Immune-mediated adverse effects may occur in patients after a pharmaceutical product is administered to treat or diagnose a disease1,2,3,4. When such reactions occur minutes to hours after systemic, intravenous product administration, they are often called infusion reactions, hypersensitivity or anaphylaxis. However, there is no uniform terminology, and the use of these terms largely depends on the field of study (clinical oncology versus immunology, for example). The World Allergy Organization classifies hypersensitivity reactions (HSRs) as immediate (occurring within minutes to hours of exposure) or delayed (requiring days before clinical manifestation)2,3. According to the World Allergy Organization Anaphylaxis Guideline, the term ‘hypersensitivity’ refers to reproducible symptoms occurring as the result of exposure to a defined stimulus; ‘allergy’ is defined as an HSR initiated by a specific immunological mechanism; and ‘anaphylaxis’ is classified as a severe, life-threatening, generalized or systemic HSR5. The European Society for Medical Oncology divides the reactions into anaphylactic and anaphylactoid, and proposes the term ‘non-allergic anaphylaxis’ instead of ‘anaphylactoid’ where there is a ‘non-immunological cause’6. The European Network for Drug Allergy’s definition is independent of the underlying immunological mechanism and categorizes the HSRs as immediate and non-immediate. Moreover, differences between European and American physicians exist in managing these reactions7. The National Cancer Institute Common Terminology Criteria for Adverse Events distinguishes between allergy, anaphylaxis and cytokine release syndrome (CRS) based on clinical manifestations8. The more commonly known Gell and Coombs classification is based on the underlying immunological mechanism and the time to manifestation, and includes four types: immediate/type I (developing within 15–30 minutes of exposure), type II (minutes to hours) and type III (3–8 hours); and delayed/type IV (48–72 hours)9.

Although many immune-mediated adverse effects exist, here we focus on IRs. These reactions are unintended and occur at therapeutic doses of various products4, including biological products (such as recombinant proteins and antibodies)10,11, therapeutic nucleic acids, low-molecular-weight drugs1,12, and complex nanotechnology-based formulations13,14,15 herein referred to as nanomedicines (Tables 1 and 2). Despite the difference in underlying mechanisms, IRs may have overlapping clinical manifestations. Patients with acute IRs experience various symptoms including, but not limited to, skin flushing or rash, chest and back pain, dyspnoea, wheezing, chills, fever and rigor (Tables 1 and 2). These adverse effects require timely and accurate assessment, and proper management to avoid severe and potentially fatal consequences. Severe IRs are rare and occur in less than 5% of patients4. The incidence may increase, however, when different drugs are used in combination (for example paclitaxel and carboplatin), when patients have a certain type of human leukocyte antigens (HLA; for example HLA-B*57:01 and abacavir) or when there is an underlying viral infection (for example Epstein–Barr virus infection and penicillin)6,13. Despite decades of research suggesting that the incidence of IRs depends on both pharmacodynamics and pharmacogenetics, it is largely unknown why some patients develop these reactions while others do not. The lack of uniform terminology and classification of the reactions further complicates the issue. As such, acute IRs cause substantial stress among patients and their families as well as care providers and regulatory agencies16. Several world-leading health authorities (the World Health Organization, the United States Food and Drug Administration17 and the European Medicines Agency) agree on the noxious nature of these adverse effects and the need for improved prevention and management6. These improvements largely depend on an understanding of mechanisms, but the molecular and cellular processes causing IRs are incompletely understood. Therefore, IRs in patients are currently managed by systemic administration of immunosuppressive, anti-pyretic and anti-inflammatory medications before the infusion, during administration or both1,4,6.

Table 1 Selected examples of nanotechnology-based drug products known to induce IR
Table 2 Selected examples of non-nanotechnology drug products known to induce infusion reactions

Although not unique to any specific drug category, IRs present yet another hurdle in the translation of nanomedicines, owing to the complexity of both their nature and the regulatory approval process. This problem warrants thorough investigation; here, we examine the issue in the context of nanomedicines. We analyse the current understanding of the mechanisms underlying IRs as well as assays and models being used to study these reactions at the preclinical stage. We identify gaps in the knowledge and propose a roadmap to fill them. We further suggest a strategy for overcoming translational barriers of nanomedicines caused by IRs.

