Efficacy of therapeutic plasma exchange in severe COVID‐19 disease: A meta‐analysis

Abstract

Background and Objectives

Therapeutic plasma exchange (TPE) has been used in severe COVID‐19 disease to eliminate the cytokine storm. This meta‐analysis aims to assess the effectiveness of TPE in reducing mortality in severe COVID‐19 disease compared to standard treatment.

Materials and Methods

A comprehensive literature search was performed in PubMed, the Cochrane database and the International Clinical Trial Registry Platform (ICTRP). The random‐effect model was used to calculate the risk ratio and standardized mean difference (SMD) as pooled effect size for the difference in mortality and length of the intensive care unit (ICU) stay. The risk of bias and publication bias were assessed in R version 4.1.0. The certainty of the evidence was calculated using the GradePro tool.

Results

The database identified 382 participants from six studies, including one randomized control trial. Egger's test did not detect any publication bias (p = 0.178). The random model analysis for mortality evaluated a risk ratio of 0.38 (95% CI: 0.28–0.52) with a significant reduction in the TPE group. The certainty of the evidence was moderate, with a risk ratio of 0.34 (95% CI: 0.24–0.49). Length of ICU stays between TPE versus standard care showed an SMD of 0.08 (95% CI: −0.38, 0.55) and was not significant.

Conclusion

The length of ICU stay in the TPE group was not different from standard care. However, this meta‐analysis revealed a significant benefit of TPE in reducing mortality in severe COVID‐19 disease compared to standard treatment.

Highlights.

• Therapeutic plasma exchange (TPE) has been used to treat diseases with cytokine storms.

• To date, definitive treatment for severe COVID‐19 disease is unavailable.

• This meta‐analysis revealed a significant benefit of TPE in reducing mortality in severe COVID‐19 disease with moderate certainty of evidence.

INTRODUCTION

The disease COVID‐19 caused by SARS‐CoV‐2, a single‐stranded RNA virus of the beta coronavirus genus, was first reported in December 2019 in Wuhan city of China [1]. The droplet‐mediated spread of novel coronavirus from human to human through the respiratory route has resulted in worldwide suffering in the form of increased mortality and morbidity. On 11 April 2022, the World Health Organization (WHO) reported 6,179,104 cumulative deaths due to COVID‐19 [2]. Many studies and trials conducted worldwide have failed to find a specific cure for the disease. This virus enters with angiotensin converting enzyme (ACE2) receptors distributed widely in the epithelium and endothelium of the lungs, kidney and gastrointestinal, and it can potentially lead to multiorgan involvement and death in some cases [3]. The severe COVID‐19 disease manifests as cytokine release syndrome, resulting in elevated interleukin (IL)‐1 and IL‐6, TNF‐ α, lactate dehydrogenase, D‐dimer, ferritin and C‐reactive protein (CRP) [4]. The condition is also associated with microvascular thrombosis and clot formation features. The severe diseased state in COVID‐19 is defined as having a respiratory rate of >30/min, breathlessness or SpO₂ of <90% on room air by the Indian Council of Medical Research‐COVID‐19‐National Task Force [5].

Many drugs like azithromycin, doxycycline, supplements like zinc and vitamin C and other drugs like hydroxychloroquine, ivermectin, favipiravir and injection remdesivir, were used to attenuate the disease severity with debatable effects [5]. It has also been observed that approximately half of the patients with cytokine storms gradually develop severe acute respiratory distress syndrome (ARDS), and mortality in the severe category range from 40% to 50% [6, 7]. Tocilizumab, an IL‐6 antagonist, was used to reduce the disease progression in the severe category but only provided conflicting results [8]. The convalescent plasma (CP) from the recovered donor also failed to provide conclusive evidence in various studies and trials [9].

