortality (all-cause) and (2) clinical response in organ-specific domains (muscle strength, dysphagia, cutaneous involvement, pulmonary endpoints). Secondary outcomes included clinical response in specific IIM subtypes (anti-MDA5/clinically amyopathic dermatomyositis (CADM), IMNM and ASyS) and safety/adverse events. Given the lack of validity as surrogates for survival and function outcomes in IIM, biomarkers including KL-6, ferritin, and autoantibody titres were analysed as exploratory endpoints, reflecting their ongoing investigation and frequent reporting in severe or ILD-associated disease, and were reported separately from clinical outcomes. Composite endpoints, where reported, were analysed only as defined by the original studies, and no composite outcomes were constructed. The time windows for outcome assessment were recorded from each study (e.g. immediate post-PE, 1–3 months, 6–12 months) and are reported in the Online Resources where available.

Handling of overlapping cohorts and heterogeneity

To reduce the risk of double counting, we checked author names, centres, recruitment periods and patient details where available. If overlap was suspected and could not be resolved, the most complete dataset was retained for quantitative analysis. Because of clinical and methodological heterogeneity (IIM subtype, PE indication and regimens, concomitant immunosuppressants, outcome definitions), quantitative pooling was restricted to clinically coherent subsets. Meta-analysis was performed only for mortality in anti-MDA5/ASyS patients with RP-ILD, where patient subtype, disease severity, PE indication, comparator groups and outcome definitions were sufficiently aligned across studies. We used random-effects meta-analysis for pooled estimates and assessed heterogeneity with the I² statistic. All other outcomes were synthesised narratively.

PE regimen variables and concomitant therapy

We extracted regimen variables where reported: number of PE sessions, interval between sessions, exchange volume per session and replacement fluid (albumin, plasma). Concomitant immunosuppressive agents (agent, dose, timing relative to PE) were extracted and categorised (steroid only; steroid + other immunosuppressive disease-modifying anti-rheumatic drug (DMARD) therapy). Where studies did not specify timing or dose, we recorded variables as ‘not reported’.

Results

Publications search overview

The literature search yielded a total of 2927 articles. The results of the search are presented in a PRISMA flow diagram (Fig. 1). Thirty-five studies enrolling a total of 473 adult patients were included; 361 patients received PE and 112 were comparators or in non-PE groups. Full details of the 35 studies are available in Online Resource 2. Of the 333 patients with available gender data, 210 (63%) were female (gender data unavailable for 28 patients). The mean age of patients was 49.8 years (95% CI 48.16, 51.46) from 33 studies. The included studies comprised one double-blind randomised controlled trial (RCT), 19 cohort or comparative observational studies, four uncontrolled intervention studies and 11 case series aggregated for descriptive analysis. By disease subtype, 112 patients were reported as anti-MDA5/CADM, 14 were reported as anti-SRP positive, 3 were reported as anti-HMGCR positive, 13 were classified as having ASyS and the remaining 219 patients who received PE were reported as DM/PM/overlap or unspecified subtype (see Online Resource 2 for study-level subtype counts).

Fig. 1.

PRISMA flow diagram of study selection. Flowchart showing the systematic review process with four stages: identification, screening, eligibility and inclusion

Mean number of PE sessions given was 9.5 (95% CI 8.66, 10.31), from 29 studies. Mean duration of follow-up from initiation of PE was 22.2 months (95% CI 18.96, 25.51), from 21 studies. PE was performed in the setting of treatment failure or refractory disease in 24 studies, involving a total of 241 patients. Mean duration from IIM treatment commencement to PE initiation was 106.8 days (95% CI 31.21, 182.40) from 15 studies. PE was used as part of induction therapy in two of these studies, and for refractory disease treatment in the remaining 13 studies. Mean duration from occurrence of IIM symptoms to initiation of PE was 913.1 days (95% CI 736.55, 1089.73), from nine studies. Of these nine studies, the indication for PE was treatment failure or refractory disease in the majority of patients in seven studies, and undefined in two studies.

Mortality outcomes of PE in IIM

There were seven studies comparing patients receiving PE with non-PE comparators; all of these specifically reported on mortality in patients with anti-MDA5 DM or ASyS with RP-ILD (Table 1) [28–35].

Table 1.

Studies reviewing PE compared to no PE

Author, year	IIM population	PE group			Non-PE group		Observation period	Quality assessment
		PE regimen	PE group: mortality (%)	Concurrent treatment	Non-PE group: mortality (%)	Concurrent treatment
Sasaki et al. [35]	Anti-MDA5 DM/CADM with RP-ILD	3/week repeated monthly; median PE sessions 9	0/9 (0%)	Steroids, CNI and CYC for all	12/31 (39%)	Steroids for all patients. All had additional CNI and/or CYC	2 years	High
Watanabe et al. [34]	Anti-MDA5 CADM with RP-ILD	2–3/week; median PE sessions 16	1/6 (17%)	Steroids, CNI and CYC for all. 1 had additional tocilizumab, 1 had additional tofacitinib	2/6 (33%)	Steroids, CNI and CYC for 5/6. 1 had additional AZA and tofacitinib	Variable- 6–73 weeks	High
Komai et al. [29]	CADM/anti-MDA5 DM with refractory ILD	1–3/week; median PE sessions 17	1/11 (9%)	Steroids, CNI and CYC for all patients. Some had additional tofacitinib, RTX, abatacept or PMX-DHP	4/8 (50%)	Steroids, CNI and CYC for all patients. Some had additional IVIG or PMX-DHP	360 days	High
Eggleston et al. [30]	ASyS/DM with RP-ILD	Daily or alternate daily PE sessions. Median PE sessions 5	0/9 (0%)	Steroids for all patients. Some had additional IVIG, RTX, CYC, mycophenolate, or tacrolimus	2/6 (33%)	83% of patients had steroids. Some had additional IVIG, RTX, CYC, mycophenolate or tacrolimus	30 days	Moderate
Bay et al. [31]	Anti-MDA5 DM with refractory ILD	Frequency not defined. Median PE sessions 5	20/25 (80%)	Combination of therapy including steroids, MTX, mycophenolate, CNI, JAKi, CYC, RTX	12/26 (46%)	Combination of therapy including steroids, MTX, mycophenolate, CNI, JAKi, CYC, RTX, AZA, HCQ	1 year	Moderate
Shirakashi et al. [32]	Anti-MDA5 DM/CADM with ILD	1–3/week; Mean PE sessions 11.4	3/8 (38%)	Steroids, CNI and CYC for all patients	5/5 (100%)	Steroids, CNI and CYC for all patients. Some had additional treatment including: IVIG, PMX-DHP, etanercept, tocilizumab, pirfenidone, sivelestat sodium hydrate	3 years	Moderate
Abe et al. [33]	Anti-MDA5 DM with refractory RP-ILD	Daily for 2 days, then 3/week. Median PE sessions 16	0/6 (0%)	Steroids, CNI and CYC for all patients. 50% had additional IVIG	3/4 (75%)	Steroids, CNI and CYC for all patients. 50% had additional IVIG	1 year	Moderate