Mechanisms of IRs in humans

According to Gell and Coombs, type I HSRs are mediated by immunoglobulin E (IgE) specific to at least one component of a drug product (Table 3)9. The main cellular contributors to this HSR type are mast cells, the activation of which leads to hay fever, allergic asthma and/or anaphylactic shock9,13. Type II HSR is cytotoxic hypersensitivity mediated by drug-specific antibodies (mainly IgG), the complement system and natural killer cells. Symptoms include pemphigus, nephritis, autoimmune haemolytic anaemia and Goodpasture’s syndrome9,13. Type III HSRs are mediated by the immune complexes formed between a drug and an antibody (either IgM or IgG), and they involve complement activation. Clinical manifestations are serum sickness, fever, glomerulonephritis and vasculitis9,13. Type IV HSRs are mediated by T-helper cells and macrophages and manifest as erythema, induration, contact dermatitis, maculopapular rash and granuloma9,13. An immediate IgE-independent HSR with symptoms resembling a type I HSR is known as a complement activation-related pseudo-allergy (CARPA)18.

Table 3 Gell and Coombs classification of allergic reactions

Type I–III reactions and CARPA are commonly recognized mechanisms associated with drug-mediated HSRs. Platinum-based formulations are notorious for type I HSRs4,6, while CARPA is the best-studied reaction to nanomaterials18,19. CARPA has been described for liposomal drugs (Doxil and Ambisome), micelles (Taxol and Taxotere) and modified dextran-coated iron oxides (Feraheme), all of which are approved for clinical use and are marketed with a black box warning of potentially life-threatening IRs. Properties of nanomaterials commonly associated with CARPA are summarized in Fig. 1a. Uncontrolled release of cytokines resulting from excessive proinflammatory stimuli, inadequate regulation of inflammation or a combination thereof is responsible for severe CRS. The clinical manifestations of CRS include erythematous or pruritic rash, hypotension, fever, malaise, tachycardia, tachypnoea, generalized swelling, altered mental status, diffuse lymphadenopathy, and enlargement of the liver and spleen20. CRS created translation barriers for immunotherapies (for example CAR-T cells), biologics (for example TGN1412) and nanotechnology-formulated therapeutic nucleic acids (for example MRX34)10,21,22,23. The consequences were severe and, in some cases, fatal23. It is important to mention that in some cases (CAR-T therapy, for example), clinical CRS manifestations are not immediate and may take a week or more, which further complicates IR definition, diagnosis and intervention. Moreover, some infections (such as influenza virus and bacteraemia) may trigger cytokine release with symptoms indistinguishable from the drug-mediated CRS24. Although various nanoformulations can induce the release of cytokines both in vitro and in vivo25, the relevance of this mechanism to various types of nanomedicine-triggered IRs remains largely unknown.

Fig. 1: Strategy and roadmap for overcoming infusion reactions to nanomedicines.
figure 1

The strategy for overcoming nanomedicine-triggered IRs relies on a mechanistic understanding of those IRs, the identification of the leading cause and the discovery of the relationship between various mechanisms. a, Some physicochemical attributes of nanomedicines that have been linked to IRs. This list is incomplete, as other potential attributes are not yet understood. The known internal properties can be fine-tuned to decrease the risk. External features cannot be controlled directly but can be addressed through the engineering of internal properties. DSPC, 1,2-distearoyl-sn-glycero-3-phosphocholine. b, Identification of reliable biomarkers, corroboration of a methodological framework and mechanistic verification serve to improve the current knowledge base of IRs to nanomedicines. This knowledge will enable improved healthcare by combining existing approaches for monitoring and managing the IRs with new ones, which are aimed at intervention at the root cause.

The IRs to drugs are often heterogeneous and involve overlapping reactions and effector cells. For example, activation of the complement system, which occurs in CARPA, is also involved in the pathogenesis of type II and type III HSRs. Conversely, the macrophages may contribute both to type IV HSRs and to CARPA. Cytokines that are produced by leukocytes are involved in CRS and can contribute to various types of HSRs, including pseudo-allergy. An excellent demonstration of the complexity of frequently overlapping mechanisms that cause IRs to nanomedicines comes from the experiences with liposomes and lipid-based nanocarriers, which, according to a recent report by the US Food and Drug Administration, dominate the current landscape of nanomedicine26. Research groups worldwide have reported that lipid-based carriers are not immunologically inert25,27,28. Preferential clearance of these materials by macrophages, activation of proinflammatory cytokines and the complement system are well-established25,27,28. However, the cause–effect relationship between the complement activation, cell uptake and cytokine release is far from being understood. Despite the general acceptance that IRs involve multiple cellular and biochemical processes, the controversy over the leading cause of this toxicity creates considerable hurdles and delays in the development of a unified strategy for predicting and overcoming the IRs to nanomedicines.

