Immune Response Resetting as a Novel Strategy to Overcome SARS-CoV-2–Induced Cytokine Storm

A cluster of patients diagnosed with severe pneumonia in Wuhan, China, in December 2019 led to the subsequent isolation of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (1). It causes coronavirus disease 2019 (COVID-19), which rapidly evolved into one of the most aggressive pandemics of modern times, affecting millions of people and claiming hundreds of thousands of lives worldwide. COVID-19 is associated with a wide clinical spectrum that encompasses asymptomatic infection, mild upper respiratory tract disease, and severe pneumonia with respiratory failure (1, 2). Sepsis is the most frequent and serious complication reported in COVID-19 patients that occurs in virtually 100% of nonsurvivors (2). It presents as a life-threatening organ dysfunction with (or without) underlying circulatory and cellular/metabolic abnormalities, in which case it is branded as septic shock (3). The second and third most common complications are respiratory failure and acute respiratory distress syndrome (ARDS) (2).

The staple treatment for SARS-CoV-2 infection is essentially symptomatic for mild cases and consists of life-supportive measures with oxygen therapy for the more-severe and critical clinical presentations (4). The latter often require attention in intensive care units (ICUs), with ventilatory assistance for sepsis with ARDS; management of sepsis-induced complications such as coagulopathy, acute cardiac/kidney injury, and secondary infections; as well as hemodynamic support for those that evolve into septic shock (5). Beyond the established approach, the therapeutic value of several novel or repurposed drugs is being examined. A nonexhaustive shortlist of repurposed medicines includes the following: chloroquine/hydroxychloroquine, antiretrovirals, antihelminthic agents, and biologics such as anti–IL-6R antagonist mAb (6). Convalescent plasma transfusion (7) and low-dose corticosteroids (8) are also under evaluation with promising preliminary results. In addition, the clinical improvement initially associated with the compassionate use of the nucleotide analogue remdesivir (9) was further studied in four randomized clinical trials (RCTs), with the combined recruitment of 2281 patients, that failed to provide a final clear verdict (10–14). Therefore, as of today, no specific antiviral therapy for COVID-19 patients had completed the entire validation cycle of RCTs (6). There are several vaccine candidates that are in advanced stage of clinical validation and hopefully will be available soon. Unfortunately, even a safe and efficacious product will entail serious hurdles concerning manufacturing infrastructure, distribution, and other logistic issues to be cleared in a massive scale, perhaps never seen before.

The severe morbidity and lethality observed in COVID-19 are caused by excessive host response to infection (15, 16). Retrospective analyses of patients infected with SARS-CoV-2 revealed that plasma levels of multiple proinflammatory cytokines, such as IL-1β, IL-17, IFN-γ, TNF-α and IL-6, were all augmented (15–18). Most importantly, patients that end up in ICUs were more likely to have higher levels of IL-2, IL-7, IL-10, G-CSF, IP-10, MCP1, MIP1A, and TNF-α (16). There was a remarkable association between IL-6 and death, with nearly twice as much higher cytokine titer in the plasma of 68 individuals who perished out of 150 COVID-19 patients (17). The resulting cytokine storm puts the immune system in overdrive at the expense of the virus-specific immune response, which is likely to underlie the most advanced COVID-19 clinical features, including sepsis-induced coagulopathy, sepsis-related multiple organ dysfunction, as well as the pathophysiological changes found in the lungs during ARDS (2, 16, 17).