ASyS, antisynthetase syndrome; AZA, azathioprine; CADM, clinically amyopathic dermatomyositis; DM, dermatomyositis; CNI, calcineurin inhibitor; CYC, cyclophosphamide; HCQ, hydroxychloroquine; JAKi, Janus kinase inhibitor; IIM, idiopathic inflammatory myopathies; ILD, interstitial lung disease; IVIG, intravenous immunoglobulin; MDA5, melanoma differentiation-associated protein 5; PE, plasma exchange; PMX-DHP, polymyxin-B direct hemoperfusion; RP-ILD, rapidly progressive interstitial lung disease; RTX, rituximab

Across the seven comparative studies included in the pooled mortality analysis, there were 25 deaths among 73 patients in the PE groups (34%) and 40 deaths among 85 patients in the non-PE groups (47%). We performed a random-effects meta-analysis restricted to these cohorts (Fig. 2) and found a pooled risk ratio for mortality of 0.41 (95% CI 0.16–1.06), with substantial heterogeneity (I² = 70%). Individual study results varied: Abe et al. reported improved 1-year survival in PE vs non-PE groups (100% vs 25%, p = 0.033) in a cohort where both arms otherwise received similar immunosuppressive regimens [33]. In contrast, Bay et al. reported higher mortality in the PE group which may reflect confounding by indication (greater baseline severity in the PE arm) [31]. Notably, the only RCT enrolled corticosteroid-resistant PM/DM patients and did not demonstrate benefit for PE versus sham apheresis on muscle strength or functional outcomes [28]. The pooled estimate did not show a statistically significant mortality benefit for PE in RP-ILD associated with anti-MDA5/ASyS in the studies available to date.

Fig. 2.

Forest plot of mortality rates in PE compared to non-PE groups. Forest plot of a meta-analysis comparing mortality rate in PE compared to non-PE groups, illustrating the results of seven studies, with individual study effect sizes as boxes and confidence intervals as lines, and a diamond representing the pooled effect estimate

Use of PE with concomitant immunosuppressive therapy

Twenty-nine studies (282 patients in the PE + concomitant therapy group) reported PE given alongside immunosuppressive agents (steroids ± DMARDs). One small study used PE as first-line monotherapy (three patients) with partial responses in all three [36]. In studies reporting concurrent steroid-only therapy (seven studies, 28 patients), 16/28 (57%) had documented clinical and/or biochemical improvement after PE; follow-up ranged from 7 months to 5 years [28, 37–42]. In the larger group receiving combination immunosuppression (steroids with at least one of the following: CYC, CNI, IVIg, RTX, JAKi), 186/254 (73%) had reported clinical and/or biochemical improvement, or survival benefit. Reporting of timing of PE, doses and exact regimens was variable and frequently incomplete (see Online Resource 2). Because PE is frequently used as salvage therapy in patients already receiving multiple agents, these observational data are at high risk of confounding by indication.

PE-associated outcomes: organ-specific domains

Effect of PE on muscular disease in IIM

Muscle strength was reported in ten studies comprising 151 patients (Supplementary Table 1, Online Resource 3). Outcome measures included summed Manual Muscle Testing-8 (MMT-8) and Medical Research Council (MRC) grading or grip strength; five studies reported no formal strength measure. Overall, 103 of the 151 patients (68%) were reported to have any improvement in muscle strength following PE. The magnitude and clinical significance of this improvement could not be consistently determined due to heterogeneous outcome reporting. Time points of assessment were inconsistently reported; where reported, were within 1–6 months of the treatment course. Serum CK was reported in nine studies (139 patients) and fell after PE in 135/139 (97%) patients. Similarly, time windows varied between studies. Only a minority of studies linked CK change to functional improvement, and where both were reported a reduction in CK did not always correlate with clinically meaningful strength gains.

Additional data on other muscular outcome measures such as dysphagia, magnetic resonance imaging (MRI) changes and electromyography (EMG) changes are available in Online Resource 3.

Effect of PE on extramuscular manifestations of IIM

Interstitial lung disease

Sixteen studies involving CADM/anti-MDA5, ASyS and mixed IIM cohorts reported ILD-related outcomes using heterogeneous measures, including high-resolution computed tomography (HRCT) findings, oxygenation metrics, pulmonary function tests (PFT) and mortality [29–35, 37, 43–50] .

Changes in chest CT findings before and after PE were reported in nine studies. Only one study [43] employed a clearly defined quantitative HRCT scoring system; in the remaining studies, imaging assessment was descriptive or incompletely specified. Ning et al. (n = 18) [43] reported mean CT score improvement from 2.41 ± 0.14 pre-PE to 1.07 ± 0.12 post-PE (p < 0.001). Across the other eight studies, 24/30 (80%) participants showed improvement in CT appearance after PE [30, 37, 44–49]. Serial PFT data were scarce (reported for ten patients undergoing PE across two studies [30, 45]) because many patients were too unwell for reliable PFTs at presentation. Where serial PFTs were available, several patients were reported to have improved forced vital capacity (FVC) and/or diffusing capacity of the lungs for carbon monoxide (DLCO) over months following PE and adjunctive immunosuppression. Oxygenation responses were reported in five studies (18 patients), with 15/18 (83%) showing improved oxygen requirements after PE [37, 44, 47–49]. Details are summarised in Table 2. 

