Role of therapeutic plasma exchange in the management of COVID-19-induced cytokine storm syndrome

The risk of mortality in patients with coronavirus disease 2019 (COVID-19) is largely related to an excessive immune response, resulting in a hyperinflammatory and hypercoagulable condition collectively referred to as cytokine storm syndrome (CSS). Management of critically ill patients with COVID-19 has included attempts to abate this process, prevent disease progression, and reduce mortality. In this context, therapeutic plasma exchange (TPE) offers an approach to eliminate inflammatory factors and cytokines, offset the pathologic coagulopathy, and reduce the CSS effects. The aim of this review is to analyze available data on the use of TPE for the treatment of CSS in patients with COVID-19. Systematic searches of PubMed, Scopus and COVID-19 Research were conducted to identify articles published between March 1, 2020 and May 26, 2021 reporting the use of TPE for the treatment of COVID-19-induced CSS. A total of 34 peer-reviewed articles (1 randomized controlled trial, 4 matched case-control series, 15 single-group case series, and 14 case reports), including 267 patients, were selected. Despite the low evidence level of the available data, TPE appeared to be a safe intervention for critically ill patients with COVID-19-induced CSS. Although inconsistencies exist between studies, they showed a general trend for decreased interleukin-6, C-reactive protein, ferritin, D-dimer, and fibrinogen levels and increased lymphocyte counts following TPE, supporting the immunomodulatory effect of this treatment. Moreover, TPE was associated with improvements in clinical outcomes in critically ill patients with COVID-19. While TPE may offer a valuable option to treat patients with COVID-19-induced CSS, high-quality randomized controlled clinical trials are needed to confirm its potential clinical benefits, feasibility, and safety. Moreover, clear criteria should be established to identify patients with CSS who might benefit from TPE.

Currently, there is no approved specific treatment for COVID-19, but various therapeutic agents (e.g., tocilizumab, steroids) showed some level of effectiveness [16]. Besides supportive/standard care, management of patients with COVID-19 might also include timely control of the CSS to prevent disease aggravation and reduce mortality [4,15]. In this context, potentially effective treatment approaches include administration of immunomodulators, cytokine antagonists, monoclonal antibodies, and anti-inflammatory drugs [4,15,17,18]. Therapeutic plasma exchange (TPE) may also be a valuable option to control CSS by removing inflammatory markers and cytokines [4,15,19,20]. The purpose of this review is to compile and analyze available data on the use of TPE for the treatment of CSS in patients with COVID-19.

Methods
A systematic literature search was conducted to identify studies using TPE in hospitalized patients with COVID-19. Systematic searches of PubMed, Scopus, and a Dialog database called COVID-19 Research were conducted for articles published between March 1, 2020 and May 26, 2021. The searches were performed with the following terms: ("plasma exchange" OR "plasmapheresis") AND ("coronavirus" OR "COVID-19 ′′ OR "SARS-CoV-2 ′′ OR "2019-nCoV"). The search was done on April 24, 2020, and weekly updates were provided thereafter.
Identified articles were screened by one reviewer. Relevant papers were selected if at least one COVID-19 patient received TPE. Articles were excluded if COVID-19 was not the reason for TPE treatment initiation, the technique was unclear, or the article was not written in English. Systematic literature reviews and meta-analyses were excluded, but their reference lists were checked for relevant articles that might have been overlooked. Subsequently, data concerning the characteristics of patients and their disease, TPE procedures and adjunct treatments, outcomes (laboratory parameters and clinical outcomes), and TPE safety were extracted from selected articles.
Methodologic classification of articles was performed using the Oxford Centre for Evidence-Based Medicine levels by two assessors, with differences resolved by consensus [21]. The severity of COVID-19-induced CSS in the selected studies was evaluated using the Penn grading scale by two assessors [22].

Patients and disease characteristics
While most TPE-treated patients had critical or life-threatening COVID-19, a few patients with severe or moderate disease were also included in the selected studies (Table 1). Using the Penn grading scale [22], we estimated that all TPE-treated patients had grade 3 or 4 CSS (except one patient with grade 2 CSS). Available Sequential Organ Failure Assessment (SOFA) scores, which are based on the degree of organ dysfunction [57] and may help to predict outcomes in critically ill patients, ranged between 2 and 15 (median ranging from 3 to 11).

