This review has three major goals: (1) to summarize conclusions about the efficacy of post-exposure smallpox vaccination against clinical disease presentation from historic epidemiological reports and modern animal studies; (2) to identify data gaps in these studies; and (3) to summarize the clinical features of orthopoxvirus-associated infections in various animal models in order to identify those models that are most useful for post-exposure vaccination studies.

Smallpox vaccination using a heterologous species of orthopoxvirus (OPXV) became common practice after Edward Jenner’s famous experiment in which he inoculated a young James Phipps with material from a cowpox (CPXV) lesion in 1796 [1]. Early reports on the effectiveness of pre-exposure vaccination were confirmed by well-documented studies performed throughout the 19th century which demonstrated significantly lower rates of smallpox mortality in geographic areas with mandatory vaccination as opposed to areas where vaccination was not required [2].

One major contribution to the eradication of smallpox was the availability of an effective live vaccinia virus (VACV) vaccine. Vaccination was utilized pre-exposure to prevent smallpox infection and post-exposure during smallpox outbreaks to vaccinate potentially exposed contacts of infected patients. This methodology coupled with the strict isolation of patients was successful in protecting those contacts from severe disease and producing a “ring” of protection that halted disease transmission. Other factors, including the lack of a reservoir for variola virus (VARV), the development of a heat-stable vaccine, the introduction of the bifurcated needle, and a disease course which allowed time for post-exposure vaccination to elicit a protective immune response, all contributed to the development and implementation of the smallpox eradication effort [3]. Because of widespread pre-exposure vaccination, serious adverse events (SAEs) of smallpox vaccination were well known by the early 20th century but the more significant threat of endemic smallpox ensured that mass vaccination campaigns remained an important defense against outbreaks [4,5]. As eradication efforts progressed, it became apparent that eradication goals could not be met until surveillance systems, systematic investigation of outbreaks, and post-exposure isolation and vaccination were all successfully implemented [6,7]. As cases of smallpox declined, the relative risk of SAEs associated with 1st generation vaccines (vaccines utilizing live VACV propagated on livestock) rose, which led to the recommendation that mandatory vaccination be halted. This was done in the United States in 1971 and worldwide by 1980, when smallpox was officially declared eradicated by the World Health Organization (WHO) [8]. The efficacy of pre-exposure vaccination using these 1st generation vaccines in preventing smallpox disease was well documented during the eradication era. However, post-exposure vaccination with 1st generation vaccines, while generally believed to be at least partially protective, remains less defined, which makes the evaluation of the efficacy of newer and future vaccines more complicated.

The cessation of mandatory prophylactic vaccination has resulted in over half of the global population being potentially naïve to OPXV threats. Despite decades of continuous research to increase vaccine safety without a loss in efficacy, and the creation of 2nd generation vaccines (live VACV propagated in cell lines), 3rd generation vaccines (attenuated VACV) [9] and subunit vaccines [10,11]; only one vaccine (Acambis 2000) has been licensed for use at this time [12]. However, the use of Acambis 2000 continues to be limited due to its adverse event profile [13]. Recommendations to vaccinate U.S. health care workers and laboratorians have previously met with low compliance rates, largely due to the known risk of SAE’s following vaccination [14]. In addition, a sizable proportion of the global population is contraindicated for vaccination with Acambis 2000 due to various health conditions [15]. The development of medical countermeasures and safer vaccines that are efficacious against OPXV is an ongoing effort – one which requires an understanding of 1st, 2nd, 3rd and subunit vaccine efficacy in both pre-and post-exposure scenarios [11].

Medical countermeasures to OPXVs are important because smallpox re-emergence through a release of VARV would be a high-consequence event (although the risk of this happening is perceived to be low), and because emerging and re-emerging zoonotic OPXV-associated diseases continue to be a public health issue. The WHO Commission to Certify Smallpox Eradication instituted an international surveillance program for smallpox-like diseases in 1971 [16], which ultimately resulted in an increased awareness of human monkeypox virus (MPXV) infection [17]. Today, MPXV infections are on the rise in the Democratic Republic of the Congo (DRC) [18], and outbreaks in Sudan and the United States indicate the potential for MPXV to spread [19]. Other OPXV infection outbreaks are routinely observed and include VACV in Brazil [20], CPXV in Europe [21], and buffalopox in India [22]. Current research also indicates that OPXV in wildlife reservoirs is more prevalent than previously thought [23–25]. Lastly, long-held concerns regarding the threat of smallpox as a weapon of bioterrorism increased after the events of September 11, 2001 and the subsequent anthrax releases [26]. Combined, these conditions make the development of medical countermeasures against OPXV-associated disease an ongoing and current research effort.