Table 2.

PE and effect on ILD

Author, year	IIM subtype	PE regimen	Use of combination immunosuppressant (yes/no)	Outcome measures	No. (%) of patients with improved outcome	Quality assessment
Hirsch et al. [45]	Anti-MDA5 CADM	Frequency not defined Median 6 cycles	Yes	CT changes FVC and DLCO Remission induction	0/2 (0%) 2/2 (100%) 2/2 (100%)	High
Umemoto et al. [46]	Anti-MDA5 DM	1–2/week for 1–2 weeks	Yes	CT changes	1/2 (50%)	High
Eggleston et al. [30]	ASyS/anti-MDA5 DM	Daily/alternate daily Median 5 sessions	Yes	CT changes	8/9 (88.9%)	Moderate
Thompson et al. [44]	IIM including anti-Jo1, anti-MDA5, anti-PL-12, anti-SSA	Daily/alternate daily 5–7 sessions total	Mixed—all had steroids, half had additional CYC	CT changes Oxygen requirements	4/5 (80%) 6/7 (85.7%)	Low
Takahashi et al. [37]	Anti-MDA5 DM	1–3/week for 11–12 weeks Median 27.5 sessions	Mixed—1 had steroids alone, 1 had combination IS	CT changes Oxygen requirements	2/2 (100%) 2/2 (100%)	High
Saito et al. [47]	Anti-MDA5 DM	Alternate daily Median 9.5 sessions	Yes	CT changes Oxygen requirements	4/5 (80%) 4/6 (66.7%)	High
Ishiwari et al. [48]	Anti-MDA5 DM	3/week 6 sessions total	Yes	CT changes Oxygen requirements	2/2 (100%) 2/2 (100%)	Moderate
Yamagata et al. [49]	Anti-MDA5 DM	1–3/week for 3–4 weeks Median 7 sessions	Yes	CT changes Oxygen requirements	3/3 (100%) 1/1 (100%)	Moderate
Ning et al. [43]	DM, CADM, PM	Daily for 3 days, then alternate daily Median 5 sessions (responsive group); 6 sessions (non-responsive group)	Yes	CT changes	Study does not comment on how many people had CT improvement. Cumulative CT score improvement 2.414 ± 0.1379 to 1.073 ± 0.1236, p < 0.001	Moderate

ASyS, antisynthetase syndrome; CADM, clinically amyopathic dermatomyositis; CT, computed tomography; CYC, cyclophosphamide; DLCO, diffusing capacity of lung for carbon monoxide; DM, dermatomyositis; FVC, forced vital capacity; IIM, idiopathic inflammatory myopathies; IS, immunosuppressant; MDA5, melanoma differentiation-associated protein 5; PE, plasma exchange; PM, polymyositis; SSA, Sjögren’s-syndrome-related antigen A

PE and cutaneous involvement

Four studies involving six patients with DM reported cutaneous outcomes following PE [40–42, 48]. Importantly, cutaneous involvement was not the primary indication for PE in most of these cases, rather, PE was initiated for other disease manifestations, most commonly refractory disease or anti-MDA5-associated ILD. Baseline severity of skin involvement and post-treatment changes were not assessed using standardised or validated outcome measures, and pre-PE cutaneous disease activity was variably described. Consequently, cutaneous outcomes were reported descriptively rather than quantitatively. Of the six patients, five were described as having improvement in cutaneous involvement following PE, including improved appearance of rash and skin induration. In a case series of patients with anti-MDA5 DM, two patients had initial worsening of skin changes including Gottron’s and heliotrope rash after the first session of PE, although subsequently had either improvement or stabilisation of skin involvement following a second PE session [48]. In an uncontrolled trial, two patients were reported to have improvement in skin induration following induction PE (six PE sessions over 2 weeks) [40]. Overall, concomitant therapy varied: two of the six patients received additional immunosuppressive agents alongside PE, while three received concurrent glucocorticoids alone, further limiting attribution of observed cutaneous outcomes specifically to PE.

PE efficacy in specific disease subtypes

Clinically amyopathic dermatomyositis (CADM), including anti-MDA5 DM

Seventeen studies involving 112 patients examined either CADM and/or anti-MDA5 IIM [29–35, 37, 43–51] (Table 3). Outcomes measured include clinical (mortality, ILD and CT appearances, oxygen requirements) and biochemical (ferritin, KL-6, anti-MDA5 titres) domains. PE was associated with improved clinical outcomes in the majority of these patients. Nine studies evaluated changes in laboratory markers following PE [29, 32, 34, 35, 37, 46–49]. Reductions in measured biomarkers (KL-6, ferritin or anti-MDA5 antibody titre) were observed in 52/55 (95%) of cases after PE, suggesting a consistent biochemical association in this population. Detailed laboratory outcomes, including the duration between outcome measures, are summarised in Supplementary Table 1 (Online Resource 4).

Table 3.