Evolution of immune-inflammatory biomarkers
Since the main objectives of TPE are to decrease pro-inflammatory cytokines levels and correct coagulopathies, dynamic monitoring of these parameters is useful to evaluate the ability of TPE to abate the CSS.
While three studies using FFP or artificial Octaplas LG (Octapharma, Manchester, UK; a pooled FFP product that has undergone pathogen inactivation) as replacement fluid showed that fibrinogen levels decreased following TPE [24,40,48], fibrinogen levels seemed stable in two studies using albumin or a mix of albumin and FFP as replacement fluid [42,44] (Table 3). Four studies showed that the activity of disintegrin and metalloproteinase with a thrombospondin type 1 motif, member 13 (ADAMTS-13; von Willebrand factor-cleaving protease) [64] increased after TPE [23,24,29,46]. Platelet count decreased in two studies [39,54], but tended to increase in two other studies [48,56]. In a fifth study, platelet count decreased in some patients but increased in others [42]. Viscosity decreased after TPE in a study in patients with COVID-19 hyperviscosity [40].
In the matched case-control study including the highest number of TPE-treated patients, lower D-dimer and IL-6 levels were observed in the TPE versus the control group, while no differences were observed in terms of ferritin, CRP, platelet, and lymphocyte count [26] (Table 3).
Similarly to mortality rates, a high variability was observed in terms of lengths of stay (LOS) in the ICU or hospital (Table 4). In the RCT, ICU LOS was shorter for patients in the TPE versus the control group (19 versus 26 days) [23]. Results of matched case-control series were inconsistent, with two studies showing increased LOS in hospital or ICU [25,27], and another study showing shorter hospitalizations [26] in TPE recipients versus matched controls.