The patients’ immune cells simultaneously enable and are shaped by the cytokine storm. Thus, SARS-CoV-2 infection is commonly associated with lymphopenia with a reported increase in the neutrophil/lymphocyte ratio in the most critical cases (2, 18). There is a profound depletion of CD3⁺ T cells, spanning both CD8⁺ and CD4⁺ subsets, with an even higher drop of the latter in severe disease (18). The resulting state of immunosuppression is reminiscent of the late stages of sepsis triggered by other pathogens in that CD3⁺ T lymphocytes tend to get hyperactivated, acquire an exhausted phenotype, and succumb to depletion by apoptosis (19). It is relevant to mention that hyperactivated CD3⁺ T cells, as indicated by HLA-DR and CD38 double positivity, were reported in the peripheral blood of a COVID-19 ARDS patient, alongside proinflammatory CCR6⁺ TH17 CD4⁺ T cells (20). In addition, the study of a cohort of 68 SARS-CoV-2–infected individuals revealed increased frequency of T lymphocytes and NK cells displaying the inhibitory receptor NKG2A (21), which has been recently described as an exhaustion marker (22). It is noteworthy that cytokines with an overt dampening effect on antiviral adaptive and innate immunity, such as IL-10, are also upregulated alongside classical proinflammatory mediators in COVID-19 (16, 18). IL-10 may induce T cell exhaustion in other viral disease models, and it would not be unreasonable to suspect a similar role here (23, 24).

A recent comprehensive analysis of the cellular immunity in COVID-19 largely confirmed previous reports, describing lymphopenia and highly activated CD4⁺/CD8⁺ T cells that express CD38, HLA-DR, Ki-67, as well as the inhibitory marker PD-1 in the acute phase of the infection (25). Unexpectedly, the same report also describes highly functional SARS-CoV-2–specific T cells in a subset of seronegative family members, asymptomatic individuals, and in mild COVID-19 patients. Other groups have also found robust virus-specific T cell immunity in mild and ICU-admitted COVID-19 patients as well as in unexposed individuals (26–29). At least part of this response seems to derive from cross-reactivity to previous exposure to other endemic coronaviruses (26–29). In this unfolding story, a recent report described a drop in antiviral IgG and neutralizing serum Abs within 2–3 mo postinfection in patients that recover from COVID-19 (30). Thus, a proper assessment of the antiviral immune response after a previous infection or immunization has to include cellular immunity to be reliable.

SARS-CoV-2 may be seen as a new (old) foe in the sense that it is clearly a new pathogen but one that causes a disease that has many familiar complications. The current pandemic creates the daunting task of learning about COVID-19 and, at the same time, proposing solutions based on the existing meager and sketchy readout of the underlying pathophysiology. The complexity of the matter is indicated by the unexpected finding that the antiviral humoral response may be short-lived, which put in check any preliminary vaccine efficacy assumption based on Ab titer. In the best scenario, natural infection and the vaccines that are being developed will induce a strong, enduring cellular antiviral immunity that ultimately confers long-term protection to those exposed or immunized. However, an unconventional threat like COVID-19 might as well require a mix of classic and unconventional solutions.

Starting from the sound premise that the cytokine storm is a dominant feature in both sepsis and COVID-19 and that the most common COVID-19 complication is sepsis itself, we will examine a few representative experimental therapies for sepsis that have been tested over the years as well as new outside-of-the-box alternatives that might have relevance in the context of the current pandemic. Treatments that have not been previously considered for sepsis were not included in this review.

Investigators have overtly borrowed a page from therapeutic trials in sepsis to address COVID-19. This section is entirely dedicated to sepsis, whereas the corresponding trials on repurposed therapies for COVID-19 are summarized in Table I. Study acronyms are defined in Table II.

There have been multiple attempts to revert what seemed to be a hyperinflammatory status that hijacked the patients’ response to the etiological insult in sepsis (36). Anti–TNF-α therapies have been extensively tested (37, 38), and although the results obtained in animals were encouraging (37, 39), the clinical experience was less forthcoming (40). It became clear that the treatment could lower cytokine plasma levels and yet did not increase survival (40). Thus, the use of the polyclonal anti–TNF-α Fab AZD9773 did not improve ventilator-free days, mortality rates, or relative risk of death (41, 42). In fact, a similar therapy with the anti–TNF-α neutralizing fusion protein TNFR:Fc even increased mortality (43). In contrast, the MONARCS RCT (n = 2634) tested the anti–TNF-α mAb afelimomab in patients with sepsis and found a modest but significant reduction in mortality in those with high IL-6 serum titer at randomization (44). Altogether, the effort to target TNF-α in sepsis so far has been underwhelming at best.