PE in CADM/anti-MDA5-positive IIM patients

Author, year	No. of patients	IIM subtype	MSA/MAA	PE regimen	PE indication	Concurrent IS	Outcome	Follow-up time	Quality assessment
Hirsch et al. [45]	2	Anti-MDA5 CADM with RP-ILD	Anti-MDA5 Anti-Ro52 in 1	Frequency not defined Median 6 cycles	RP-ILD	GC, CNI, IVIG, CYC	Clinical remission in both cases, improvement in PFTs	863 days (542, 1184 days)	High
Guo et al. [51]	6	CADM	Not defined	Daily/alternate daily sessions Median 5 sessions	Critically unwell in ICU	Not defined	4/6 in remission, 2/6 no response, including 1 death	Not defined	Moderate
Sasaki et al. [35]	9 (vs 31 without PE)	Anti-MDA5 DM/CADM with RP-ILD	Anti-MDA5	3/week, repeated monthly Median 9 sessions	RP-ILD	GC, CNI, CYC	9/9 patients in the PE group had improved respiratory symptoms and were alive, compared to non-PE group where 12/31 died (100 vs 61%, p = 0.037). 2-year survival significantly better in PE group compared to non-PE. Significant decrease in KL-6 and ferritin levels in PE group, compared to non-PE group	2 years	High
Watanabe et al. [34]	6 (vs 6 without PE)	Anti-MDA5 CADM with RP-ILD	Anti-MDA5	2–3/week Median 16 sessions	Severe RP-ILD patients who required nasal or mask oxygenation despite intensive immunosuppression therapy	GC, CNI, CYC	5/6 patients treated with PE survived with or without home oxygen therapy, compared to 4/6 in non-PE group. No relapse during observation period, in surviving patients Reduction in ferritin levels of 5/6 patients post PE. Improvement in degree of ILD on HRCT	23.5 ± 15.1 weeks	High
Umemoto et al. [46]	2	Anti-MDA5 DM with ILD	Anti-MDA5	1–2/week for 1–2 weeks	Worsening ground glass on CT	GC, CYC, CNI	Rapid attenuation in symptoms in 1 patient. Reduction in ferritin and anti-MDA5 antibody titre in 2 patients. Reduction in KL-6 in 1 patient Improved CT appearances in both patients. No flares for up to 24 days post PE for patient 1, 123 days for patient 2	92 days for patient 1, 141 days for patient 2	High
Shirai et al. [50]	8 in the PE group	Anti-MDA5 DM with RP-ILD	Anti-MDA5	3/week	Anti-MDA5-positive RP-ILD with multiple poor prognostic factors	GC, CNI	6/8 (75%) patients survived (all previously deemed to have fatal disease)	36 months after development of initial symptoms of DM	Moderate
Komai et al. [29]	11 (vs 8 without PE)	CADM/anti-MDA5 DM complicated by refractory ILD	Anti-MDA5 * apart from 2 patients. Anti-Ro in 4 patients	1–3/week Median 17 sessions	Inadequate response to the triple intensive immunosuppressive therapy	GC, CNI	Combination therapy with PE was associated with an improved survival rate at 360 days. The survival rate for treatment with PE was higher than for treatment without PE (91% and 50%, respectively, p < 0.05) Among PE-treated patients, anti-MDA-5 antibody titre, ferritin and KL-6 as serological activity markers were sustainably reduced only after initiating PE Therapeutic intervention with PE reduced the frequency of exacerbation of ILD requiring methylprednisolone pulse therapy	360 days	High
Eggleston et al. [30]	9 (vs 6 without PE)	ASyS or DM and RP-ILD	3 anti-MDA5 positive; 3 anti-Ro-52 positive; 3 anti-Jo-1 positive	Daily/alternate daily Median 5 sessions	Development of acute hypoxemic respiratory failure, which was refractory to 1 or multiple immunosuppressive agents, and patient’s rapidly deteriorating clinical status	GC, IVIG, RTX, CYC, Mycophenolate, CNI	Primary outcome was 30-day mortality. Zero death occurred at 30 days in PE group compared to 2 deaths in non-PE group (33%, p = 0.14) Median peak oxygen requirement was met later during hospitalization for patients who received PE (9 days [IQR, 4–12 days]) than for those who did not receive PE (6 days [IQR, 5–16 days]) (p = 0.84) Two patients (22%) in PE group required invasive mechanical ventilation because of hypoxemic respiratory failure compared with 3 patients (50%) in the non-PE group (p = 0.33) Statistically significant improvement in 2 composite end points including 30-day mortality or need for lung transplant (0% in PE group vs 50% without PE, p = 0.04) and 1-year mortality or need for lung transplant or hospital readmission for RP-ILD in those receiving PE (22% vs 83%, p = 0.04)	Median 11 months (IQR 4.5–31.0 months)	Moderate
Thompson et al. [44]	8	IIM with treatment refractory ILD	3 anti-Jo-1 positive; 2 anti-MDA5 positive; 1 anti-PL-12 positive; 1 anti-SSA positive	Daily/alternate daily 5–7 sessions total	Treatment refractory ILD	GC, CYC	7/8 (88%) patients improved over the course of 1–6 PEs, over 1–10 days, with reduction in oxygen requirements. 1 patient died. 4 patients who had follow-up CT showed improvement	Insufficient information	Low
Bay et al. [31]	25 (vs 26 without PE)	Anti-MDA5 DM with RP-ILD	Anti-MDA5	Median 5 sessions	Nine patients (36%) received PLEX as first-line therapy whereas the remaining received PLEX during the subsequent lines of treatments	GC, MTX, Mycophenolate, CNI, JAKi, CYC, RTX, IVIG	1-year transplant free survival in PE group vs non-PE group was 20% vs 54%, p = 0.01. The Kaplan–Meier estimated a higher probability of 1-year transplant-free survival in non-PE compared to PE patients (p = 0.05) 1-year transplant-free survival for patients requiring mechanical ventilation was not different in PE vs non-PE group (5% vs 23%, p = 0.3) The use of PE was not associated with a better outcome, but they were given to sicker patients	7 (IQR 1–18) months	Moderate
Takahashi et al. [37]	2	Anti-MDA5 DM with RP-ILD	Anti-MDA5	1–3/week for 11–12 weeks Median 27.5 sessions	Patient 1: initiated after development of pseudomonas aeruginosa pneumonia leading to discontinuation of immunosuppressive drugs other than prednisolone. Patient 2: initiated after worsening of RP-ILD	GC, CNI, CYC, RTX	Patients had reduction in anti-MDA5 antibody titres, AaDO2 levels, oxygen requirements and improvement in CT findings post PE	7.5 months (7, 8 months)	High
Saito et al. [47]	6	Anti-MDA5 CADM with RP-ILD	Anti-MDA5	Alternate daily Median 9.5 sessions	Refractory—PE considered after worsening of respiratory conditions following primary therapy	GC, CYC, CNI, IVIG, Mycophenolate, RTX, PMX-DHP	Four patients improved, with reduction in oxygen requirements and improvement in CT appearances. One maintained remission at 6 months and 1 at 12 months. Two died due to respiratory failure. The survivors had lower ferritin, KL-6 and anti-MDA5-ab levels and fewer lung lesions post PE. Three patients had improvement in vital capacity predicted	60.5 weeks (IQR 15–119)	High
Ishiwari et al. [48]	2	Anti-MDA5 DM with RP-ILD	Anti-MDA5	3/week 6 sessions total	PE initiated after decline in respiratory function and/or skin lesions	GC, CNI, CYC	Improvement in titres of anti-MDA5 Ab and serum KL-6 and ferritin values following PE. Clinical symptoms were fully recovered, and serum KL-6 and ferritin decreased to the normal range. Consolidation and GGO completely resolved on chest HRCT in both cases. Both patients were discharged from hospital without any symptoms. One patient remained in remission for more than 1 year	One year in patient 1, 90 days in patient 2	Moderate
Yamagata et al. [49]	3	Anti-MDA5 CADM-RP interstitial pneumonia	Anti-MDA5	1–3/week for 3–4 weeks Median 7 sessions	Refractory—PE considered only after refractory to conventional immunosuppressive therapy	CYC, CNI, GC	IgG, fibrinogen, anti-MDA5 Ab, titre and KL-6 levels were significantly decreased following PE. Serum albumin level was not significantly changed Serum ferritin level showed statistically non-significant trend toward decreasing over the course of PE. Improvement in chest shadows on HRCT All patients survived	Median 14 months (range 13–30)	Moderate
Shirakashi et al. [32]	8 (vs 5 without PE)	Anti-MDA5-DM/CADM-ILD	Anti-MDA5	1–3/week for 3–15 consecutive weeks 11.4 ± 3.4 times for 60.9 ± 6.2 days	Relapse treatment—only following deterioration of respiratory function	GC, CNI, CYC	Cumulative survival for 3 years was 62.5 and 0% in those with PE and without PE, respectively (p = 0.001). In the PE group, 5/8 (63%) survived. In the non-PE group, 0/5 (0%) survived Reduction in ferritin—median 1620.0 (931.8–4842.6) ng/ml just before PE to 167.0 (57.8–297.1) 6 months after PE (for the 5 survivors) KL-6 1535.0 (942.3–1942.5) U/ml to 876.0 (312.5–1654.5). Anti-MDA5 U/ml 159.0 (144.1–215.2) to 36.8 (18.8–56.6)	4.7 ± 3.0 years (for entire cohort including non-PE patients)	Moderate
Abe et al. [33]	6 (vs 4 without PE)	Anti-MDA5 DM with refractory RP-ILD	Anti-MDA5; 2 had anti-Ro/SS-A ab	2 consecutive days, then 3/week Median 16 (IQR 14–22) sessions	Patients were treated with PE or conventional therapy as determined by their primary physicians	GC, CYC, CNI, IVIG	Primary outcome was non-specific mortality. The 1-year survival rate in the PE and non-PE groups was 100% and 25% (p = 0.033), respectively. The non-PE group tended to be diagnosed with RP-ILD earlier (p = 0.067)	Mean 13 months (range 1–47 months) (inclusive of non-PE group)	Moderate
Ning et al. [43]	18	PM/DM-ILD resistant to conventional therapies, admitted to ICU for ILD aggravation post failure of intensive treatment. 5 DM (28%), 11 CADM (61%), 2 PM (11%)	11 with anti Ro52, 12 with Anti-SSA, 1 with anti-Jo1. Anti-MDA5 and other myositis-specific antibodies not tested	Daily for 3 days, then alternate daily Median 5 sessions (responsive group); 6 sessions (non-responsive group), p = 0.724	Refractory—patients were resistant to conventional therapies and admitted to ICU for ILD aggravation post failure of intensive treatment	GC, CYC, CNI, IVIG, HCQ, MTX, Thalidomide	Seven patients (39%) died from respiratory failure after PE Eleven patients (61%) who survived showed great improvement in lung involvement, significant reduced HRCT scores, and their conditions were not life-threatening after treatment Patients divided into responsive (n = 11) and non-responsive (n = 7) groups. Responsiveness was defined as improved or controlled lung involvement and rescue from life-threating complications, whereas non-responsiveness was defined as aggressive lung involvement and death	Not defined	Moderate