Discussion
The pathophysiology of sepsis involves a complex interaction of inflammation, endothelial dysfunction, and pathologic activation of coagulation [68]. These dysregulations appear common to sepsis from multiple inciting pathogens, including SARS-CoV-2, and much of the morbidity is due to the abnormal host response rather than the infection itself [69,70]. In contrast to many therapies that are targeting different components of this pathway [68,69], TPE offers a potential non-specific therapeutic modality. Although evidence for TPE efficacy in sepsis is not robust, available data suggested potential clinical efficacy and safety [71][72][73][74]. Based on these data, the American Society for Apheresis (ASFA) issued a Category III (optimum role of TPE is not established and   decision making should be individualized), Grade 2B (weak recommendation based on moderate-quality evidence) recommendation for TPE to improve organ function by removing inflammatory and antifibrinolytic mediators and replenishing anticoagulant proteins, to reverse the pathobiological derangement, and to restore hemostasis in patients with sepsis with MODS, allowing for individual consideration on a case by case basis [75].
Considering the similar pathophysiology of severe COVID-19 and sepsis, TPE may be beneficial in patients with fulminant COVID-19 infection and it was utilized in selected cases since the onset of the pandemic [20,76,77]. Our literature review identified 267 patients included in 34 studies with published results. In these studies, TPE was almost exclusively utilized as rescue or adjunct therapy in patients with critical or life-threatening COVID-19 disease. While limited by the largely retrospective nature of available data, TPE was shown to be feasible, safe, and often clinically efficacious for these patients.
In the identified studies, the main reasons to initiate TPE were the presence of septic shock, MODS, and/or ARDS [23,[27][28][29][30][31][33][34][35]37,38,41,42,44,46,48,49,52]. The earliest reports of successful use of TPE to treat severe COVID-19 infections included case reports or small case series in patients with MODS, in line with the ASFA indications of sepsis with multiple organ failure. Keith et al. reported an early case of COVID-19-induced pneumonia complicated by ARDS, sepsis with vasopressor-dependent hypotension, acute renal failure, and viral cardiomyopathy, who responded to one TPE session using FFP [49]. Shi et al. reported another case of a patient with severe COVID-19, respiratory failure and vasopressor-dependent hypotension who had resolution of shock and organ failures after four TPE sessions [54]. In a large case series, 11 severely ill patients with COVID-19 (ARDS, severe pneumonia, septic shock, and/or MODS) responded favorably to TPE compared with patients who received standard care [27]. Patients receiving TPE experienced higher extubation rates (73% versus 20%; P = 0.018) and lower all-cause 28-day mortality rates (0 vs. 35%; P = 0.033). A single prospective RCT in 87 patients with life-threatening COVID-19 showed that the addition of TPE to standard care was associated with a statistically insignificant decrease in 35-day mortality (20.9% versus 34.1%; P = 0.09) and statistically significant decreases in number of days on mechanical ventilation and ICU LOS [23]. This study also supported the feasibility and safety of TPE in this setting. Other studies have shown variable clinical responses, as summarized in Table 4.
While ASFA criteria must be met (sepsis with MODS) to perform TPE    in the United States [75], criteria are more variable and the decision is often at the discretion of the physician in other jurisdictions. In several studies, increases in serum markers of inflammation and coagulation, which are indicators of CSS, were used as triggers to initiate TPE [23-26, 29,31,39,44,48,51,52,56]. In a pilot study, ten COVID-19 patients meeting criteria for Penn class 3 or 4 CSS were identified as candidates for TPE and showed rapid improvements in oxygenation and significant reductions in biomarkers of cytokine load [32]. Kamran et al. retrospectively analyzed the clinical and biochemical effects of TPE in 90 patients with COVID-19-induced CSS (defined by specific biomarker levels) using propensity score matching [26]. TPE recipients demonstrated statistically significantly improved 28-day survival (91.1% versus 61.5%), shorter hospital LOS (10 versus 15 days), and shorter time to CSS resolution (6 versus 12 days). In the single prospective RCT, CSS-associated biomarkers decreased significantly with TPE [23]. Several other studies also reported an immunomodulatory effect of TPE through decreases in IL-6, CRP, ferritin, and D-dimer levels, and elevations in lymphocyte counts, even if these findings were not observed in all reports [23][24][25]27,29,[31][32][33][34][35][36][37][38][39][40][41][42][43][44][46][47][48]51,52,55,56]. While the time to CSS resolution was evaluated in one study [26], other studies focused on patient outcomes, symptom improvements, and evolution of immune-inflammatory markers, highlighting the need for standardized definitions of CSS resolution.
The strong systemic cytokine release in severely ill patients with COVID-19 generates numerous phenotypes that look similar to other diseases, often collectively referred to as cytokine storms [78][79][80]. These include macrophage activation syndrome (MAS), secondary hemophagocytic lymphohistiocytosis (sHLH), and thrombocytopenia-associated multiple organ failure (TAMOF) [78,81]. Many of these diseases that COVID-19 can mimic were shown to improve with TPE and may be considered as separate entities [80,[82][83][84][85][86]. Although Gluck et al. utilized the Penn grading scale for CSS to identify patients eligible for TPE treatment [32], scales evaluating CSS severity [5,22] were not used in the other studies to guide the therapeutic strategy for critically ill patients with COVID-19 due to need for quick treatment decision, lack of knowledge of these scales by clinicians, and their absence in international guidelines. The Penn grading scale is based on diagnostic and clinical aspects and distinguishes among mild, moderate, severe, and life-threatening CSS [22]. When we applied this scale to the other studies, we found that almost all TPE-treated patients met criteria for Penn class 3 or 4 CSS. Because the Penn grading scale or other grading scales are not specific for COVID-19-induced CSS, they may potentially be used to identify patients with CSS (caused by any condition) who could benefit from TPE [22].
While clinical and biochemical responses to TPE were often favorable, legitimate concerns were voiced. Many clinicians are worried that the removal of anti-SARS-CoV-2 neutralizing antibodies and other host defenses may be clinically detrimental [87]. The net effect on the host immune response cannot be interpreted through available data in this analysis, but very few adverse events were attributed to TPE and none were considered life-threatening. While these data do not directly address the concerns of those skeptical of the intervention, they reaffirm the safety of TPE in the context of sepsis.
Logistics of the TPE treatment(s) were highly variable. While a single TPE session was performed in some studies, the number of TPE sessions most often ranged from three to five [23,25-30,32,33,36,41,42,44-48, 50,51,53,54,56]. A daily frequency seemed optimal considering the short half-lives of cytokines, and TPE sessions were mainly performed daily or every other day [23,24,26-32,38-42,44-48,50,51,54,56]. In general, the volume of exchanged plasma was based on the total plasma volume of patients, and although different methods were used to determine the volume, it generally ranged between 0.75 and 1.5 plasma volume [23,26,29,31,32,38,40,42,44,46,48]. These observations are consistent with ASFA guidelines for the treatment of sepsis with MODS, where daily TPE sessions for 1-14 days, or until the resolution of symptoms, are recommended with an exchanged volume of 1-1.5 plasma volume [75]. The most frequently used replacement fluid were FFP or artificial Octaplas [23,24,27,29,31,32,34,35,37,38,40,41,43,[46][47][48][49][50]54]. These replacement fluids offer potential superiority over albumin solutions based on the pathways previously described, manifesting as endothelial injury and microthromboses in multiple organs [29,88]. When using FFP as replacement fluid, large, prothrombotic multimers are removed along with antibodies to ADAMTS-13, ADAMTS-13 is replenished, microthrombosis risk is theoretically reduced, and tissue perfusion is improved [29,89]. CCP has also been used as partial replacement fluid to compensate the removal of anti-SARS-Cov-2 neutralizing antibodies [33,42,52,53]. While the use of albumin may result in depletion of procoagulant factors and increased bleeding risk, some providers implement 5% albumin as replacement fluid (or a mixture of albumin and FFP) to avoid the replenishment of immune response effectors, such as complement, cytokines, and chemokines, and the decreases in coagulation factors [77]. A recent publication has reported an immunomodulatory effect of albumin through interaction with endosomal Toll-like receptors in leukocytes from patients with cirrhosis [90].
While promising, results from available studies are difficult to interpret due to multiple limitations. The biggest limitation is the retrospective nature of nearly all available data even if we attempted to ensure inclusion of only higher-quality reports. Our review may also be limited by the fact that the search, screening, and article selection were performed by one author. Interpretation is further limited because in the absence of a universally established standard of care for patients with COVID-19, TPE-treated patients often received other drugs, and treatment regimens were heterogeneous. A further limitation is the fact that positive results are more frequently reported in publications, leading to a risk of underreporting of data on unsuccessful interventions. Nevertheless, it is important to note that almost all studies consistently reported feasibility and safety while observing clinical and biochemical efficacy of TPE, despite geographical variations and discrepancies in terms of treatment regimen, study endpoints, and eligibility criteria. These observations lay the foundation and confirm the need for welldesigned RCTs to evaluate the utility of TPE for the treatment of COVID-19-induced CSS.

