As an expanded number of coronavirus vaccines enter human clinical trials, in addition
to understanding their efficacy in preventing severe SARS-CoV-2-related disease, a
key outcome that will be receiving outsized scrutiny will be whether these vaccines
contribute to lung immunopathology upon natural viral infection. Since the emergence
of life-threatening severe acute respiratory syndrome (SARS) almost 20 years ago and
subsequently Middle East respiratory syndrome (MERS) in 2012, numerous vaccines have
been developed and tested in experimental animals to combat these lethal coronavirus-associated
respiratory syndromes. An unexpected and concerning feature of several of these is
the appearance of lung immunopathology that is seen in animals receiving certain types
of vaccines. This result is especially concerning given that vaccine associated enhanced
respiratory disease (VAERD) was seen in human vaccine trials against the ubiquitous
airway pathogen respiratory syncytial virus (RSV). Although we must be mindful of
potentially disappointing outcomes, a careful assessment of vaccine design, immunobiology,
and clinical and experimental outcomes published thus far suggests that VAERD may
not represent a major threat to ongoing vaccination efforts.
Lung immunopathology refers to exaggerated inflammation that envelopes the gas-exchanging
units of the lung after viral infection and which may interfere with oxygenation.
It is a concern because it can lead to worse disease than what would normally be seen
after virus infection in the complete absence of vaccination. Clinical VAERD was first
seen in human infants with RSV infection after receiving a formalin-inactivated vaccine
against RSV in the 1960s that led to markedly worse respiratory disease as compared
to non-vaccinated infants, in two cases leading to death [1]. The type of inflammation
observed in RSV VAERD was also qualitatively different from that seen in natural infection.
In suitable animal models of disease, RSV-related VAERD is characterized as a pulmonary
“Arthus reaction” - infiltration of the lungs with neutrophils and lymphocytes as
observed in a cotton rat model [2], or eosinophils observed in a Balb/c mouse model
[3]. Histopathologic autopsy findings from an infant who died potentially of VAERD
linked to RSV included monocytic pulmonary inflammation together with eosinophils
[4].
Eosinophils are a type of infection-fighting cell of the immune system that are normally
seen in parasitic and fungal infections or in unrelated non-communicable diseases
such as asthma and inflammatory bowel disease. Although not proven, causal associations,
eosinophilic lung immunopathology has been linked to multiple factors including 1)
formalin alteration of vaccine antigens [1]; 2) complement activation [5]; and 3)
T helper type 2 (Th2) and Th17 cell-predominant immune responses that coordinately
drive the production and recruitment of eosinophils [6].
Because the immunopathology seen in experimental SARS and MERS coronavirus-related
VAERD models was also eosinophilic, investigators have rightly raised concerns about
the safety of coronavirus vaccines that will soon be tested in humans against COVID-19.
However, beyond the fact that RSV is genetically distinct from coronaviruses, there
are several additional differences between the vaccine-related VAERD that was seen
in human RSV infection and that seen after experimental SARS and MERS vaccines. First,
lethal vaccine-related immunopathology has only been seen in infants, who have immature
immune systems that are less capable of mounting robust type 1 (i.e., interferon-dominated)
immune responses as compared to adults. In general, type 1 immunity is required to
overcome most viral infections and is readily generated in more mature individuals.
Thus, RSV vaccine-related immunopathology may have had more to do with the immaturity
of the infants’ immune system and less to do with vaccine-specific toxicity. This
is supported by studies showing that older children do not experience immunopathology
after RSV vaccinations [7, 8], a study demonstrating that some RSV vaccines fail to
induce antibody affinity maturation due to inadequate B cell activation, again a potential
consequence of immaturity of the immune system [9], and studies of SARS vaccines in
mature rodents. Regarding these latter studies, despite the emergence of eosinophilic
immunopathology following infection, the animals all survived, in contrast to unvaccinated
controls that all succumbed [10, 11].
Second, eosinophilic immunopathology due to SARS infection occurred in vaccinated
rodents despite their having abundant titers of neutralizing antibodies that, when
present, normally preclude active infection [12]. One possible explanation for this
paradoxical outcome is that experimental models of SARS infections as used in these
studies involved viral exposures that likely far exceed natural exposures. Thus, in
experimental contexts, viral exposures could be overwhelming vaccine-induced protective
immunity, leading to an initial infection that, while inducing pathology, cannot propagate
beyond a few rounds of viral reproduction and thus is ultimately self-limited. If
this is true, then lung viral loads should be lower in vaccinated as compared to unvaccinated
animals. In fact, mice receiving SARS vaccines that exhibited eosinophilic lung immunopathology
demonstrated significantly lower lung viral titers within the first week of infection
as compared to unvaccinated controls [10, 12, 13].
A third observation is that “immunopathology” as seen in experimental animals given
different vaccine formulations appears to be quantitatively similar, although qualitatively
dissimilar based on whether or not eosinophils predominate in the lungs. Although
the adjuvant factor alum has been implicated in eosinophilic immunopathology, in fact
this complication is seen with coronavirus vaccines both with and without alum; moreover,
addition of alum appears to actually protect from eosinophilic lung pathology [12,
14]. Regardless, these observations do not indicate that eosinophilia per se is harmful
in this context.
While it is difficult to compare vaccines across their many different platforms and
formulations, species tested in, and eras in which they were studied, a consistent,
critically important issue appears to be the quality of the antibodies produced after
vaccination. Early RSV vaccines failed to consistently induce neutralizing antibody
responses [1] and careful follow-up studies now indicate that poor outcomes related
to early RSV vaccines were indeed due to inadequate generation of neutralizing antibodies
[9]. Moreover, it is clear from animal studies that vaccination leads to survival
regardless of the type of immunopathology as long as neutralizing antibodies are produced
[[10], [11], [12]].
These observations give us hope that naturally occurring COVID-19 infections, typically
involving fewer virions initially acquired as compared to experimental infections,
will be short-lived and rapidly controlled in properly vaccinated individuals. Such
individuals may in fact remain asymptomatic and never know they were infected. It
is furthermore possible that the fate of naturally acquired SARS-CoV-2 virus in properly
vaccinated individuals will simply be neutralization, with the virus never initiating
either infection or immunopathology. This is based on the robust protection against
SARS-CoV reinfection afforded mice receiving a recombinant protein-based vaccine [15]
and rhesus macaques that received an inactivated SARS-CoV-2 vaccine formulated with
alum [16]. Additional findings confirm that SARS-CoV-2 vaccine-induced protection
in rhesus macaques correlates with the generation of high titers of neutralizing antibodies
[17]. We should always be prepared to find and avoid vaccine-related complications
such as lung immunopathology. Nonetheless, the available data indicate that the best
way to avoid this complication-and defeat SARS-related coronaviruses-is through vaccines
that generate robust neutralizing antibodies.