AaDO2, alveolar-arterial oxygen partial pressure difference; ASyS, antisynthetase syndrome; CADM, clinically amyopathic dermatomyositis; DM, dermatomyositis; CNI, calcineurin inhibitor; CYC, cyclophosphamide; GC, glucocorticoid; GGO, ground-glass opacification/opacity; HCQ, hydroxychloroquine; HRCT, high-resolution computed tomography; IgG, immunoglobulin G; JAKi, Janus kinase inhibitor; KL-6, Krebs von den Lungen 6; ICU, intensive care unit; IIM, idiopathic inflammatory myopathies; ILD, interstitial lung disease; IVIG, intravenous immunoglobulin; MDA5, melanoma differentiation-associated protein 5; MTX, methotrexate; PE, plasma exchange; PMX-DHP, polymyxin-B direct hemoperfusion; PM, polymyositis; RP, rapidly progressive; RP-ILD, rapidly progressive interstitial lung disease; RTX, rituximab; SSA, Sjögren’s-syndrome-related antige

Anti-synthetase syndrome (ASyS)

Three studies involving 13 patients included patients with ASyS and/or positive anti-synthetase antibodies (Supplementary Table 2, Online Resource 4) [28, 30, 44]. Individual patient data were obtainable from only one study [44]. In this retrospective single group study by Thompson et al., three patients with ASyS (two with anti-Jo-1 and one with anti-PL-12 positivity) complicated by ILD underwent five to seven sessions of PE. All patients were on steroids, and some had additional CYC without improvement prior to commencement of PE. Following PE, all had clinical improvement in oxygenation and lung function.