Therapeutic plasma exchange (TPE) is recommended to address the hyperinflammation in sepsis with multiorgan failure and haemophagocytic syndrome in haemophagocytic histiocytosis for the removal of cytokines and immune complexes involved in deranged coagulation and clot formation [10]. Various studies and trials conducted across the world find some favourable evidence [11], and this meta‐analysis was planned to generate evidence from published observational studies and trials exploring the efficacy of TPE in attenuating disease progression in severe COVID‐19, which will eventually help in framing the therapeutic guideline.

MATERIALS AND METHODS

Search strategy

A systematic literature search was performed independently by four review authors (S.P., S.M., A.S. and S.S.R.) using PubMed, the Cochrane database and the International Clinical Trial Registry Platform (ICTRP). The literature search included prospective clinical trials and observational studies on the efficacy of TPE procedures in reducing morbidity (length of intensive care unit [ICU] stay) and mortality in severe COVID‐19 patients. The search strategy was not restricted by the date of publication and the clinical or genetic variants of COVID‐19. However, articles published in the English language only were included for meta‐analysis. Population, intervention, control and outcomes (PICO) scheme was followed for reporting inclusion criteria. The key elements used in our search using MESH terms are the ‘P’ (COVID‐19, SARS‐CoV‐2), the ‘I’ (Plasma exchange, Plasmapheresis, TPE), the ‘C’ (Coronavirus infections/therapy, Oxygenation) and the ‘O’ (Efficacy), that is, mortality in TPE group versus non‐TPE control group, ICU stay, inflammatory markers, such as IL‐6, CRP, D‐dimer and Ferritin. The adverse events associated with the TPE procedure, for example, hypocalcaemia, hypotension and vasovagal attack, were also included for review.

The meta‐analysis research protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) vide registration number CRD42021273748 and approved by the institutional ethics committee (IEC), All India Institute of Medical Sciences, Bhubaneswar [12, 13].

Study selection criteria

Types of studies

Prospective clinical trials, observational studies and retrospective studies that had evaluated the effect of plasma exchange on mortality as a primary outcome were included in this meta‐analysis. Review articles, letters to the editor, comments, case reports and studies where it was impossible to retrieve or calculate data of interest were excluded from this review.

Types of participants

Severe COVID‐19 patients who had undergone TPE were compared with patients on conventional therapy and not undergone the TPE procedure.

Types of interventions

• Experimental: TPE procedure, in addition to conventional therapy performed on COVID‐19 patients with ARDS in different clinical conditions.

• Control: COVID‐19 patients with ARDS on conventional therapy and have not undergone TPE.

Outcome measures

Primary outcome

1. Effect of TPE in reducing mortality.

Secondary outcomes

1. Length of ICU stay (in days)

2. Change in the cytokine level of IL‐6 and CRP after TPE.

3. TPE procedure‐related adverse events, for example, vasovagal syncope, hypocalcaemia, hypotension, etc.

Data extraction and risk of bias statement

For this meta‐analysis, four review authors (S.P., S.M., A.S. and S.S.R.) independently screened the titles, abstracts and keywords of all references retrieved. The authors then obtained and assessed the full text from all selected studies and assessed the quality using guidelines published by the Cochrane Collaboration [14]. The risk of bias assessment of included studies was separately performed for randomized controlled trials and observational studies with the ‘ROB2’ and ‘ROBINS‐I’ tools of the ‘robvis’ package in R programming using the Cochrane Risk of Bias Assessment tool and judged them as low, moderate, serious, critical and low, some concern and high, respectively. Any disagreement between the review authors was resolved by consensus or consultation with the clinical pharmacologist cum statistical advisor (R.M.). The authors converted the median and interquartile range from the studies into mean and standard deviation with the help of an online calculator formulated and described by Luo et al.[15] and Wan et al. [16]. Extracted data includes:

• Publication type and source

• The trial design includes timing, follow‐up, sequence generation and allocation concealment

• Setting including country, level of care

• Participants including selection criteria, number of dropouts

• Interventions, that is, TPE and related adverse events

• Outcome measures include mortality, ICU stays (in days), inflammatory cytokine markers before and after TPE, and drugs used for standard care.

Statistical analysis

This meta‐analysis was conducted using the various package in the R programming language (version 4.1.0) [17].