During the global eradication of smallpox, a wealth of epidemiologic data was collected. These data have subsequently informed public health practices regarding response strategies to outbreaks of OPXV-associated diseases. However, in the absence of smallpox disease, evaluating the efficacy of newer medical countermeasures – to include 2nd and 3rd generation as well as subunit vaccines – will likely depend on information derived from laboratory, rather than clinical, research [16,27]. The development and utilization of surrogate models will play an important role in the testing of newer vaccines for efficacy in both prophylactic and post-exposure/post-challenge scenarios against systemic OPXV infections [28,29].

The data presented in this review can be used to identify those surrogate models that are most useful for post-exposure vaccination studies, inform policy and practice, identify additional research needs and offer insight into the use of post-exposure vaccination to control zoonotic OPXV outbreaks and potential VARV release.

Historical reports on outbreaks of variola major offer insight to the use of post-exposure vaccination as a medical countermeasure to OPXV exposure. As early as 1904, it was well known that vaccination of persons exposed to smallpox helped protect these persons from smallpox disease [30], but the literature is rife with conflicting information about the timing of effective post-exposure vaccination. A summary review of the epidemiological literature which specifically mentions post-exposure vaccination efficacy in the context of variola major outbreaks [31–42] highlights historic estimations of the efficacy of vaccination at various timepoints post-exposure (Table 1). This literature is complicated to interpret due to various data gaps, which have also been detailed in Table 1.

Historical data in the early 20th century indicated to some researchers that post-exposure vaccination is not effective in preventing disease in immunologically naïve persons when administered at any timepoint post-exposure [43], while other researchers argued that primary vaccination administered post-exposure was completely protective if administered <1 day post-exposure, partially protective at up to 3 days post-exposure, but not protective after 3 days post-exposure [44]. This latter view is supported by more recent retrospective analyses. A DELPHI analysis of expert opinion [45] suggested post-exposure vaccination within 24 h would be 90% effective, and within 1–3 days post-exposure would be 80% effective, in disease prevention. In those who did develop disease, disease severity was also estimated to be minimized. Vaccination up to a week post-exposure was also estimated to have partial benefit. A review of four British clinical summaries of post-exposure vaccinations provided from the late 1800s through the early 1900s which did not prevent smallpox disease do indicate that post-exposure vaccination within 3 days post-exposure is effective at decreasing disease severity in those not previously vaccinated [46]. Fig. 1 is a generalized summary of data from the epidemiological reports reviewed in Table 1 and is designed to visually chart the longest period post-exposure where vaccination which was reported to increase survival and/or to reduce disease severity in each report. As can be seen in Fig. 1, these data suggest that the administration of post-exposure vaccination to contacts decreases mortality and/or reduces morbidity in 100% (12/12), 83% (10/12), 75% (9/12), 67% (8/12), 58% (7/12), 42% (5/12), 33% (4/12) and 25% (3/12) of reports when vaccine is given at <1, <2, <3, <5, <7, <9, <10 and <12 days post-exposure, respectively. Due to the data gaps described below, the data summary in Fig. 1 is a qualitative representation of these reports.

In general, the epidemiological data from variola major outbreaks before and during the eradication era appear to support the current view that vaccination prior to 3 days post-exposure will provide benefit in preventing smallpox disease in exposed persons regardless of prior vaccination status. In addition, it appears that post-exposure vaccination given prior to the appearance of rash affords clinical or survival benefit, but these benefits diminish when vaccine is administered greater than one-week post-exposure. However, it is important to note that these data are limited in many ways.

During this period only 1st generation vaccines were used; prior immunity in individuals within the population (whether from previous vaccination or exposure) may be underestimated; underlying immune issues of patients are unknown; specific descriptions (i.e. Rao’s classification) of the clinical presentation of index patients and contacts that develop disease are not included; vaccine quality and administration differences may be present as vaccine quality prior to 1971 was largely uncontrolled and the bifurcated needle was not recommended until 1968 [47,48]. Additionally, patient demographics as well as social/cultural/economic differences are also not always identified. Any one of these missing pieces of information could help explain the variability among – and affect interpretation of – these studies. Consequentially, our conclusions are tempered with the understanding that these patients are presumed to be naïve when given post-exposure vaccination, are presumed to be immunocompetent, are presumed to be exposed to sufficient VARV to cause infection and disease, and are presumed to have been given a quality vaccine using appropriate administration routes. These presumptions and data gaps make it difficult to interpret the true efficacy of post-exposure vaccination. In the absence of human smallpox disease, surrogate models offer our best chance of defining when post-exposure vaccination will provide a survival benefit. The next section of this review will examine the limited literature regarding post-exposure vaccination studies in surrogate models.