Immune-mediated necrotising myopathies (IMNM)

Four studies included patients with IMNM or who had positive anti-signal recognition particle (SRP) or anti-3-hydroxy-3-methyl-glutaryl-coenzyme A reductase (HMGCR) antibodies (Supplementary Table 3, Online Resource 4) [28, 52–54]. Of these, three were anti-HMGCR positive and 14 were anti-SRP positive. The three patients positive for anti-HMGCR antibodies had prior statin exposure and prior immunosuppressive therapy including AZA, glucocorticoids, IVIg, MTX, mycophenolate, RTX, tacrolimus and tofacitinib. Following five to 11 sessions of PE, all were reported to have improvement in CK, clinical stabilisation and/or improvement. In the study by Kruse et al., two patients had serial anti-HMGCR antibody titres reported, both of whom demonstrated marked reduction following PE, with reported values decreasing from > 200 to ≤ 33; however, assay units were not specified [53].

Among the 14 anti-SRP positive patients, individual patient data was available for 12 patients. All 12 had prior immunosuppressant therapy including glucocorticoid, IVIg, CNI, AZA, CYC, MTX and mycophenolate prior to PE. The number of PE sessions ranged from one to 15. Two patients required a second course of PE, in combination with RTX and glucocorticoids. All except one patient were reported to have clinical and biochemical improvement.

Safety of PE

Safety outcomes were reported heterogeneously across included studies, with variable definitions, denominators and reporting units, precluding quantitative pooling. Overall, PE was generally well tolerated, with adverse events predominantly mild to moderate in severity. Importantly, no deaths were explicitly attributed to PE across the 20 studies reporting safety outcomes.

Adverse events temporally associated with PE included vascular access-related events (such as catheter infection and venous thrombosis), transient hypotension, hypocalcaemia and allergic reactions, including rare anaphylaxis related to plasma replacement. In contrast, infectious complications, including cytomegalovirus reactivation and opportunistic infections, were observed primarily in patients with anti-MDA5 RP-ILD receiving intensive concomitant immunosuppressive therapy, and were not consistently attributed to PE itself. Across comparative studies, there was no consistent signal of excess infection risk attributable to PE. Full details including absolute numbers of reported adverse events, where available, can be found in Supplementary Tables 1–3 in Online Resource 5.

Quality assessment and risk of bias

The only RCT included in this review scored 12 out of 14 using the NIH quality assessment tool for controlled intervention studies [28]. Uncontrolled intervention studies (n = 4) had a mean score of 4 ± 2 (range 2–6) on a scale of 0–9 using the JBI critical appraisal tool for quasi-experimental studies [38, 40, 42, 55–58]. Cohort studies (n = 7) had a mean score of 7 ± 1 (range 6–9) on a scale of 0–11, using the JBI critical appraisal checklist for cohort studies [29–33, 35, 50]. Single group studies (n = 12) had a mean score of 5 ± 2 (range 1–8) on a scale of 0–12 using the NIH quality assessment tool for before-after (pre-post) studies with no control group [36, 43, 44, 49, 51, 52, 59–64]. Case series (n = 11) had a mean score of 7 ± 1 (range 5–9) on a scale of 0–10, using the JBI checklist for case series [34, 37, 39, 41, 45–48, 53, 54, 65]. Overall study quality was deemed high in ten studies, moderate in 16 and low in nine. The detailed scores of each study are shown in Online Resource 5.

Discussion

In this systematic review of 35 studies including 473 adult patients, PE was frequently used as an adjunct to immunosuppressive therapy across multiple IIM phenotypes and was associated with reported improvements in a variety of clinical and biochemical outcomes in observational series. Pooled data for mortality in anti-MDA5/ASyS patients with RP-ILD did not demonstrate a statistically significant reduction in mortality (RR 0.41, 95% CI 0.16–1.06; I² = 70%), and heterogeneity and confounding by indication limit causal inference. The single small RCT that evaluated PE as monotherapy in steroid-resistant PM/DM did not demonstrate benefit over sham apheresis [28]. Nonetheless, multiple cohort studies and case series, particularly in anti-MDA5 RP-ILD and in selected IMNM cases, report clinical and biochemical improvements after PE when used alongside combination immunosuppressive therapy. Interpretation of these findings should be undertaken in the context of the quality and design of the underlying evidence. Most included studies were observational and lacked control groups, and PE was frequently used as rescue therapy alongside escalation of immunosuppression. Consequently, observed improvements should not be interpreted as evidence of independent therapeutic efficacy of PE and may reflect confounding by indication, concurrent treatment effects, survivor bias, or regression to the mean, contributing to the overall low certainty of evidence.

When to consider PE

Based on synthesis of the included studies, interpreted in the context of expert consensus statements and reviews, including international apheresis guidelines, PE may be considered in selected clinical scenarios. In the 2023 ASFA guidelines, PE for IIM (ASyS, CADM and IMNM) is classified as a category III indication with a grade 2B recommendation, indicating that its optimal role is not definitively established and that individualised clinical decision-making is required [18]. Within this context, clinical scenarios in which PE can be considered include: (1) rapidly progressive, life-threatening ILD (e.g. anti-MDA5 RP-ILD) where rapid removal of circulating mediators may stabilise respiratory failure; (2) severe dysphagia or rapidly progressive disease refractory to standard immunosuppression and (3) seropositive IIM where there is growing evidence for the pathogenicity of certain myositis-specific autoantibodies (MSAs) and correlation of antibody titres with disease activity [66–70]. Across the included studies, PE was almost exclusively used as an adjunctive therapy rather than as a stand-alone intervention, with reported clinical benefit occurring in the context of combination immunosuppressive treatment. Timing, optimal exchange schedule, replacement fluid and the precise combination of concomitant immunosuppression remain unanswered practical questions. Available data suggest that PE is most often employed as salvage or rescue therapy in patients with rapidly progressive or refractory disease, particularly in anti-MDA5 RP-ILD. However, delayed initiation may limit potential benefit if irreversible lung injury or muscle damage is already established. Conversely, the absence of evidence supporting PE as monotherapy, together with its transient effect on circulating mediators, argues against its routine early use in all patients with IIM.