Controversy surrounding mechanisms

Activation of pulmonary intravascular macrophages (PIMs) by PEGylated liposomes (where PEG is polyethylene glycol) was suggested as a key effector arm of CARPA in pigs29. The proposed double-hit scenario in this mechanism implies that both the complement and macrophage activation trigger the IRs symptoms29. Recently, a study of polystyrene beads in the same model concluded that PIMs activation is the leading cause of HSRs to nanomaterials and challenged the role of the complement30. This mechanism caused robust debate31,32. On the one hand, there is ample evidence that the pig model is sensitive to the detection of nanomedicine-induced CARPA32,33. On the other hand, PIMs are not present in the lungs of humans and animals commonly used in preclinical research (mice, rats, dogs and non-human primates)34,35,36. In these animal species, the liver- and spleen-resident macrophages are primarily responsible for nanoparticle clearance. PIMs are induced in humans under certain pathological conditions, such as liver failure34,35. In contrast, PIMs are common in the lungs of pigs, sheep and horses, not broadly recognized as preclinical models34,35. Therefore, one area that requires close attention is the understanding of whether liver- and spleen-resident macrophages in humans play the same role as PIMs in pigs during IRs to nanomedicines. The hypothesis that hepatopulmonary macrophage migration triggers cardiorespiratory symptoms in humans exposed to nanomedicines37 is attractive but requires verification. The pivotal role of PIMs in nanoparticle-mediated IRs in the porcine model will benefit from confirmation by various research groups and with various types of nanomaterials. For example, a recent study demonstrated that during the infusion reaction triggered in pigs by carboxylated, hydrophobic, highly anionic polystyrene nanoparticles, complement activation-related opsonization coincided with the peak of pulmonary distress38. The relative contribution of the complement and PIMs, as well as the cause–effect relationship, may vary between different types of nanoparticles. Therefore, studies with clinically relevant well-characterized nanomedicines (for example liposomes, micelles, iron-oxides) are essential, while research-grade nanomaterials with poorly understood physicochemical properties (such as polystyrene beads) may not provide clinically relevant answers unless these particles are thoroughly characterized. The nature of a condition (cancer versus inflammatory disorder, for example) treated with a nanodrug should also be considered when analysing the mechanism of IRs to the nanomedicine. Other recently proposed mechanisms, such as those involving platelets as an effector arm39,40, should also be fully investigated in the context of nanoparticle physicochemical properties and interaction with other mechanisms.

Although complement activation in CARPA is commonly verified by assessing the complement split products, there is no universal agreement regarding which macrophage activation markers are relevant to the IRs. Models relevant to the prediction of IR in humans are also being debated3133,41,42. Therefore, there is currently a critical need to verify the biomarkers and models that are necessary for identifying the potential of nanomedicines to cause IRs in patients.

Biomarkers and models of IRs

The current selection of biomarkers is not straightforward, owing to the controversy over both the definition of IRs43 and the mechanism(s) responsible for them31,32. A recent approach proposed for therapeutic antibody-mediated IRs based on the long usage of these products in the clinic could serve as a starting point for nanomedicines44. In this approach, clinical symptoms of IRs12, also called Sampson criteria44, are reviewed first to diagnose anaphylaxis44. Next, the assessment of drug-specific IgEs (a skin test and enzyme-linked immunosorbent assay, ELISA) and markers of mast cell degranulation (histamine and tryptase) are used to verify the anaphylactic nature of the reaction. Additionally, cytokines (for example interleukin 6 (IL-6), IL-8, interferon γ, tumour necrosis factor α, IL-2 and IL-10), complement split products (C3a, C5a, sC5b-9), and complement consumption (for example CH50) are used to verify IgE-independent pathogenesis of IRs44. Re-challenge or avoidance of the allergen is finally used to confirm the IR44. Because anaphylactoid reactions to proteins may also increase the levels of tryptase and cytokines, these biomarkers are debated as suboptimal for diagnosing anaphylaxis to protein-based therapeutics18.