Although both pre-exposure vaccination and post-exposure antiviral therapy studies in surrogate models contribute to our understanding of outcomes following OPXV challenge, for the purposes of this review, only those surrogate model studies that have provided information regarding the efficacy of post-exposure vaccination are addressed. The data from these few reports are highly variable depending on the surrogate model used. In contrast to the human epidemiologic literature, where “post-exposure” may not necessarily mean post-infection, in the animal models, all post-exposure data on vaccine efficacy is “post-infection”. Studies that did not show a significant survival benefit for post-exposure vaccination include a study where 2.5 × 10⁵ TCID₅₀ of VACV-Elstree (intracutaneous) was administered 1 day post-exposure in cynomologous macaques infected with 10⁷ pfu of MPXV (intratracheal) without any significant survival benefit when compared to control animals [49]. In a second study, 10⁶ pfu of Elstree (scarification) or 10⁸ IU of Modified Vaccinia Ankara (MVA) (intramuscular) were administered post-exposure to BALB/C mice which were challenged with 10⁴ or 10⁶ pfu (VACV-Western Reserve) (respiratory) respectively. The authors report that although MVA administration within 3 h of challenge protected the mice from death, these animals still manifest substantial disease symptoms. There was no significant survival benefit with MVA vaccination on day 1, 2, 3 or 4 and none with VACV-Elstree vaccination on day 0, 1 or 2 post-challenge [50].

Studies that did show a survival benefit for post-exposure vaccination used a murine model with an intranasal (respiratory) challenge using ectromelia virus (ECTV). This model has a longer disease course (mean time to death 10 days vs. 6 days) and longer incubation period (time to initial weight loss 7 days vs. 2–3 days) than VACV-WR intranasal (respiratory) infection of BALB/C or C57BL/6 mice. Using C57BL/6 mice lethally challenged with ECTV at 5 × LD50 (respiratory), 100% survival was observed after 10⁶ pfu VACV-Lister (intramuscular) vaccination at 0 and 1 day post-exposure and a smaller survival benefit of 40% was observed with vaccination 2 or 3 days post-exposure. Intramuscular administration of 1 × 10⁸ pfu of MVA on day 0, 1 or 2 post-exposure demonstrated 100% protection while day 3 vaccination resulted in 80% survival benefit. In this same study, the use of BALB/c mice in a 3 × LD50 challenge with ECTV (respiratory), coupled with vaccination with 10⁶ pfu VACV-Lister (intradermal tail scarification) resulted in 100%, 83% and 16% survival when administered on day 0, 1 or 2 post-exposure. In a second experiment, 10⁶ pfu VACV-Lister (intramuscular) given on days 0, 1, 2, 3, or 4 post-exposure, survival benefit was 100%, 80%, 40%, 20% and 20%. MVA proved more effective with 10⁸ pfu of MVA (intramuscular) resulting in survival benefits of 100%, 100%, 100%, 60% and 20% when given on day 0, 1, 2, 3 or 4 post challenge. When C57BL/6 mice were used in the 3× LD50 ECTV challenge, all animals (100%) survived when 1 × 10⁸ pfu MVA vaccination (intranasal) was given on day 0 or day 1 with day 2 post-exposure vaccination providing 50% survival [51]. In a second study, C57BL/6 mice were infected with a lethal dose of 3 × 10⁴ TCID50 ECTV (intranasal) and then vaccinated on day 2, 3, or 5 with 5 × 10⁷ TCID50 of MVA (intranasal). The survival benefit was 30% for day 2 vaccination and 0% for day 3 and 5 vaccination. However, 100% survival was seen with 5 × 10⁷ TCID₅₀ MVA (intravenous) vaccination on day 2 and day 3 and 20% survival was seen with vaccination on day 5 [52]. In the final study reviewed here, mice with a deficient innate immune response (TLR9^−/−) were lethally challenged with 100 TCID50 ECTV (respiratory) and received 1 × 10⁸ TCID₅₀ MVA (intranasal) 24 h and 48 h post-exposure with 100% survival benefit. The same study using 72 h post-exposure vaccination showed a 30% survival benefit [53]. Fig. 2 graphs the combined survival benefits shown in each of the aforementioned experiments by day of post-exposure vaccination.