The risk ratio and standardized mean difference (SMD) were calculated to estimate the effect size for mortality and length of ICU stays (in days), respectively. The forest plot was prepared using the random‐effect model for between‐group analyses. The outcome has been depicted as a point estimate with a 95% confidence interval. The chi‐square test was used to assess whether observed differences in results are compatible with chance alone. I ² statistics, an estimate due to heterogeneity, was done to quantify inconsistency. A prediction interval was also calculated to predict the outcome probability of a new observation later.

The sensitivity analysis was performed to test the robustness of the results in case of significant heterogeneity in ICU stays. We constructed funnel plots and performed the Eggers' regression test as a quantitative test for publication bias. Standard Cochrane methodology and the GRADE Working Group guidance were followed to create the ‘Summary of findings’ table. The five grade considerations (risk of bias, consistency, imprecision, indirectness and publication bias) of the included studies were considered to conclude the certainty of the evidence for the outcome [18].

RESULTS

The database identified 410 publications and reduced them to 117 after removing duplicates, trials in progress, only protocol published and irrelevant studies manually. Further screening of the remaining studies and excluding case reports, reviews, letters to editors, comments, and studies in the Chinese language, 17 studies sought eligibility [19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35]. Finally, we included six studies, as shown in the PRISMA flow chart (Figure 1). One was a randomized trial among six studies, and the remaining five were either prospective or retrospective observational studies [30, 31, 32, 33, 34, 35]. The details of the included and excluded studies are described in Table 1 and Table 2, respectively. We could not observe any publication bias in the studies. The funnel plot is not asymmetrical, and Eggers' test showed a p‐value of 0.178(95% CI: −2.66, 0.24; Figure 2). The risk of bias plot has been produced by ‘robvis’ package via the ‘rob_traffic_light’ function separately for the randomized trial and observational study, as shown in Figure 3a,b.

FIGURE 1.

Study identification and selection process as per PRISMA guideline 2020

TABLE 1.

Characteristics of included studies comparing mortality in the intervention versus standard care

Author name, year and location	Study type	No. of participants	Intervention/apheresis machine	Outcome (mortality)
				Intervention arm	Standard care
Dai et al., 2020, China	Prospective case control	101	TPE/not mentioned	16% (8/50)	50.98% (26/51)
Faqihi et al., 2021, Saudi Arabia	Open‐label randomized trial	87 in intervention and 43 in standard care	TPE/centrifugal ‘Spectra optia’	20.9% (9/43)	34.1% (15/44)
Gucyetmez et al., 2020, Turkey	Retrospective	24	TPE/not mentioned	8.3% (1/12)	58.3% (7/12)
Kamran et al., 2021, Pakistan	Retrospective (propensity‐matched control)	90	TPE/centrifugal ‘Cobe Spectra’	8.9% (5/45)	38.5% (18/45)
Khamis et al., 2020, Oman	Prospective (with hypothetical control; medical records)	31	TPE/centrifugal ‘Spectra Optia’	9.1% (1/11)	45% (9/20)
Nusshag et al., 2021, Germany	Retrospective case control	49	TPE/centrifugal ‘Comtech’	39.28% (11/28)	95.23% (20/21)

Abbreviation: TPE, therapeutic plasma exchange.

TABLE 2.

Characteristics of excluded studies comparing mortality in the intervention versus standard care