Practical considerations are also central to the clinical application of PE. PE is a resource-intensive intervention, requiring specialised equipment, trained personnel, vascular access and repeated procedures over a short time frame [14]. Availability may be limited to tertiary or selected centres with appropriate logistical set-ups, reinforcing the need for selective rather than routine use.

Biologic rationale and interpretation

The proposed mechanism by which PE could benefit certain IIM phenotypes is chiefly removal of circulating pathogenic autoantibodies and immune complexes (for example anti-MDA5, anti-HMGCR, anti-SRP), as well as potential modulation of inflammatory cytokines [15]. However, an important caveat is that the pathogenicity of some myositis autoantibodies is incompletely understood: while anti-HMGCR and anti-SRP autoantibodies are supported by experimental and clinical evidence of direct pathogenicity, the pathogenic role of anti-MDA5 is less clearly established and may in part reflect broader immune activation, including type I interferon-mediated pathways [66–71]. Thus, observed reductions in autoantibody titres after PE should be interpreted as biologically plausible but not definitively causal. Furthermore, PE does not address ongoing autoantibody production, therefore supporting the case for combination with agents that reduce antibody generation (e.g. RTX, CYC or mycophenolate).

PE in combination immunosuppressive regimens

In our review, outcomes appeared more frequently favourable when PE was used with other DMARDs, as opposed to with steroids alone, with 186/254 (73%) showing clinical or biochemical improvement in the former compared to 16/28 (57%) in the steroid-alone group. It has been hypothesised that the bulk removal of immunoglobulin by PE might lead to the removal of negative feedback on the antibody-producing cells, resulting in rebound hyperproduction of immunoglobulin, either from proliferation of antibody-producing cells and/or an increase in synthetic activity of these cells [15]. This could increase the susceptibility of these cells to immunosuppressant and chemotherapeutic agents.

In an RCT by Miller et al., PE was shown to be ineffective when used as monotherapy for corticosteroid-resistant PM or DM in 39 patients [28]. In this trial, patients demonstrated no improvement or deterioration in clinical status following PE, with outcomes similar to comparator groups who had either sham apheresis or leukapheresis. In contrast, early single-centre experience reported by Dau et al. described clinical improvement in 46 of 48 adult patients with PM, DM or CTD-associated IIM, characterised by increased strength, reduced CK or improved muscle-biopsy findings [55–57]. In these reports, PE was administered alongside immunosuppressive therapy, including daily low-dose CYC or chlorambucil. Based on individual patient data presented in the publication, patients underwent a median of 12 PE sessions (IQR 8.5–16.5; range 4–39) per treatment course [55]. These publications represent related reports from the same institution with partial cohort overlap. Although predating contemporary immunosuppressive strategies and standardised outcome measures, this early experience provides insight into historical clinical practice and the feasibility of PE in IIM.

More recently, a 2023 retrospective cohort study by Eggleston et al. involving anti-MDA5 DM patients with ILD demonstrated a statistically significant improvement in two composite endpoints including 30-day mortality or need for lung transplant (0% vs 50% in PE compared to non-PE groups respectively, p = 0.04) and 1-year mortality or need for lung transplant or hospital readmission for RP-ILD in those receiving PE (22% vs 83%, p = 0.04) [30]. Eight of the nine patients in the PE group were on adjunctive immunosuppressive therapy besides glucocorticoids, including IVIg in five (56%), RTX in four (44%) and CYC in five (56%), compared to IVIg in three (50%), RTX in one (17%) and CYC in three (50%) in the non-PE group. Similarly, in 2020, Shirakashi et al. reported on a cohort of eight patients who received PE for treatment-refractory anti-MDA5 DM with RP-ILD [32]. Compared with a similar cohort of five patients who did not receive PE, PE was associated with a significant survival benefit (1-year survival 63% vs 0% respectively). All patients had prior high-dose GC, CNI and CYC. Additional therapy used in the non-PE group included etanercept (n = 1), tocilizumab (n = 2), IVIg (n = 1) and CYC (n = 1). Collectively, these studies suggest that in similar comparator groups, the addition of PE to combination immunosuppressive therapy appears to have added benefit.

PE use in ILD and specific IIM subtypes

In IIM patients with RP-ILD, we found no association of PE with mortality reduction. Our results did not reach statistical significance in our meta-analysis (RR 0.41, 95% CI 0.16–1.06, I² = 70%). Our results do not echo the findings of two recently published meta-analyses. A meta-analysis by Yang et al. found that 6-month survival rate was significantly better in the PE group compared with non-PE group. The review included a study by Vuillard et al., which we excluded due to insufficient information regarding the PE group [72]. Tangborwornweerakul et al. found that patients receiving PE in addition to other therapies demonstrated a greater 1-year survival rate (RR 1.59, 95% CI 0.96–2.65, I² = 52%) than did patients without PE [73]. The review included one fewer study than ours, as this study was published after their literature search concluded in August 2023 [34]. Notably, significance was reached in a sensitivity analysis only after excluding one outlier (RR 1.71, 95% CI 1.30–2.25; I² = 0%) [31]. This study, which we included in our meta-analysis, reported a contrary result, showing lower 1-year survival rate in the PE group [31]. In this retrospective cohort study involving anti-MDA5 DM with RP-ILD, patients received a combination of glucocorticoids and immunosuppressants with or without PE. However, notable differences existed in baseline characteristics between the treatment arms including symptom duration, rate of mechanical ventilation and extracorporeal life support and immunosuppressant use. These factors suggest greater disease severity in the PE group and presence of a selection bias and could account for the higher mortality observed in the PE group.

Our review found that PE was associated with reductions in KL-6, ferritin and anti-MDA5 antibody levels. Both KL‐6 and hyper-ferritinaemia are associated with increased mortality and a poorer prognosis in ILD and are useful biomarkers for the evaluation of the response to treatment and the status of ILD in patients with anti-MDA5-positive DM [74–77]. In our report, the reduction in KL-6 and ferritin levels was temporally associated with clinical improvement and survival in the majority of patients in whom serial measurements were available. The pathogenicity of anti-MDA5 antibodies has not been fully elucidated, although research suggests that anti-MDA5 antibodies are involved in the overproduction of type I interferons, leading to inflammation and tissue damage. In our review, patients who had reduction in anti-MDA5 antibody titres post-PE demonstrated clinical improvement. The removal of these antibodies by PE and subsequent suppression of inflammatory response may represent a biologically plausible mechanism underlying observed associations.