Conclusion
Although the evidence level was low and treatment regimens were heterogeneous in the selected studies, available data suggest that TPE alone or in combination with other drugs should be considered as a safe and valuable option for the treatment of critically ill patients with COVID-19-induced CSS. While high-quality RCTs are needed to confirm the clinical benefits of this treatment, available data suggest that CSS should be considered as a standalone pathological manifestation caused by multiple underlying diseases. Therefore, clear criteria should be defined to classify patients with CSS and to facilitate the identification of those eligible for TPE treatment. CI, confidence interval; COVID-19, coronavirus disease 2019; CSS, cytokine storm syndrome; CT, computed tomography; DVT, deep vein thrombosis; ECMO, extracorporeal membrane oxygenation; GCS, Glasgow Coma Scale; HFOT, high-flow oxygen therapy; ICU, intensive care unit; IQR, interquartile range; IMV, invasive mechanical ventilation; LOS, length of stay; MV, mechanical ventilation; NA, not applicable; NIMV, non-invasive mechanical ventilation; NR, not reported; OI, oxygenation index; PaO2:FiO2, pressure of arterial oxygen to fractional inspired oxygen concentration; SOFA, Sequential Organ Failure Assessment; SD, standard deviation; SP02, partial arterial pressure of oxygen; TPE, therapeutic plasma exchange; UK, United Kingdom; USA, United States of America. * Statistically significant difference.