Although IL-6 plays a prominent role in the cytokine storm observed in the acute phase of sepsis, it has not been part of any major RCT focused on this condition, excluding those now ongoing for COVID-19 (45). Instead, there are many studies performed with IL-1 blockade in critically ill patients (46). The IL-1R antagonist (IL-1ra, anakinra) generated optimism in murine sepsis models and in a phase II RCT, which found survival benefit for sepsis patients (47). Disappointingly, two larger phase III RCTs (n = 893 and 696) failed to confirm the survival increase attributed to IL-1ra in patients with sepsis and septic shock (48, 49). However, secondary retrospective analyses of the data from the first of these trials suggested a dose-related increase in survival for patients with organ dysfunction (48). Also, a subset reanalysis of the second RCT identified a significant reduction in mortality for the patients with sepsis and concurrent hepatobiliary dysfunction and disseminated intravascular coagulation that were treated with IL-1ra (50). Accepting its obvious caveats, this second look at the old data has opened the door for further trials (51). In fact, PROVIDE is an ongoing phase II study that will randomize patients with severe infections and sepsis to receive IL-1ra, rIFN-γ, or placebo according to hyper- or hypoinflammatory profiles (NCT03332225).

The term immunothrombosis defines the interdependency of endothelial and platelet biology, coagulation, and inflammation (52). Many attempts have been made in the past to alter the course of immunothrombosis in sepsis (53, 54). Among them, a large phase III study randomized 1261 patients with sepsis to receive recombinant platelet-activating factor acetylhydrolase or placebo and failed to show decrease in mortality (55). Given that complement acts as an archetypical danger sensor, its inhibition was also considered to limit inflammation and thrombotic complications of sepsis. The SCIENS phase II RCT (n = 72) tested the anti-C5a mAb (CaCP29) in patients with septic organ dysfunction, but the results are pending (NCT02246595).

The endothelium is the setting for immunothrombosis in septic patients (52, 56). Adrenomedullin (ADM) is a peptide hormone that stabilizes the endothelial barrier and keeps vascular integrity (57, 58). The nonneutralizing ADM-binding mAb adrecizumab traps ADM within the vascular lumen where it binds to receptors on the endothelial cells, reducing vascular leakage (58, 59). Preclinical tests revealed an improved vascular barrier function and survival in rodent sepsis models (59, 60). The phase II AdrenOSS-2 RCT studied the effect of adrecizumab in 301 patients with septic shock and elevated ADM concentration (NCT03085758). The results have not yet been published.

Four large clinical trials illustrate the difficulty of treating sepsis-induced coagulopathy: OPTIMIST (n = 1955) with tissue factor pathway inhibitor tifacogin, KyberSept (n = 2314) with antithrombin III, PROWESS-SHOCK (n = 1697) with activated protein C, and SCARLET (n = 800) with thrombomodulin—all failed to show survival benefit for patients with sepsis or septic shock (61–64).

The rationale for the passive administration of IVIg in sepsis is multifaceted and goes beyond the scope of this review (65). Besides potentially neutralizing and/or opsonizing pathogens, IVIg may also deplete noxious products, such as upregulated cytokines and chemokines (66). More than 40 RCTs on the use of IVIg in sepsis have been conducted over the years, including the SBITS phase III study (n = 653) that did not show mortality reduction (67). Despite lack of consensus (68, 69), one of the most comprehensive meta-analyses found lack of evidence of survival benefit of this therapy in pediatric patients and unclear benefit for adults (70). However, another meta-analysis of 19 smaller studies found that IgM-enriched Ig (IVIgM) reduced mortality in adults with sepsis (71). A trial to test IVIgM in 200 peritonitis/sepsis patients was initiated, but its status is currently suspended (NCT03334006) (72).