In summary, these results indicate no protective effect of post-exposure vaccination with intradermal administration of 1st generation, or subcutaneous administration of 3rd generation, vaccines for either the MPXV (intratracheal) macaque model or the VACV-WR (respiratory) challenged mouse model – both of which have rapid onset of symptomatic illness. In contrast, protection by post-exposure vaccination was seen in the ECTV (respiratory) challenged BALB/C or C57BL/6 mouse models. Protection was beneficial for a longer period post-exposure if 1st generation smallpox vaccines were administered intramuscularly rather than intradermally (tail scarification), or when 3rd generation vaccines were given via intranasal, intramuscular or intravenous routes of administration.

The studies above also provide data regarding immune responses related to short-term pre-exposure and post-exposure vaccination in the ECTV model. One study utilized Rag-1^−/− and IFNAR^−/− mice vaccinated with 1 × 10⁸ pfu MVA (intranasal) 2 days prior to ECTV challenge to demonstrate that Type IIFN is important but not essential for MVA-induced protection, whereas adaptive immunity is essential. In animals given either 1 × 10⁶ or 1 × 10⁸ pfu of VACV-Lister or MVA (intramuscular), a higher dose of either vaccine resulted in more antibodies being made earlier. This study also demonstrated that the titer of antibodies necessary for passive protection (3 × 10⁴) is higher than the titer induced by vaccination in this model [51]. In an additional study, ECTV was shown to depend on TLR9 recognition in order to stimulate innate immunity, whereas MVA could stimulate innate immunity without TLR9 [53]. C57BL/6 mice challenged with a 14× LD50 dose of ECTV and vaccinated with MVA (intranasal and intravenous) on day 3 post-exposure were only protected with intravenous vaccination. In these protected animals activation of innate and cell mediated immunity shown by increased cytokine production and activation of NK and T cells was higher than in non-protected animals. In animals with intravenous vaccination IgG neutralizing titers could be detected earlier (day 6) and had higher titers (~10³) than in intranasally vaccinated animals (day 9) and (~10²). The similarity of results between unvaccinated and intranasally vaccinated animals indicates that the immune response (both innate and adaptive) to the original ECTV challenge was not enhanced by intranasal vaccination [52].

As the results above demonstrate, multiple variables contribute to the difficulty in interpreting surrogate model studies. These variables include differences in disease course and immune responses - which are, in turn, induced by different challenge viruses, vaccines, routes of infection, routes of vaccine administration, doses of virus and doses of vaccine. In the studies reviewed above, there is variability among models and even within the ECTV infected mouse model – results are consistent for day 0 and day 1 vaccination but not day 2, 3, 4 and 5 vaccination. This may be due to differences in BALB/C and C57BL/6 mice or may reflect differences in vaccine routes of administration. The differences between VACV or ECTV disease in mice and human systemic OPXV disease do not permit a full evaluation of the potential effect of post-exposure vaccination on human disease. Lastly, intranasal or intravenous administration of post-exposure vaccination has not been studied in humans making the applicability of these studies difficult to ascertain. Therefore, while these studies have been the first to address the issue of post-exposure vaccination and certainly inform the field on this issue, additional animal models – preferably those models with an extended incubation period – and diseases courses, which better mimic human disease – will enable better predictions per the human experience. The next section of this review covers those surrogate models that are currently in use that would be most suitable for future post-exposure vaccination studies.

Because of the data gaps identified above in both historical epidemiology reports and current surrogate model data, more laboratory studies in different surrogate models may help to define the window of time in which post-exposure vaccination offers a survival benefit. Because there is no non-human reservoir for VARV [54], it is not possible to use a naturally infected species as a surrogate for human systemic OPXV infection [55]. Fortunately, scientists have been developing, testing, evaluating and improving surrogate models that use non-VARV OPXVs to challenge numerous animal species since the beginning of the 20th century [56,57]. While much progress has been made, many models are difficult to compare to human disease and most do not fulfill all the requirements of the Federal Drug Administration’s (FDA) animal rule, which provides the current standards that surrogate models must meet in order to replace human efficacy trials for FDA licensure [58]. The major requirements of this regulation are that there is (1) a reasonably well understood mechanism for the pathogenesis of the challenge virus, (2) an understanding of the correlates of protection for vaccination and (3) that these are demonstrated in more than one animal species. While many surrogate models share similarities with systemic OPXV disease in humans, none of the individual models fully recapitulates smallpox disease in humans. The majority of models have very short incubation times and rapid disease courses, which make them less desirable for post-exposure vaccination studies, therefore we have limited our review to those models that offer an incubation period of at least 5 days and in which previous studies have indicated vaccine efficacy of some kind [59–79].