Author name, year and location	Study type	Total no. participants	Intervention	Outcome (mortality)	Reasons for exclusion
Adeli et al., 2020, Iran	Prospective	8	TPE	12.5% (1/8)	No comparator
Zhang et al., 2020, China	Case series	3	TPE	Nil	Comparator not available
Dogan et al., 2020, Turkey	Case series	6	TPE	16.6% (1/6)	No comparator
Fernandez et al., 2020, Spain	Case series	4	TPE	Nil	No comparator
Gluck et al., 2020, USA	Single‐arm trial without placebo randomization	6 of 10	TPE on Spectra Optia apheresis system	Nil	No comparator
Faqihi et al., 2020, Saudi Arabia	A pilot study of a randomized trial	10	TPE	(1/10) 10%	No comparator and (main study trial included for meta‐analysis)
Hashemian et al., 2020, Tehran	Prospective	15	Plasmapheresis with haemodialysis machine	40% (6/15)	No comparator
Jaiswal et al., 2021, United Arab Emirates	Prospective	14	TPE followed by CP transfusion	28.6% (4/14)	No comparator
Morath et al., 2020, Germany	Retrospective	5	Plasma exchange	40% (2/5)	No comparator
Matsushita et al., 2021, Japan	Retrospective	5	TPE and haemodiafiltration	60% (3/5)	No comparator
Roshandel et al., 2021, Iran	Prospective	5	TPE followed by CP transfusion	20% (1/5)	No comparator

Abbreviation: TPE, therapeutic plasma exchange.

FIGURE 2.

Contour‐enhanced funnel plot of the effect estimate. TPE, therapeutic plasma exchange

FIGURE 3.

Risk of bias graph, clinical trial (a), observational study (b)

The primary outcome, that is, reducing mortality in the intervention arm (TPE), was compared with standard care among six included studies. The test of heterogeneity was not significant (heterogeneity: χ2 = 0, I ² = 0%, p = 0.55). The random model analysis of the studies evaluated a risk ratio of 0.38 (95% CI: 0.28–0.52; Figure 4a) with a significant reduction in mortality in the TPE group (p < 0.01).

FIGURE 4.

Forest plot (a): Analysing the risk ratio of mortality in TPE (intervention) versus no TPE (standard care) group, Forest plot (b): Illustrating the difference in length of ICU stay (in days) between TPE (intervention) and No TPE (standard care) group. CI, confidence interval; ICU, intensive care unit; RR, risk ratio; SD, standard deviation; SMD, standardized mean difference; TPE, therapeutic plasma exchange

The length of ICU stay in included six studies showed significant heterogeneity (heterogeneity: I ²  = 57%, p = 0.04). The random model analysis showed a SMD of 0.08 (95% CI: −0.38, 0.55; Figure 4b). The sensitivity analysis was performed because of significant heterogeneity by excluding studies sequentially. However, changes in heterogeneity were minimal, and the overall p‐value was not significant, as summarized in Table 3.

TABLE 3.

Sensitivity analysis by excluding studies one by one with no change in overall effect (ICU stay in days)

Study excluded	SMD	95% CI	p (overall effect)	I 2	P (I 2)
Dai et al.	0.18	−0.40, 0.76	0.43	60%	0.04
Faqihi et al.	0.18	−0.40, 0.75	0.44	61%	0.04
Gucyetmez et al.	−0.01	−0.50, 0.47	0.94	51%	0.08
Kamran et al.	0.17	−0.42, 0.76	0.47	63%	0.03
Khamis et al.	−0.02	−0.51, 0.48	0.93	52%	0.08
Nusshag et al.	0.01	−0.55, 0.58	0.96	55%	0.06

Abbreviations: CI, confidence interval; ICU, intensive care unit; SMD, standardized mean difference.

We could not analyse other secondary outcome measures like change in cytokine level, sequential organ failure assessment (SOFA) score, and adverse events as data obtained from these studies were not adequate. Only a few studies documented hypotension and allergic reaction, and any serious adverse events were not encountered. The evidence was graded using online GradePro software. There were five observational and only one randomized trial, and separate grading was recommended for various types of study. The Grade assessment for the primary outcome was calculated for five observational studies, including 295 participants, and the risk ratio was significant with a moderate grade of evidence shown in Table 4.

TABLE 4.