Another way to reduce proinflammatory mediators in sepsis is through the administration of mesenchymal stem cells (MSC) (73). The latter have immunosuppressive effect by favoring the M2 phenotype of macrophages, which produce IL-10 and TGF-β and support tissue repair (74, 75). In addition, MSC inhibit the inflammasome, thereby reducing IL-1β and IL-18 production (73) and, indirectly, lowering the systemic level of other cytokines, such as IL-6. Some of this immune modulatory activity can also be present in derived exosomes (76). There is little clinical experience with cellular immunotherapy in sepsis, but two first-in-human studies showed safety (77, 78). Although the CISS trial was underpowered to access changes induced in the plasma levels of cytokines, a patient-specific, dose-dependent, transient early dampening of several mediators (IL-1β, IL-2, IL-6, IL-8, and MCP-1) was identified (77). The other phase I study (RUMСESS) found improvement in 28-d but not in 90-d survival of neutropenic patients with septic shock (78). The phase II randomized trial CHOCMSC is under way to evaluate the efficacy of this form of cellular immunotherapy for septic shock (NCT02883803).

Corticosteroids produced by adrenocortical cells and their synthetic analogues have multiple biological effects, including immune modulation (78, 79). Several clinical trials conducted over the years provided discordant results (80–83). Among them, ADRENAL stands as one of the largest RCTs (n = 3800), which showed that hydrocortisone did not decrease mortality in septic shock patients who needed mechanical ventilation (83). Nevertheless, a very recent Cochrane review of 61 trials with 12,192 participants found moderate evidence that corticosteroids may reduce mortality but identified high evidence that they may decrease ICU and hospital stay (80). There is also some renewed interest in the combination of hydrocortisone, ascorbic acid, and thiamine for the treatment of sepsis, with the establishment of several RCTs, such as the ACTS trial (NCT03389555) (84).

Hyperinflammation is a central feature of sepsis, but it comes alongside immunosuppression, with lymphopenia being one of the most consistent findings (19, 85). Thus, therapeutic strategies designed solely to dampen inflammation, such as most anticytokine approaches, may have serious limitations in yielding a full, sequelae-free, long-term recovery (86). Immune stimulation with IL-7, GM-CSF, or IFN-γ as well as checkpoint inhibition might be part of the solution. Indeed, ex vivo studies revealed that IL-7 could rescue lymphocytes derived from septic shock patients from apoptosis and improved proliferation (87–89). In addition, the IRIS-7-B phase IIb RCT (n = 27) showed that the glycosylated recombinant human IL-7 (CYT107) was well tolerated, did not increase the cytokine storm, and reversed lymphopenia in patients with septic shock (NCT02640807) (90). These very encouraging results need now to be confirmed in a larger study.

There are some 50 reported cases of sepsis that received rIFN-γ, and the treatment appears to have been well tolerated (91, 92). In a recent prospective case series, IFN-γ was associated with an increase in monocytic HLA-DR as well as a decrease in IL-6 and IL-10 plasma levels (92). Nevertheless, RCTs are needed to ascertain the putative benefit of IFN-γ on sepsis survival: one phase II trial is ongoing (NCT03332225), and one phase III study was completed with results not yet published (NCT01649921).

GM-CSF is a hematopoietic growth factor that was shown to upregulate HLA-DR expression in monocytes and their capacity to release proinflammatory cytokines in ex vivo blood cultures from patients with sepsis (93, 94). A phase II RCT (n = 38) found that rGM-CSF treatment was safe and could reverse the monocyte phenotypic and functional deactivation in patients with sepsis and septic shock (NCT00252915) (95). The combined results from several other trials generally point to some benefit as regards recovery from infection and other clinical endpoints but also to failure in improving survival (93). The results are pending for the GRID phase III RCT (n = 166) that tested GM-CSF in patients with sepsis and septic shock (NCT02361528).

T cells tend to exhibit an exhausted phenotype in sepsis with higher expression of inhibitory receptors and have lower effector function (96). Blockade of the PD-1/PD-L1 pathway is an attractive means of reverting exhaustion (97). The anti–PD-1 mAb nivolumab was well tolerated in two early phase clinical trials for sepsis (NCT02960854 and JapicCTI-173600) (98, 99). Similarly, the anti–PD-L1 mAb BMS-936559 was well tolerated and did not induce hypercytokinemia in a phase Ib study (NCT02576457E) (100). Although checkpoint inhibition holds promise, its safety and efficacy in sepsis has yet to be fully defined.