In the field of nanomedicine, assessing the complement split products (C3a, C5a, sC5b-9) and consumption (CH-50) is used to establish CARPA in both humans and animal models. However, earlier studies with nanomedicines indicated that CARPA response towards liposomes is accompanied by the release of many secondary messengers (for example thromboxane, leukotrienes, eicosanoids, histamines, cytokines and tryptase)18. In pigs, thromboxane A2 is recognized as the main mediator of pulmonary symptoms commonly seen during IRs. However, the use of this and other secondary messengers for predicting IRs to nanomedicines in humans remains unknown. Although IgG and IgM have been implicated in type II and type III HSRs to proteins and low-molecular-weight drugs41, no data exist about such responses to nanomedicines. However, several studies described naturally existing IgM and IgG that can bind to various components commonly present in nanomedicines (cholesterol, phospholipids and PEG)45,46. The accelerated blood clearance of PEGylated liposomes due to the anti-PEG IgM was reported in animals47. However, the functional significance and relevance of these and other pre-existing antibodies to nanomedicine-triggered IRs in patients as well as their diagnostic utility require thorough investigation.

Currently, identifying biomarkers that are relevant to IRs depends heavily on in vitro and in vivo models, both of which have advantages and limitations in their relevance to human patients (Table 4). As there is limited information on nanomaterial-mediated type I–IV HSRs, our discussion will focus on models applicable to the two established mechanisms (cytokines and CARPA) and the alternative mechanism (PIMs).

Table 4 Available animal models

Rodent models reproduce the hypotension observed in humans during IRs. However, the doses required to induce HSRs in mice and rats are several orders of magnitude higher than those needed to trigger reactions in humans42,48. Therefore, rodent models are suitable for mechanistic studies but irrelevant for screening materials for reactogenicity.

The dog model reproduces some symptoms of human HSRs, such as hypotension, fainting and other disturbances of vegetative functions, at dose levels relevant to those in humans. However, studies of reactogenicity in dogs require large numbers of animals because of the high inter-animal variability in the response42. Therefore, screening for reactogenicity in the dog model may underestimate the toxicity, particularly if the number of tested animals is low.

The pig model allows reproduction of the cardiopulmonary distress typically observed in humans reacting to the infusion of nanomedicines15. However, it requires a better characterization and understanding of the role of PIMs34,35. Despite differences in the mechanism(s) underlying IRs, the pig model can be used to predict the reactogenicity of nanoparticles at low, clinically relevant doses using a reasonably low number of animals. Other animal models are not well-established to study nanomedicine-induced HSRs.

In vitro studies using patient serum or plasma are often considered for prediction of CARPA to nanomedicines. Likewise, cultures of human peripheral blood mononuclear cells can be used to estimate the risk of CRS induction. The correlation between in vitro screening and in vivo studies has been confirmed for CARPA49 and CRS25. However, current experience with these tests suggests that the positive response in complement or cytokine assays can predict the risk, but neither the incidence nor the magnitude, of IRs. This observation is consistent with the multicausality of IRs even when it is evident that complement activation or cytokines are primarily responsible. The utility of basophil activation, mast cell degranulation, leukocyte oxidative burst50 and other common laboratory allergy tests51 for nanomedicines requires careful investigation.

Strategy and roadmap for addressing IRs to nanomedicines

Translational hurdles due to drug-mediated IRs are not unique to nanotechnology-formulated drug products. Therefore, one way to address the problem in nanomedicine is to use the knowledge and lessons learned from the clinical use of other drug products. For this to happen, the issue must be approached in a systematic way, beginning with the identification of relevant biomarkers in patients, establishing appropriate models and understanding the mechanisms of IRs.

In the first step, reliable biomarkers need to be identified and assessed for correlation to clinical outcomes (Fig. 1b). One approach is to use retrospective clinical trials to identify patients who are sensitive to IRs when administered with nanomedicines. Their sera could be retrieved and tested in vitro. Obtaining viable cells from archived specimens of such patients could be problematic, however, because common preservation techniques protect plasma proteins better than cells. The timeline between specimen collection and nanomedicine administration may also affect the assay outcome because blood composition is dynamic and reflects the physiological status of the patient at the time of collection.