Grade of evidence for the primary outcome (mortality)

Summary of findings:
[Therapeutic plasma exchange] compared to [Standard care] for [Severe COVID‐19 disease patients]
Patient or population: [Severe COVID‐19 disease patients] Setting: Intervention: [Therapeutic plasma exchange] Comparison: [Standard care]
Outcomes	Anticipated absolute effects a (95% CI)		Relative effect (95% CI)	No. of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with [Standard care]	Risk with [Therapeutic plasma exchange]
Mortality	537 per 1000	183 per 1000 (129–263)	RR 0.34 (0.24–0.49)	295 (5 observational studies)	⨁⨁⨁◯ Moderate b

Note: GRADE Working Group grades of evidence: High certainty, we are very confident that the true effect lies close to that of the estimate of the effect; Moderate certainty, we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different; Low certainty, our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect; Very low certainty, we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Abbreviations: CI, confidence interval; RR, risk ratio.

^a The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

^b The sample size of a few studies was not sufficient.

DISCUSSION

TPE is one of the life‐saving modalities in diseases with underlying cytokine storms such as sepsis and haemophagocytic syndrome. In the initial phase after SARS‐CoV‐2 infection, lymphocytosis ensues, resulting in a rise in the inflammatory and chemotactic cytokines like TNF‐ α, IL‐6, IL‐1β and MCP‐1. This hypercytokinaemia was found to lower the lymphocyte count, thus further reducing viral clearance. Immunomodulators like steroids, tocilizumab, cyclooxygenase inhibitors, ACE inhibitors and extracorporeal therapy were used to treat hypercytokinaemia. The direct removal of these inflammatory mediators by TPE may help to prevent T‐cell exhaustion [36]. The favourable change in laboratory profile after TPE was evident in many studies. However, few authors opposed the benefits of TPE in reducing mortality and disagree with the assumption of improving survival by simply decreasing cytokine levels of IL‐6, CRP and TNF‐ α [37, 38]. This meta‐analysis was planned to see the clinical efficacy of TPE in reducing mortality in severe COVID‐19 disease.

The meta‐analysis shows a significant reduction in mortality in severe COVID‐19 patients despite no change in the ICU stays. The random model analysis reflects no heterogeneity, and overall effects were significantly in favour of the TPE procedure in reducing the mortality in severe COVID‐19 patients. The benefit of TPE in reducing mortality was evident clearly with the prediction interval in favour of the intervention group. The insight of improving survival by removing harmful inflammatory mediators and antibodies and replacing the deficient blood components in the TPE procedure was found useful in many studies. The COVID‐19‐related coagulopathy was associated with a decrease in antithrombin, disintegrin, metalloproteinases like ADAMTS‐13, and the increase in von‐Willebrand multimers (vWF) and D‐dimers level [39, 40]. The benefit of reducing mortality by TPE in patients with elevated D‐dimer levels was shown in the study of Gucyetmez et al. [35]. TPE, thus, may be a rational therapeutic approach by replacing these deficient components and correcting coagulopathy. As sought in some case reports, the removal of inflammatory mediators and harmful antibodies to interferon‐I was also found protective [41]. The TPE was proposed by a few authors to cause immune paralysis by reducing IL‐10 levels, an immunosuppressive cytokine. However, an elevated level of IL‐10 was associated with a decrease in lymphocyte count and increased neutrophil‐to‐lymphocyte ratio (NLR). Previous studies also quoted decreased lymphocyte count and increased NLR as a poor prognostic marker in COVID‐19 [42, 43]. Again, the TPE procedure may help eliminate the elevated IL‐10 levels, reverse the NLR and improve the cellular profile. The change in these inflammatory mediators also improved the SOFA score, organ dysfunction and overall survival and was not just a change in laboratory markers, as suggested by a few authors. The claim of immune paralysis by removing protecting antibodies through TPE may also be refuted as the trial on CP fails to provide any mortality benefits in various age groups and severe categories of patients with COVID‐19. This evidence further disagrees with the use of CP as a replacement fluid for the TPE procedure [34]. The use of fresh frozen plasma (FFP) as replacement fluid was also debatable due to its hypercoagulable state. However, deficient metalloproteinase and an increase in vWF multimers were also found to cause a hypercoagulable state in COVID‐19. Thus, the use of FFP as a replacement fluid may be beneficial. Five out of six studies [30, 31, 32, 33, 34] used plasma as the replacement fluid, and one study [35] has not mentioned the type of replacement fluid used for the TPE procedure. Few participants in this meta‐analysis were also receiving tocilizumab, and a claim of a decrease in cytokines due to this IL‐6 inhibitor was proposed. However, an instant reduction in IL‐6 and other cytokines was observed post‐TPE procedure in patients who did not receive the tocilizumab [33]. Specific drugs like tocilizumab were not much beneficial in controlling inflammation in COVID‐19, and it may be due to a diverse array of immunopathology in COVID‐19 [44].