Many of the strategies discussed above are now being tested in COVID-19 as illustrated in Table I.

The reasons for failure to translate sepsis preclinical research into valid treatment options are multiple and complex (36, 101). Albeit the laboratory mouse has undeniable value for studying immune responses in general, mimicking certain aspects of the immunopathology of human disease in murine models is far from perfect (102). Thus, for instance, immunothrombosis does not unfold in the same way in mice and humans, largely because in the former, neutrophils tend to undergo NETosis more often and platelets are more numerous (36). Perhaps even more important, ignoring the impact that the natural microbiome has in shaping the immune response might have induced serious underestimation of sepsis severity in the past (103, 104). Indeed, most preclinical studies used mice kept in specific pathogen-free (SPF) conditions (104–106). Now we know that the host’s cumulative exposure to commensal and pathogenic microbes potentiates the cytokine storm in polymicrobial sepsis, and the magnitude of this response is somewhat muted in SPF animals (103). Incidentally, the experiments that had previously shown protection from death by an anti–TNF-α Ab and the TNFR:Fc fusion protein in SPF mouse-based sepsis models could not be reproduced in animals that had reacquired their natural wild-type microbiome (104). Provided we learn from our mistakes, the above context gives hope that perhaps some of the previous attempts to treat sepsis might be rescuable to address COVID-19. The use of microbially experienced mice will be particularly helpful to test some of these ideas before moving into clinical trials (103, 104). Nevertheless, given the long list of previous treatment failures, one should be open-minded to embrace a bench-to-bedside-to-bench bidirectional flow (107) as well as new technologies that heighten the ability to mine for valuable information in already available clinical and omics datasets (108–110). Thinking outside the box is a must to effectively address the cytokine storm and sepsis that dominate the most serious presentations of COVID-19 and that are largely responsible for the paralysis that this pandemic is causing globally.

The adaptive immune system is classically associated with the development of protective memory responses against pathogens, which are characterized by heightened effector function and high specificity upon a second encounter (111). There is mounting experimental evidence suggesting that the innate immune system also possesses the ability to “remember” a previous encounter so as to influence its future reaction against the same or even unrelated pathogens (112). This process—defined as trained immunity—is a consequence of an intricate interplay of epigenetic alterations and metabolic reprograming (113). Thus, vaccination with the bacillus Calmette–Guérin (BCG) leads to genome-wide changes, such as the acetylation of histone 3 lysine 27 (H3K27ac) in monocytes (114). These are epigenetic activity marks that are found in the vicinity of several genes, including those encoding cytokines, growth factors (e.g., EGFR), and signaling pathway kinases (e.g., MAPKs, PI3K). Of particular importance is the marking of genes involved in the PI3K/AKT/mTOR pathway that have a central role in the metabolic shift that enhances aerobic glycolysis and favors cytokine production during induction of trained immunity (112, 113). IL-1 seems to be a key player in implementing the effect of BCG on reprogramming granulocyte-monocyte progenitor cells in the bone marrow (114).

Well before the mechanistic basis of trained immunity was envisaged, there were epidemiological cues that pointed toward a possible nonspecific cross-protection of vaccines against pathogens other than those against which they were generated (115, 116). Thus, type 2 attenuated poliovirus vaccine protected against type 1 poliomyelitis during a pandemic in Singapore in 1959 (115). Similarly, the analysis of 10 cohort studies on measles immunization conducted in developing countries between 1974 and 1991 revealed a reduction in childhood mortality of 30–86% (116). Obviously, the specific protection against measles could not account for an effect of such a magnitude, and it was speculated that the measles vaccine would activate the immune system in a nonspecific way, thereby protecting against other infections (116). There is additional evidence that suggests the heterologous protection of vaccines, including reports that BCG given at birth induces a nonspecific protection against neonatal sepsis and respiratory infection (117, 118) as well as a recent report that shows that BCG vaccination protects adults against subsequent experimental infection with an attenuated yellow fever vaccine virus (114).