The purpose of our analysis is to identify those models that appear to be most relevant for extrapolation to humans for the development of public health policy and research in the absence of pre-exposure vaccination and human VARV disease. Within the subgroup of currently used surrogate models that have an extended incubation period and previous vaccine efficacy data, we have given priority to those models that (1) utilize an etiological agent that is a human pathogen – in order to better compare mechanisms of pathogenesis between the model and humans, (2) exhibit rash illness and ~30% mortality – as rash illness is an important marker of morbidity and that coupled with realistic mortality allows for the fine analysis of the immune components of partial protection and (3) can be vaccinated with current vaccines at similar doses and routes of administration which have been tested in human trials - again, in order to clarify the comparisons between models and humans. Table 2 details the clinical features of the incubation, prodrome, rash and resolution stages of OPXV infections in selected models. A comparative analysis of the disease course of smallpox in humans to the disease course of OPXV infections in these surrogate models allows for the identification of models that can be utilized most effectively for post-exposure vaccination efficacy studies.

Based on epidemiological data from the smallpox era, post-exposure vaccination is likely to be most efficacious when administered during the incubation period. As stated above, we have chosen only to consider those models demonstrating incubation periods of more than 5 days that have also previously been utilized to evaluate either pre- or post-exposure vaccination. This selection criteria led to the exclusion of the Rabbitpox – New Zealand White Rabbits (aerosol) or Variola – cynomologous macaque (intravenous) models due to short incubation periods [80–82]. The recently developed Calpox – marmoset (intranasal) [83,84] model shows promise for post-exposure vaccination modeling but was excluded from this review because no vaccine testing has occurred in this model at this time.

Each of the models in this table has weaknesses that limit their effectiveness, suggesting that post-exposure vaccination studies should be conducted in multiple surrogate models. The rhesus and cynomologous macaque model of VARV infection (aerosol) is a model with very low mortality, thus making it difficult to show survival benefit of post-exposure vaccination. The cynomologous macaque model of MPXV infection (aerosol) has differences in disease presentation when compared to human infections and has an abbreviated incubation period and high mortality (100%) when compared to the cynomologous macaque model of MPXV infection (intratracheal microspray aerosol) which has 33% mortality and a more similar disease course to humans with systemic OPXV infections. While the non-human primate model is the most studied of the smallpox surrogate models because these animals have a close evolutionary relationship to humans and reagents are readily available, there are several practical constraints to widespread use. For example, the expense, training and facility requirements associated with non-human primate studies limits the number of institutions that can perform research using these models and can occasionally result in underpowered experiments. While this concern can be partially offset by the use of the aforementioned highly lethal non-human primate models, these typically have a rapid disease onset and do not allow for an understanding of the kinetics of the immune response generated by post-exposure vaccination. In addition, regulations stipulate that potential therapies be proven efficacious in multiple animal models, which suggests the need for small animal models to complement non-human primate models.

Immunological reagents are also commercially available for mice and the availability and relatively low cost of mouse studies make them an attractive small animal alternative. The ectromelia virus (ECTV) (intranasal) infected mouse model has high lethality and dissimilar disease presentation when compared with human infections and ECTV is not virulent in humans. A better mouse model appears to be the MPXV infected (intranasal) CAST/EiJ model that has an extended incubation period when compared to the ECTV model. Unfortunately, this model also has a disease presentation that differs from human systemic OPXV infections in that animals do not demonstrate rash illness, which is an important marker for determining the effect of post-exposure vaccination on morbidity. Proportional mortality and presence of rash illness are especially important when attempting a fine analysis of the immune components of partial protection, vital for evaluation of smallpox vaccines in immunocompromised humans.

Lastly, the outbreak of monkeypox in the United States in 2003 offered a unique opportunity to evaluate exotic surrogate models, which are relatively inexpensive, outbred, natural hosts of MPXV. However, the lack of commercially available immunological reagents is a major weakness of these models. The MPXV infected (intranasal) prairie dog model has the advantage of an extended incubation period, similar disease presentation to infected humans to include rash illness and 30–50% mortality. This is superior to the ground squirrel model, which has high lethality and atypical disease presentation. Therefore, although each of the models in Table 2 are suitable for post-exposure vaccination studies, the MPXV infected (intratracheal microspray aerosol) cynomologous macaque, the MPXV infected (intranasal) CAST/EiJ mouse and the MPXV infected (intranasal) prairie dog offer distinct advantages in terms of disease presentation, challenge virus used and extended incubation periods for post-exposure vaccination testing.