An alternative is through prospective studies that enrol patients prescribed with nanomedicines and collect their fresh blood before and after administration of a nano-drug. Whole-blood samples collected before treatment could be exposed to a nanomedicine in vitro, and the various endpoints, including immune cell, complement and clotting cascade activation markers, could be studied. Prospective studies are potentially limited by patients’ premedication with immunosuppressive drugs, which are classically used to prevent the incidence of IRs, or by a slow infusion rate, typically used to avoid anaphylaxis. Despite these limitations, some markers (such as complement split products or cytokines) can still be detected, even in the absence of clinical signs of HSRs. For example, in the case of Doxil and Taxol infused at high rates, HSR symptoms were detected in patients only when high levels (more than fivefold above the baseline) of the terminal complex (sC5b-9) were detected in the blood52,53. However, complement split products at physiologically significant levels (at least twofold above the baseline) were detectable in the absence of clinical symptoms of HSRs19.

The second step requires the establishment of in vivo and in vitro models and their relevance to human patients. One way is to take advantage of existing models. To verify their relevance to humans, one must compare the underlying mechanism(s) between these models and patients. Although several mechanisms of IRs have been proposed for nanomedicines, CARPA is the most well-studied and understood32. The relevance of other mechanisms to nanomedicines remains unclear, given the current lack of human data. This work, therefore, is critical and interdependent on biomarker selection.

The third step is mechanistic verification of the selected biomarkers. Here we propose the use of inhibitors in patients when approved drugs are available. For example, a clinically approved C5 inhibitory antibody, eculizumab (Soliris); a plasma-derived C1 inhibitor (C1INH); and a small-molecule inhibitor of factor D could be used to understand the contribution of the complement in humans54. Because it is not yet known what inhibitor would work best for preventing nanoparticle-induced complement activation in patients, non-clinical-grade inhibitors of C3 convertase (compstatin, APT070 or APL-2), soluble complement receptors (sCD35) and decay accelerating factor (CD55), chimeric receptors (CAB-2) or their animal counterparts could be investigated in the animal models. For example, sCD35 was effective at inhibiting liposome-triggered IRs in the pig model of CARPA55.

The contribution of other mechanisms (CRS, platelets, and direct macrophage activation, for example) could be verified by inhibitors of cyclooxygenase (such as indomethacin) and cytokines (such as neutralizing antibodies). For example, indomethacin efficiently blocked liposome-mediated IRs in pigs, suggesting cooperation between macrophage- and complement-mediated mechanisms55. Special consideration should be given to the type of the nanomedicine tested (for example PEGylated liposome or dextran-coated iron oxides), and the category of human subjects enrolled in the clinical trial. For example, the consequences of administering a complement inhibitor to cancer patients prescribed with Doxil are uncertain because the implications of complement in tumour growth are poorly understood. Likewise, inhibiting complement in patients with chronic kidney disease who had been prescribed Feraheme may increase the risk of infections because the immune system in these patients is already weakened, and complement-mediated protection from pathogens is further reduced by the inhibitor. The administration of empty carriers in combination with complement inhibitors can be much less dangerous and more ethically justified in healthy volunteers.

Desensitization strategy, commonly used in the field of protein-based allergens, is not well investigated in the field of nanomedicine. The only known preclinical example involves injecting Doxebo (a placebo PEGylated liposome) to reduce the IRs to subsequently administered Doxil (a drug-loaded PEGylated liposome) in a pig model29. Clinical investigation of this and other desensitization strategies would further benefit the field.

The current approach of slowing down the infusion rate is a powerful tool in reducing IRs to nanomedicines19, and it would further benefit from an understanding of the underlying mechanism. The improved knowledge base will also allow researchers to unravel the complex relationship between IRs, other immune-mediated adverse effects, and long-term or tissue-specific toxicities similar to those described in rats with CARPA induced by a high dose of cholesterol-rich liposomes56.


The foundation of the strategy to overcome IRs to nanomedicines comprises a mechanistic understanding of those IRs and the identification of the leading cause and relationship between various mechanisms, as well as critical attributes of the nanomedicines that are responsible for triggering IRs. The nanomedicine community must clarify the role of the complement, cytokines, macrophages, platelets and other mechanisms in the context of the physicochemical attributes of the nanoparticles. In the long run, this information can be used to understand the potential role of IRs in tissue-specific and long-term toxicities. Furthermore, the community needs to harmonize methods, models and biomarkers for predicting IRs in patients. Finally, the improved knowledge should be used to combine existing strategies (which focus on management of the symptoms) with new ones (which focus on intervening at the root of the cause) to overcome translational barriers caused by IRs.