The accumulated epidemiological evidence seems to be in line with recent reports of a correlation between BCG vaccination and COVID-19–related mortality (Ref. 119 and A. Miller, M.J. Reandelar, K. Fasciglione, V. Roumenova, Y. Li, and G.H. Otazu, manuscript posted on medRxiv, DOI: 10.1101/2020.03.24.20042937). One intriguing implication of these studies is that a country’s vaccination policy could determine how its population fares through the pandemic: those with long-standing BCG immunization programs (e.g., Japan) would have low mortality, as opposed to those without universal BCG vaccination (e.g., Italy) with high mortality rates. The discussion of the merits and limitations of these studies on BCG and COVID-19 would have to take into account many confounding issues, including the bacillus strain heterogeneity, asynchronous evolution of the pandemic in different countries, access to testing, and quality of the health system, and go beyond the scope of this review. Nevertheless, as a whole, the body of epidemiological evidence on the heterologous effect of BCG and other vaccines looks solid. Therefore, it is not surprising that there have been suggestions for the use of BCG (119) and oral poliovirus vaccine (120) to induce nonspecific protection against SARS-CoV-2 infection. There are indeed eight ongoing clinical studies already testing this possibility and a similar number about to open enrollment, including three BCG trials in healthcare workers: BRACE (n = 10078, NCT04327206), BCG-CORONA (n = 1500, NCT04328441), and BADAS (n = 1800, NCT04348370).

Both immune response resetting (usually particulate Ags) and trained immunity (often live or attenuated microbes) allow antigenic stimuli to generate broader protection, but they also differ in how this effect is achieved. Trained immunity relies on epigenetic/metabolic reprograming of innate immune cells, whereas immune response resetting achieves off-target protection through memory T cell reactivation (113, 121). Thus, the nature of the heterologous protection may be innate induced or adaptive induced, depending on how the antigenic stimulation occurs. However, in the context of a secondary adaptive immune response, the memory lymphocytes are dominant in resetting innate and adaptive cells to be more effective, shifting any concurrent primary immune response to a low inflammatory profile (122–129). These recall conditions are likely to be less permissive to inflammasome- or IL-1–driven hyperinflammation.

In a nutshell, immune response resetting could be achieved by systemic and repeated reactivation of established memory to a pool of well-defined, unrelated Ags against which most humans have preformed immunity (121). Instead of one new agent against which the immune system might occasionally mount a disorganized, innate-biased, hyperinflammatory response, it now likely “sees” the pathogen diluted in a pool of multiple and diverse attackers for which there are circulating T cell memory precursors that may be promptly called from “retirement.” In the case of sepsis, the strong secondary activation and memory response would generate a new hypoinflammatory context that takes over the antipathogen response and is more conducive to disease resolution. Indeed, we have recently demonstrated that repeated reactivation of established T cell memory to a pool of diverse and mostly unrelated pathogens may reset unfavorable immune responses in a mouse model of polymicrobial high-grade sepsis induced by cecum ligation and puncture (121). This off-target revaccination produced >5-fold higher sepsis cure rate when employed in combination with imipenem as compared with the survival achieved with the use of this antibiotic alone. The resulting immune response resetting in our sepsis model produced a contraction of the unspecific CD3⁺ T cell population without compromising the recovery of naive subsets. Remarkably, it also reduced the hyperactivation state of all memory T cell populations, notably in the CD8⁺ subset. Additionally, the molecular signatures of the cytokine storm and the proapoptotic environment were reverted. This outcome was not dependent on eliciting a pathogen-specific new immune response but rather on the repeated reactivation of T cell memory to a diverse Ag pool capable of modulating the already ongoing response against sepsis (121).