Further, the broad‐spectrum oral and inhaled steroid was shown to reduce the severity, morbidity and mortality of COVID‐19. TPE treatment also appears to act in a wide spectrum by removing many pathogenic substances and replacing deficient components. Moreover, like many other autoimmune diseases and diseases with inflammatory components, TPE had shown a good response in patients refractory to steroid treatment. Thus, TPE may be used as life‐saving last resort therapy in severe COVID‐19 refractory to steroid and standard care.

The higher mortality in ICU was associated with a longer length of stay in the previous study. Prolonged immunosuppression, nosocomial infections and protein‐energy malnutrition causing myopathy were responsible for poor outcomes with increased ICU stay, irrespective of mechanical ventilation [45]. In this meta‐analysis, the length of ICU stay was not different between the intervention and standard care arm. The heterogeneity was high, and the p‐value of heterogeneity was significant (Figure 4b). However, sensitivity analysis by sequentially excluding the study fails to get an overall significant difference in length of ICU to stay in TPE and non‐TPE groups. (Table 3) TPE procedure was performed in 2–5 cycles on a daily or alternate day basis and was responsible for increasing the day of ICU admission. Further, no difference in ICU stay can be explained due to the increased day of admission with associated comorbidities like chronic kidney disease patients on dialysis support, hypertension and diabetes. Random inclusion of more seriously ill patients in the TPE group also contributed to increased days of ICU admission [30, 33].

The TPE procedure's safety has already been established, and the procedure itself had minimal adverse events. Most of the adverse events in TPE are hypotension, allergic reaction to replacement fluid and citrate‐related side effects [46]. All these adverse events can easily be managed by medical intervention. In this meta‐analysis, mild hypotension and allergic reaction were encountered by a few authors and managed by saline fluid bolus and supportive management. Serious adverse events, including mortality related to the TPE procedure itself, were not found in any of the participants in the intervention arm (TPE). Recent studies mentioned the change in the level of cytokines before and after the TPE procedure, and the change was significant for IL‐6, CRP, SOFA score, LDH and ferritin levels [23, 26, 30, 32, 35]. However, we could not conduct a meta‐analysis of a few secondary outcomes like adverse events to TPE procedures and the change in the level of various cytokines due to inadequate data in both the intervention and standard care arm.

We have a few limitations in our meta‐analysis. This meta‐analysis included only one randomized trial and a few observational and retrospective studies. The chance of selection bias, the inadequate sample size in a few studies and other unclear risks of bias may be the drawback of these studies, which could hamper the suitable comparison between intervention and standard care groups. However, most of the included studies in this meta‐analysis performed the matching of participants for age, gender, demography, clinical and laboratory parameters. Thus, selection bias may be compensated.

In conclusion, the evidence generated with this meta‐analysis suggests TPE as a promising and safe therapeutic approach for a subset of patients with severe COVID‐19. However, the certainty of evidence in reducing mortality is moderate. Hence, further high‐quality large randomized controlled trials are desirable to attain factual inference on the efficacy of TPE in severe COVID‐19.

CONFLICT OF INTEREST

The authors declare that there is no conflict of interest.

Prakash S, Sahu A, Routray SS, Maiti R, Mitra JK, Mukherjee S. Efficacy of therapeutic plasma exchange in severe COVID‐19 disease: A meta‐analysis. Vox Sang. 2023;118:49–58.