Our review found that PE was associated with clinical improvement in both ASyS and IMNM patients. In the limited number of patients with ASyS included in the review, all three showed clinical improvement in ILD following PE. The small number of patients makes this observation underpowered to draw firm conclusions. Similarly, PE was associated with clinical improvement in all three anti-HMGCR antibody-positive patients and was associated with reductions in anti-HMGCR antibody titres. Experimental and clinical data support a direct pathogenic role for anti-HMGCR antibodies, with in vitro and ex vivo evidence of complement-mediated muscle injury and impaired muscle regeneration [66, 67, 69]. Although serial anti-SRP antibody titre testing is not routinely performed, there is some evidence that anti-SRP antibodies are pathogenic, with available data demonstrating associations between autoantibody titres, muscle injury and disease activity in patients with necrotising myopathy [67, 68, 70].

Collectively, these data support consideration of PE as an adjunctive or rescue therapy in selected IIM subtypes, enabling rapid removal of pathogenic antibodies and immune complexes. However, despite its mechanistic rationale, contemporary practice often favours B-cell–directed therapies such as RTX due to greater practicality, ease of use and sustained immunological effect, with PE considered primarily in severe, refractory or rapidly progressive presentations. This evolving treatment paradigm represents an important limitation, as it restricts the routine applicability of PE and necessitates careful patient selection.

Safety of PE

Overall, PE appeared to be generally well tolerated across the included studies, with serious procedure-related adverse events reported infrequently and no procedure-related mortality reported. However, interpretation of safety data is limited by heterogeneous and non-standardised adverse event reporting and by the frequent use of PE in critically ill patients receiving intensive concomitant immunosuppression. Safety data from other autoimmune and neurological indications provide useful contextual information, with large observational series reporting low rates of serious procedure-related complications [78, 79]. Korkmaz et al. reviewed the efficacy and safety of PE in 265 patients and 1274 procedures over an 11-year period [80]. Observed adverse events were recorded in 33 (13%) patients, the most common being allergic reaction, seen in 15 (6%) patients, followed by catheter-related complications, seen in seven (3%) patients. While these data support the procedural safety of PE in experienced centres, extrapolation to IIM should be undertaken cautiously given differences in disease severity and background immunosuppression. Taken together, available data suggest that PE is generally well tolerated when used selectively, but limitations in attribution and reporting preclude definitive conclusions regarding procedure-specific risk in IIM.

Limitations

There were several limitations of this systematic review. First, the majority of included studies were observational (retrospective cohorts, single-group interventions and case series), frequently small and underpowered, resulting in low-certainty evidence vulnerable to selection bias and confounding by indication and incomplete outcome reporting. Only a small number of comparative studies were available, limiting causal inference regarding the effectiveness of PE. Second, there was marked clinical and methodological heterogeneity across studies, including disease phenotypes, PE regimens (number of exchanges, timing, exchange volume), concomitant immunosuppression, outcome definitions and assessment time points. Variable outcome reporting limited assessment of the clinical significance or magnitude of observed effects. This heterogeneity precluded meaningful quantitative synthesis for most outcomes and restricted meta-analysis to a single clinically coherent subgroup (mortality in anti-MDA5/ASyS RP-ILD). Third, diagnostic and classification criteria varied across studies. The majority of studies included IIM patients diagnosed by the EULAR/ACR Classification Criteria for Adult and Juvenile IIM [24] or Peter and Bohan Criteria (1975) [22, 23]. Criticisms of these criteria include lack of specification of how to exclude other forms of myopathy and nonexplicit definition of inclusion criteria, as well as the lack of inclusion of MSAs, which has been a relatively recent development [81]. This complicates phenotype stratification and subgroup interpretation, particularly for antibody-mediated disease subsets where PE might be biologically most relevant. Fourth, outcome assessment windows were inconsistently reported; and for several outcomes (e.g. muscle strength or CK), temporal relationships to PE could not be standardised, limiting interpretability of treatment effects. Finally, publication bias is possible because small positive series are more likely to be published than negative reports. Collectively, these limitations result in low to very low certainty of evidence for most clinically important outcomes and underscore the need for prospective studies with standardised outcome measures.

Conclusion

In adult IIM, the available evidence from 35 studies encompassing 473 patients supports a biologically plausible but clinically selective role for PE. PE has been reported to be associated with clinical and biochemical improvements in selected patients, especially when used as an adjunct to combination immunosuppressive therapy in severe phenotypes such as CADM/anti-MDA5 RP-ILD and in some cases of IMNM and ASyS. Current evidence is derived predominantly from observational studies with substantial heterogeneity and high risk of bias, limiting causal inference and resulting in a low certainty of evidence. Pooled data from seven studies in anti-MDA5/ASyS-associated RP-ILD did not show a statistically significant mortality benefit. PE should therefore be considered a potential adjunctive therapy for carefully selected, severely ill patients, rather than a standard routine component of IIM management, with its potential benefits weighed against procedural risks, resource intensity and the availability of alternative immunomodulatory therapies with more sustained effects. Future research should prioritise prospective, multicentre registries and controlled studies in well-defined severe disease contexts, incorporating standardised PE regimens, clear pre-specified outcome measures, harmonised safety reporting and systematic documentation of assessment time points and concomitant therapies to better define optimal patient selection, timing and dosing for PE in IIM.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Kimberly Loh, Gowri Manoharan and Kirsten Richards. The first draft of the manuscript was written by Kimberly Loh, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

Open access funding provided by South Metropolitan Health Service, WA Health Libraries Network Consortium.

Data availability

All data are available in the manuscript and supplementary materials.

Declarations

Statement of ethics and consent

Ethics approval was not required for this systematic review. This study was conducted in accordance with the ethical stands laid down in the 1964 Declaration of Helsinki and its later amendments.

Disclosures

None.