Our hypothesis is that one could revert sepsis and the cytokine storm in COVID-19 by using off-target revaccination before ARDS and multiorgan irreversible damage set in (Fig. 1). Repeated stimulation with unrelated Ags would change the way SARS-CoV-2 would be perceived by the immune system. Instead of the overblown, innate-dominated response with little antiviral effectiveness, the action blueprints provided by the T cell memory precursor “retirees” would hopefully induce a dominant hypoinflammatory, controlled response that could lead to viral clearance. Interestingly, the most dramatic effect of immune response resetting on reverting hyperactivation in our previous work was observed in the CD8⁺ T cell compartment (121)—arguably the most important effector population in viral clearance. In the case of SARS-CoV-2, one should aim at multiple unrelated antigenic targets against which there should be preexisting memory with preferably large precursor frequency due to natural priming or previous vaccination. The use of sterile particulate antigenic preparations would be advisable in the context of immunosuppression. Ideally, this approach should be combined with other therapeutic strategies directed at reducing the pathogen load. In our earlier work, imipenem had this role (121). Perhaps remdesivir, or any other antiviral drug that proves to be effective, could play this part in the context of SARS-CoV-2 infection. This hypothesis has yet to be tested preclinically, and it must respect the most stringent ethical guidelines should it graduate into the clinical setting.

As the pandemic continues to advance with overwhelming pace throughout the world, a classic vaccine remains the gold standard solution for COVID-19. The faster we get a safe and effective vaccine, the better. There are several experimental therapies that have been considered for sepsis that deserve a shot for COVID-19. The IL-1ra anakinra and the anti–IL-6 Ab tocilizumab are good examples that already showed some initial hope alongside therapies that have been tried directly for COVID-19, such as the anti–GM-CSFR Ab mavrilimumab (Table I and Refs. 31–35). In fact, some failed strategies may actually work: against most therapeutic guidelines (130), impressive results have recently been reported for low-dose dexamethasone in COVID-19 (8). However, both sepsis and COVID-19 are complex diseases that we comprehend poorly. Thus, any therapy for both conditions must ideally be guided by detailed clinical and immunological stratification. In addition, assuming that there will be a better picture of safety and efficacy for each individual therapeutic agent, concurrent strategies such as the use of IL-1 blockade plus IL-7 stimulation (anakinra plus CYT107) might also be evaluated. A combination containing an agent that reduces immunothrombosis (e.g., nafamostat) might also be an option.

Moreover, we believe that the rationale for the use of heterologous vaccination (e.g., BCG and/or oral poliovirus vaccine) as a means to reduce the impact of SARS-CoV-2 infection is sound. If employed, trained immunity could probably make the first encounter with the virus less likely to overcome innate immunity and overwhelm the unfolding virus-specific response. Thus, it might help to mitigate the virus threat in the absence of a viable specific vaccine.

Differently from the preventive scenario described above, we believe that immune resetting triggered by the recall of memory T lymphocytes would be much more efficient in regaining control of the immune response in the context of a developing SARS-CoV-2 infection. Also, we believe that the recall should involve multiple and diverse Ags to amplify a large number of circulating memory T cell precursors. This means that Ag pools from different pathogens should be used to reach an effective secondary activation state. Moreover, the immune resetting strategy implies removing the pathogen from the driver’s seat, defusing the cytokine storm, and letting the adaptive recall induced by the antigenic pool overtake guidance to a proper response.

Thus, it is reasonable to assume that the possible heterologous protection against SARS-CoV-2 primary infection provided by trained (innate) immunity might be more prophylactic, whereas that given by (adaptive memory-initiated) immune response resetting might be more therapeutic in nature. It is worth remembering that memory T cells are arguably the most powerful known natural component of the immune system readily available for therapeutic use that is capable of turning off the inflammasome (125), thereby dampening the cytokine storm and hyperinflammation, which seem to dominate the COVID-19 pathophysiology (15, 16).

The proposed off-target revaccination obviously is not intended to replace a standard vaccine, but it may provide an immediate real-time alternative solution for the treatment of severe disease presentations. Putting it simply, the idea is to let the immune system learn by example how to exit the virus-induced overdrive (through repeated reactivation of established memory to unrelated pathogens).

A.E.N. holds a patent on Immune Response Shifter (IRSh) (US patent application no. US 15/431,329 and international patent no. PCT/BR2018/000004) and has filed a second provisional application before the U.S. Patent and Trademark Office. The other author has no financial conflicts of interest.